Commit 9df95765368e7863e836348714709c09a0c5757c - libde265

Imported Upstream version 1.0.2 Joachim Bauch 8 years ago

163 changed file(s) with 31933 addition(s) and 3559 deletion(s). Raw diff Collapse all Expand all

+36

-16

.travis.yml less more

4	4	- gcc
5	5	env:
6	6	- HOST= WINE= DECODESTREAMS=
7		- HOST= WINE= DECODESTREAMS=libde265-teststreams-nolf
8		- HOST= WINE= DECODESTREAMS=libde265-teststreams-sao
9		- HOST= WINE= DECODESTREAMS=libde265-teststreams-tiles
10		- HOST= WINE= DECODESTREAMS=libde265-teststreams-tiles-nolf
11		- HOST= WINE= DECODESTREAMS=libde265-teststreams-weighted
12		- HOST= WINE= DECODESTREAMS=libde265-teststreams-wpp-nolf
	7	- HOST= WINE= DECODESTREAMS=libde265-teststreams-fuzzing THREADING=
	8	- HOST= WINE= DECODESTREAMS=libde265-teststreams-fuzzing THREADING=--single-threaded
	9	- HOST= WINE= DECODESTREAMS=libde265-teststreams-nolf THREADING=
	10	- HOST= WINE= DECODESTREAMS=libde265-teststreams-nolf THREADING=--single-threaded
	11	- HOST= WINE= DECODESTREAMS=libde265-teststreams-sao THREADING=
	12	- HOST= WINE= DECODESTREAMS=libde265-teststreams-sao THREADING=--single-threaded
	13	- HOST= WINE= DECODESTREAMS=libde265-teststreams-tiles THREADING=
	14	- HOST= WINE= DECODESTREAMS=libde265-teststreams-tiles THREADING=--single-threaded
	15	- HOST= WINE= DECODESTREAMS=libde265-teststreams-tiles-nolf THREADING=
	16	- HOST= WINE= DECODESTREAMS=libde265-teststreams-tiles-nolf THREADING=--single-threaded
	17	- HOST= WINE= DECODESTREAMS=libde265-teststreams-weighted THREADING=
	18	- HOST= WINE= DECODESTREAMS=libde265-teststreams-weighted THREADING=--single-threaded
	19	- HOST= WINE= DECODESTREAMS=libde265-teststreams-wpp-nolf THREADING=
	20	- HOST= WINE= DECODESTREAMS=libde265-teststreams-wpp-nolf THREADING=--single-threaded
13	21	- HOST=i686-w64-mingw32 WINE=wine DECODESTREAMS=
14	22	- HOST=x86_64-w64-mingw32 WINE=wine64 DECODESTREAMS=
	23	- HOST=arm-linux-gnueabihf WINE= DECODESTREAMS=
	24	- HOST=cmake WINE= DECODESTREAMS=
15	25
16	26	matrix:
17	27	include:

22	32	- sh -c "if [ ! -z '$DECODESTREAMS' ]; then sudo add-apt-repository -y ppa:strukturag/libde265; fi"
23	33	- sudo apt-get update -qq
24	34	- sh -c "if [ -z '$HOST' ]; then sudo apt-get install -qq valgrind libsdl-dev libqt4-dev libswscale-dev; fi"
25		- sh -c "if [ ! -z '$HOST' ]; then sudo apt-get install -qq wine; fi"
	35	- sh -c "if [ -z '$HOST' ] && [ -z '$DECODESTREAMS' ]; then sudo apt-get install -qq devscripts; fi"
	36	- sh -c "if [ ! -z '$WINE' ]; then sudo apt-get install -qq wine; fi"
26	37	- sh -c "if [ '$WINE' = 'wine' ]; then sudo apt-get install -qq gcc-mingw-w64-i686 g++-mingw-w64-i686 binutils-mingw-w64-i686 mingw-w64-dev; fi"
27	38	- sh -c "if [ '$WINE' = 'wine64' ]; then sudo apt-get install -qq gcc-mingw-w64-x86-64 g++-mingw-w64-x86-64 binutils-mingw-w64-x86-64 mingw-w64-dev; fi"
	39	- sh -c "if ( echo '$HOST' \| grep -q '^arm' ); then sudo apt-get install -qq g++-arm-linux-gnueabihf gcc-arm-linux-gnueabihf qemu-user; fi"
28	40	- sh -c "if [ ! -z '$DECODESTREAMS' ]; then sudo apt-get install $DECODESTREAMS; fi"
29	41
30	42	install:
31	43	- git clone https://github.com/strukturag/libde265-data.git
32	44
33	45	before_script:
34		- ./autogen.sh
35		- if [ ! -z "$HOST" ]; then unset CC; fi
36		- if [ ! -z "$HOST" ]; then unset CXX; fi
37		- ./configure --host=$HOST
	46	- if [ "$HOST" != "cmake" ]; then ./autogen.sh; fi
	47	- if [ ! -z "$HOST" ] && [ "$HOST" != "cmake" ]; then unset CC; fi
	48	- if [ ! -z "$HOST" ] && [ "$HOST" != "cmake" ]; then unset CXX; fi
	49	- if [ "$HOST" != "cmake" ]; then ./configure --host=$HOST; fi
	50	- if [ "$HOST" = "cmake" ]; then cmake .; fi
38	51
39	52	script:
	53	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then ./scripts/check_licenses.sh; fi"
40	54	- make
	55	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then make dist && mkdir dist-test && cd dist-test && tar xzf ../libde265-.tar.gz && cd libde265- && ./configure && make; fi"
	56	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then make dist && mkdir dist-cmake-test && cd dist-cmake-test && tar xzf ../libde265-.tar.gz && cd libde265- && cmake . && make; fi"
41	57	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then LD_LIBRARY_PATH=./libde265/.libs/ valgrind --tool=memcheck --quiet --error-exitcode=1 ./dec265/.libs/dec265 -q -c -f 100 ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
42	58	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then LD_LIBRARY_PATH=./libde265/.libs/ valgrind --tool=memcheck --quiet --error-exitcode=1 ./dec265/.libs/dec265 -t 4 -q -c -f 100 ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
43	59	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then LD_LIBRARY_PATH=./libde265/.libs/ valgrind --tool=memcheck --quiet --error-exitcode=1 ./dec265/.libs/dec265 -q -c -f 100 ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
44	60	- sh -c "if [ -z "$HOST" ] && [ -z "$DECODESTREAMS" ]; then LD_LIBRARY_PATH=./libde265/.libs/ valgrind --tool=memcheck --quiet --error-exitcode=1 ./dec265/.libs/dec265 -t 4 -q -c -f 100 ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
45		- sh -c "if [ ! -z "$HOST" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -q -c ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
46		- sh -c "if [ ! -z "$HOST" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -t 4 -q -c ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
47		- sh -c "if [ ! -z "$HOST" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -q -c ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
48		- sh -c "if [ ! -z "$HOST" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -t 4 -q -c ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
49		- sh -c "if [ ! -z '$DECODESTREAMS' ]; then python scripts/decodestreams.py /var/lib/libde265-teststreams; fi"
	61	- sh -c "if [ ! -z "$WINE" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -q -c ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
	62	- sh -c "if [ ! -z "$WINE" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -t 4 -q -c ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
	63	- sh -c "if [ ! -z "$WINE" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -q -c ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
	64	- sh -c "if [ ! -z "$WINE" ]; then WINEPREFIX=`pwd`/$WINE WINEPATH=/usr/lib/gcc/$HOST/4.6/ $WINE ./dec265/dec265.exe -t 4 -q -c ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
	65	- sh -c "if ( echo '$HOST' \| grep -q '^arm' ); then LD_LIBRARY_PATH=`pwd`/libde265/.libs/ qemu-arm -L /usr/$HOST ./dec265/.libs/dec265 -q -c ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
	66	#- sh -c "if ( echo '$HOST' \| grep -q '^arm' ); then LD_LIBRARY_PATH=`pwd`/libde265/.libs/ qemu-arm -L /usr/$HOST ./dec265/.libs/dec265 -t 4 -q -c ./libde265-data/IDR-only/paris-352x288-intra.bin; fi"
	67	- sh -c "if ( echo '$HOST' \| grep -q '^arm' ); then LD_LIBRARY_PATH=`pwd`/libde265/.libs/ qemu-arm -L /usr/$HOST ./dec265/.libs/dec265 -q -c ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
	68	#- sh -c "if ( echo '$HOST' \| grep -q '^arm' ); then LD_LIBRARY_PATH=`pwd`/libde265/.libs/ qemu-arm -L /usr/$HOST ./dec265/.libs/dec265 -t 4 -q -c ./libde265-data/RandomAccess/paris-ra-wpp.bin; fi"
	69	- sh -c "if [ ! -z '$DECODESTREAMS' ]; then python scripts/decodestreams.py $THREADING /var/lib/libde265-teststreams; fi"

+68

-0

CMakeLists.txt less more

	0	project (libde265)
	1	cmake_minimum_required (VERSION 2.8)
	2
	3	# The version number.
	4	set (NUMERIC_VERSION 0x01000000)
	5	set (PACKAGE_VERSION 1.0.0)
	6
	7	include (${CMAKE_ROOT}/Modules/CheckCCompilerFlag.cmake)
	8	include (${CMAKE_ROOT}/Modules/CheckIncludeFile.cmake)
	9	include (${CMAKE_ROOT}/Modules/FindSDL.cmake)
	10	include (${CMAKE_ROOT}/Modules/FindThreads.cmake)
	11
	12	CHECK_INCLUDE_FILE(malloc.h HAVE_MALLOC_H)
	13	CHECK_INCLUDE_FILE(stdint.h HAVE_STDINT_H)
	14	CHECK_INCLUDE_FILE(stdbool.h HAVE_STDBOOL_H)
	15
	16	if (HAVE_MALLOC_H)
	17	add_definitions(-DHAVE_MALLOC_H)
	18	endif()
	19	if (HAVE_STDINT_H)
	20	add_definitions(-DHAVE_STDINT_H)
	21	endif()
	22	if (HAVE_STDBOOL_H)
	23	add_definitions(-DHAVE_STDBOOL_H)
	24	endif()
	25
	26	configure_file (
	27	"${PROJECT_SOURCE_DIR}/libde265/de265-version.h.in"
	28	"${PROJECT_BINARY_DIR}/libde265/de265-version.h"
	29	)
	30
	31	if(CMAKE_COMPILER_IS_GNUCXX)
	32	set(GCC 1)
	33	add_definitions(-Wall)
	34	set(CMAKE_CXX_FLAGS "-std=gnu++0x ${CMAKE_CXX_FLAGS}")
	35	elseif("${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang")
	36	add_definitions(-Wall)
	37	set(CMAKE_CXX_FLAGS "-std=c++11 ${CMAKE_CXX_FLAGS}")
	38	endif()
	39
	40	option(DISABLE_SSE "Disable SSE optimizations")
	41	if(NOT ${DISABLE_SSE} EQUAL OFF)
	42	if(MSVC)
	43	set(SUPPORTS_SSE4_1 1)
	44	else()
	45	CHECK_C_COMPILER_FLAG(-msse4.1 SUPPORTS_SSE4_1)
	46	endif()
	47	endif()
	48
	49	include_directories ("${PROJECT_SOURCE_DIR}")
	50	include_directories ("${PROJECT_BINARY_DIR}")
	51	include_directories ("${PROJECT_SOURCE_DIR}/libde265")
	52	if(MSVC)
	53	include_directories ("${PROJECT_SOURCE_DIR}/extra")
	54	add_definitions(-DHAVE_STDINT_H)
	55	add_definitions(-DHAVE_STDBOOL_H)
	56	add_definitions(-DNOMINMAX)
	57	endif()
	58
	59	if(UNIX)
	60	set(LIBDE265_LIBRARY_NAME de265)
	61	else()
	62	set(LIBDE265_LIBRARY_NAME libde265)
	63	endif()
	64
	65	add_subdirectory (libde265)
	66	add_subdirectory (dec265)
	67	add_subdirectory (enc265)

-1

Makefile.am less more

7	7	SUBDIRS+=dec265
8	8	endif
9	9
	10	SUBDIRS+=enc265
	11
	12	SUBDIRS+=tools
	13
10	14	if ENABLE_SHERLOCK265
11	15	SUBDIRS+=sherlock265
12	16	endif

15	19	autogen.sh \
16	20	build.bat \
17	21	m4/m4_ax_check_compile_flag.m4 \
18		build \
19	22	Makefile.vc7 \
	23	CMakeLists.txt \
20	24	README.md \
21	25	libde265.png \
22	26	*/COPYING

+13

-8

Makefile.in less more

84	84	DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
85	85	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
86	86	$(top_srcdir)/configure $(am__configure_deps) \
87		$(srcdir)/config.h.in $(srcdir)/libde265.pc.in COPYING compile \
88		config.guess config.sub depcomp install-sh missing ltmain.sh
	87	$(srcdir)/config.h.in $(srcdir)/libde265.pc.in COPYING TODO \
	88	compile config.guess config.sub depcomp install-sh missing \
	89	ltmain.sh
89	90	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
90	91	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	92	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
91	93	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
92	94	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
93	95	$(top_srcdir)/m4/lt~obsolete.m4 \
94	96	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
95		$(top_srcdir)/m4/visibility.m4 $(top_srcdir)/configure.ac
	97	$(top_srcdir)/configure.ac
96	98	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
97	99	$(ACLOCAL_M4)
98	100	am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \

186	188	ETAGS = etags
187	189	CTAGS = ctags
188	190	CSCOPE = cscope
189		DIST_SUBDIRS = libde265 dec265 sherlock265
	191	DIST_SUBDIRS = libde265 dec265 enc265 tools sherlock265
190	192	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
191	193	distdir = $(PACKAGE)-$(VERSION)
192	194	top_distdir = $(distdir)

230	232	\| sed 's\|^\./\|$(prefix)/\|' \| grep -v '$(infodir)/dir$$'
231	233	distcleancheck_listfiles = find . -type f -print
232	234	ACLOCAL = @ACLOCAL@
	235	ALLOCA = @ALLOCA@
233	236	AMTAR = @AMTAR@
234	237	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
235	238	AR = @AR@

238	241	AUTOMAKE = @AUTOMAKE@
239	242	AWK = @AWK@
240	243	CC = @CC@
	244	CCAS = @CCAS@
	245	CCASDEPMODE = @CCASDEPMODE@
	246	CCASFLAGS = @CCASFLAGS@
241	247	CCDEPMODE = @CCDEPMODE@
242	248	CFLAGS = @CFLAGS@
243		CFLAG_VISIBILITY = @CFLAG_VISIBILITY@
244	249	CPP = @CPP@
245	250	CPPFLAGS = @CPPFLAGS@
246	251	CXX = @CXX@

260	265	EXEEXT = @EXEEXT@
261	266	FGREP = @FGREP@
262	267	GREP = @GREP@
263		HAVE_VISIBILITY = @HAVE_VISIBILITY@
	268	HAVE_CXX11 = @HAVE_CXX11@
264	269	INSTALL = @INSTALL@
265	270	INSTALL_DATA = @INSTALL_DATA@
266	271	INSTALL_PROGRAM = @INSTALL_PROGRAM@

371	376	top_build_prefix = @top_build_prefix@
372	377	top_builddir = @top_builddir@
373	378	top_srcdir = @top_srcdir@
374		SUBDIRS = libde265 $(am__append_1) $(am__append_2)
	379	SUBDIRS = libde265 $(am__append_1) enc265 tools $(am__append_2)
375	380	ACLOCAL_AMFLAGS = -I m4
376	381	EXTRA_DIST = .travis.yml \
377	382	autogen.sh \
378	383	build.bat \
379	384	m4/m4_ax_check_compile_flag.m4 \
380		build \
381	385	Makefile.vc7 \
	386	CMakeLists.txt \
382	387	README.md \
383	388	libde265.png \
384	389	*/COPYING

-0

Makefile.vc7 less more

0	0	clean all:
1	1	cd libde265 && $(MAKE) -f Makefile.vc7 $*
2	2	cd dec265 && $(MAKE) -f Makefile.vc7 $*
	3	cd enc265 && $(MAKE) -f Makefile.vc7 $*

+19

-2

README.md less more

76	76
77	77	You can disable building of the example programs by running `./configure` with
78	78	<pre>
79		--disable-dec265 Do not dec265 decoder program.
80		--disable-sherlock265 Do not build sherlock265 visual inspection program.
	79	--disable-dec265 Do not build the dec265 decoder program.
	80	--disable-sherlock265 Do not build the sherlock265 visual inspection program.
81	81	</pre>
82	82
83	83	Additional logging information can be turned on and off using these `./configure` flags:

86	86	--enable-log-info turn on logging at info level (default=no)
87	87	--enable-log-trace turn on logging at trace level (default=no)
88	88	</pre>
	89
	90
	91	Build using cmake
	92	=================
	93
	94	cmake scripts to build libde265 and the sample scripts `dec265` and `enc265` are
	95	included and can be compiled using these commands:
	96
	97	```
	98	mkdir build
	99	cd build
	100	cmake ..
	101	make
	102	```
	103
	104	See the [cmake documentation](http://www.cmake.org) for further information on
	105	using cmake on other platforms.
89	106
90	107
91	108	Prebuilt binaries

+13

-0

TODO less more

	0	/mnt/temp/dirk/yuv/flower_garden_422_720x486_30fps_simple_intra.bin
	1
	2	diff at byte 00196800
	3	frame 3
	4	channel: 0
	5	pixel position: x=384;y=125
	6	file A: 56
	7	file B: 86
	8
	9
	10	- API: request IDR-frame
	11	- API: return SPS header infos
	12	- rate-control: specify bit-rate

+21

-1

aclocal.m4 less more

214	214	m4_ifndef([AC_AUTOCONF_VERSION],
215	215	[m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
216	216	_AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
	217
	218	# Figure out how to run the assembler. -- Autoconf --
	219
	220	# Copyright (C) 2001-2013 Free Software Foundation, Inc.
	221	#
	222	# This file is free software; the Free Software Foundation
	223	# gives unlimited permission to copy and/or distribute it,
	224	# with or without modifications, as long as this notice is preserved.
	225
	226	# AM_PROG_AS
	227	# ----------
	228	AC_DEFUN([AM_PROG_AS],
	229	[# By default we simply use the C compiler to build assembly code.
	230	AC_REQUIRE([AC_PROG_CC])
	231	test "${CCAS+set}" = set \|\| CCAS=$CC
	232	test "${CCASFLAGS+set}" = set \|\| CCASFLAGS=$CFLAGS
	233	AC_ARG_VAR([CCAS], [assembler compiler command (defaults to CC)])
	234	AC_ARG_VAR([CCASFLAGS], [assembler compiler flags (defaults to CFLAGS)])
	235	_AM_IF_OPTION([no-dependencies],, [_AM_DEPENDENCIES([CCAS])])dnl
	236	])
217	237
218	238	# AM_AUX_DIR_EXPAND -- Autoconf --
219	239

1307	1327	]) # _AM_PROG_TAR
1308	1328
1309	1329	m4_include([m4/ax_compare_version.m4])
	1330	m4_include([m4/ax_cxx_compile_stdcxx_11.m4])
1310	1331	m4_include([m4/libtool.m4])
1311	1332	m4_include([m4/ltoptions.m4])
1312	1333	m4_include([m4/ltsugar.m4])
1313	1334	m4_include([m4/ltversion.m4])
1314	1335	m4_include([m4/lt~obsolete.m4])
1315	1336	m4_include([m4/m4_ax_check_compile_flag.m4])
1316		m4_include([m4/visibility.m4])

+20

-0

autogen.sh less more

0	0	#!/bin/sh
	1	set -eu
	2	#
	3	# H.265 video codec.
	4	# Copyright (c) 2013 struktur AG, Joachim Bauch <bauch@struktur.de>
	5	#
	6	# This file is part of libde265.
	7	#
	8	# libde265 is free software: you can redistribute it and/or modify
	9	# it under the terms of the GNU General Public License as published by
	10	# the Free Software Foundation, either version 3 of the License, or
	11	# (at your option) any later version.
	12	#
	13	# libde265 is distributed in the hope that it will be useful,
	14	# but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	# GNU General Public License for more details.
	17	#
	18	# You should have received a copy of the GNU General Public License
	19	# along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	#
1	21	if [ -x "`which autoreconf 2>/dev/null`" ] ; then
2	22	exec autoreconf -ivf
3	23	fi

-6

~~build/vc9-x86/make-solutions.bat~~ less more

0		@echo off
1		::
2		:: run this batch file to create a Visual Studion solution file for this project.
3		:: See the cmake documentation for other generator targets
4		::
5		cmake -G "Visual Studio 9 2008" ..\..\dec265 && cmake-gui ..\..\dec265

-0

build.bat less more

19	19	mkdir bin_x86\lib
20	20	)
21	21	copy /y dec265\dec265.exe bin_x86\
	22	copy /y enc265\enc265.exe bin_x86\
22	23	copy /y libde265\libde265.dll bin_x86\
23	24	copy /y libde265\libde265.lib bin_x86\lib\
24	25	copy /y libde265\libde265.exp bin_x86\lib\

32	33	mkdir bin_x64\lib
33	34	)
34	35	copy /y dec265\dec265.exe bin_x64\
	36	copy /y enc265\enc265.exe bin_x64\
35	37	copy /y libde265\libde265.dll bin_x64\
36	38	copy /y libde265\libde265.lib bin_x64\lib\
37	39	copy /y libde265\libde265.exp bin_x64\lib\

+81

-4

config.h.in less more

0	0	/* config.h.in. Generated from configure.ac by autoheader. */
	1
	2	/* Define to one of `_getb67', `GETB67', `getb67' for Cray-2 and Cray-YMP
	3	systems. This function is required for `alloca.c' support on those systems.
	4	*/
	5	#undef CRAY_STACKSEG_END
	6
	7	/* Define to 1 if using `alloca.c'. */
	8	#undef C_ALLOCA
	9
	10	/* Define to 1 if you have the `alarm' function. */
	11	#undef HAVE_ALARM
	12
	13	/* Define to 1 if you have `alloca', as a function or macro. */
	14	#undef HAVE_ALLOCA
	15
	16	/* Define to 1 if you have <alloca.h> and it should be used (not on Ultrix).
	17	*/
	18	#undef HAVE_ALLOCA_H
	19
	20	/* Support ARM instructions */
	21	#undef HAVE_ARM
	22
	23	/* define if the compiler supports basic C++11 syntax */
	24	#undef HAVE_CXX11
1	25
2	26	/* Define to 1 if you have the <dlfcn.h> header file. */
3	27	#undef HAVE_DLFCN_H
4	28
	29	/* Define to 1 if you have the `gettimeofday' function. */
	30	#undef HAVE_GETTIMEOFDAY
	31
5	32	/* Define to 1 if you have the <inttypes.h> header file. */
6	33	#undef HAVE_INTTYPES_H
7	34

23	50	/* Define to 1 if you have the `memset' function. */
24	51	#undef HAVE_MEMSET
25	52
	53	/* Support ARM NEON instructions */
	54	#undef HAVE_NEON
	55
26	56	/* Define to 1 if you have the `posix_memalign' function. */
27	57	#undef HAVE_POSIX_MEMALIGN
28	58
	59	/* Define to 1 if you have the `pow' function. */
	60	#undef HAVE_POW
	61
	62	/* Define to 1 if the system has the type `ptrdiff_t'. */
	63	#undef HAVE_PTRDIFF_T
	64
29	65	/* Whether libsdl was found. */
30	66	#undef HAVE_SDL
31	67
	68	/* Define to 1 if you have the <setjmp.h> header file. */
	69	#undef HAVE_SETJMP_H
	70
	71	/* Define to 1 if you have the <signal.h> header file. */
	72	#undef HAVE_SIGNAL_H
	73
	74	/* Define to 1 if you have the `sqrt' function. */
	75	#undef HAVE_SQRT
	76
32	77	/* Support SSSE4.1 (Streaming SIMD Extensions 4.1) instructions */
33	78	#undef HAVE_SSE4_1
34	79
35	80	/* Define to 1 if stdbool.h conforms to C99. */
36	81	#undef HAVE_STDBOOL_H
37	82
	83	/* Define to 1 if you have the <stddef.h> header file. */
	84	#undef HAVE_STDDEF_H
	85
38	86	/* Define to 1 if you have the <stdint.h> header file. */
39	87	#undef HAVE_STDINT_H
40	88
41	89	/* Define to 1 if you have the <stdlib.h> header file. */
42	90	#undef HAVE_STDLIB_H
43	91
	92	/* Define to 1 if you have the `strchr' function. */
	93	#undef HAVE_STRCHR
	94
44	95	/* Define to 1 if you have the <strings.h> header file. */
45	96	#undef HAVE_STRINGS_H
46	97
47	98	/* Define to 1 if you have the <string.h> header file. */
48	99	#undef HAVE_STRING_H
49	100
	101	/* Define to 1 if you have the `strrchr' function. */
	102	#undef HAVE_STRRCHR
	103
50	104	/* Whether libswscale was found. */
51	105	#undef HAVE_SWSCALE
52	106
53	107	/* Define to 1 if you have the <sys/stat.h> header file. */
54	108	#undef HAVE_SYS_STAT_H
55	109
	110	/* Define to 1 if you have the <sys/time.h> header file. */
	111	#undef HAVE_SYS_TIME_H
	112
56	113	/* Define to 1 if you have the <sys/types.h> header file. */
57	114	#undef HAVE_SYS_TYPES_H
58	115

61	118
62	119	/* Whether libvideogfx was found. */
63	120	#undef HAVE_VIDEOGFX
64
65		/* Define to 1 or 0, depending whether the compiler supports simple visibility
66		declarations. */
67		#undef HAVE_VISIBILITY
68	121
69	122	/* Define to 1 if the system has the type `_Bool'. */
70	123	#undef HAVE__BOOL

103	156	/* Define to the version of this package. */
104	157	#undef PACKAGE_VERSION
105	158
	159	/* If using the C implementation of alloca, define if you know the
	160	direction of stack growth for your system; otherwise it will be
	161	automatically deduced at runtime.
	162	STACK_DIRECTION > 0 => grows toward higher addresses
	163	STACK_DIRECTION < 0 => grows toward lower addresses
	164	STACK_DIRECTION = 0 => direction of growth unknown */
	165	#undef STACK_DIRECTION
	166
106	167	/* Define to 1 if you have the ANSI C header files. */
107	168	#undef STDC_HEADERS
	169
	170	/* Define to 1 if you can safely include both <sys/time.h> and <time.h>. */
	171	#undef TIME_WITH_SYS_TIME
108	172
109	173	/* Version number of package */
110	174	#undef VERSION

124	188	#define below would cause a syntax error. */
125	189	#undef _UINT8_T
126	190
	191	/* Define to `__inline__' or `__inline' if that's what the C compiler
	192	calls it, or to nothing if 'inline' is not supported under any name. */
	193	#ifndef __cplusplus
	194	#undef inline
	195	#endif
	196
127	197	/* Define to the type of a signed integer type of width exactly 16 bits if
128	198	such a type exists and the standard includes do not define it. */
129	199	#undef int16_t

131	201	/* Define to the type of a signed integer type of width exactly 32 bits if
132	202	such a type exists and the standard includes do not define it. */
133	203	#undef int32_t
	204
	205	/* Define to the type of a signed integer type of width exactly 64 bits if
	206	such a type exists and the standard includes do not define it. */
	207	#undef int64_t
134	208
135	209	/* Define to the type of a signed integer type of width exactly 8 bits if such
136	210	a type exists and the standard includes do not define it. */
137	211	#undef int8_t
138	212
	213	/* Define to `unsigned int' if <sys/types.h> does not define. */
	214	#undef size_t
	215
139	216	/* Define to the type of an unsigned integer type of width exactly 16 bits if
140	217	such a type exists and the standard includes do not define it. */
141	218	#undef uint16_t

+1323

-165

configure less more

0	0	#! /bin/sh
1	1	# Guess values for system-dependent variables and create Makefiles.
2		# Generated by GNU Autoconf 2.69 for libde265 0.9.
	2	# Generated by GNU Autoconf 2.69 for libde265 1.0.2.
3	3	#
4	4	# Report bugs to <farin@struktur.de>.
5	5	#

589	589	# Identity of this package.
590	590	PACKAGE_NAME='libde265'
591	591	PACKAGE_TARNAME='libde265'
592		PACKAGE_VERSION='0.9'
593		PACKAGE_STRING='libde265 0.9'
	592	PACKAGE_VERSION='1.0.2'
	593	PACKAGE_STRING='libde265 1.0.2'
594	594	PACKAGE_BUGREPORT='farin@struktur.de'
595	595	PACKAGE_URL=''
596	596

631	631	# include <unistd.h>
632	632	#endif"
633	633
	634	ac_header_list=
	635	ac_func_list=
634	636	ac_subst_vars='am__EXEEXT_FALSE
635	637	am__EXEEXT_TRUE
636	638	LTLIBOBJS
637		LIBOBJS
638	639	ENABLE_SHERLOCK265_FALSE
639	640	ENABLE_SHERLOCK265_TRUE
640	641	ENABLE_DEC265_FALSE

658	659	PKG_CONFIG_LIBDIR
659	660	PKG_CONFIG_PATH
660	661	PKG_CONFIG
	662	ENABLE_ARM_THUMB_FALSE
	663	ENABLE_ARM_THUMB_TRUE
	664	ENABLE_NEON_OPT_FALSE
	665	ENABLE_NEON_OPT_TRUE
	666	ENABLE_ARM_OPT_FALSE
	667	ENABLE_ARM_OPT_TRUE
661	668	ENABLE_SSE_OPT_FALSE
662	669	ENABLE_SSE_OPT_TRUE
663	670	MINGW_FALSE
664	671	MINGW_TRUE
	672	LIBOBJS
	673	ALLOCA
665	674	HAVE_VISIBILITY_FALSE
666	675	HAVE_VISIBILITY_TRUE
667		HAVE_VISIBILITY
668		CFLAG_VISIBILITY
	676	HAVE_CXX11
669	677	AM_BACKSLASH
670	678	AM_DEFAULT_VERBOSITY
671	679	AM_DEFAULT_V

676	684	am__fastdepCC_FALSE
677	685	am__fastdepCC_TRUE
678	686	CCDEPMODE
679		am__nodep
680		AMDEPBACKSLASH
681		AMDEP_FALSE
682		AMDEP_TRUE
683		am__quote
684		am__include
685		DEPDIR
686	687	am__untar
687	688	am__tar
688	689	AMTAR
689		am__leading_dot
690	690	SET_MAKE
691	691	mkdir_p
692	692	MKDIR_P

708	708	ac_ct_CXX
709	709	CXXFLAGS
710	710	CXX
	711	am__fastdepCCAS_FALSE
	712	am__fastdepCCAS_TRUE
	713	CCASDEPMODE
	714	am__nodep
	715	AMDEPBACKSLASH
	716	AMDEP_FALSE
	717	AMDEP_TRUE
	718	am__quote
	719	am__include
	720	DEPDIR
	721	am__leading_dot
	722	CCASFLAGS
	723	CCAS
711	724	CPP
712	725	OTOOL64
713	726	OTOOL

806	819	enable_dependency_tracking
807	820	enable_silent_rules
808	821	enable_sse
	822	enable_arm
	823	enable_thumb
809	824	enable_log_error
810	825	enable_log_info
811	826	enable_log_debug

822	837	LIBS
823	838	CPPFLAGS
824	839	CPP
	840	CCAS
	841	CCASFLAGS
825	842	CXX
826	843	CXXFLAGS
827	844	CCC

1377	1394	# Omit some internal or obsolete options to make the list less imposing.
1378	1395	# This message is too long to be a string in the A/UX 3.1 sh.
1379	1396	cat <<_ACEOF
1380		\`configure' configures libde265 0.9 to adapt to many kinds of systems.
	1397	\`configure' configures libde265 1.0.2 to adapt to many kinds of systems.
1381	1398
1382	1399	Usage: $0 [OPTION]... [VAR=VALUE]...
1383	1400

1448	1465
1449	1466	if test -n "$ac_init_help"; then
1450	1467	case $ac_init_help in
1451		short \| recursive ) echo "Configuration of libde265 0.9:";;
	1468	short \| recursive ) echo "Configuration of libde265 1.0.2:";;
1452	1469	esac
1453	1470	cat <<\_ACEOF
1454	1471

1468	1485	--enable-silent-rules less verbose build output (undo: "make V=1")
1469	1486	--disable-silent-rules verbose build output (undo: "make V=0")
1470	1487	--disable-sse disable SSE optimizations (default=no)
	1488	--disable-arm disable ARM optimizations (default=no)
	1489	--enable-thumb disable ARM THUMB instructions (default=no)
1471	1490	--enable-log-error turn on logging at error level (default=yes)
1472	1491	--enable-log-info turn on logging at info level (default=no)
1473	1492	--enable-log-debug turn on logging at debug level (default=no)

1493	1512	CPPFLAGS (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
1494	1513	you have headers in a nonstandard directory <include dir>
1495	1514	CPP C preprocessor
	1515	CCAS assembler compiler command (defaults to CC)
	1516	CCASFLAGS assembler compiler flags (defaults to CFLAGS)
1496	1517	CXX C++ compiler command
1497	1518	CXXFLAGS C++ compiler flags
1498	1519	CXXCPP C++ preprocessor

1580	1601	test -n "$ac_init_help" && exit $ac_status
1581	1602	if $ac_init_version; then
1582	1603	cat <<\_ACEOF
1583		libde265 configure 0.9
	1604	libde265 configure 1.0.2
1584	1605	generated by GNU Autoconf 2.69
1585	1606
1586	1607	Copyright (C) 2012 Free Software Foundation, Inc.

2254	2275	This file contains any messages produced by compilers while
2255	2276	running configure, to aid debugging if configure makes a mistake.
2256	2277
2257		It was created by libde265 $as_me 0.9, which was
	2278	It was created by libde265 $as_me 1.0.2, which was
2258	2279	generated by GNU Autoconf 2.69. Invocation command line was
2259	2280
2260	2281	$ $0 $@

2534	2555	>$cache_file
2535	2556	fi
2536	2557
	2558	as_fn_append ac_header_list " sys/time.h"
	2559	as_fn_append ac_header_list " unistd.h"
	2560	as_fn_append ac_func_list " alarm"
2537	2561	# Check that the precious variables saved in the cache have kept the same
2538	2562	# value.
2539	2563	ac_cache_corrupted=false

2606	2630	ac_config_headers="$ac_config_headers config.h"
2607	2631
2608	2632
2609		NUMERIC_VERSION=0x00090000 # Numeric representation of the version
	2633	NUMERIC_VERSION=0x01000200 # Numeric representation of the version (A.B.C[.D] = 0xAABBCCDD)
2610	2634
2611	2635
2612	2636	LIBDE265_CURRENT=0
2613		LIBDE265_REVISION=8
	2637	LIBDE265_REVISION=10
2614	2638	LIBDE265_AGE=0
2615	2639
2616	2640	# ---------------------------------------------------------------------------

11182	11206
11183	11207
11184	11208	# Checks for programs.
	11209	rm -rf .tst 2>/dev/null
	11210	mkdir .tst 2>/dev/null
	11211	if test -d .tst; then
	11212	am__leading_dot=.
	11213	else
	11214	am__leading_dot=_
	11215	fi
	11216	rmdir .tst 2>/dev/null
	11217
	11218	DEPDIR="${am__leading_dot}deps"
	11219
	11220	ac_config_commands="$ac_config_commands depfiles"
	11221
	11222
	11223	am_make=${MAKE-make}
	11224	cat > confinc << 'END'
	11225	am__doit:
	11226	@echo this is the am__doit target
	11227	.PHONY: am__doit
	11228	END
	11229	# If we don't find an include directive, just comment out the code.
	11230	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
	11231	$as_echo_n "checking for style of include used by $am_make... " >&6; }
	11232	am__include="#"
	11233	am__quote=
	11234	_am_result=none
	11235	# First try GNU make style include.
	11236	echo "include confinc" > confmf
	11237	# Ignore all kinds of additional output from 'make'.
	11238	case `$am_make -s -f confmf 2> /dev/null` in #(
	11239	the\ am__doit\ target)
	11240	am__include=include
	11241	am__quote=
	11242	_am_result=GNU
	11243	;;
	11244	esac
	11245	# Now try BSD make style include.
	11246	if test "$am__include" = "#"; then
	11247	echo '.include "confinc"' > confmf
	11248	case `$am_make -s -f confmf 2> /dev/null` in #(
	11249	the\ am__doit\ target)
	11250	am__include=.include
	11251	am__quote="\""
	11252	_am_result=BSD
	11253	;;
	11254	esac
	11255	fi
	11256
	11257
	11258	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
	11259	$as_echo "$_am_result" >&6; }
	11260	rm -f confinc confmf
	11261
	11262	# Check whether --enable-dependency-tracking was given.
	11263	if test "${enable_dependency_tracking+set}" = set; then :
	11264	enableval=$enable_dependency_tracking;
	11265	fi
	11266
	11267	if test "x$enable_dependency_tracking" != xno; then
	11268	am_depcomp="$ac_aux_dir/depcomp"
	11269	AMDEPBACKSLASH='\'
	11270	am__nodep='_no'
	11271	fi
	11272	if test "x$enable_dependency_tracking" != xno; then
	11273	AMDEP_TRUE=
	11274	AMDEP_FALSE='#'
	11275	else
	11276	AMDEP_TRUE='#'
	11277	AMDEP_FALSE=
	11278	fi
	11279
	11280
	11281	# By default we simply use the C compiler to build assembly code.
	11282
	11283	test "${CCAS+set}" = set \|\| CCAS=$CC
	11284	test "${CCASFLAGS+set}" = set \|\| CCASFLAGS=$CFLAGS
	11285
	11286
	11287
	11288	depcc="$CCAS" am_compiler_list=
	11289
	11290	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
	11291	$as_echo_n "checking dependency style of $depcc... " >&6; }
	11292	if ${am_cv_CCAS_dependencies_compiler_type+:} false; then :
	11293	$as_echo_n "(cached) " >&6
	11294	else
	11295	if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
	11296	# We make a subdir and do the tests there. Otherwise we can end up
	11297	# making bogus files that we don't know about and never remove. For
	11298	# instance it was reported that on HP-UX the gcc test will end up
	11299	# making a dummy file named 'D' -- because '-MD' means "put the output
	11300	# in D".
	11301	rm -rf conftest.dir
	11302	mkdir conftest.dir
	11303	# Copy depcomp to subdir because otherwise we won't find it if we're
	11304	# using a relative directory.
	11305	cp "$am_depcomp" conftest.dir
	11306	cd conftest.dir
	11307	# We will build objects and dependencies in a subdirectory because
	11308	# it helps to detect inapplicable dependency modes. For instance
	11309	# both Tru64's cc and ICC support -MD to output dependencies as a
	11310	# side effect of compilation, but ICC will put the dependencies in
	11311	# the current directory while Tru64 will put them in the object
	11312	# directory.
	11313	mkdir sub
	11314
	11315	am_cv_CCAS_dependencies_compiler_type=none
	11316	if test "$am_compiler_list" = ""; then
	11317	am_compiler_list=`sed -n 's/^#$[a-zA-Z0-9]$)$/\1/p' < ./depcomp`
	11318	fi
	11319	am__universal=false
	11320
	11321
	11322	for depmode in $am_compiler_list; do
	11323	# Setup a source with many dependencies, because some compilers
	11324	# like to wrap large dependency lists on column 80 (with \), and
	11325	# we should not choose a depcomp mode which is confused by this.
	11326	#
	11327	# We need to recreate these files for each test, as the compiler may
	11328	# overwrite some of them when testing with obscure command lines.
	11329	# This happens at least with the AIX C compiler.
	11330	: > sub/conftest.c
	11331	for i in 1 2 3 4 5 6; do
	11332	echo '#include "conftst'$i'.h"' >> sub/conftest.c
	11333	# Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
	11334	# Solaris 10 /bin/sh.
	11335	echo '/* dummy */' > sub/conftst$i.h
	11336	done
	11337	echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
	11338
	11339	# We check with '-c' and '-o' for the sake of the "dashmstdout"
	11340	# mode. It turns out that the SunPro C++ compiler does not properly
	11341	# handle '-M -o', and we need to detect this. Also, some Intel
	11342	# versions had trouble with output in subdirs.
	11343	am__obj=sub/conftest.${OBJEXT-o}
	11344	am__minus_obj="-o $am__obj"
	11345	case $depmode in
	11346	gcc)
	11347	# This depmode causes a compiler race in universal mode.
	11348	test "$am__universal" = false \|\| continue
	11349	;;
	11350	nosideeffect)
	11351	# After this tag, mechanisms are not by side-effect, so they'll
	11352	# only be used when explicitly requested.
	11353	if test "x$enable_dependency_tracking" = xyes; then
	11354	continue
	11355	else
	11356	break
	11357	fi
	11358	;;
	11359	msvc7 \| msvc7msys \| msvisualcpp \| msvcmsys)
	11360	# This compiler won't grok '-c -o', but also, the minuso test has
	11361	# not run yet. These depmodes are late enough in the game, and
	11362	# so weak that their functioning should not be impacted.
	11363	am__obj=conftest.${OBJEXT-o}
	11364	am__minus_obj=
	11365	;;
	11366	none) break ;;
	11367	esac
	11368	if depmode=$depmode \
	11369	source=sub/conftest.c object=$am__obj \
	11370	depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
	11371	$SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
	11372	>/dev/null 2>conftest.err &&
	11373	grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
	11374	grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
	11375	grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
	11376	${MAKE-make} -s -f confmf > /dev/null 2>&1; then
	11377	# icc doesn't choke on unknown options, it will just issue warnings
	11378	# or remarks (even with -Werror). So we grep stderr for any message
	11379	# that says an option was ignored or not supported.
	11380	# When given -MP, icc 7.0 and 7.1 complain thusly:
	11381	# icc: Command line warning: ignoring option '-M'; no argument required
	11382	# The diagnosis changed in icc 8.0:
	11383	# icc: Command line remark: option '-MP' not supported
	11384	if (grep 'ignoring option' conftest.err \|\|
	11385	grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
	11386	am_cv_CCAS_dependencies_compiler_type=$depmode
	11387	break
	11388	fi
	11389	fi
	11390	done
	11391
	11392	cd ..
	11393	rm -rf conftest.dir
	11394	else
	11395	am_cv_CCAS_dependencies_compiler_type=none
	11396	fi
	11397
	11398	fi
	11399	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CCAS_dependencies_compiler_type" >&5
	11400	$as_echo "$am_cv_CCAS_dependencies_compiler_type" >&6; }
	11401	CCASDEPMODE=depmode=$am_cv_CCAS_dependencies_compiler_type
	11402
	11403	if
	11404	test "x$enable_dependency_tracking" != xno \
	11405	&& test "$am_cv_CCAS_dependencies_compiler_type" = gcc3; then
	11406	am__fastdepCCAS_TRUE=
	11407	am__fastdepCCAS_FALSE='#'
	11408	else
	11409	am__fastdepCCAS_TRUE='#'
	11410	am__fastdepCCAS_FALSE=
	11411	fi
	11412
	11413
11185	11414	ac_ext=cpp
11186	11415	ac_cpp='$CXXCPP $CPPFLAGS'
11187	11416	ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'

15684	15913	SET_MAKE="MAKE=${MAKE-make}"
15685	15914	fi
15686	15915
15687		rm -rf .tst 2>/dev/null
15688		mkdir .tst 2>/dev/null
15689		if test -d .tst; then
15690		am__leading_dot=.
15691		else
15692		am__leading_dot=_
15693		fi
15694		rmdir .tst 2>/dev/null
15695
15696		DEPDIR="${am__leading_dot}deps"
15697
15698		ac_config_commands="$ac_config_commands depfiles"
15699
15700
15701		am_make=${MAKE-make}
15702		cat > confinc << 'END'
15703		am__doit:
15704		@echo this is the am__doit target
15705		.PHONY: am__doit
15706		END
15707		# If we don't find an include directive, just comment out the code.
15708		{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
15709		$as_echo_n "checking for style of include used by $am_make... " >&6; }
15710		am__include="#"
15711		am__quote=
15712		_am_result=none
15713		# First try GNU make style include.
15714		echo "include confinc" > confmf
15715		# Ignore all kinds of additional output from 'make'.
15716		case `$am_make -s -f confmf 2> /dev/null` in #(
15717		the\ am__doit\ target)
15718		am__include=include
15719		am__quote=
15720		_am_result=GNU
15721		;;
15722		esac
15723		# Now try BSD make style include.
15724		if test "$am__include" = "#"; then
15725		echo '.include "confinc"' > confmf
15726		case `$am_make -s -f confmf 2> /dev/null` in #(
15727		the\ am__doit\ target)
15728		am__include=.include
15729		am__quote="\""
15730		_am_result=BSD
15731		;;
15732		esac
15733		fi
15734
15735
15736		{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
15737		$as_echo "$_am_result" >&6; }
15738		rm -f confinc confmf
15739
15740		# Check whether --enable-dependency-tracking was given.
15741		if test "${enable_dependency_tracking+set}" = set; then :
15742		enableval=$enable_dependency_tracking;
15743		fi
15744
15745		if test "x$enable_dependency_tracking" != xno; then
15746		am_depcomp="$ac_aux_dir/depcomp"
15747		AMDEPBACKSLASH='\'
15748		am__nodep='_no'
15749		fi
15750		if test "x$enable_dependency_tracking" != xno; then
15751		AMDEP_TRUE=
15752		AMDEP_FALSE='#'
15753		else
15754		AMDEP_TRUE='#'
15755		AMDEP_FALSE=
15756		fi
15757
15758
15759	15916	# Check whether --enable-silent-rules was given.
15760	15917	if test "${enable_silent_rules+set}" = set; then :
15761	15918	enableval=$enable_silent_rules;

15817	15974
15818	15975	# Define the identity of the package.
15819	15976	PACKAGE='libde265'
15820		VERSION='0.9'
	15977	VERSION='1.0.2'
15821	15978
15822	15979
15823	15980	cat >>confdefs.h <<_ACEOF

16167	16324	fi
16168	16325
16169	16326	CFLAGS+=" -std=c99"
	16327	CXXFLAGS+=" -Werror=return-type -Werror=unused-result -Werror=reorder"
	16328	ax_cxx_compile_cxx11_required=true
	16329	ac_ext=cpp
	16330	ac_cpp='$CXXCPP $CPPFLAGS'
	16331	ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
	16332	ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
	16333	ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
	16334	ac_success=no
	16335	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features by default" >&5
	16336	$as_echo_n "checking whether $CXX supports C++11 features by default... " >&6; }
	16337	if ${ax_cv_cxx_compile_cxx11+:} false; then :
	16338	$as_echo_n "(cached) " >&6
	16339	else
	16340	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	16341	/* end confdefs.h. */
	16342
	16343	template <typename T>
	16344	struct check
	16345	{
	16346	static_assert(sizeof(int) <= sizeof(T), "not big enough");
	16347	};
	16348
	16349	struct Base {
	16350	virtual void f() {}
	16351	};
	16352	struct Child : public Base {
	16353	virtual void f() {} // DiFa: override {} # override not supported in gcc 4.6
	16354	};
	16355
	16356	typedef check<check<bool>> right_angle_brackets;
	16357
	16358	int a;
	16359	decltype(a) b;
	16360
	16361	typedef check<int> check_type;
	16362	check_type c;
	16363	check_type&& cr = static_cast<check_type&&>(c);
	16364
	16365	auto d = a;
	16366	auto l = [](){};
	16367
	16368	_ACEOF
	16369	if ac_fn_cxx_try_compile "$LINENO"; then :
	16370	ax_cv_cxx_compile_cxx11=yes
	16371	else
	16372	ax_cv_cxx_compile_cxx11=no
	16373	fi
	16374	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
	16375	fi
	16376	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_cxx_compile_cxx11" >&5
	16377	$as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
	16378	if test x$ax_cv_cxx_compile_cxx11 = xyes; then
	16379	ac_success=yes
	16380	fi
	16381
	16382	if test x$ac_success = xno; then
	16383	for switch in -std=gnu++11 -std=gnu++0x; do
	16384	cachevar=`$as_echo "ax_cv_cxx_compile_cxx11_$switch" \| $as_tr_sh`
	16385	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features with $switch" >&5
	16386	$as_echo_n "checking whether $CXX supports C++11 features with $switch... " >&6; }
	16387	if eval \${$cachevar+:} false; then :
	16388	$as_echo_n "(cached) " >&6
	16389	else
	16390	ac_save_CXXFLAGS="$CXXFLAGS"
	16391	CXXFLAGS="$CXXFLAGS $switch"
	16392	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	16393	/* end confdefs.h. */
	16394
	16395	template <typename T>
	16396	struct check
	16397	{
	16398	static_assert(sizeof(int) <= sizeof(T), "not big enough");
	16399	};
	16400
	16401	struct Base {
	16402	virtual void f() {}
	16403	};
	16404	struct Child : public Base {
	16405	virtual void f() {} // DiFa: override {} # override not supported in gcc 4.6
	16406	};
	16407
	16408	typedef check<check<bool>> right_angle_brackets;
	16409
	16410	int a;
	16411	decltype(a) b;
	16412
	16413	typedef check<int> check_type;
	16414	check_type c;
	16415	check_type&& cr = static_cast<check_type&&>(c);
	16416
	16417	auto d = a;
	16418	auto l = [](){};
	16419
	16420	_ACEOF
	16421	if ac_fn_cxx_try_compile "$LINENO"; then :
	16422	eval $cachevar=yes
	16423	else
	16424	eval $cachevar=no
	16425	fi
	16426	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
	16427	CXXFLAGS="$ac_save_CXXFLAGS"
	16428	fi
	16429	eval ac_res=\$$cachevar
	16430	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
	16431	$as_echo "$ac_res" >&6; }
	16432	if eval test x\$$cachevar = xyes; then
	16433	CXXFLAGS="$CXXFLAGS $switch"
	16434	ac_success=yes
	16435	break
	16436	fi
	16437	done
	16438	fi
	16439
	16440	if test x$ac_success = xno; then
	16441	for switch in -std=c++11 -std=c++0x; do
	16442	cachevar=`$as_echo "ax_cv_cxx_compile_cxx11_$switch" \| $as_tr_sh`
	16443	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features with $switch" >&5
	16444	$as_echo_n "checking whether $CXX supports C++11 features with $switch... " >&6; }
	16445	if eval \${$cachevar+:} false; then :
	16446	$as_echo_n "(cached) " >&6
	16447	else
	16448	ac_save_CXXFLAGS="$CXXFLAGS"
	16449	CXXFLAGS="$CXXFLAGS $switch"
	16450	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	16451	/* end confdefs.h. */
	16452
	16453	template <typename T>
	16454	struct check
	16455	{
	16456	static_assert(sizeof(int) <= sizeof(T), "not big enough");
	16457	};
	16458
	16459	struct Base {
	16460	virtual void f() {}
	16461	};
	16462	struct Child : public Base {
	16463	virtual void f() {} // DiFa: override {} # override not supported in gcc 4.6
	16464	};
	16465
	16466	typedef check<check<bool>> right_angle_brackets;
	16467
	16468	int a;
	16469	decltype(a) b;
	16470
	16471	typedef check<int> check_type;
	16472	check_type c;
	16473	check_type&& cr = static_cast<check_type&&>(c);
	16474
	16475	auto d = a;
	16476	auto l = [](){};
	16477
	16478	_ACEOF
	16479	if ac_fn_cxx_try_compile "$LINENO"; then :
	16480	eval $cachevar=yes
	16481	else
	16482	eval $cachevar=no
	16483	fi
	16484	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
	16485	CXXFLAGS="$ac_save_CXXFLAGS"
	16486	fi
	16487	eval ac_res=\$$cachevar
	16488	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
	16489	$as_echo "$ac_res" >&6; }
	16490	if eval test x\$$cachevar = xyes; then
	16491	CXXFLAGS="$CXXFLAGS $switch"
	16492	ac_success=yes
	16493	break
	16494	fi
	16495	done
	16496	fi
	16497	ac_ext=c
	16498	ac_cpp='$CPP $CPPFLAGS'
	16499	ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
	16500	ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
	16501	ac_compiler_gnu=$ac_cv_c_compiler_gnu
	16502
	16503	if test x$ax_cxx_compile_cxx11_required = xtrue; then
	16504	if test x$ac_success = xno; then
	16505	as_fn_error $? "*** A compiler with support for C++11 language features is required." "$LINENO" 5
	16506	fi
	16507	else
	16508	if test x$ac_success = xno; then
	16509	HAVE_CXX11=0
	16510	{ $as_echo "$as_me:${as_lineno-$LINENO}: No compiler with C++11 support was found" >&5
	16511	$as_echo "$as_me: No compiler with C++11 support was found" >&6;}
	16512	else
	16513	HAVE_CXX11=1
	16514
	16515	$as_echo "#define HAVE_CXX11 1" >>confdefs.h
	16516
	16517	fi
	16518
	16519
	16520	fi
	16521
16170	16522
16171	16523	if test "x$GCC" = "xyes"; then
16172	16524	case " $CFLAGS " in

16175	16527	esac
16176	16528	fi
16177	16529
16178
16179
16180		CFLAG_VISIBILITY=
16181		HAVE_VISIBILITY=0
16182		if test -n "$GCC"; then
16183		{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for simple visibility declarations" >&5
16184		$as_echo_n "checking for simple visibility declarations... " >&6; }
16185		if ${gl_cv_cc_visibility+:} false; then :
16186		$as_echo_n "(cached) " >&6
16187		else
16188
16189		gl_save_CFLAGS="$CFLAGS"
16190		CFLAGS="$CFLAGS -fvisibility=hidden"
16191		cat confdefs.h - <<_ACEOF >conftest.$ac_ext
16192		/* end confdefs.h. */
16193		extern __attribute__((__visibility__("hidden"))) int hiddenvar;
16194		extern __attribute__((__visibility__("default"))) int exportedvar;
16195		extern __attribute__((__visibility__("hidden"))) int hiddenfunc (void);
16196		extern __attribute__((__visibility__("default"))) int exportedfunc (void);
16197		int
16198		main ()
16199		{
16200
16201		;
16202		return 0;
16203		}
16204		_ACEOF
16205		if ac_fn_c_try_compile "$LINENO"; then :
16206		gl_cv_cc_visibility=yes
16207		else
16208		gl_cv_cc_visibility=no
16209		fi
16210		rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
16211		CFLAGS="$gl_save_CFLAGS"
16212		fi
16213
16214		{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $gl_cv_cc_visibility" >&5
16215		$as_echo "$gl_cv_cc_visibility" >&6; }
16216		if test $gl_cv_cc_visibility = yes; then
16217		CFLAG_VISIBILITY="-fvisibility=hidden"
16218		HAVE_VISIBILITY=1
16219		fi
16220		fi
16221
16222
16223
16224		cat >>confdefs.h <<_ACEOF
16225		#define HAVE_VISIBILITY $HAVE_VISIBILITY
16226		_ACEOF
16227
16228
	16530	HAVE_VISIBILITY=0
16229	16531	if test "x$HAVE_VISIBILITY" != "x0"; then
16230	16532	HAVE_VISIBILITY_TRUE=
16231	16533	HAVE_VISIBILITY_FALSE='#'

16236	16538
16237	16539
16238	16540	# Checks for header files.
16239		for ac_header in stdint.h stdlib.h string.h malloc.h
	16541	for ac_header in stdint.h stdlib.h string.h malloc.h signal.h setjmp.h stddef.h sys/time.h
16240	16542	do :
16241	16543	as_ac_Header=`$as_echo "ac_cv_header_$ac_header" \| $as_tr_sh`
16242	16544	ac_fn_c_check_header_mongrel "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default"

16344	16646
16345	16647	fi
16346	16648
	16649	ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
	16650	if test "x$ac_cv_type_size_t" = xyes; then :
	16651
	16652	else
	16653
	16654	cat >>confdefs.h <<_ACEOF
	16655	#define size_t unsigned int
	16656	_ACEOF
	16657
	16658	fi
	16659
	16660	ac_fn_c_find_intX_t "$LINENO" "8" "ac_cv_c_int8_t"
	16661	case $ac_cv_c_int8_t in #(
	16662	no\|yes) ;; #(
	16663	*)
	16664
	16665	cat >>confdefs.h <<_ACEOF
	16666	#define int8_t $ac_cv_c_int8_t
	16667	_ACEOF
	16668	;;
	16669	esac
	16670
16347	16671	ac_fn_c_find_intX_t "$LINENO" "16" "ac_cv_c_int16_t"
16348	16672	case $ac_cv_c_int16_t in #(
16349	16673	no\|yes) ;; #(

16366	16690	;;
16367	16691	esac
16368	16692
16369		ac_fn_c_find_intX_t "$LINENO" "8" "ac_cv_c_int8_t"
16370		case $ac_cv_c_int8_t in #(
	16693	ac_fn_c_find_intX_t "$LINENO" "64" "ac_cv_c_int64_t"
	16694	case $ac_cv_c_int64_t in #(
16371	16695	no\|yes) ;; #(
16372	16696	*)
16373	16697
16374	16698	cat >>confdefs.h <<_ACEOF
16375		#define int8_t $ac_cv_c_int8_t
	16699	#define int64_t $ac_cv_c_int64_t
16376	16700	_ACEOF
16377	16701	;;
16378	16702	esac
	16703
	16704	ac_fn_c_find_uintX_t "$LINENO" "8" "ac_cv_c_uint8_t"
	16705	case $ac_cv_c_uint8_t in #(
	16706	no\|yes) ;; #(
	16707	*)
	16708
	16709	$as_echo "#define _UINT8_T 1" >>confdefs.h
	16710
	16711
	16712	cat >>confdefs.h <<_ACEOF
	16713	#define uint8_t $ac_cv_c_uint8_t
	16714	_ACEOF
	16715	;;
	16716	esac
16379	16717
16380	16718	ac_fn_c_find_uintX_t "$LINENO" "16" "ac_cv_c_uint16_t"
16381	16719	case $ac_cv_c_uint16_t in #(

16417	16755	;;
16418	16756	esac
16419	16757
16420		ac_fn_c_find_uintX_t "$LINENO" "8" "ac_cv_c_uint8_t"
16421		case $ac_cv_c_uint8_t in #(
16422		no\|yes) ;; #(
	16758	ac_fn_c_check_type "$LINENO" "ptrdiff_t" "ac_cv_type_ptrdiff_t" "$ac_includes_default"
	16759	if test "x$ac_cv_type_ptrdiff_t" = xyes; then :
	16760
	16761	cat >>confdefs.h <<_ACEOF
	16762	#define HAVE_PTRDIFF_T 1
	16763	_ACEOF
	16764
	16765
	16766	fi
	16767
	16768	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for inline" >&5
	16769	$as_echo_n "checking for inline... " >&6; }
	16770	if ${ac_cv_c_inline+:} false; then :
	16771	$as_echo_n "(cached) " >&6
	16772	else
	16773	ac_cv_c_inline=no
	16774	for ac_kw in inline __inline__ __inline; do
	16775	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	16776	/* end confdefs.h. */
	16777	#ifndef __cplusplus
	16778	typedef int foo_t;
	16779	static $ac_kw foo_t static_foo () {return 0; }
	16780	$ac_kw foo_t foo () {return 0; }
	16781	#endif
	16782
	16783	_ACEOF
	16784	if ac_fn_c_try_compile "$LINENO"; then :
	16785	ac_cv_c_inline=$ac_kw
	16786	fi
	16787	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
	16788	test "$ac_cv_c_inline" != no && break
	16789	done
	16790
	16791	fi
	16792	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_inline" >&5
	16793	$as_echo "$ac_cv_c_inline" >&6; }
	16794
	16795	case $ac_cv_c_inline in
	16796	inline \| yes) ;;
16423	16797	*)
16424
16425		$as_echo "#define _UINT8_T 1" >>confdefs.h
16426
16427
16428		cat >>confdefs.h <<_ACEOF
16429		#define uint8_t $ac_cv_c_uint8_t
	16798	case $ac_cv_c_inline in
	16799	no) ac_val=;;
	16800	*) ac_val=$ac_cv_c_inline;;
	16801	esac
	16802	cat >>confdefs.h <<_ACEOF
	16803	#ifndef __cplusplus
	16804	#define inline $ac_val
	16805	#endif
16430	16806	_ACEOF
16431		;;
16432		esac
	16807	;;
	16808	esac
16433	16809
16434	16810
16435	16811	# Checks for library functions.

16445	16821	fi
16446	16822	done
16447	16823
	16824
	16825	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for library containing pow" >&5
	16826	$as_echo_n "checking for library containing pow... " >&6; }
	16827	if ${ac_cv_search_pow+:} false; then :
	16828	$as_echo_n "(cached) " >&6
	16829	else
	16830	ac_func_search_save_LIBS=$LIBS
	16831	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	16832	/* end confdefs.h. */
	16833
	16834	/* Override any GCC internal prototype to avoid an error.
	16835	Use char because int might match the return type of a GCC
	16836	builtin and then its argument prototype would still apply. */
	16837	#ifdef __cplusplus
	16838	extern "C"
	16839	#endif
	16840	char pow ();
	16841	int
	16842	main ()
	16843	{
	16844	return pow ();
	16845	;
	16846	return 0;
	16847	}
	16848	_ACEOF
	16849	for ac_lib in '' m; do
	16850	if test -z "$ac_lib"; then
	16851	ac_res="none required"
	16852	else
	16853	ac_res=-l$ac_lib
	16854	LIBS="-l$ac_lib $ac_func_search_save_LIBS"
	16855	fi
	16856	if ac_fn_c_try_link "$LINENO"; then :
	16857	ac_cv_search_pow=$ac_res
	16858	fi
	16859	rm -f core conftest.err conftest.$ac_objext \
	16860	conftest$ac_exeext
	16861	if ${ac_cv_search_pow+:} false; then :
	16862	break
	16863	fi
	16864	done
	16865	if ${ac_cv_search_pow+:} false; then :
	16866
	16867	else
	16868	ac_cv_search_pow=no
	16869	fi
	16870	rm conftest.$ac_ext
	16871	LIBS=$ac_func_search_save_LIBS
	16872	fi
	16873	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_search_pow" >&5
	16874	$as_echo "$ac_cv_search_pow" >&6; }
	16875	ac_res=$ac_cv_search_pow
	16876	if test "$ac_res" != no; then :
	16877	test "$ac_res" = "none required" \|\| LIBS="$ac_res $LIBS"
	16878
	16879	fi
16448	16880
16449	16881	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for library containing sqrt" >&5
16450	16882	$as_echo_n "checking for library containing sqrt... " >&6; }

16559	16991	fi
16560	16992
16561	16993
	16994	for ac_func in gettimeofday
	16995	do :
	16996	ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
	16997	if test "x$ac_cv_func_gettimeofday" = xyes; then :
	16998	cat >>confdefs.h <<_ACEOF
	16999	#define HAVE_GETTIMEOFDAY 1
	17000	_ACEOF
	17001
	17002	fi
	17003	done
	17004
	17005	for ac_func in pow sqrt
	17006	do :
	17007	as_ac_var=`$as_echo "ac_cv_func_$ac_func" \| $as_tr_sh`
	17008	ac_fn_c_check_func "$LINENO" "$ac_func" "$as_ac_var"
	17009	if eval test \"x\$"$as_ac_var"\" = x"yes"; then :
	17010	cat >>confdefs.h <<_ACEOF
	17011	#define `$as_echo "HAVE_$ac_func" \| $as_tr_cpp` 1
	17012	_ACEOF
	17013
	17014	fi
	17015	done
	17016
	17017	for ac_func in strchr strrchr
	17018	do :
	17019	as_ac_var=`$as_echo "ac_cv_func_$ac_func" \| $as_tr_sh`
	17020	ac_fn_c_check_func "$LINENO" "$ac_func" "$as_ac_var"
	17021	if eval test \"x\$"$as_ac_var"\" = x"yes"; then :
	17022	cat >>confdefs.h <<_ACEOF
	17023	#define `$as_echo "HAVE_$ac_func" \| $as_tr_cpp` 1
	17024	_ACEOF
	17025
	17026	fi
	17027	done
	17028
	17029
	17030	# The Ultrix 4.2 mips builtin alloca declared by alloca.h only works
	17031	# for constant arguments. Useless!
	17032	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for working alloca.h" >&5
	17033	$as_echo_n "checking for working alloca.h... " >&6; }
	17034	if ${ac_cv_working_alloca_h+:} false; then :
	17035	$as_echo_n "(cached) " >&6
	17036	else
	17037	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17038	/* end confdefs.h. */
	17039	#include <alloca.h>
	17040	int
	17041	main ()
	17042	{
	17043	char p = (char ) alloca (2 * sizeof (int));
	17044	if (p) return 0;
	17045	;
	17046	return 0;
	17047	}
	17048	_ACEOF
	17049	if ac_fn_c_try_link "$LINENO"; then :
	17050	ac_cv_working_alloca_h=yes
	17051	else
	17052	ac_cv_working_alloca_h=no
	17053	fi
	17054	rm -f core conftest.err conftest.$ac_objext \
	17055	conftest$ac_exeext conftest.$ac_ext
	17056	fi
	17057	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_working_alloca_h" >&5
	17058	$as_echo "$ac_cv_working_alloca_h" >&6; }
	17059	if test $ac_cv_working_alloca_h = yes; then
	17060
	17061	$as_echo "#define HAVE_ALLOCA_H 1" >>confdefs.h
	17062
	17063	fi
	17064
	17065	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for alloca" >&5
	17066	$as_echo_n "checking for alloca... " >&6; }
	17067	if ${ac_cv_func_alloca_works+:} false; then :
	17068	$as_echo_n "(cached) " >&6
	17069	else
	17070	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17071	/* end confdefs.h. */
	17072	#ifdef __GNUC__
	17073	# define alloca __builtin_alloca
	17074	#else
	17075	# ifdef _MSC_VER
	17076	# include <malloc.h>
	17077	# define alloca _alloca
	17078	# else
	17079	# ifdef HAVE_ALLOCA_H
	17080	# include <alloca.h>
	17081	# else
	17082	# ifdef _AIX
	17083	#pragma alloca
	17084	# else
	17085	# ifndef alloca /* predefined by HP cc +Olibcalls */
	17086	void *alloca (size_t);
	17087	# endif
	17088	# endif
	17089	# endif
	17090	# endif
	17091	#endif
	17092
	17093	int
	17094	main ()
	17095	{
	17096	char p = (char ) alloca (1);
	17097	if (p) return 0;
	17098	;
	17099	return 0;
	17100	}
	17101	_ACEOF
	17102	if ac_fn_c_try_link "$LINENO"; then :
	17103	ac_cv_func_alloca_works=yes
	17104	else
	17105	ac_cv_func_alloca_works=no
	17106	fi
	17107	rm -f core conftest.err conftest.$ac_objext \
	17108	conftest$ac_exeext conftest.$ac_ext
	17109	fi
	17110	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_func_alloca_works" >&5
	17111	$as_echo "$ac_cv_func_alloca_works" >&6; }
	17112
	17113	if test $ac_cv_func_alloca_works = yes; then
	17114
	17115	$as_echo "#define HAVE_ALLOCA 1" >>confdefs.h
	17116
	17117	else
	17118	# The SVR3 libPW and SVR4 libucb both contain incompatible functions
	17119	# that cause trouble. Some versions do not even contain alloca or
	17120	# contain a buggy version. If you still want to use their alloca,
	17121	# use ar to extract alloca.o from them instead of compiling alloca.c.
	17122
	17123	ALLOCA=\${LIBOBJDIR}alloca.$ac_objext
	17124
	17125	$as_echo "#define C_ALLOCA 1" >>confdefs.h
	17126
	17127
	17128	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether \`alloca.c' needs Cray hooks" >&5
	17129	$as_echo_n "checking whether \`alloca.c' needs Cray hooks... " >&6; }
	17130	if ${ac_cv_os_cray+:} false; then :
	17131	$as_echo_n "(cached) " >&6
	17132	else
	17133	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17134	/* end confdefs.h. */
	17135	#if defined CRAY && ! defined CRAY2
	17136	webecray
	17137	#else
	17138	wenotbecray
	17139	#endif
	17140
	17141	_ACEOF
	17142	if (eval "$ac_cpp conftest.$ac_ext") 2>&5 \|
	17143	$EGREP "webecray" >/dev/null 2>&1; then :
	17144	ac_cv_os_cray=yes
	17145	else
	17146	ac_cv_os_cray=no
	17147	fi
	17148	rm -f conftest*
	17149
	17150	fi
	17151	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_os_cray" >&5
	17152	$as_echo "$ac_cv_os_cray" >&6; }
	17153	if test $ac_cv_os_cray = yes; then
	17154	for ac_func in _getb67 GETB67 getb67; do
	17155	as_ac_var=`$as_echo "ac_cv_func_$ac_func" \| $as_tr_sh`
	17156	ac_fn_c_check_func "$LINENO" "$ac_func" "$as_ac_var"
	17157	if eval test \"x\$"$as_ac_var"\" = x"yes"; then :
	17158
	17159	cat >>confdefs.h <<_ACEOF
	17160	#define CRAY_STACKSEG_END $ac_func
	17161	_ACEOF
	17162
	17163	break
	17164	fi
	17165
	17166	done
	17167	fi
	17168
	17169	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking stack direction for C alloca" >&5
	17170	$as_echo_n "checking stack direction for C alloca... " >&6; }
	17171	if ${ac_cv_c_stack_direction+:} false; then :
	17172	$as_echo_n "(cached) " >&6
	17173	else
	17174	if test "$cross_compiling" = yes; then :
	17175	ac_cv_c_stack_direction=0
	17176	else
	17177	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17178	/* end confdefs.h. */
	17179	$ac_includes_default
	17180	int
	17181	find_stack_direction (int *addr, int depth)
	17182	{
	17183	int dir, dummy = 0;
	17184	if (! addr)
	17185	addr = &dummy;
	17186	*addr = addr < &dummy ? 1 : addr == &dummy ? 0 : -1;
	17187	dir = depth ? find_stack_direction (addr, depth - 1) : 0;
	17188	return dir + dummy;
	17189	}
	17190
	17191	int
	17192	main (int argc, char **argv)
	17193	{
	17194	return find_stack_direction (0, argc + !argv + 20) < 0;
	17195	}
	17196	_ACEOF
	17197	if ac_fn_c_try_run "$LINENO"; then :
	17198	ac_cv_c_stack_direction=1
	17199	else
	17200	ac_cv_c_stack_direction=-1
	17201	fi
	17202	rm -f core .core core.conftest. gmon.out bb.out conftest$ac_exeext \
	17203	conftest.$ac_objext conftest.beam conftest.$ac_ext
	17204	fi
	17205
	17206	fi
	17207	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_stack_direction" >&5
	17208	$as_echo "$ac_cv_c_stack_direction" >&6; }
	17209	cat >>confdefs.h <<_ACEOF
	17210	#define STACK_DIRECTION $ac_cv_c_stack_direction
	17211	_ACEOF
	17212
	17213
	17214	fi
	17215
	17216	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for error_at_line" >&5
	17217	$as_echo_n "checking for error_at_line... " >&6; }
	17218	if ${ac_cv_lib_error_at_line+:} false; then :
	17219	$as_echo_n "(cached) " >&6
	17220	else
	17221	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17222	/* end confdefs.h. */
	17223	#include <error.h>
	17224	int
	17225	main ()
	17226	{
	17227	error_at_line (0, 0, "", 0, "an error occurred");
	17228	;
	17229	return 0;
	17230	}
	17231	_ACEOF
	17232	if ac_fn_c_try_link "$LINENO"; then :
	17233	ac_cv_lib_error_at_line=yes
	17234	else
	17235	ac_cv_lib_error_at_line=no
	17236	fi
	17237	rm -f core conftest.err conftest.$ac_objext \
	17238	conftest$ac_exeext conftest.$ac_ext
	17239	fi
	17240	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_error_at_line" >&5
	17241	$as_echo "$ac_cv_lib_error_at_line" >&6; }
	17242	if test $ac_cv_lib_error_at_line = no; then
	17243	case " $LIBOBJS " in
	17244	" error.$ac_objext " ) ;;
	17245	*) LIBOBJS="$LIBOBJS error.$ac_objext"
	17246	;;
	17247	esac
	17248
	17249	fi
	17250
	17251	# Checking for malloc breaks building on ARM for us. A similar issue is described
	17252	# here: http://nerdland.net/unstumping-the-internet/malloc-has-not-been-declared/
	17253	# AC_FUNC_MALLOC
	17254	# AC_FUNC_REALLOC
	17255	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether time.h and sys/time.h may both be included" >&5
	17256	$as_echo_n "checking whether time.h and sys/time.h may both be included... " >&6; }
	17257	if ${ac_cv_header_time+:} false; then :
	17258	$as_echo_n "(cached) " >&6
	17259	else
	17260	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17261	/* end confdefs.h. */
	17262	#include <sys/types.h>
	17263	#include <sys/time.h>
	17264	#include <time.h>
	17265
	17266	int
	17267	main ()
	17268	{
	17269	if ((struct tm *) 0)
	17270	return 0;
	17271	;
	17272	return 0;
	17273	}
	17274	_ACEOF
	17275	if ac_fn_c_try_compile "$LINENO"; then :
	17276	ac_cv_header_time=yes
	17277	else
	17278	ac_cv_header_time=no
	17279	fi
	17280	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
	17281	fi
	17282	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_time" >&5
	17283	$as_echo "$ac_cv_header_time" >&6; }
	17284	if test $ac_cv_header_time = yes; then
	17285
	17286	$as_echo "#define TIME_WITH_SYS_TIME 1" >>confdefs.h
	17287
	17288	fi
	17289
	17290
	17291
	17292
	17293	for ac_header in $ac_header_list
	17294	do :
	17295	as_ac_Header=`$as_echo "ac_cv_header_$ac_header" \| $as_tr_sh`
	17296	ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
	17297	"
	17298	if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
	17299	cat >>confdefs.h <<_ACEOF
	17300	#define `$as_echo "HAVE_$ac_header" \| $as_tr_cpp` 1
	17301	_ACEOF
	17302
	17303	fi
	17304
	17305	done
	17306
	17307
	17308
	17309
	17310
	17311
	17312
	17313
	17314	for ac_func in $ac_func_list
	17315	do :
	17316	as_ac_var=`$as_echo "ac_cv_func_$ac_func" \| $as_tr_sh`
	17317	ac_fn_c_check_func "$LINENO" "$ac_func" "$as_ac_var"
	17318	if eval test \"x\$"$as_ac_var"\" = x"yes"; then :
	17319	cat >>confdefs.h <<_ACEOF
	17320	#define `$as_echo "HAVE_$ac_func" \| $as_tr_cpp` 1
	17321	_ACEOF
	17322
	17323	fi
	17324	done
	17325
	17326
	17327
	17328
	17329
	17330	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for working mktime" >&5
	17331	$as_echo_n "checking for working mktime... " >&6; }
	17332	if ${ac_cv_func_working_mktime+:} false; then :
	17333	$as_echo_n "(cached) " >&6
	17334	else
	17335	if test "$cross_compiling" = yes; then :
	17336	ac_cv_func_working_mktime=no
	17337	else
	17338	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17339	/* end confdefs.h. */
	17340	/* Test program from Paul Eggert and Tony Leneis. */
	17341	#ifdef TIME_WITH_SYS_TIME
	17342	# include <sys/time.h>
	17343	# include <time.h>
	17344	#else
	17345	# ifdef HAVE_SYS_TIME_H
	17346	# include <sys/time.h>
	17347	# else
	17348	# include <time.h>
	17349	# endif
	17350	#endif
	17351
	17352	#include <limits.h>
	17353	#include <stdlib.h>
	17354
	17355	#ifdef HAVE_UNISTD_H
	17356	# include <unistd.h>
	17357	#endif
	17358
	17359	#ifndef HAVE_ALARM
	17360	# define alarm(X) /* empty */
	17361	#endif
	17362
	17363	/* Work around redefinition to rpl_putenv by other config tests. */
	17364	#undef putenv
	17365
	17366	static time_t time_t_max;
	17367	static time_t time_t_min;
	17368
	17369	/* Values we'll use to set the TZ environment variable. */
	17370	static const char *tz_strings[] = {
	17371	(const char *) 0, "TZ=GMT0", "TZ=JST-9",
	17372	"TZ=EST+3EDT+2,M10.1.0/00:00:00,M2.3.0/00:00:00"
	17373	};
	17374	#define N_STRINGS (sizeof (tz_strings) / sizeof (tz_strings[0]))
	17375
	17376	/* Return 0 if mktime fails to convert a date in the spring-forward gap.
	17377	Based on a problem report from Andreas Jaeger. */
	17378	static int
	17379	spring_forward_gap ()
	17380	{
	17381	/* glibc (up to about 1998-10-07) failed this test. */
	17382	struct tm tm;
	17383
	17384	/* Use the portable POSIX.1 specification "TZ=PST8PDT,M4.1.0,M10.5.0"
	17385	instead of "TZ=America/Vancouver" in order to detect the bug even
	17386	on systems that don't support the Olson extension, or don't have the
	17387	full zoneinfo tables installed. */
	17388	putenv ((char*) "TZ=PST8PDT,M4.1.0,M10.5.0");
	17389
	17390	tm.tm_year = 98;
	17391	tm.tm_mon = 3;
	17392	tm.tm_mday = 5;
	17393	tm.tm_hour = 2;
	17394	tm.tm_min = 0;
	17395	tm.tm_sec = 0;
	17396	tm.tm_isdst = -1;
	17397	return mktime (&tm) != (time_t) -1;
	17398	}
	17399
	17400	static int
	17401	mktime_test1 (time_t now)
	17402	{
	17403	struct tm *lt;
	17404	return ! (lt = localtime (&now)) \|\| mktime (lt) == now;
	17405	}
	17406
	17407	static int
	17408	mktime_test (time_t now)
	17409	{
	17410	return (mktime_test1 (now)
	17411	&& mktime_test1 ((time_t) (time_t_max - now))
	17412	&& mktime_test1 ((time_t) (time_t_min + now)));
	17413	}
	17414
	17415	static int
	17416	irix_6_4_bug ()
	17417	{
	17418	/* Based on code from Ariel Faigon. */
	17419	struct tm tm;
	17420	tm.tm_year = 96;
	17421	tm.tm_mon = 3;
	17422	tm.tm_mday = 0;
	17423	tm.tm_hour = 0;
	17424	tm.tm_min = 0;
	17425	tm.tm_sec = 0;
	17426	tm.tm_isdst = -1;
	17427	mktime (&tm);
	17428	return tm.tm_mon == 2 && tm.tm_mday == 31;
	17429	}
	17430
	17431	static int
	17432	bigtime_test (int j)
	17433	{
	17434	struct tm tm;
	17435	time_t now;
	17436	tm.tm_year = tm.tm_mon = tm.tm_mday = tm.tm_hour = tm.tm_min = tm.tm_sec = j;
	17437	now = mktime (&tm);
	17438	if (now != (time_t) -1)
	17439	{
	17440	struct tm *lt = localtime (&now);
	17441	if (! (lt
	17442	&& lt->tm_year == tm.tm_year
	17443	&& lt->tm_mon == tm.tm_mon
	17444	&& lt->tm_mday == tm.tm_mday
	17445	&& lt->tm_hour == tm.tm_hour
	17446	&& lt->tm_min == tm.tm_min
	17447	&& lt->tm_sec == tm.tm_sec
	17448	&& lt->tm_yday == tm.tm_yday
	17449	&& lt->tm_wday == tm.tm_wday
	17450	&& ((lt->tm_isdst < 0 ? -1 : 0 < lt->tm_isdst)
	17451	== (tm.tm_isdst < 0 ? -1 : 0 < tm.tm_isdst))))
	17452	return 0;
	17453	}
	17454	return 1;
	17455	}
	17456
	17457	static int
	17458	year_2050_test ()
	17459	{
	17460	/* The correct answer for 2050-02-01 00:00:00 in Pacific time,
	17461	ignoring leap seconds. */
	17462	unsigned long int answer = 2527315200UL;
	17463
	17464	struct tm tm;
	17465	time_t t;
	17466	tm.tm_year = 2050 - 1900;
	17467	tm.tm_mon = 2 - 1;
	17468	tm.tm_mday = 1;
	17469	tm.tm_hour = tm.tm_min = tm.tm_sec = 0;
	17470	tm.tm_isdst = -1;
	17471
	17472	/* Use the portable POSIX.1 specification "TZ=PST8PDT,M4.1.0,M10.5.0"
	17473	instead of "TZ=America/Vancouver" in order to detect the bug even
	17474	on systems that don't support the Olson extension, or don't have the
	17475	full zoneinfo tables installed. */
	17476	putenv ((char*) "TZ=PST8PDT,M4.1.0,M10.5.0");
	17477
	17478	t = mktime (&tm);
	17479
	17480	/* Check that the result is either a failure, or close enough
	17481	to the correct answer that we can assume the discrepancy is
	17482	due to leap seconds. */
	17483	return (t == (time_t) -1
	17484	\|\| (0 < t && answer - 120 <= t && t <= answer + 120));
	17485	}
	17486
	17487	int
	17488	main ()
	17489	{
	17490	time_t t, delta;
	17491	int i, j;
	17492
	17493	/* This test makes some buggy mktime implementations loop.
	17494	Give up after 60 seconds; a mktime slower than that
	17495	isn't worth using anyway. */
	17496	alarm (60);
	17497
	17498	for (;;)
	17499	{
	17500	t = (time_t_max << 1) + 1;
	17501	if (t <= time_t_max)
	17502	break;
	17503	time_t_max = t;
	17504	}
	17505	time_t_min = - ((time_t) ~ (time_t) 0 == (time_t) -1) - time_t_max;
	17506
	17507	delta = time_t_max / 997; /* a suitable prime number */
	17508	for (i = 0; i < N_STRINGS; i++)
	17509	{
	17510	if (tz_strings[i])
	17511	putenv ((char*) tz_strings[i]);
	17512
	17513	for (t = 0; t <= time_t_max - delta; t += delta)
	17514	if (! mktime_test (t))
	17515	return 1;
	17516	if (! (mktime_test ((time_t) 1)
	17517	&& mktime_test ((time_t) (60 * 60))
	17518	&& mktime_test ((time_t) (60 * 60 * 24))))
	17519	return 1;
	17520
	17521	for (j = 1; ; j <<= 1)
	17522	if (! bigtime_test (j))
	17523	return 1;
	17524	else if (INT_MAX / 2 < j)
	17525	break;
	17526	if (! bigtime_test (INT_MAX))
	17527	return 1;
	17528	}
	17529	return ! (irix_6_4_bug () && spring_forward_gap () && year_2050_test ());
	17530	}
	17531	_ACEOF
	17532	if ac_fn_c_try_run "$LINENO"; then :
	17533	ac_cv_func_working_mktime=yes
	17534	else
	17535	ac_cv_func_working_mktime=no
	17536	fi
	17537	rm -f core .core core.conftest. gmon.out bb.out conftest$ac_exeext \
	17538	conftest.$ac_objext conftest.beam conftest.$ac_ext
	17539	fi
	17540
	17541	fi
	17542	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_func_working_mktime" >&5
	17543	$as_echo "$ac_cv_func_working_mktime" >&6; }
	17544	if test $ac_cv_func_working_mktime = no; then
	17545	case " $LIBOBJS " in
	17546	" mktime.$ac_objext " ) ;;
	17547	*) LIBOBJS="$LIBOBJS mktime.$ac_objext"
	17548	;;
	17549	esac
	17550
	17551	fi
	17552
	17553
16562	17554	if expr $host : '.*-mingw' >/dev/null 2>&1; then
16563	17555	MINGW_TRUE=
16564	17556	MINGW_FALSE='#'

16568	17560	fi
16569	17561
16570	17562
	17563	# Check if "__STRICT_ANSI__" is required.
	17564	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking if __STRICT_ANSI__ is required" >&5
	17565	$as_echo_n "checking if __STRICT_ANSI__ is required... " >&6; }
	17566	ac_ext=cpp
	17567	ac_cpp='$CXXCPP $CPPFLAGS'
	17568	ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
	17569	ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
	17570	ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
	17571
	17572	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17573	/* end confdefs.h. */
	17574
	17575	#include <vector>
	17576
	17577	int
	17578	main ()
	17579	{
	17580
	17581	;
	17582	return 0;
	17583	}
	17584	_ACEOF
	17585	if ac_fn_cxx_try_compile "$LINENO"; then :
	17586	need_strict_ansi=no
	17587	else
	17588	need_strict_ansi=yes
	17589	fi
	17590	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext;
	17591	ac_ext=c
	17592	ac_cpp='$CPP $CPPFLAGS'
	17593	ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
	17594	ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
	17595	ac_compiler_gnu=$ac_cv_c_compiler_gnu
	17596
	17597	if eval "test x$need_strict_ansi = xyes"; then
	17598	CFLAGS+=" -D__STRICT_ANSI__"
	17599	CXXFLAGS+=" -D__STRICT_ANSI__"
	17600	fi
	17601	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $need_strict_ansi" >&5
	17602	$as_echo "$need_strict_ansi" >&6; }
16571	17603
16572	17604	# --- machine dependent optimizations ---
16573	17605

16647	17679
16648	17680	# CFLAGS+=$SIMD_FLAGS
16649	17681	# CFLAGS+=" -march=x86-64"
	17682
	17683	case $target_cpu in
	17684	arm*)
	17685	# Check whether --enable-arm was given.
	17686	if test "${enable_arm+set}" = set; then :
	17687	enableval=$enable_arm; disable_arm=yes
	17688	else
	17689	disable_arm=no
	17690	fi
	17691
	17692
	17693	if test x"$disable_arm" != x"yes"; then
	17694
	17695	$as_echo "#define HAVE_ARM 1" >>confdefs.h
	17696
	17697
	17698	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C compiler accepts -mfpu=neon" >&5
	17699	$as_echo_n "checking whether C compiler accepts -mfpu=neon... " >&6; }
	17700	if ${ax_cv_check_cflags___mfpu_neon+:} false; then :
	17701	$as_echo_n "(cached) " >&6
	17702	else
	17703
	17704	ax_check_save_flags=$CFLAGS
	17705	CFLAGS="$CFLAGS -mfpu=neon"
	17706	cat confdefs.h - <<_ACEOF >conftest.$ac_ext
	17707	/* end confdefs.h. */
	17708
	17709	int
	17710	main ()
	17711	{
	17712
	17713	;
	17714	return 0;
	17715	}
	17716	_ACEOF
	17717	if ac_fn_c_try_compile "$LINENO"; then :
	17718	ax_cv_check_cflags___mfpu_neon=yes
	17719	else
	17720	ax_cv_check_cflags___mfpu_neon=no
	17721	fi
	17722	rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
	17723	CFLAGS=$ax_check_save_flags
	17724	fi
	17725	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cflags___mfpu_neon" >&5
	17726	$as_echo "$ax_cv_check_cflags___mfpu_neon" >&6; }
	17727	if test x"$ax_cv_check_cflags___mfpu_neon" = xyes; then :
	17728
	17729
	17730	$as_echo "#define HAVE_NEON 1" >>confdefs.h
	17731
	17732	ax_cv_support_neon_ext=yes
	17733	else
	17734	:
	17735	fi
	17736
	17737
	17738	# Check whether --enable-thumb was given.
	17739	if test "${enable_thumb+set}" = set; then :
	17740	enableval=$enable_thumb; enable_thumb=yes
	17741	else
	17742	enable_thumb=no
	17743	fi
	17744
	17745	fi
	17746	;;
	17747
	17748	*)
	17749	disable_arm=yes
	17750	;;
	17751
	17752	esac
	17753
	17754	if test x"$disable_arm" != x"yes"; then
	17755	ENABLE_ARM_OPT_TRUE=
	17756	ENABLE_ARM_OPT_FALSE='#'
	17757	else
	17758	ENABLE_ARM_OPT_TRUE='#'
	17759	ENABLE_ARM_OPT_FALSE=
	17760	fi
	17761
	17762	if test x"$ax_cv_support_neon_ext" = x"yes"; then
	17763	ENABLE_NEON_OPT_TRUE=
	17764	ENABLE_NEON_OPT_FALSE='#'
	17765	else
	17766	ENABLE_NEON_OPT_TRUE='#'
	17767	ENABLE_NEON_OPT_FALSE=
	17768	fi
	17769
	17770	if test x"$enable_thumb" != x"no"; then
	17771	ENABLE_ARM_THUMB_TRUE=
	17772	ENABLE_ARM_THUMB_FALSE='#'
	17773	else
	17774	ENABLE_ARM_THUMB_TRUE='#'
	17775	ENABLE_ARM_THUMB_FALSE=
	17776	fi
16650	17777
16651	17778
16652	17779	# --- additional logging ---

17449	18576
17450	18577	ac_config_files="$ac_config_files libde265/Makefile"
17451	18578
	18579	ac_config_files="$ac_config_files libde265/arm/Makefile"
	18580
17452	18581	ac_config_files="$ac_config_files libde265/x86/Makefile"
17453	18582
	18583	ac_config_files="$ac_config_files libde265/encoder/Makefile"
	18584
	18585	ac_config_files="$ac_config_files libde265/encoder/algo/Makefile"
	18586
17454	18587	ac_config_files="$ac_config_files libde265/de265-version.h"
17455	18588
17456	18589	ac_config_files="$ac_config_files dec265/Makefile"
17457	18590
	18591	ac_config_files="$ac_config_files enc265/Makefile"
	18592
17458	18593	ac_config_files="$ac_config_files sherlock265/Makefile"
	18594
	18595	ac_config_files="$ac_config_files tools/Makefile"
17459	18596
17460	18597	ac_config_files="$ac_config_files libde265.pc"
17461	18598

17568	18705	LTLIBOBJS=$ac_ltlibobjs
17569	18706
17570	18707
	18708	if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
	18709	as_fn_error $? "conditional \"AMDEP\" was never defined.
	18710	Usually this means the macro was only invoked conditionally." "$LINENO" 5
	18711	fi
	18712	if test -z "${am__fastdepCCAS_TRUE}" && test -z "${am__fastdepCCAS_FALSE}"; then
	18713	as_fn_error $? "conditional \"am__fastdepCCAS\" was never defined.
	18714	Usually this means the macro was only invoked conditionally." "$LINENO" 5
	18715	fi
17571	18716	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
17572	18717	$as_echo_n "checking that generated files are newer than configure... " >&6; }
17573	18718	if test -n "$am_sleep_pid"; then

17576	18721	fi
17577	18722	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
17578	18723	$as_echo "done" >&6; }
17579		if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
17580		as_fn_error $? "conditional \"AMDEP\" was never defined.
17581		Usually this means the macro was only invoked conditionally." "$LINENO" 5
17582		fi
17583	18724	if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
17584	18725	as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
17585	18726	Usually this means the macro was only invoked conditionally." "$LINENO" 5

17606	18747	fi
17607	18748	if test -z "${ENABLE_SSE_OPT_TRUE}" && test -z "${ENABLE_SSE_OPT_FALSE}"; then
17608	18749	as_fn_error $? "conditional \"ENABLE_SSE_OPT\" was never defined.
	18750	Usually this means the macro was only invoked conditionally." "$LINENO" 5
	18751	fi
	18752	if test -z "${ENABLE_ARM_OPT_TRUE}" && test -z "${ENABLE_ARM_OPT_FALSE}"; then
	18753	as_fn_error $? "conditional \"ENABLE_ARM_OPT\" was never defined.
	18754	Usually this means the macro was only invoked conditionally." "$LINENO" 5
	18755	fi
	18756	if test -z "${ENABLE_NEON_OPT_TRUE}" && test -z "${ENABLE_NEON_OPT_FALSE}"; then
	18757	as_fn_error $? "conditional \"ENABLE_NEON_OPT\" was never defined.
	18758	Usually this means the macro was only invoked conditionally." "$LINENO" 5
	18759	fi
	18760	if test -z "${ENABLE_ARM_THUMB_TRUE}" && test -z "${ENABLE_ARM_THUMB_FALSE}"; then
	18761	as_fn_error $? "conditional \"ENABLE_ARM_THUMB\" was never defined.
17609	18762	Usually this means the macro was only invoked conditionally." "$LINENO" 5
17610	18763	fi
17611	18764	if test -z "${HAVE_VIDEOGFX_TRUE}" && test -z "${HAVE_VIDEOGFX_FALSE}"; then

18025	19178	# report actual input values of CONFIG_FILES etc. instead of their
18026	19179	# values after options handling.
18027	19180	ac_log="
18028		This file was extended by libde265 $as_me 0.9, which was
	19181	This file was extended by libde265 $as_me 1.0.2, which was
18029	19182	generated by GNU Autoconf 2.69. Invocation command line was
18030	19183
18031	19184	CONFIG_FILES = $CONFIG_FILES

18091	19244	cat >>$CONFIG_STATUS <<_ACEOF \|\| ac_write_fail=1
18092	19245	ac_cs_config="`$as_echo "$ac_configure_args" \| sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
18093	19246	ac_cs_version="\\
18094		libde265 config.status 0.9
	19247	libde265 config.status 1.0.2
18095	19248	configured by $0, generated by GNU Autoconf 2.69,
18096	19249	with options \\"\$ac_cs_config\\"
18097	19250

18600	19753	"depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
18601	19754	"Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
18602	19755	"libde265/Makefile") CONFIG_FILES="$CONFIG_FILES libde265/Makefile" ;;
	19756	"libde265/arm/Makefile") CONFIG_FILES="$CONFIG_FILES libde265/arm/Makefile" ;;
18603	19757	"libde265/x86/Makefile") CONFIG_FILES="$CONFIG_FILES libde265/x86/Makefile" ;;
	19758	"libde265/encoder/Makefile") CONFIG_FILES="$CONFIG_FILES libde265/encoder/Makefile" ;;
	19759	"libde265/encoder/algo/Makefile") CONFIG_FILES="$CONFIG_FILES libde265/encoder/algo/Makefile" ;;
18604	19760	"libde265/de265-version.h") CONFIG_FILES="$CONFIG_FILES libde265/de265-version.h" ;;
18605	19761	"dec265/Makefile") CONFIG_FILES="$CONFIG_FILES dec265/Makefile" ;;
	19762	"enc265/Makefile") CONFIG_FILES="$CONFIG_FILES enc265/Makefile" ;;
18606	19763	"sherlock265/Makefile") CONFIG_FILES="$CONFIG_FILES sherlock265/Makefile" ;;
	19764	"tools/Makefile") CONFIG_FILES="$CONFIG_FILES tools/Makefile" ;;
18607	19765	"libde265.pc") CONFIG_FILES="$CONFIG_FILES libde265.pc" ;;
18608	19766
18609	19767	*) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;

+78

-7

configure.ac less more

1	1	# Process this file with autoconf to produce a configure script.
2	2
3	3	AC_PREREQ([2.68])
4		AC_INIT([libde265], [0.9], [farin@struktur.de])
	4	AC_INIT([libde265], [1.0.2], [farin@struktur.de])
5	5	AC_CONFIG_SRCDIR([libde265/de265.cc])
6	6	AC_CONFIG_HEADERS([config.h])
7	7
8		NUMERIC_VERSION=0x00090000 # Numeric representation of the version
	8	NUMERIC_VERSION=0x01000200 # Numeric representation of the version (A.B.C[.D] = 0xAABBCCDD)
9	9	AC_SUBST(NUMERIC_VERSION)
10	10
11	11	LIBDE265_CURRENT=0
12		LIBDE265_REVISION=8
	12	LIBDE265_REVISION=10
13	13	LIBDE265_AGE=0
14	14
15	15	# ---------------------------------------------------------------------------

25	25	AC_CONFIG_MACRO_DIR([m4])
26	26
27	27	# Checks for programs.
	28	AM_PROG_AS
28	29	AC_PROG_CXX
29	30	AC_PROG_CC
30	31	AC_PROG_INSTALL

35	36	AM_INIT_AUTOMAKE
36	37
37	38	CFLAGS+=" -std=c99"
	39	CXXFLAGS+=" -Werror=return-type -Werror=unused-result -Werror=reorder"
	40	AX_CXX_COMPILE_STDCXX_11()
38	41
39	42	dnl Use -Wall if we have gcc.
40	43	changequote(,)dnl

46	49	fi
47	50	changequote([,])dnl
48	51
49		gl_VISIBILITY
	52	dnl gl_VISIBILITY
	53	dnl : In encoder branch, we still export all library symbols :
	54	HAVE_VISIBILITY=0
50	55	AM_CONDITIONAL([HAVE_VISIBILITY], [test "x$HAVE_VISIBILITY" != "x0"])
51	56
52	57	# Checks for header files.
53		AC_CHECK_HEADERS([stdint.h stdlib.h string.h malloc.h])
	58	AC_CHECK_HEADERS([stdint.h stdlib.h string.h malloc.h signal.h setjmp.h stddef.h sys/time.h])
54	59
55	60	# Checks for typedefs, structures, and compiler characteristics.
56	61	AC_HEADER_STDBOOL
	62	AC_TYPE_SIZE_T
	63	AC_TYPE_INT8_T
57	64	AC_TYPE_INT16_T
58	65	AC_TYPE_INT32_T
59		AC_TYPE_INT8_T
	66	AC_TYPE_INT64_T
	67	AC_TYPE_UINT8_T
60	68	AC_TYPE_UINT16_T
61	69	AC_TYPE_UINT32_T
62	70	AC_TYPE_UINT64_T
63		AC_TYPE_UINT8_T
	71	AC_CHECK_TYPES([ptrdiff_t])
	72	AC_C_INLINE
64	73
65	74	# Checks for library functions.
66	75	AC_CHECK_FUNCS([malloc memmove memset __malloc_hook memalign posix_memalign __mingw_aligned_malloc __mingw_aligned_free])
67	76
	77	AC_SEARCH_LIBS([pow], [m])
68	78	AC_SEARCH_LIBS([sqrt], [m])
69	79	AC_SEARCH_LIBS([pthread_create], [pthread])
70	80
	81	AC_CHECK_FUNCS([gettimeofday])
	82	AC_CHECK_FUNCS([pow sqrt])
	83	AC_CHECK_FUNCS([strchr strrchr])
	84
	85	AC_FUNC_ALLOCA
	86	AC_FUNC_ERROR_AT_LINE
	87	# Checking for malloc breaks building on ARM for us. A similar issue is described
	88	# here: http://nerdland.net/unstumping-the-internet/malloc-has-not-been-declared/
	89	# AC_FUNC_MALLOC
	90	# AC_FUNC_REALLOC
	91	AC_FUNC_MKTIME
	92
71	93	AM_CONDITIONAL(MINGW, expr $host : '.*-mingw' >/dev/null 2>&1)
72	94
	95	# Check if "__STRICT_ANSI__" is required.
	96	AC_MSG_CHECKING([if __STRICT_ANSI__ is required])
	97	AC_LANG_PUSH(C++)
	98	AC_TRY_COMPILE([
	99	#include <vector>
	100	],[],[need_strict_ansi=no],[need_strict_ansi=yes]);
	101	AC_LANG_POP(C++)
	102	if eval "test x$need_strict_ansi = xyes"; then
	103	CFLAGS+=" -D__STRICT_ANSI__"
	104	CXXFLAGS+=" -D__STRICT_ANSI__"
	105	fi
	106	AC_MSG_RESULT([$need_strict_ansi])
73	107
74	108	# --- machine dependent optimizations ---
75	109

104	138	# CFLAGS+=$SIMD_FLAGS
105	139	# CFLAGS+=" -march=x86-64"
106	140
	141	case $target_cpu in
	142	arm*)
	143	AC_ARG_ENABLE(arm,
	144	[AS_HELP_STRING([--disable-arm],
	145	[disable ARM optimizations (default=no)])],
	146	[disable_arm=yes],
	147	[disable_arm=no])
	148
	149	if test x"$disable_arm" != x"yes"; then
	150	AC_DEFINE(HAVE_ARM, 1, [Support ARM instructions])
	151
	152	AX_CHECK_COMPILE_FLAG(-mfpu=neon, [
	153	AC_DEFINE(HAVE_NEON, 1, [Support ARM NEON instructions])
	154	ax_cv_support_neon_ext=yes], [])
	155
	156	AC_ARG_ENABLE(thumb,
	157	[AS_HELP_STRING([--enable-thumb],
	158	[disable ARM THUMB instructions (default=no)])],
	159	[enable_thumb=yes],
	160	[enable_thumb=no])
	161	fi
	162	;;
	163
	164	*)
	165	disable_arm=yes
	166	;;
	167
	168	esac
	169
	170	AM_CONDITIONAL([ENABLE_ARM_OPT], [test x"$disable_arm" != x"yes"])
	171	AM_CONDITIONAL([ENABLE_NEON_OPT], [test x"$ax_cv_support_neon_ext" = x"yes"])
	172	AM_CONDITIONAL([ENABLE_ARM_THUMB], [test x"$enable_thumb" != x"no"])
107	173
108	174	# --- additional logging ---
109	175

230	296
231	297	AC_CONFIG_FILES([Makefile])
232	298	AC_CONFIG_FILES([libde265/Makefile])
	299	AC_CONFIG_FILES([libde265/arm/Makefile])
233	300	AC_CONFIG_FILES([libde265/x86/Makefile])
	301	AC_CONFIG_FILES([libde265/encoder/Makefile])
	302	AC_CONFIG_FILES([libde265/encoder/algo/Makefile])
234	303	AC_CONFIG_FILES([libde265/de265-version.h])
235	304	AC_CONFIG_FILES([dec265/Makefile])
	305	AC_CONFIG_FILES([enc265/Makefile])
236	306	AC_CONFIG_FILES([sherlock265/Makefile])
	307	AC_CONFIG_FILES([tools/Makefile])
237	308	AC_CONFIG_FILES([libde265.pc])
238	309	AC_OUTPUT

+33

-124

dec265/CMakeLists.txt less more

0		set(CMAKE_CONFIGURATION_TYPES "Debug;Release;RelWithDebInfo" CACHE STRING "Configurations" FORCE)
	0	set (dec265_sources
	1	dec265.cc
	2	)
1	3
2		if(NOT CMAKE_BUILD_TYPE)
3		# default to Release build for GCC builds
4		set(CMAKE_BUILD_TYPE Release CACHE STRING
5		"Choose the type of build, options are: None(CMAKE_CXX_FLAGS or CMAKE_C_FLAGS used) Debug Release."
6		FORCE)
	4	set (hdrcopy_sources
	5	hdrcopy.cc
	6	)
	7
	8	if(MSVC)
	9	set (dec265_sources
	10	${dec265_sources}
	11	../extra/getopt.c
	12	../extra/getopt_long.c
	13	)
	14	set (hdrcopy_sources
	15	${hdrcopy_sources}
	16	../extra/getopt.c
	17	../extra/getopt_long.c
	18	)
7	19	endif()
8	20
9		project (dec265)
10		cmake_minimum_required (VERSION 2.8)
11
12		SET(CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/../cmake" "${CMAKE_MODULE_PATH}")
13
14		if ("${CMAKE_SIZEOF_VOID_P}" MATCHES 8)
15		set(X64 1)
16		add_definitions(-DX86_64)
	21	if(SDL_FOUND)
	22	add_definitions(-DHAVE_SDL)
	23	include_directories ("${SDL_INCLUDE_DIR}")
	24	set (dec265_sources
	25	${dec265_sources}
	26	sdl.cc
	27	)
17	28	endif()
18	29
19		# Enforce coding standards. Full warnings and warnings as errors
20		if(MSVC)
21		add_definitions(/W4 /D_CRT_SECURE_NO_WARNINGS)
22		add_definitions(/Ob2) # always inline
23		add_definitions(/Oi) # enable intrinsics
24		# add_definitions(/MP) # multithreaded build
25		# add_definitions(/TP) # all files as cpp
	30	add_executable (dec265 ${dec265_sources})
26	31
27		include_directories(../extra)
28		endif(MSVC)
	32	target_link_libraries (dec265 ${LIBDE265_LIBRARY_NAME} ${SDL_LIBRARY})
29	33
30		if("$ENV{CXX}" STREQUAL "icpc")
31		set(GCC 1)
32		add_definitions(-Wall -Wextra -Wunused-variable -Wunused-function -Wshadow -no-vec -msse4)
33		elseif(CMAKE_COMPILER_IS_GNUCXX)
34		set(GCC 1)
35		add_definitions(-Wall -Wextra -Wunused-variable -Wunused-function -Wshadow -msse4)
	34
	35	if(NOT MSVC)
	36	# hdrcopy uses internal APIs that are not available when compiled for Windows
	37	add_executable (hdrcopy ${hdrcopy_sources})
	38
	39	target_link_libraries (hdrcopy ${LIBDE265_LIBRARY_NAME})
36	40	endif()
37
38		option(USE_ASM "Use SSE41 SIMD optimize" ON)
39		if(USE_ASM)
40		add_definitions(-DUSE_ASM=ASM_SSE4)
41		add_definitions(-DHAVE_SSE4_1)
42		else(USE_ASM)
43		add_definitions(-DUSE_ASM=ASM_NONE)
44		endif(USE_ASM)
45
46		option(USE_VIDEOGFX "Use of VIDEOGFX" OFF)
47		if(USE_VIDEOGFX)
48		add_definitions(-D HAVE_VIDEOGFX=1)
49		else(USE_VIDEOGFX)
50		add_definitions(-D HAVE_VIDEOGFX=0)
51		endif(USE_VIDEOGFX)
52
53		option(BUILD_STATIC "Static Lib" ON)
54		if(BUILD_STATIC)
55		add_definitions(-DLIBDE265_STATIC_BUILD)
56		endif(BUILD_STATIC)
57
58		option(DE265_LOG_ERROR "Enable DE265_LOG_ERROR" OFF)
59		if(DE265_LOG_ERROR)
60		add_definitions(-DDE265_LOG_ERROR)
61		endif(DE265_LOG_ERROR)
62
63		option(DE265_LOG_INFO "Enable DE265_LOG_INFO" OFF)
64		if(DE265_LOG_INFO)
65		add_definitions(-DDE265_LOG_INFO)
66		endif(DE265_LOG_INFO)
67
68		option(DE265_LOG_DEBUG "Enable DE265_LOG_DEBUG" OFF)
69		if(DE265_LOG_DEBUG)
70		add_definitions(-DDE265_LOG_DEBUG)
71		endif(DE265_LOG_DEBUG)
72
73		option(DE265_LOG_TRACE "Enable DE265_LOG_TRACE" OFF)
74		if(DE265_LOG_TRACE)
75		add_definitions(-DDE265_LOG_TRACE)
76		endif(DE265_LOG_TRACE)
77
78		include_directories(../ ../libde265)
79
80		file(GLOB LIBSRC ../libde265/.cc ../extra/.c)
81		file(GLOB LIBINC ../libde265/.h ../extra/.h)
82		file(GLOB APPSRC dec265.cc)
83		file(GLOB ASMSRC0 ../libde265/x86/sse.cc ../libde265/x86/sse-dct.cc)
84		file(GLOB ASMSRC1 ../libde265/x86/sse-motion.cc)
85		file(GLOB ASMINC ../libde265/x86/*.h)
86
87		source_group(INC FILES ${LIBINC})
88		source_group(SRC FILES ${LIBSRC})
89		source_group(APP FILES ${APPSRC})
90		source_group(ASM FILES ${ASMSRC0} ${ASMSRC1} ${ASMINC})
91
92		if(USE_ASM)
93		SET(LIBSRC ${LIBSRC} ${ASMSRC0} ${ASMSRC1})
94		SET(LIBINC ${LIBINC} ${ASMINC})
95
96		# disable uninitialize check on VC, because x1 is not assign, remove later
97		#if(MSVC)
98		# SET_SOURCE_FILES_PROPERTIES(${ASMSRC1} PROPERTIES COMPILE_FLAGS "/RTCs")
99		#endif(MSVC)
100		endif(USE_ASM)
101
102		if(UNIX)
103		SET(PLATFORM_LIBS pthread rt m)
104		SET_SOURCE_FILES_PROPERTIES(../extra/win32thread.c PROPERTIES HEADER_FILE_ONLY TRUE)
105		endif(UNIX)
106
107		# Main CLI application
108		if(MSVC)
109		add_definitions(/wd4244) # type conversion, possible loss of data
110		add_definitions(/wd4100) # unreferenced formal parameter
111		add_definitions(/wd4505) # unreferenced local function has been removed
112		add_definitions(/wd4701) # potentially uninitialized local variable
113		add_definitions(/wd4127) # conditional expression is constant
114		# add_definitions(/wd4018) # signed/unsigned mismatch
115		add_definitions(/wd4189) # local variable is initialized but not referenced
116		add_definitions(/wd4715) # not all control paths return a value
117		add_definitions(/wd4324) # structure was padded due to __declspec(align())
118
119		set_source_files_properties(${LIBSRC} PROPERTIES LANGUAGE CXX)
120		endif(MSVC)
121		if(GCC)
122		add_definitions(-Wno-sign-compare)
123		add_definitions(-Wno-unused-parameter)
124		endif(GCC)
125
126		add_executable(dec265
127		${APPSRC} ${EXTRAS} ${LIBSRC} ${LIBINC}
128		)
129
130		target_link_libraries(dec265 ${PLATFORM_LIBS})
131

-1

dec265/Makefile.am less more

0	0
1		bin_PROGRAMS = dec265
	1	bin_PROGRAMS = dec265 hdrcopy
2	2
3	3	AM_CPPFLAGS = -I../libde265
4	4

7	7	dec265_LDFLAGS =
8	8	dec265_LDADD = ../libde265/libde265.la -lstdc++
9	9	dec265_SOURCES = dec265.cc
	10
	11	hdrcopy_DEPENDENCIES = ../libde265/libde265.la
	12	hdrcopy_CXXFLAGS =
	13	hdrcopy_LDFLAGS =
	14	hdrcopy_LDADD = ../libde265/libde265.la -lstdc++
	15	hdrcopy_SOURCES = hdrcopy.cc
10	16
11	17	if HAVE_VIDEOGFX
12	18	dec265_CXXFLAGS += $(VIDEOGFX_CFLAGS)

+39

-6

dec265/Makefile.in less more

78	78	build_triplet = @build@
79	79	host_triplet = @host@
80	80	target_triplet = @target@
81		bin_PROGRAMS = dec265$(EXEEXT)
	81	bin_PROGRAMS = dec265$(EXEEXT) hdrcopy$(EXEEXT)
82	82	@HAVE_VIDEOGFX_TRUE@am__append_1 = $(VIDEOGFX_CFLAGS)
83	83	@HAVE_VIDEOGFX_TRUE@am__append_2 = $(VIDEOGFX_LIBS)
84	84	@HAVE_SDL_TRUE@am__append_3 = $(SDL_CFLAGS)

90	90	$(top_srcdir)/depcomp COPYING
91	91	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
92	92	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	93	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
93	94	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
94	95	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
95	96	$(top_srcdir)/m4/lt~obsolete.m4 \
96	97	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
97		$(top_srcdir)/m4/visibility.m4 $(top_srcdir)/configure.ac
	98	$(top_srcdir)/configure.ac
98	99	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
99	100	$(ACLOCAL_M4)
100	101	mkinstalldirs = $(install_sh) -d

114	115	dec265_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
115	116	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(dec265_CXXFLAGS) \
116	117	$(CXXFLAGS) $(dec265_LDFLAGS) $(LDFLAGS) -o $@
	118	am_hdrcopy_OBJECTS = hdrcopy-hdrcopy.$(OBJEXT)
	119	hdrcopy_OBJECTS = $(am_hdrcopy_OBJECTS)
	120	hdrcopy_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	121	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(hdrcopy_CXXFLAGS) \
	122	$(CXXFLAGS) $(hdrcopy_LDFLAGS) $(LDFLAGS) -o $@
117	123	AM_V_P = $(am__v_P_@AM_V@)
118	124	am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
119	125	am__v_P_0 = false

166	172	am__v_CCLD_ = $(am__v_CCLD_@AM_DEFAULT_V@)
167	173	am__v_CCLD_0 = @echo " CCLD " $@;
168	174	am__v_CCLD_1 =
169		SOURCES = $(dec265_SOURCES)
170		DIST_SOURCES = $(am__dec265_SOURCES_DIST)
	175	SOURCES = $(dec265_SOURCES) $(hdrcopy_SOURCES)
	176	DIST_SOURCES = $(am__dec265_SOURCES_DIST) $(hdrcopy_SOURCES)
171	177	am__can_run_installinfo = \
172	178	case $$AM_UPDATE_INFO_DIR in \
173	179	n\|no\|NO) false;; \

194	200	CTAGS = ctags
195	201	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
196	202	ACLOCAL = @ACLOCAL@
	203	ALLOCA = @ALLOCA@
197	204	AMTAR = @AMTAR@
198	205	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
199	206	AR = @AR@

202	209	AUTOMAKE = @AUTOMAKE@
203	210	AWK = @AWK@
204	211	CC = @CC@
	212	CCAS = @CCAS@
	213	CCASDEPMODE = @CCASDEPMODE@
	214	CCASFLAGS = @CCASFLAGS@
205	215	CCDEPMODE = @CCDEPMODE@
206	216	CFLAGS = @CFLAGS@
207		CFLAG_VISIBILITY = @CFLAG_VISIBILITY@
208	217	CPP = @CPP@
209	218	CPPFLAGS = @CPPFLAGS@
210	219	CXX = @CXX@

224	233	EXEEXT = @EXEEXT@
225	234	FGREP = @FGREP@
226	235	GREP = @GREP@
227		HAVE_VISIBILITY = @HAVE_VISIBILITY@
	236	HAVE_CXX11 = @HAVE_CXX11@
228	237	INSTALL = @INSTALL@
229	238	INSTALL_DATA = @INSTALL_DATA@
230	239	INSTALL_PROGRAM = @INSTALL_PROGRAM@

341	350	dec265_LDFLAGS = $(am__append_2) $(am__append_4) $(am__append_6)
342	351	dec265_LDADD = ../libde265/libde265.la -lstdc++
343	352	dec265_SOURCES = dec265.cc $(am__append_5)
	353	hdrcopy_DEPENDENCIES = ../libde265/libde265.la
	354	hdrcopy_CXXFLAGS =
	355	hdrcopy_LDFLAGS =
	356	hdrcopy_LDADD = ../libde265/libde265.la -lstdc++
	357	hdrcopy_SOURCES = hdrcopy.cc
344	358	EXTRA_DIST = Makefile.vc7 \
345	359	CMakeLists.txt \
346	360	../extra/getopt.c \

435	449	@rm -f dec265$(EXEEXT)
436	450	$(AM_V_CXXLD)$(dec265_LINK) $(dec265_OBJECTS) $(dec265_LDADD) $(LIBS)
437	451
	452	hdrcopy$(EXEEXT): $(hdrcopy_OBJECTS) $(hdrcopy_DEPENDENCIES) $(EXTRA_hdrcopy_DEPENDENCIES)
	453	@rm -f hdrcopy$(EXEEXT)
	454	$(AM_V_CXXLD)$(hdrcopy_LINK) $(hdrcopy_OBJECTS) $(hdrcopy_LDADD) $(LIBS)
	455
438	456	mostlyclean-compile:
439	457	-rm -f *.$(OBJEXT)
440	458

443	461
444	462	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/dec265-dec265.Po@am__quote@
445	463	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/dec265-sdl.Po@am__quote@
	464	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/hdrcopy-hdrcopy.Po@am__quote@
446	465
447	466	.cc.o:
448	467	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<

492	511	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sdl.cc' object='dec265-sdl.obj' libtool=no @AMDEPBACKSLASH@
493	512	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
494	513	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(dec265_CXXFLAGS) $(CXXFLAGS) -c -o dec265-sdl.obj `if test -f 'sdl.cc'; then $(CYGPATH_W) 'sdl.cc'; else $(CYGPATH_W) '$(srcdir)/sdl.cc'; fi`
	514
	515	hdrcopy-hdrcopy.o: hdrcopy.cc
	516	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(hdrcopy_CXXFLAGS) $(CXXFLAGS) -MT hdrcopy-hdrcopy.o -MD -MP -MF $(DEPDIR)/hdrcopy-hdrcopy.Tpo -c -o hdrcopy-hdrcopy.o `test -f 'hdrcopy.cc' \|\| echo '$(srcdir)/'`hdrcopy.cc
	517	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/hdrcopy-hdrcopy.Tpo $(DEPDIR)/hdrcopy-hdrcopy.Po
	518	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='hdrcopy.cc' object='hdrcopy-hdrcopy.o' libtool=no @AMDEPBACKSLASH@
	519	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	520	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(hdrcopy_CXXFLAGS) $(CXXFLAGS) -c -o hdrcopy-hdrcopy.o `test -f 'hdrcopy.cc' \|\| echo '$(srcdir)/'`hdrcopy.cc
	521
	522	hdrcopy-hdrcopy.obj: hdrcopy.cc
	523	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(hdrcopy_CXXFLAGS) $(CXXFLAGS) -MT hdrcopy-hdrcopy.obj -MD -MP -MF $(DEPDIR)/hdrcopy-hdrcopy.Tpo -c -o hdrcopy-hdrcopy.obj `if test -f 'hdrcopy.cc'; then $(CYGPATH_W) 'hdrcopy.cc'; else $(CYGPATH_W) '$(srcdir)/hdrcopy.cc'; fi`
	524	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/hdrcopy-hdrcopy.Tpo $(DEPDIR)/hdrcopy-hdrcopy.Po
	525	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='hdrcopy.cc' object='hdrcopy-hdrcopy.obj' libtool=no @AMDEPBACKSLASH@
	526	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	527	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(hdrcopy_CXXFLAGS) $(CXXFLAGS) -c -o hdrcopy-hdrcopy.obj `if test -f 'hdrcopy.cc'; then $(CYGPATH_W) 'hdrcopy.cc'; else $(CYGPATH_W) '$(srcdir)/hdrcopy.cc'; fi`
495	528
496	529	mostlyclean-libtool:
497	530	-rm -f *.lo

-1

dec265/Makefile.vc7 less more

1	1	# Makefile for Microsoft Visual Studio 2003
2	2	#
3	3	CFLAGS=/I.. /I..\libde265 /I..\extra
4		CC=cl /nologo
	4	CC=cl /nologo
5	5	LINK=link /nologo /subsystem:console
6	6	DEFINES=/DWIN32
7	7
8	8	CFLAGS=$(CFLAGS) /MT /Ob2 /Oi /W4 /EHsc
9	9	CFLAGS=$(CFLAGS) $(DEFINES)
	10
	11	# unreferenced formal parameter
	12	CFLAGS=$(CFLAGS) /wd4100
10	13
11	14	OBJS=\
12	15	..\extra\getopt_long.obj \

+290

-22

dec265/dec265.cc less more

37	37	#include <unistd.h>
38	38	#endif
39	39
	40	#include "libde265/quality.h"
	41
40	42	#if HAVE_VIDEOGFX
41	43	#include <libvideogfx.hh>
42	44	using namespace videogfx;

52	54
53	55	int nThreads=0;
54	56	bool nal_input=false;
55		bool quiet=false;
	57	int quiet=0;
56	58	bool check_hash=false;
57	59	bool show_help=false;
58	60	bool dump_headers=false;

64	66	uint32_t max_frames=UINT32_MAX;
65	67	bool write_bytestream=false;
66	68	const char *bytestream_filename;
	69	bool measure_quality=false;
	70	bool show_ssim_map=false;
	71	bool show_psnr_map=false;
	72	const char* reference_filename;
	73	FILE* reference_file;
67	74	int highestTID = 100;
68	75	int verbosity=0;
69	76	int disable_deblocking=0;

83	90	{"help", no_argument, 0, 'h' },
84	91	{"noaccel", no_argument, 0, '0' },
85	92	{"write-bytestream", required_argument,0, 'B' },
	93	{"measure", required_argument, 0, 'm' },
	94	{"ssim", no_argument, 0, 's' },
	95	{"errmap", no_argument, 0, 'e' },
86	96	{"highest-TID", required_argument, 0, 'T' },
87	97	{"verbose", no_argument, 0, 'v' },
88	98	{"disable-deblocking", no_argument, &disable_deblocking, 1 },

97	107	static FILE* fh = NULL;
98	108	if (fh==NULL) { fh = fopen(output_filename, "wb"); }
99	109
100
	110
101	111
102	112	for (int c=0;c<3;c++) {
103	113	int stride;
104	114	const uint8_t* p = de265_get_image_plane(img, c, &stride);
105	115	int width = de265_get_image_width(img,c);
106	116
107		for (int y=0;y<de265_get_image_height(img,c);y++) {
108		fwrite(p + y*stride, width, 1, fh);
109		}
110		}
111
	117	if (de265_get_bits_per_pixel(img,c)<=8) {
	118	// --- save 8 bit YUV ---
	119
	120	for (int y=0;y<de265_get_image_height(img,c);y++) {
	121	fwrite(p + y*stride, width, 1, fh);
	122	}
	123	}
	124	else {
	125	// --- save 16 bit YUV ---
	126
	127	uint8_t* buf = new uint8_t[width*2];
	128	uint16_t* p16 = (uint16_t*)p;
	129
	130	for (int y=0;y<de265_get_image_height(img,c);y++) {
	131	for (int x=0;x<width;x++) {
	132	uint16_t pixel_value = (p16+y*stride)[x];
	133	buf[2*x+0] = pixel_value & 0xFF;
	134	buf[2*x+1] = pixel_value >> 8;
	135	}
	136
	137	fwrite(buf, width*2, 1, fh);
	138	}
	139
	140	delete[] buf;
	141	}
	142	}
	143
112	144	fflush(fh);
113	145	}
114	146

134	166
135	167	int width = de265_get_image_width(img,0);
136	168	int height = de265_get_image_height(img,0);
	169	de265_chroma chroma = de265_get_chroma_format(img);
	170
	171	ChromaFormat vgfx_chroma;
	172	Colorspace vgfx_cs = Colorspace_YUV;
	173
	174	switch (chroma) {
	175	case de265_chroma_420: vgfx_chroma = Chroma_420; break;
	176	case de265_chroma_422: vgfx_chroma = Chroma_422; break;
	177	case de265_chroma_444: vgfx_chroma = Chroma_444; break;
	178	case de265_chroma_mono: vgfx_cs = Colorspace_Greyscale; break;
	179	}
137	180
138	181	Image<Pixel> visu;
139		visu.Create(width, height, Colorspace_YUV, Chroma_420);
140
141		for (int ch=0;ch<3;ch++) {
	182	visu.Create(width, height, vgfx_cs, vgfx_chroma);
	183
	184	int nChannels = 3;
	185	if (chroma == de265_chroma_mono) {
	186	nChannels = 1;
	187	}
	188
	189	for (int ch=0;ch<nChannels;ch++) {
142	190	const uint8_t* data;
143	191	int stride;
144	192

146	194	width = de265_get_image_width(img,ch);
147	195	height = de265_get_image_height(img,ch);
148	196
149		for (int y=0;y<height;y++) {
150		memcpy(visu.AskFrame((BitmapChannel)ch)[y], data + y*stride, width);
	197	int bit_depth = de265_get_bits_per_pixel(img,ch);
	198
	199	if (bit_depth==8) {
	200	for (int y=0;y<height;y++) {
	201	memcpy(visu.AskFrame((BitmapChannel)ch)[y], data + y*stride, width);
	202	}
	203	}
	204	else {
	205	const uint16_t* data16 = (const uint16_t*)data;
	206	for (int y=0;y<height;y++) {
	207	for (int x=0;x<width;x++) {
	208	visu.AskFrame((BitmapChannel)ch)[y][x] = (data16 + ystride +x) >> (bit_depth-8);
	209	}
	210	}
151	211	}
152	212	}
153	213

155	215	win.WaitForKeypress();
156	216	}
157	217	#endif
	218
	219	static uint8_t* convert_to_8bit(const uint8_t* data, int width, int height, int stride, int bit_depth)
	220	{
	221	const uint16_t* data16 = (const uint16_t*)data;
	222	uint8_t* out = new uint8_t[stride*height];
	223
	224	for (int y=0;y<height;y++) {
	225	for (int x=0;x<width;x++) {
	226	out[ystride + x] = (data16 + y*stride +x) >> (bit_depth-8);
	227	}
	228	}
	229
	230	return out;
	231	}
	232
158	233
159	234	#if HAVE_SDL
160	235	SDL_YUV_Display sdlWin;

162	237
163	238	bool display_sdl(const struct de265_image* img)
164	239	{
	240	int width = de265_get_image_width(img,0);
	241	int height = de265_get_image_height(img,0);
	242
	243	int chroma_width = de265_get_image_width(img,1);
	244	int chroma_height = de265_get_image_height(img,1);
	245
	246	de265_chroma chroma = de265_get_chroma_format(img);
	247
165	248	if (!sdl_active) {
166		int width = de265_get_image_width(img,0);
167		int height = de265_get_image_height(img,0);
168
169	249	sdl_active=true;
170		sdlWin.init(width,height);
	250	enum SDL_YUV_Display::SDL_Chroma sdlChroma;
	251	switch (chroma) {
	252	case de265_chroma_420: sdlChroma = SDL_YUV_Display::SDL_CHROMA_420; break;
	253	case de265_chroma_422: sdlChroma = SDL_YUV_Display::SDL_CHROMA_422; break;
	254	case de265_chroma_444: sdlChroma = SDL_YUV_Display::SDL_CHROMA_444; break;
	255	case de265_chroma_mono: sdlChroma = SDL_YUV_Display::SDL_CHROMA_MONO; break;
	256	}
	257
	258	sdlWin.init(width,height, sdlChroma);
171	259	}
172	260
173	261	int stride,chroma_stride;

175	263	const uint8_t* cb =de265_get_image_plane(img,1,&chroma_stride);
176	264	const uint8_t* cr =de265_get_image_plane(img,2,NULL);
177	265
	266	uint8_t* y16 = NULL;
	267	uint8_t* cb16 = NULL;
	268	uint8_t* cr16 = NULL;
	269	int bd;
	270
	271	if ((bd=de265_get_bits_per_pixel(img, 0)) > 8) {
	272	y16 = convert_to_8bit(y, width,height,stride,bd); y=y16;
	273	}
	274
	275	if (chroma != de265_chroma_mono) {
	276	if ((bd=de265_get_bits_per_pixel(img, 1)) > 8) {
	277	cb16 = convert_to_8bit(cb, chroma_width,chroma_height,chroma_stride,bd); cb=cb16;
	278	}
	279	if ((bd=de265_get_bits_per_pixel(img, 2)) > 8) {
	280	cr16 = convert_to_8bit(cr, chroma_width,chroma_height,chroma_stride,bd); cr=cr16;
	281	}
	282	}
	283
178	284	sdlWin.display(y,cb,cr, stride, chroma_stride);
	285
	286	delete[] y16;
	287	delete[] cb16;
	288	delete[] cr16;
179	289
180	290	return sdlWin.doQuit();
181	291	}

232	342	}
233	343
234	344	return stop;
	345	}
	346
	347
	348	static double mse_y=0.0, mse_cb=0.0, mse_cr=0.0;
	349	static int mse_frames=0;
	350
	351	static double ssim_y=0.0;
	352	static int ssim_frames=0;
	353
	354	void measure(const de265_image* img)
	355	{
	356	// --- compute PSNR ---
	357
	358	int width = de265_get_image_width(img,0);
	359	int height = de265_get_image_height(img,0);
	360
	361	uint8_t* p = (uint8_t)malloc(widthheight*3/2);
	362	if (p == NULL) {
	363	return;
	364	}
	365
	366	size_t toread = widthheight3/2;
	367	if (fread(p,1,toread,reference_file) != toread) {
	368	free(p);
	369	return;
	370	}
	371
	372	int stride, cstride;
	373	const uint8_t* yptr = de265_get_image_plane(img,0, &stride);
	374	const uint8_t* cbptr = de265_get_image_plane(img,1, &cstride);
	375	const uint8_t* crptr = de265_get_image_plane(img,2, &cstride);
	376
	377	double img_mse_y = MSE( yptr, stride, p, width, width, height);
	378	double img_mse_cb = MSE(cbptr, cstride, p+width*height, width/2, width/2,height/2);
	379	double img_mse_cr = MSE(crptr, cstride, p+widthheight5/4, width/2, width/2,height/2);
	380
	381	mse_frames++;
	382
	383	mse_y += img_mse_y;
	384	mse_cb += img_mse_cb;
	385	mse_cr += img_mse_cr;
	386
	387
	388
	389	// --- compute SSIM ---
	390
	391	double ssimSum = 0.0;
	392
	393	#if HAVE_VIDEOGFX
	394	Bitmap<Pixel> ref, coded;
	395	ref .Create(width, height); // reference image
	396	coded.Create(width, height); // coded image
	397
	398	const uint8_t* data;
	399	data = de265_get_image_plane(img,0,&stride);
	400
	401	for (int y=0;y<height;y++) {
	402	memcpy(coded[y], data + y*stride, width);
	403	memcpy(ref[y], p + y*stride, width);
	404	}
	405
	406	SSIM ssimAlgo;
	407	Bitmap<float> ssim = ssimAlgo.calcSSIM(ref,coded);
	408
	409	Bitmap<Pixel> ssimMap;
	410	ssimMap.Create(width,height);
	411
	412	for (int y=0;y<height;y++)
	413	for (int x=0;x<width;x++)
	414	{
	415	float v = ssim[y][x];
	416	ssimSum += v;
	417	v = v*v;
	418	v = 255*v; //pow(v, 20);
	419
	420	//assert(v<=255.0);
	421	ssimMap[y][x] = v;
	422	}
	423
	424	ssimSum /= width*height;
	425
	426
	427	Bitmap<Pixel> error_map = CalcErrorMap(ref, coded, TransferCurve_Sqrt);
	428
	429
	430	// display PSNR error map
	431
	432	if (show_psnr_map) {
	433	static X11Win win;
	434	static bool first=true;
	435
	436	if (first) {
	437	first=false;
	438	win.Create(de265_get_image_width(img,0),
	439	de265_get_image_height(img,0),
	440	"psnr output");
	441	}
	442
	443	win.Display(MakeImage(error_map));
	444	}
	445
	446
	447	// display SSIM error map
	448
	449	if (show_ssim_map) {
	450	static X11Win win;
	451	static bool first=true;
	452
	453	if (first) {
	454	first=false;
	455	win.Create(de265_get_image_width(img,0),
	456	de265_get_image_height(img,0),
	457	"ssim output");
	458	}
	459
	460	win.Display(MakeImage(ssimMap));
	461	}
	462	#endif
	463
	464	ssim_frames++;
	465	ssim_y += ssimSum;
	466
	467	printf("%5d %6f %6f %6f %6f\n",
	468	framecnt,
	469	PSNR(img_mse_y), PSNR(img_mse_cb), PSNR(img_mse_cr),
	470	ssimSum);
	471
	472	free(p);
235	473	}
236	474
237	475

298	536	#endif
299	537	#endif
300	538
	539
301	540	int main(int argc, char** argv)
302	541	{
303	542	while (1) {
304	543	int option_index = 0;
305	544
306		int c = getopt_long(argc, argv, "qt:chf:o:dLB:n0vT:"
	545	int c = getopt_long(argc, argv, "qt:chf:o:dLB:n0vT:m:se"
307	546	#if HAVE_VIDEOGFX && HAVE_SDL
308	547	"V"
309	548	#endif

312	551	break;
313	552
314	553	switch (c) {
315		case 'q': quiet=true; break;
	554	case 'q': quiet++; break;
316	555	case 't': nThreads=atoi(optarg); break;
317	556	case 'c': check_hash=true; break;
318	557	case 'f': max_frames=atoi(optarg); break;

324	563	case 'L': logging=false; break;
325	564	case '0': no_acceleration=true; break;
326	565	case 'B': write_bytestream=true; bytestream_filename=optarg; break;
	566	case 'm': measure_quality=true; reference_filename=optarg; break;
	567	case 's': show_ssim_map=true; break;
	568	case 'e': show_psnr_map=true; break;
327	569	case 'T': highestTID=atoi(optarg); break;
328	570	case 'v': verbosity++; break;
329	571	}

347	589	fprintf(stderr," -V, --videogfx output with videogfx instead of SDL\n");
348	590	#endif
349	591	fprintf(stderr," -0, --noaccel do not use any accelerated code (SSE)\n");
	592	fprintf(stderr," -v, --verbose increase verbosity level (up to 3 times)\n");
350	593	fprintf(stderr," -L, --no-logging disable logging\n");
351	594	fprintf(stderr," -B, --write-bytestream FILENAME write raw bytestream (from NAL input)\n");
	595	fprintf(stderr," -m, --measure YUV compute PSNRs relative to reference YUV\n");
	596	#if HAVE_VIDEOGFX
	597	fprintf(stderr," -s, --ssim show SSIM-map (only when -m active)\n");
	598	fprintf(stderr," -e, --errmap show error-map (only when -m active)\n");
	599	#endif
352	600	fprintf(stderr," -T, --highest-TID select highest temporal sublayer to decode\n");
353	601	fprintf(stderr," --disable-deblocking disable deblocking filter\n");
354	602	fprintf(stderr," --disable-sao disable sample-adaptive offset filter\n");

394	642
395	643	de265_set_limit_TID(ctx, highestTID);
396	644
	645
	646	if (measure_quality) {
	647	reference_file = fopen(reference_filename, "rb");
	648	}
397	649
398	650
399	651	FILE* fh = fopen(argv[optind], "rb");

483	735
484	736	err = de265_decode(ctx, &more);
485	737	if (err != DE265_OK) {
	738	// if (quiet<=1) fprintf(stderr,"ERROR: %s\n", de265_get_error_text(err));
	739
486	740	if (check_hash && err == DE265_ERROR_CHECKSUM_MISMATCH)
487	741	stop = 1;
488	742	more = 0;

493	747
494	748	const de265_image* img = de265_get_next_picture(ctx);
495	749	if (img) {
	750	if (measure_quality) {
	751	measure(img);
	752	}
	753
496	754	stop = output_image(img);
497	755	if (stop) more=0;
498	756	else more=1;

506	764	break;
507	765	}
508	766
509		fprintf(stderr,"WARNING: %s\n", de265_get_error_text(warning));
	767	if (quiet<=1) fprintf(stderr,"WARNING: %s\n", de265_get_error_text(warning));
510	768	}
511	769	}
512	770	}

517	775	fclose(bytestream_fh);
518	776	}
519	777
	778	if (measure_quality) {
	779	printf("#total %6f %6f %6f %6f\n",
	780	PSNR(mse_y /mse_frames),
	781	PSNR(mse_cb/mse_frames),
	782	PSNR(mse_cr/mse_frames),
	783	ssim_y/ssim_frames);
	784
	785	fclose(reference_file);
	786	}
	787
520	788	de265_free_decoder(ctx);
521	789
522	790	struct timeval tv_end;
523	791	gettimeofday(&tv_end, NULL);
524	792
525	793	if (err != DE265_OK) {
526		fprintf(stderr,"decoding error: %s (code=%d)\n", de265_get_error_text(err), err);
	794	if (quiet<=1) fprintf(stderr,"decoding error: %s (code=%d)\n", de265_get_error_text(err), err);
527	795	}
528	796
529	797	double secs = tv_end.tv_sec-tv_start.tv_sec;
530	798	secs += (tv_end.tv_usec - tv_start.tv_usec)0.0010.001;
531	799
532		fprintf(stderr,"nFrames decoded: %d (%dx%d @ %5.2f fps)\n",framecnt,
533		width,height,framecnt/secs);
	800	if (quiet<=1) fprintf(stderr,"nFrames decoded: %d (%dx%d @ %5.2f fps)\n",framecnt,
	801	width,height,framecnt/secs);
534	802
535	803
536	804	return err==DE265_OK ? 0 : 10;

+118

-0

dec265/hdrcopy.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20
	21	#include "libde265/nal-parser.h"
	22	#include "libde265/decctx.h"
	23	#include <assert.h>
	24
	25	error_queue errqueue;
	26
	27	video_parameter_set vps;
	28	seq_parameter_set sps;
	29	pic_parameter_set pps;
	30
	31	CABAC_encoder_bitstream writer;
	32
	33
	34	void process_nal(NAL_unit* nal)
	35	{
	36	de265_error err = DE265_OK;
	37
	38	bitreader reader;
	39	bitreader_init(&reader, nal->data(), nal->size());
	40
	41	nal_header nal_hdr;
	42	nal_hdr.read(&reader);
	43	writer.write_startcode();
	44	nal_hdr.write(writer);
	45
	46	printf("NAL: 0x%x 0x%x - unit type:%s temporal id:%d\n",
	47	nal->data()[0], nal->data()[1],
	48	get_NAL_name(nal_hdr.nal_unit_type),
	49	nal_hdr.nuh_temporal_id);
	50
	51
	52	if (nal_hdr.nal_unit_type<32) {
	53	//err = read_slice_NAL(reader, nal, nal_hdr);
	54	}
	55	else switch (nal_hdr.nal_unit_type) {
	56	case NAL_UNIT_VPS_NUT:
	57	vps.read(&errqueue, &reader);
	58	vps.dump(1);
	59	vps.write(&errqueue, writer);
	60	writer.flush_VLC();
	61	break;
	62
	63	case NAL_UNIT_SPS_NUT:
	64	sps.read(&errqueue, &reader);
	65	sps.dump(1);
	66	sps.write(&errqueue, writer);
	67	writer.flush_VLC();
	68	break;
	69
	70	case NAL_UNIT_PPS_NUT:
	71	//err = read_pps_NAL(reader);
	72	break;
	73
	74	case NAL_UNIT_PREFIX_SEI_NUT:
	75	case NAL_UNIT_SUFFIX_SEI_NUT:
	76	//err = read_sei_NAL(reader, nal_hdr.nal_unit_type==NAL_UNIT_SUFFIX_SEI_NUT);
	77	break;
	78
	79	case NAL_UNIT_EOS_NUT:
	80	//ctx->FirstAfterEndOfSequenceNAL = true;
	81	break;
	82	}
	83	}
	84
	85
	86	int main(int argc, char** argv)
	87	{
	88	NAL_Parser nal_parser;
	89
	90	FILE* fh = fopen(argv[1],"rb");
	91	unsigned char buf[1024];
	92
	93	writer.write_bits(0,8); // because HM has an extra byte at the beginning
	94
	95	while(!feof(fh))
	96	{
	97	int n = fread(buf,1,1024,fh);
	98	if (n>0) {
	99	nal_parser.push_data(buf,n, 0);
	100	}
	101
	102	if (nal_parser.get_NAL_queue_length()>0) {
	103	NAL_unit* nal = nal_parser.pop_from_NAL_queue();
	104	assert(nal);
	105	process_nal(nal);
	106	nal_parser.free_NAL_unit(nal);
	107	}
	108	}
	109
	110	fclose(fh);
	111
	112	fh = fopen("out.bin","wb");
	113	fwrite(writer.data(), 1,writer.size(), fh);
	114	fclose(fh);
	115
	116	return 0;
	117	}

+144

-9

dec265/sdl.cc less more

18	18	*/
19	19
20	20	#include "sdl.hh"
21
22
23		bool SDL_YUV_Display::init(int frame_width, int frame_height)
	21	#include <assert.h>
	22
	23
	24	bool SDL_YUV_Display::init(int frame_width, int frame_height, enum SDL_Chroma chroma)
24	25	{
25	26	// reduce image size to a multiple of 8 (apparently required by YUV overlay)
26	27
27	28	frame_width &= ~7;
28	29	frame_height &= ~7;
29	30
	31	mChroma = chroma;
30	32
31	33	if (SDL_Init(SDL_INIT_VIDEO \| SDL_INIT_NOPARACHUTE) < 0 ) {
32	34	printf("SDL_Init() failed: %s\n", SDL_GetError( ) );

61	63	return false;
62	64	}
63	65
64		mYUVOverlay = SDL_CreateYUVOverlay(frame_width, frame_height, SDL_YV12_OVERLAY, mScreen);
	66	uint32_t pixelFormat;
	67	switch (mChroma) {
	68	case SDL_CHROMA_MONO: pixelFormat = SDL_YV12_OVERLAY; break;
	69	case SDL_CHROMA_420: pixelFormat = SDL_YV12_OVERLAY; break;
	70	case SDL_CHROMA_422: pixelFormat = SDL_YUY2_OVERLAY; break;
	71	case SDL_CHROMA_444: pixelFormat = SDL_YV12_OVERLAY; break;
	72	//case SDL_CHROMA_444: pixelFormat = SDL_YUY2_OVERLAY; break;
	73	}
	74
	75	mYUVOverlay = SDL_CreateYUVOverlay(frame_width, frame_height, pixelFormat, mScreen);
65	76	if (mYUVOverlay == NULL ) {
66	77	printf("SDL: Couldn't create SDL YUV overlay: %s",SDL_GetError());
67	78	SDL_Quit();

86	97	if (!mWindowOpen) return;
87	98	if (SDL_LockYUVOverlay(mYUVOverlay) < 0) return;
88	99
	100	if (mChroma == SDL_CHROMA_420) {
	101	display420(Y,U,V,stride,chroma_stride);
	102	}
	103	else if (mChroma == SDL_CHROMA_422) {
	104	display422(Y,U,V,stride,chroma_stride);
	105	}
	106	else if (mChroma == SDL_CHROMA_444) {
	107	display444as420(Y,U,V,stride,chroma_stride);
	108	//display444as422(Y,U,V,stride,chroma_stride);
	109	}
	110	else if (mChroma == SDL_CHROMA_MONO) {
	111	display400(Y,stride);
	112	}
	113
	114	SDL_UnlockYUVOverlay(mYUVOverlay);
	115
	116	SDL_DisplayYUVOverlay(mYUVOverlay, &rect);
	117	}
	118
	119
	120	void SDL_YUV_Display::display420(const unsigned char *Y,
	121	const unsigned char *U,
	122	const unsigned char *V,
	123	int stride, int chroma_stride)
	124	{
89	125	if (stride == rect.w && chroma_stride == rect.w/2) {
90	126
91	127	// fast copy

108	144	memcpy(mYUVOverlay->pixels[1]+yrect.w/2, V+chroma_stridey, rect.w/2);
109	145	}
110	146	}
111
112		SDL_UnlockYUVOverlay(mYUVOverlay);
113
114		SDL_DisplayYUVOverlay(mYUVOverlay, &rect);
115		}
	147	}
	148
	149
	150	void SDL_YUV_Display::display400(const unsigned char *Y, int stride)
	151	{
	152	if (stride == rect.w) {
	153
	154	// fast copy
	155
	156	memcpy(mYUVOverlay->pixels[0], Y, rect.w * rect.h);
	157	}
	158	else {
	159	// copy line by line, because sizes are different
	160
	161	for (int y=0;y<rect.h;y++)
	162	{
	163	memcpy(mYUVOverlay->pixels[0]+yrect.w, Y+stridey, rect.w);
	164	}
	165	}
	166
	167	// clear chroma planes
	168
	169	memset(mYUVOverlay->pixels[1], 0x80, rect.w * rect.h / 4);
	170	memset(mYUVOverlay->pixels[2], 0x80, rect.w * rect.h / 4);
	171	}
	172
	173
	174	void SDL_YUV_Display::display422(const unsigned char *Y,
	175	const unsigned char *U,
	176	const unsigned char *V,
	177	int stride, int chroma_stride)
	178	{
	179	for (int y=0;y<rect.h;y++)
	180	{
	181	unsigned char* p = mYUVOverlay->pixels[0] + yrect.w 2;
	182
	183	const unsigned char* Yp = Y + y*stride;
	184	const unsigned char* Up = U + y*chroma_stride;
	185	const unsigned char* Vp = V + y*chroma_stride;
	186
	187	for (int x=0;x<rect.w;x+=2) {
	188	*p++ = Yp[x];
	189	*p++ = Up[x/2];
	190	*p++ = Yp[x+1];
	191	*p++ = Vp[x/2];
	192	}
	193	}
	194	}
	195
	196
	197	/* This converts down 4:4:4 input to 4:2:2 for display, as SDL does not support
	198	any 4:4:4 pixel format.
	199	*/
	200	void SDL_YUV_Display::display444as422(const unsigned char *Y,
	201	const unsigned char *U,
	202	const unsigned char *V,
	203	int stride, int chroma_stride)
	204	{
	205	for (int y=0;y<rect.h;y++)
	206	{
	207	unsigned char* p = mYUVOverlay->pixels[0] + yrect.w 2;
	208
	209	const unsigned char* Yp = Y + y*stride;
	210	const unsigned char* Up = U + y*chroma_stride;
	211	const unsigned char* Vp = V + y*chroma_stride;
	212
	213	for (int x=0;x<rect.w;x+=2) {
	214	*p++ = Yp[x];
	215	*p++ = Up[x];
	216	*p++ = Yp[x+1];
	217	*p++ = Vp[x];
	218	}
	219	}
	220	}
	221
	222
	223	void SDL_YUV_Display::display444as420(const unsigned char *Y,
	224	const unsigned char *U,
	225	const unsigned char *V,
	226	int stride, int chroma_stride)
	227	{
	228	for (int y=0;y<rect.h;y++)
	229	{
	230	unsigned char* p = mYUVOverlay->pixels[0] + y*rect.w;
	231	memcpy(p, Y+y*stride, rect.w);
	232	}
	233
	234	for (int y=0;y<rect.h;y+=2)
	235	{
	236	unsigned char* u = mYUVOverlay->pixels[2] + y/2*rect.w/2;
	237	unsigned char* v = mYUVOverlay->pixels[1] + y/2*rect.w/2;
	238
	239	for (int x=0;x<rect.w;x+=2) {
	240	u[x/2] = (U[ y chroma_stride + x] + U[ y chroma_stride + x +1] +
	241	U[(y+1)chroma_stride + x] + U[(y+1)chroma_stride + x +1])/4;
	242	v[x/2] = (V[ y chroma_stride + x] + V[ y chroma_stride + x +1] +
	243	V[(y+1)chroma_stride + x] + V[(y+1)chroma_stride + x +1])/4;
	244
	245	//u[x/2] = U[y*chroma_stride + x];
	246	//v[x/2] = V[y*chroma_stride + x];
	247	}
	248	}
	249	}
	250
116	251
117	252	bool SDL_YUV_Display::doQuit() const
118	253	{

+29

-1

dec265/sdl.hh less more

24	24	{
25	25	public:
26	26
27		bool init(int frame_width, int frame_height);
	27	enum SDL_Chroma {
	28	SDL_CHROMA_MONO=400,
	29	SDL_CHROMA_420 =420,
	30	SDL_CHROMA_422 =422,
	31	SDL_CHROMA_444 =444
	32	};
	33
	34	bool init(int frame_width, int frame_height, enum SDL_Chroma chroma = SDL_CHROMA_420);
28	35	void display(const unsigned char Y, const unsigned char U, const unsigned char *V,
29	36	int stride, int chroma_stride);
30	37	void close();

38	45	SDL_Overlay *mYUVOverlay;
39	46	SDL_Rect rect;
40	47	bool mWindowOpen;
	48
	49	SDL_Chroma mChroma;
	50
	51	void display400(const unsigned char *Y,
	52	int stride);
	53	void display420(const unsigned char *Y,
	54	const unsigned char *U,
	55	const unsigned char *V,
	56	int stride, int chroma_stride);
	57	void display422(const unsigned char *Y,
	58	const unsigned char *U,
	59	const unsigned char *V,
	60	int stride, int chroma_stride);
	61	void display444as422(const unsigned char *Y,
	62	const unsigned char *U,
	63	const unsigned char *V,
	64	int stride, int chroma_stride);
	65	void display444as420(const unsigned char *Y,
	66	const unsigned char *U,
	67	const unsigned char *V,
	68	int stride, int chroma_stride);
41	69	};

+15

-0

enc265/CMakeLists.txt less more

	0	set (enc265_sources
	1	enc265.cc
	2	)
	3
	4	if(MSVC)
	5	set (enc265_sources
	6	${enc265_sources}
	7	../extra/getopt.c
	8	../extra/getopt_long.c
	9	)
	10	endif()
	11
	12	add_executable (enc265 ${enc265_sources})
	13
	14	target_link_libraries (enc265 ${LIBDE265_LIBRARY_NAME})

+19

-0

enc265/Makefile.am less more

	0
	1	bin_PROGRAMS = enc265
	2
	3	AM_CPPFLAGS = -I../libde265
	4
	5	enc265_DEPENDENCIES = ../libde265/libde265.la
	6	enc265_CXXFLAGS =
	7	enc265_LDFLAGS =
	8	enc265_LDADD = ../libde265/libde265.la -lstdc++
	9	enc265_SOURCES = enc265.cc
	10
	11	if HAVE_VIDEOGFX
	12	enc265_CXXFLAGS += $(VIDEOGFX_CFLAGS)
	13	enc265_LDFLAGS += $(VIDEOGFX_LIBS)
	14	endif
	15
	16	EXTRA_DIST = \
	17	CMakeLists.txt \
	18	Makefile.vc7

+674

-0

enc265/Makefile.in less more

	0	# Makefile.in generated by automake 1.14.1 from Makefile.am.
	1	# @configure_input@
	2
	3	# Copyright (C) 1994-2013 Free Software Foundation, Inc.
	4
	5	# This Makefile.in is free software; the Free Software Foundation
	6	# gives unlimited permission to copy and/or distribute it,
	7	# with or without modifications, as long as this notice is preserved.
	8
	9	# This program is distributed in the hope that it will be useful,
	10	# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
	11	# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
	12	# PARTICULAR PURPOSE.
	13
	14	@SET_MAKE@
	15
	16	VPATH = @srcdir@
	17	am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
	18	am__make_running_with_option = \
	19	case $${target_option-} in \
	20	?) ;; \
	21	*) echo "am__make_running_with_option: internal error: invalid" \
	22	"target option '$${target_option-}' specified" >&2; \
	23	exit 1;; \
	24	esac; \
	25	has_opt=no; \
	26	sane_makeflags=$$MAKEFLAGS; \
	27	if $(am__is_gnu_make); then \
	28	sane_makeflags=$$MFLAGS; \
	29	else \
	30	case $$MAKEFLAGS in \
	31	\\[\ \ ]) \
	32	bs=\\; \
	33	sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
	34	\| sed "s/$$bs$$bs[$$bs $$bs ]*//g"`;; \
	35	esac; \
	36	fi; \
	37	skip_next=no; \
	38	strip_trailopt () \
	39	{ \
	40	flg=`printf '%s\n' "$$flg" \| sed "s/$$1.*$$//"`; \
	41	}; \
	42	for flg in $$sane_makeflags; do \
	43	test $$skip_next = yes && { skip_next=no; continue; }; \
	44	case $$flg in \
	45	=\|--*) continue;; \
	46	-*I) strip_trailopt 'I'; skip_next=yes;; \
	47	-I?) strip_trailopt 'I';; \
	48	-*O) strip_trailopt 'O'; skip_next=yes;; \
	49	-O?) strip_trailopt 'O';; \
	50	-*l) strip_trailopt 'l'; skip_next=yes;; \
	51	-l?) strip_trailopt 'l';; \
	52	-[dEDm]) skip_next=yes;; \
	53	-[JT]) skip_next=yes;; \
	54	esac; \
	55	case $$flg in \
	56	$$target_option) has_opt=yes; break;; \
	57	esac; \
	58	done; \
	59	test $$has_opt = yes
	60	am__make_dryrun = (target_option=n; $(am__make_running_with_option))
	61	am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
	62	pkgdatadir = $(datadir)/@PACKAGE@
	63	pkgincludedir = $(includedir)/@PACKAGE@
	64	pkglibdir = $(libdir)/@PACKAGE@
	65	pkglibexecdir = $(libexecdir)/@PACKAGE@
	66	am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
	67	install_sh_DATA = $(install_sh) -c -m 644
	68	install_sh_PROGRAM = $(install_sh) -c
	69	install_sh_SCRIPT = $(install_sh) -c
	70	INSTALL_HEADER = $(INSTALL_DATA)
	71	transform = $(program_transform_name)
	72	NORMAL_INSTALL = :
	73	PRE_INSTALL = :
	74	POST_INSTALL = :
	75	NORMAL_UNINSTALL = :
	76	PRE_UNINSTALL = :
	77	POST_UNINSTALL = :
	78	build_triplet = @build@
	79	host_triplet = @host@
	80	target_triplet = @target@
	81	bin_PROGRAMS = enc265$(EXEEXT)
	82	@HAVE_VIDEOGFX_TRUE@am__append_1 = $(VIDEOGFX_CFLAGS)
	83	@HAVE_VIDEOGFX_TRUE@am__append_2 = $(VIDEOGFX_LIBS)
	84	subdir = enc265
	85	DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/Makefile.am \
	86	$(top_srcdir)/depcomp
	87	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
	88	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	89	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
	90	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
	91	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
	92	$(top_srcdir)/m4/lt~obsolete.m4 \
	93	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
	94	$(top_srcdir)/configure.ac
	95	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
	96	$(ACLOCAL_M4)
	97	mkinstalldirs = $(install_sh) -d
	98	CONFIG_HEADER = $(top_builddir)/config.h
	99	CONFIG_CLEAN_FILES =
	100	CONFIG_CLEAN_VPATH_FILES =
	101	am__installdirs = "$(DESTDIR)$(bindir)"
	102	PROGRAMS = $(bin_PROGRAMS)
	103	am_enc265_OBJECTS = enc265-enc265.$(OBJEXT)
	104	enc265_OBJECTS = $(am_enc265_OBJECTS)
	105	AM_V_lt = $(am__v_lt_@AM_V@)
	106	am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
	107	am__v_lt_0 = --silent
	108	am__v_lt_1 =
	109	enc265_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	110	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(enc265_CXXFLAGS) \
	111	$(CXXFLAGS) $(enc265_LDFLAGS) $(LDFLAGS) -o $@
	112	AM_V_P = $(am__v_P_@AM_V@)
	113	am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
	114	am__v_P_0 = false
	115	am__v_P_1 = :
	116	AM_V_GEN = $(am__v_GEN_@AM_V@)
	117	am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
	118	am__v_GEN_0 = @echo " GEN " $@;
	119	am__v_GEN_1 =
	120	AM_V_at = $(am__v_at_@AM_V@)
	121	am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
	122	am__v_at_0 = @
	123	am__v_at_1 =
	124	DEFAULT_INCLUDES = -I.@am__isrc@ -I$(top_builddir)
	125	depcomp = $(SHELL) $(top_srcdir)/depcomp
	126	am__depfiles_maybe = depfiles
	127	am__mv = mv -f
	128	CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
	129	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
	130	LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	131	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
	132	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	133	$(AM_CXXFLAGS) $(CXXFLAGS)
	134	AM_V_CXX = $(am__v_CXX_@AM_V@)
	135	am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
	136	am__v_CXX_0 = @echo " CXX " $@;
	137	am__v_CXX_1 =
	138	CXXLD = $(CXX)
	139	CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	140	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
	141	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
	142	AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
	143	am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
	144	am__v_CXXLD_0 = @echo " CXXLD " $@;
	145	am__v_CXXLD_1 =
	146	SOURCES = $(enc265_SOURCES)
	147	DIST_SOURCES = $(enc265_SOURCES)
	148	am__can_run_installinfo = \
	149	case $$AM_UPDATE_INFO_DIR in \
	150	n\|no\|NO) false;; \
	151	*) (install-info --version) >/dev/null 2>&1;; \
	152	esac
	153	am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
	154	# Read a list of newline-separated strings from the standard input,
	155	# and print each of them once, without duplicates. Input order is
	156	# not preserved.
	157	am__uniquify_input = $(AWK) '\
	158	BEGIN { nonempty = 0; } \
	159	{ items[$$0] = 1; nonempty = 1; } \
	160	END { if (nonempty) { for (i in items) print i; }; } \
	161	'
	162	# Make sure the list of sources is unique. This is necessary because,
	163	# e.g., the same source file might be shared among _SOURCES variables
	164	# for different programs/libraries.
	165	am__define_uniq_tagged_files = \
	166	list='$(am__tagged_files)'; \
	167	unique=`for i in $$list; do \
	168	if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
	169	done \| $(am__uniquify_input)`
	170	ETAGS = etags
	171	CTAGS = ctags
	172	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
	173	ACLOCAL = @ACLOCAL@
	174	ALLOCA = @ALLOCA@
	175	AMTAR = @AMTAR@
	176	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
	177	AR = @AR@
	178	AUTOCONF = @AUTOCONF@
	179	AUTOHEADER = @AUTOHEADER@
	180	AUTOMAKE = @AUTOMAKE@
	181	AWK = @AWK@
	182	CC = @CC@
	183	CCAS = @CCAS@
	184	CCASDEPMODE = @CCASDEPMODE@
	185	CCASFLAGS = @CCASFLAGS@
	186	CCDEPMODE = @CCDEPMODE@
	187	CFLAGS = @CFLAGS@
	188	CPP = @CPP@
	189	CPPFLAGS = @CPPFLAGS@
	190	CXX = @CXX@
	191	CXXCPP = @CXXCPP@
	192	CXXDEPMODE = @CXXDEPMODE@
	193	CXXFLAGS = @CXXFLAGS@
	194	CYGPATH_W = @CYGPATH_W@
	195	DEFS = @DEFS@
	196	DEPDIR = @DEPDIR@
	197	DLLTOOL = @DLLTOOL@
	198	DSYMUTIL = @DSYMUTIL@
	199	DUMPBIN = @DUMPBIN@
	200	ECHO_C = @ECHO_C@
	201	ECHO_N = @ECHO_N@
	202	ECHO_T = @ECHO_T@
	203	EGREP = @EGREP@
	204	EXEEXT = @EXEEXT@
	205	FGREP = @FGREP@
	206	GREP = @GREP@
	207	HAVE_CXX11 = @HAVE_CXX11@
	208	INSTALL = @INSTALL@
	209	INSTALL_DATA = @INSTALL_DATA@
	210	INSTALL_PROGRAM = @INSTALL_PROGRAM@
	211	INSTALL_SCRIPT = @INSTALL_SCRIPT@
	212	INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
	213	LD = @LD@
	214	LDFLAGS = @LDFLAGS@
	215	LIBDE265_AGE = @LIBDE265_AGE@
	216	LIBDE265_CURRENT = @LIBDE265_CURRENT@
	217	LIBDE265_REVISION = @LIBDE265_REVISION@
	218	LIBOBJS = @LIBOBJS@
	219	LIBS = @LIBS@
	220	LIBTOOL = @LIBTOOL@
	221	LIPO = @LIPO@
	222	LN_S = @LN_S@
	223	LTLIBOBJS = @LTLIBOBJS@
	224	MAKEINFO = @MAKEINFO@
	225	MANIFEST_TOOL = @MANIFEST_TOOL@
	226	MKDIR_P = @MKDIR_P@
	227	NM = @NM@
	228	NMEDIT = @NMEDIT@
	229	NUMERIC_VERSION = @NUMERIC_VERSION@
	230	OBJDUMP = @OBJDUMP@
	231	OBJEXT = @OBJEXT@
	232	OTOOL = @OTOOL@
	233	OTOOL64 = @OTOOL64@
	234	PACKAGE = @PACKAGE@
	235	PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
	236	PACKAGE_NAME = @PACKAGE_NAME@
	237	PACKAGE_STRING = @PACKAGE_STRING@
	238	PACKAGE_TARNAME = @PACKAGE_TARNAME@
	239	PACKAGE_URL = @PACKAGE_URL@
	240	PACKAGE_VERSION = @PACKAGE_VERSION@
	241	PATH_SEPARATOR = @PATH_SEPARATOR@
	242	PKG_CONFIG = @PKG_CONFIG@
	243	PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
	244	PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
	245	QTCHOOSER = @QTCHOOSER@
	246	QTMOC = @QTMOC@
	247	QT_CFLAGS = @QT_CFLAGS@
	248	QT_LIBS = @QT_LIBS@
	249	RANLIB = @RANLIB@
	250	SDL_CFLAGS = @SDL_CFLAGS@
	251	SDL_LIBS = @SDL_LIBS@
	252	SED = @SED@
	253	SET_MAKE = @SET_MAKE@
	254	SHELL = @SHELL@
	255	STRIP = @STRIP@
	256	SWSCALE_CFLAGS = @SWSCALE_CFLAGS@
	257	SWSCALE_LIBS = @SWSCALE_LIBS@
	258	VERSION = @VERSION@
	259	VIDEOGFX_CFLAGS = @VIDEOGFX_CFLAGS@
	260	VIDEOGFX_LIBS = @VIDEOGFX_LIBS@
	261	abs_builddir = @abs_builddir@
	262	abs_srcdir = @abs_srcdir@
	263	abs_top_builddir = @abs_top_builddir@
	264	abs_top_srcdir = @abs_top_srcdir@
	265	ac_ct_AR = @ac_ct_AR@
	266	ac_ct_CC = @ac_ct_CC@
	267	ac_ct_CXX = @ac_ct_CXX@
	268	ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
	269	am__include = @am__include@
	270	am__leading_dot = @am__leading_dot@
	271	am__quote = @am__quote@
	272	am__tar = @am__tar@
	273	am__untar = @am__untar@
	274	bindir = @bindir@
	275	build = @build@
	276	build_alias = @build_alias@
	277	build_cpu = @build_cpu@
	278	build_os = @build_os@
	279	build_vendor = @build_vendor@
	280	builddir = @builddir@
	281	datadir = @datadir@
	282	datarootdir = @datarootdir@
	283	docdir = @docdir@
	284	dvidir = @dvidir@
	285	exec_prefix = @exec_prefix@
	286	host = @host@
	287	host_alias = @host_alias@
	288	host_cpu = @host_cpu@
	289	host_os = @host_os@
	290	host_vendor = @host_vendor@
	291	htmldir = @htmldir@
	292	includedir = @includedir@
	293	infodir = @infodir@
	294	install_sh = @install_sh@
	295	libdir = @libdir@
	296	libexecdir = @libexecdir@
	297	localedir = @localedir@
	298	localstatedir = @localstatedir@
	299	mandir = @mandir@
	300	mkdir_p = @mkdir_p@
	301	oldincludedir = @oldincludedir@
	302	pdfdir = @pdfdir@
	303	prefix = @prefix@
	304	program_transform_name = @program_transform_name@
	305	psdir = @psdir@
	306	sbindir = @sbindir@
	307	sharedstatedir = @sharedstatedir@
	308	srcdir = @srcdir@
	309	sysconfdir = @sysconfdir@
	310	target = @target@
	311	target_alias = @target_alias@
	312	target_cpu = @target_cpu@
	313	target_os = @target_os@
	314	target_vendor = @target_vendor@
	315	top_build_prefix = @top_build_prefix@
	316	top_builddir = @top_builddir@
	317	top_srcdir = @top_srcdir@
	318	AM_CPPFLAGS = -I../libde265
	319	enc265_DEPENDENCIES = ../libde265/libde265.la
	320	enc265_CXXFLAGS = $(am__append_1)
	321	enc265_LDFLAGS = $(am__append_2)
	322	enc265_LDADD = ../libde265/libde265.la -lstdc++
	323	enc265_SOURCES = enc265.cc
	324	EXTRA_DIST = \
	325	CMakeLists.txt \
	326	Makefile.vc7
	327
	328	all: all-am
	329
	330	.SUFFIXES:
	331	.SUFFIXES: .cc .lo .o .obj
	332	$(srcdir)/Makefile.in: $(srcdir)/Makefile.am $(am__configure_deps)
	333	@for dep in $?; do \
	334	case '$(am__configure_deps)' in \
	335	$$dep) \
	336	( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
	337	&& { if test -f $@; then exit 0; else break; fi; }; \
	338	exit 1;; \
	339	esac; \
	340	done; \
	341	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu enc265/Makefile'; \
	342	$(am__cd) $(top_srcdir) && \
	343	$(AUTOMAKE) --gnu enc265/Makefile
	344	.PRECIOUS: Makefile
	345	Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
	346	@case '$?' in \
	347	config.status) \
	348	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
	349	*) \
	350	echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
	351	cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
	352	esac;
	353
	354	$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
	355	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	356
	357	$(top_srcdir)/configure: $(am__configure_deps)
	358	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	359	$(ACLOCAL_M4): $(am__aclocal_m4_deps)
	360	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	361	$(am__aclocal_m4_deps):
	362	install-binPROGRAMS: $(bin_PROGRAMS)
	363	@$(NORMAL_INSTALL)
	364	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" \|\| list=; \
	365	if test -n "$$list"; then \
	366	echo " $(MKDIR_P) '$(DESTDIR)$(bindir)'"; \
	367	$(MKDIR_P) "$(DESTDIR)$(bindir)" \|\| exit 1; \
	368	fi; \
	369	for p in $$list; do echo "$$p $$p"; done \| \
	370	sed 's/$(EXEEXT)$$//' \| \
	371	while read p p1; do if test -f $$p \
	372	\|\| test -f $$p1 \
	373	; then echo "$$p"; echo "$$p"; else :; fi; \
	374	done \| \
	375	sed -e 'p;s,.*/,,;n;h' \
	376	-e 's\|.*\|.\|' \
	377	-e 'p;x;s,.*/,,;s/$(EXEEXT)$$//;$(transform);s/$$/$(EXEEXT)/' \| \
	378	sed 'N;N;N;s,\n, ,g' \| \
	379	$(AWK) 'BEGIN { files["."] = ""; dirs["."] = 1 } \
	380	{ d=$$3; if (dirs[d] != 1) { print "d", d; dirs[d] = 1 } \
	381	if ($$2 == $$4) files[d] = files[d] " " $$1; \
	382	else { print "f", $$3 "/" $$4, $$1; } } \
	383	END { for (d in files) print "f", d, files[d] }' \| \
	384	while read type dir files; do \
	385	if test "$$dir" = .; then dir=; else dir=/$$dir; fi; \
	386	test -z "$$files" \|\| { \
	387	echo " $(INSTALL_PROGRAM_ENV) $(LIBTOOL) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=install $(INSTALL_PROGRAM) $$files '$(DESTDIR)$(bindir)$$dir'"; \
	388	$(INSTALL_PROGRAM_ENV) $(LIBTOOL) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=install $(INSTALL_PROGRAM) $$files "$(DESTDIR)$(bindir)$$dir" \|\| exit $$?; \
	389	} \
	390	; done
	391
	392	uninstall-binPROGRAMS:
	393	@$(NORMAL_UNINSTALL)
	394	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" \|\| list=; \
	395	files=`for p in $$list; do echo "$$p"; done \| \
	396	sed -e 'h;s,^.*/,,;s/$(EXEEXT)$$//;$(transform)' \
	397	-e 's/$$/$(EXEEXT)/' \
	398	`; \
	399	test -n "$$list" \|\| exit 0; \
	400	echo " ( cd '$(DESTDIR)$(bindir)' && rm -f" $$files ")"; \
	401	cd "$(DESTDIR)$(bindir)" && rm -f $$files
	402
	403	clean-binPROGRAMS:
	404	@list='$(bin_PROGRAMS)'; test -n "$$list" \|\| exit 0; \
	405	echo " rm -f" $$list; \
	406	rm -f $$list \|\| exit $$?; \
	407	test -n "$(EXEEXT)" \|\| exit 0; \
	408	list=`for p in $$list; do echo "$$p"; done \| sed 's/$(EXEEXT)$$//'`; \
	409	echo " rm -f" $$list; \
	410	rm -f $$list
	411
	412	enc265$(EXEEXT): $(enc265_OBJECTS) $(enc265_DEPENDENCIES) $(EXTRA_enc265_DEPENDENCIES)
	413	@rm -f enc265$(EXEEXT)
	414	$(AM_V_CXXLD)$(enc265_LINK) $(enc265_OBJECTS) $(enc265_LDADD) $(LIBS)
	415
	416	mostlyclean-compile:
	417	-rm -f *.$(OBJEXT)
	418
	419	distclean-compile:
	420	-rm -f *.tab.c
	421
	422	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/enc265-enc265.Po@am__quote@
	423
	424	.cc.o:
	425	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	426	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	427	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	428	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	429	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
	430
	431	.cc.obj:
	432	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
	433	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	434	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	435	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	436	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
	437
	438	.cc.lo:
	439	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	440	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Plo
	441	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
	442	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	443	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
	444
	445	enc265-enc265.o: enc265.cc
	446	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(enc265_CXXFLAGS) $(CXXFLAGS) -MT enc265-enc265.o -MD -MP -MF $(DEPDIR)/enc265-enc265.Tpo -c -o enc265-enc265.o `test -f 'enc265.cc' \|\| echo '$(srcdir)/'`enc265.cc
	447	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/enc265-enc265.Tpo $(DEPDIR)/enc265-enc265.Po
	448	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='enc265.cc' object='enc265-enc265.o' libtool=no @AMDEPBACKSLASH@
	449	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	450	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(enc265_CXXFLAGS) $(CXXFLAGS) -c -o enc265-enc265.o `test -f 'enc265.cc' \|\| echo '$(srcdir)/'`enc265.cc
	451
	452	enc265-enc265.obj: enc265.cc
	453	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(enc265_CXXFLAGS) $(CXXFLAGS) -MT enc265-enc265.obj -MD -MP -MF $(DEPDIR)/enc265-enc265.Tpo -c -o enc265-enc265.obj `if test -f 'enc265.cc'; then $(CYGPATH_W) 'enc265.cc'; else $(CYGPATH_W) '$(srcdir)/enc265.cc'; fi`
	454	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/enc265-enc265.Tpo $(DEPDIR)/enc265-enc265.Po
	455	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='enc265.cc' object='enc265-enc265.obj' libtool=no @AMDEPBACKSLASH@
	456	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	457	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(enc265_CXXFLAGS) $(CXXFLAGS) -c -o enc265-enc265.obj `if test -f 'enc265.cc'; then $(CYGPATH_W) 'enc265.cc'; else $(CYGPATH_W) '$(srcdir)/enc265.cc'; fi`
	458
	459	mostlyclean-libtool:
	460	-rm -f *.lo
	461
	462	clean-libtool:
	463	-rm -rf .libs _libs
	464
	465	ID: $(am__tagged_files)
	466	$(am__define_uniq_tagged_files); mkid -fID $$unique
	467	tags: tags-am
	468	TAGS: tags
	469
	470	tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	471	set x; \
	472	here=`pwd`; \
	473	$(am__define_uniq_tagged_files); \
	474	shift; \
	475	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
	476	test -n "$$unique" \|\| unique=$$empty_fix; \
	477	if test $$# -gt 0; then \
	478	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	479	"$$@" $$unique; \
	480	else \
	481	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	482	$$unique; \
	483	fi; \
	484	fi
	485	ctags: ctags-am
	486
	487	CTAGS: ctags
	488	ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	489	$(am__define_uniq_tagged_files); \
	490	test -z "$(CTAGS_ARGS)$$unique" \
	491	\|\| $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
	492	$$unique
	493
	494	GTAGS:
	495	here=`$(am__cd) $(top_builddir) && pwd` \
	496	&& $(am__cd) $(top_srcdir) \
	497	&& gtags -i $(GTAGS_ARGS) "$$here"
	498	cscopelist: cscopelist-am
	499
	500	cscopelist-am: $(am__tagged_files)
	501	list='$(am__tagged_files)'; \
	502	case "$(srcdir)" in \
	503	[\\/]* \| ?:[\\/]*) sdir="$(srcdir)" ;; \
	504	*) sdir=$(subdir)/$(srcdir) ;; \
	505	esac; \
	506	for i in $$list; do \
	507	if test -f "$$i"; then \
	508	echo "$(subdir)/$$i"; \
	509	else \
	510	echo "$$sdir/$$i"; \
	511	fi; \
	512	done >> $(top_builddir)/cscope.files
	513
	514	distclean-tags:
	515	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
	516
	517	distdir: $(DISTFILES)
	518	@srcdirstrip=`echo "$(srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	519	topsrcdirstrip=`echo "$(top_srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	520	list='$(DISTFILES)'; \
	521	dist_files=`for file in $$list; do echo $$file; done \| \
	522	sed -e "s\|^$$srcdirstrip/\|\|;t" \
	523	-e "s\|^$$topsrcdirstrip/\|$(top_builddir)/\|;t"`; \
	524	case $$dist_files in \
	525	/) $(MKDIR_P) `echo "$$dist_files" \| \
	526	sed '/\//!d;s\|^\|$(distdir)/\|;s,/[^/]*$$,,' \| \
	527	sort -u` ;; \
	528	esac; \
	529	for file in $$dist_files; do \
	530	if test -f $$file \|\| test -d $$file; then d=.; else d=$(srcdir); fi; \
	531	if test -d $$d/$$file; then \
	532	dir=`echo "/$$file" \| sed -e 's,/[^/]*$$,,'`; \
	533	if test -d "$(distdir)/$$file"; then \
	534	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	535	fi; \
	536	if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
	537	cp -fpR $(srcdir)/$$file "$(distdir)$$dir" \|\| exit 1; \
	538	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	539	fi; \
	540	cp -fpR $$d/$$file "$(distdir)$$dir" \|\| exit 1; \
	541	else \
	542	test -f "$(distdir)/$$file" \
	543	\|\| cp -p $$d/$$file "$(distdir)/$$file" \
	544	\|\| exit 1; \
	545	fi; \
	546	done
	547	check-am: all-am
	548	check: check-am
	549	all-am: Makefile $(PROGRAMS)
	550	installdirs:
	551	for dir in "$(DESTDIR)$(bindir)"; do \
	552	test -z "$$dir" \|\| $(MKDIR_P) "$$dir"; \
	553	done
	554	install: install-am
	555	install-exec: install-exec-am
	556	install-data: install-data-am
	557	uninstall: uninstall-am
	558
	559	install-am: all-am
	560	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
	561
	562	installcheck: installcheck-am
	563	install-strip:
	564	if test -z '$(STRIP)'; then \
	565	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	566	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	567	install; \
	568	else \
	569	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	570	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	571	"INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
	572	fi
	573	mostlyclean-generic:
	574
	575	clean-generic:
	576
	577	distclean-generic:
	578	-test -z "$(CONFIG_CLEAN_FILES)" \|\| rm -f $(CONFIG_CLEAN_FILES)
	579	-test . = "$(srcdir)" \|\| test -z "$(CONFIG_CLEAN_VPATH_FILES)" \|\| rm -f $(CONFIG_CLEAN_VPATH_FILES)
	580
	581	maintainer-clean-generic:
	582	@echo "This command is intended for maintainers to use"
	583	@echo "it deletes files that may require special tools to rebuild."
	584	clean: clean-am
	585
	586	clean-am: clean-binPROGRAMS clean-generic clean-libtool mostlyclean-am
	587
	588	distclean: distclean-am
	589	-rm -rf ./$(DEPDIR)
	590	-rm -f Makefile
	591	distclean-am: clean-am distclean-compile distclean-generic \
	592	distclean-tags
	593
	594	dvi: dvi-am
	595
	596	dvi-am:
	597
	598	html: html-am
	599
	600	html-am:
	601
	602	info: info-am
	603
	604	info-am:
	605
	606	install-data-am:
	607
	608	install-dvi: install-dvi-am
	609
	610	install-dvi-am:
	611
	612	install-exec-am: install-binPROGRAMS
	613
	614	install-html: install-html-am
	615
	616	install-html-am:
	617
	618	install-info: install-info-am
	619
	620	install-info-am:
	621
	622	install-man:
	623
	624	install-pdf: install-pdf-am
	625
	626	install-pdf-am:
	627
	628	install-ps: install-ps-am
	629
	630	install-ps-am:
	631
	632	installcheck-am:
	633
	634	maintainer-clean: maintainer-clean-am
	635	-rm -rf ./$(DEPDIR)
	636	-rm -f Makefile
	637	maintainer-clean-am: distclean-am maintainer-clean-generic
	638
	639	mostlyclean: mostlyclean-am
	640
	641	mostlyclean-am: mostlyclean-compile mostlyclean-generic \
	642	mostlyclean-libtool
	643
	644	pdf: pdf-am
	645
	646	pdf-am:
	647
	648	ps: ps-am
	649
	650	ps-am:
	651
	652	uninstall-am: uninstall-binPROGRAMS
	653
	654	.MAKE: install-am install-strip
	655
	656	.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean \
	657	clean-binPROGRAMS clean-generic clean-libtool cscopelist-am \
	658	ctags ctags-am distclean distclean-compile distclean-generic \
	659	distclean-libtool distclean-tags distdir dvi dvi-am html \
	660	html-am info info-am install install-am install-binPROGRAMS \
	661	install-data install-data-am install-dvi install-dvi-am \
	662	install-exec install-exec-am install-html install-html-am \
	663	install-info install-info-am install-man install-pdf \
	664	install-pdf-am install-ps install-ps-am install-strip \
	665	installcheck installcheck-am installdirs maintainer-clean \
	666	maintainer-clean-generic mostlyclean mostlyclean-compile \
	667	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
	668	tags tags-am uninstall uninstall-am uninstall-binPROGRAMS
	669
	670
	671	# Tell versions [3.59,3.63) of GNU make to not export all variables.
	672	# Otherwise a system limit (for SysV at least) may be exceeded.
	673	.NOEXPORT:

+37

-0

enc265/Makefile.vc7 less more

	0	#
	1	# Makefile for Microsoft Visual Studio 2003
	2	#
	3	CFLAGS=/I.. /I..\libde265 /I..\extra
	4	CC=cl /nologo
	5	LINK=link /nologo /subsystem:console
	6	DEFINES=/DWIN32
	7
	8	CFLAGS=$(CFLAGS) /MT /Ob2 /Oi /W4 /EHsc
	9	CFLAGS=$(CFLAGS) $(DEFINES)
	10
	11	# unreferenced formal parameter
	12	CFLAGS=$(CFLAGS) /wd4100
	13
	14
	15	OBJS=\
	16	..\extra\getopt_long.obj \
	17	..\extra\getopt.obj \
	18	enc265.obj
	19
	20	all: enc265.exe
	21
	22	enc265.obj: enc265.cc
	23	$(CC) /c $.cc /Fo$.obj /TP $(CFLAGS)
	24
	25	.c.obj:
	26	$(CC) /c $.c /Fo$.obj $(CFLAGS)
	27
	28	.cc.obj:
	29	$(CC) /c $.cc /Fo$.obj $(CFLAGS)
	30
	31	enc265.exe: $(OBJS) ..\libde265\libde265.lib
	32	$(LINK) /out:enc265.exe $** ..\libde265\libde265.lib
	33
	34	clean:
	35	del enc265.exe
	36	del $(OBJS)

+322

-0

enc265/enc265.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "libde265/en265.h" //coder-context.h"
	21
	22	#include "libde265/configparam.h"
	23	#include "libde265/image-io.h"
	24	#include "libde265/encoder/analyze.h"
	25	#include "libde265/util.h"
	26
	27	#include <getopt.h>
	28
	29
	30
	31	#if HAVE_VIDEOGFX
	32	#include <libvideogfx.hh>
	33	using namespace videogfx;
	34
	35	void debug_show_image_libvideogfx(const de265_image* input, int slot)
	36	{
	37	static X11Win debugwin;
	38	static bool opened=false;
	39	int w = input->get_width();
	40	int h = input->get_height();
	41	if (!opened) {
	42	opened=true;
	43	debugwin.Create(w,h, "debug");
	44	}
	45
	46	Image<Pixel> img;
	47	img.Create(w,h,Colorspace_YUV, Chroma_420);
	48
	49	for (int y=0;y<h;y++)
	50	memcpy(img.AskFrameY()[y], input->get_image_plane_at_pos(0,0,y), w);
	51
	52	for (int y=0;y<h/2;y++) {
	53	memcpy(img.AskFrameU()[y], input->get_image_plane_at_pos(1,0,y), w/2);
	54	memcpy(img.AskFrameV()[y], input->get_image_plane_at_pos(2,0,y), w/2);
	55	}
	56
	57	debugwin.Display(img);
	58	//debugwin.WaitForKeypress();
	59	}
	60	#endif
	61
	62
	63
	64	int show_help=false;
	65	int verbosity=0;
	66
	67	static struct option long_options[] = {
	68	{"help", no_argument, &show_help, 1 },
	69	{"verbose", no_argument, 0, 'v' },
	70	{0, 0, 0, 0 }
	71	};
	72
	73
	74	struct inout_params
	75	{
	76	inout_params();
	77
	78	// input
	79
	80	option_int first_frame;
	81	option_int max_number_of_frames;
	82
	83	option_string input_yuv;
	84	option_int input_width;
	85	option_int input_height;
	86
	87	// output
	88
	89	option_string output_filename;
	90
	91	// debug
	92
	93	option_string reconstruction_yuv;
	94
	95
	96	void register_params(config_parameters& config);
	97	};
	98
	99
	100	inout_params::inout_params()
	101	{
	102	input_yuv.set_ID("input"); input_yuv.set_short_option('i');
	103	input_yuv.set_default("paris_cif.yuv");
	104
	105	output_filename.set_ID("output"); output_filename.set_short_option('o');
	106	output_filename.set_default("out.bin");
	107
	108	reconstruction_yuv.set_ID("input");
	109	reconstruction_yuv.set_default("recon.yuv");
	110
	111	first_frame.set_ID("first-frame");
	112	first_frame.set_default(0);
	113	first_frame.set_minimum(0);
	114
	115	max_number_of_frames.set_ID("frames");
	116	max_number_of_frames.set_short_option('f');
	117	max_number_of_frames.set_minimum(1);
	118	//max_number_of_frames.set_default(INT_MAX);
	119
	120	input_width.set_ID("width"); input_width.set_short_option('w');
	121	input_width.set_minimum(1); input_width.set_default(352);
	122
	123	input_height.set_ID("height"); input_height.set_short_option('h');
	124	input_height.set_minimum(1); input_height.set_default(288);
	125	}
	126
	127
	128	void inout_params::register_params(config_parameters& config)
	129	{
	130	config.add_option(&input_yuv);
	131	config.add_option(&output_filename);
	132	config.add_option(&first_frame);
	133	config.add_option(&max_number_of_frames);
	134	config.add_option(&input_width);
	135	config.add_option(&input_height);
	136	}
	137
	138
	139	void test_parameters_API(en265_encoder_context* ectx)
	140	{
	141	const char** param = en265_list_parameters(ectx);
	142	if (param) {
	143	for (int i=0; param[i]; i++) {
	144	printf("\|%s\| ",param[i]);
	145
	146	enum en265_parameter_type type = en265_get_parameter_type(ectx, param[i]);
	147	const char* type_name="unknown";
	148	switch (type) {
	149	case en265_parameter_int: type_name="int"; break;
	150	case en265_parameter_bool: type_name="bool"; break;
	151	case en265_parameter_string: type_name="string"; break;
	152	case en265_parameter_choice: type_name="choice"; break;
	153	}
	154
	155	printf("(%s)",type_name);
	156
	157	if (type==en265_parameter_choice) {
	158	const char** choices = en265_list_parameter_choices(ectx, param[i]);
	159	if (choices) {
	160	for (int k=0; choices[k]; k++) {
	161	printf(" %s",choices[k]);
	162	}
	163	}
	164	}
	165
	166	printf("\n");
	167	}
	168	}
	169
	170	// en265_set_parameter_int(ectx, "min-tb-size", 8);
	171	}
	172
	173
	174	extern int skipTBSplit, noskipTBSplit;
	175	extern int zeroBlockCorrelation[6][2][5];
	176
	177	int main(int argc, char** argv)
	178	{
	179	de265_init();
	180
	181	en265_encoder_context* ectx = en265_new_encoder();
	182
	183
	184	bool cmdline_errors = false;
	185
	186	// --- in/out parameters ---
	187
	188	struct inout_params inout_params;
	189	config_parameters inout_param_config;
	190	inout_params.register_params(inout_param_config);
	191
	192	int first_idx=1;
	193	if (!inout_param_config.parse_command_line_params(&argc,argv, &first_idx, true)) {
	194	cmdline_errors = true;
	195	}
	196
	197
	198	// --- read encoder parameters ---
	199
	200	if (en265_parse_command_line_parameters(ectx, &argc, argv) != DE265_OK) {
	201	cmdline_errors = true;
	202	}
	203
	204
	205
	206	while (1) {
	207	int option_index = 0;
	208
	209	int c = getopt_long(argc, argv, "v"
	210	, long_options, &option_index);
	211	if (c == -1)
	212	break;
	213
	214	switch (c) {
	215	case 'v': verbosity++; break;
	216	}
	217	}
	218
	219
	220	// --- show usage information ---
	221
	222	if (optind != argc \|\| cmdline_errors \|\| show_help) {
	223	fprintf(stderr," enc265 v%s\n", de265_get_version());
	224	fprintf(stderr,"--------------\n");
	225	fprintf(stderr,"usage: enc265 [options]\n");
	226	fprintf(stderr,"The video file must be a raw YUV file\n");
	227	fprintf(stderr,"\n");
	228	fprintf(stderr,"options:\n");
	229	fprintf(stderr," --help show help\n");
	230	fprintf(stderr," -v, --verbose increase verbosity level (up to 3 times)\n");
	231
	232	inout_param_config.print_params();
	233	fprintf(stderr,"\n");
	234	en265_show_parameters(ectx);
	235
	236	exit(show_help ? 0 : 5);
	237	}
	238
	239
	240
	241	de265_set_verbosity(verbosity);
	242	#if HAVE_VIDEOGFX
	243	//debug_set_image_output(debug_show_image_libvideogfx);
	244	#endif
	245
	246	//test_parameters_API(ectx);
	247
	248
	249	ImageSink_YUV reconstruction_sink;
	250	if (strlen(inout_params.reconstruction_yuv.get().c_str()) != 0) {
	251	reconstruction_sink.set_filename(inout_params.reconstruction_yuv.get().c_str());
	252	//ectx.reconstruction_sink = &reconstruction_sink;
	253	}
	254
	255	ImageSource_YUV image_source;
	256	image_source.set_input_file(inout_params.input_yuv.get().c_str(),
	257	inout_params.input_width,
	258	inout_params.input_height);
	259
	260	PacketSink_File packet_sink;
	261	packet_sink.set_filename(inout_params.output_filename.get().c_str());
	262
	263
	264	// --- run encoder ---
	265
	266	image_source.skip_frames( inout_params.first_frame );
	267
	268	en265_start_encoder(ectx, 0);
	269
	270	int maxPoc = INT_MAX;
	271	if (inout_params.max_number_of_frames.is_defined()) {
	272	maxPoc = inout_params.max_number_of_frames;
	273	}
	274
	275	bool eof = false;
	276	for (int poc=0; poc<maxPoc && !eof ;poc++)
	277	{
	278	// push one image into the encoder
	279
	280	de265_image* input_image = image_source.get_image();
	281	if (input_image==NULL) {
	282	en265_push_eof(ectx);
	283	eof=true;
	284	}
	285	else {
	286	en265_push_image(ectx, input_image);
	287	}
	288
	289
	290
	291	// encode images while more are available
	292
	293	en265_encode(ectx);
	294
	295
	296	// write all pending packets
	297
	298	for (;;) {
	299	en265_packet* pck = en265_get_packet(ectx,0);
	300	if (pck==NULL)
	301	break;
	302
	303	packet_sink.send_packet(pck->data, pck->length);
	304
	305	en265_free_packet(ectx,pck);
	306	}
	307	}
	308
	309
	310
	311	// --- print statistics ---
	312
	313	en265_print_logging((encoder_context*)ectx, "tb-split", NULL);
	314
	315
	316	en265_free_encoder(ectx);
	317
	318	de265_free();
	319
	320	return 0;
	321	}

-3

extra/win32cond.c less more

50	50
51	51	int win32_cond_wait(win32_cond_t cv, HANDLE external_mutex)
52	52	{
	53	int last_waiter;
	54
53	55	// Avoid race conditions.
54	56	EnterCriticalSection (&cv->waiters_count_lock_);
55	57	cv->waiters_count_++;

67	69	cv->waiters_count_--;
68	70
69	71	// Check to see if we're the last waiter after <pthread_cond_broadcast>.
70		int last_waiter = cv->was_broadcast_ && cv->waiters_count_ == 0;
	72	last_waiter = cv->was_broadcast_ && cv->waiters_count_ == 0;
71	73
72	74	LeaveCriticalSection (&cv->waiters_count_lock_);
73	75

86	88
87	89	int win32_cond_signal(win32_cond_t *cv)
88	90	{
	91	int have_waiters;
	92
89	93	EnterCriticalSection (&cv->waiters_count_lock_);
90		int have_waiters = cv->waiters_count_ > 0;
	94	have_waiters = cv->waiters_count_ > 0;
91	95	LeaveCriticalSection (&cv->waiters_count_lock_);
92	96
93	97	// If there aren't any waiters, then this is a no-op.

98	102
99	103	int win32_cond_broadcast(win32_cond_t *cv)
100	104	{
	105	int have_waiters = 0;
	106
101	107	// This is needed to ensure that <waiters_count_> and <was_broadcast_> are
102	108	// consistent relative to each other.
103	109	EnterCriticalSection (&cv->waiters_count_lock_);
104		int have_waiters = 0;
105	110
106	111	if (cv->waiters_count_ > 0) {
107	112	// We are broadcasting, even if there is just one waiter...

+84

-0

libde265/CMakeLists.txt less more

	0	set (libde265_sources
	1	bitstream.cc
	2	cabac.cc
	3	de265.cc
	4	deblock.cc
	5	decctx.cc
	6	nal-parser.cc
	7	nal-parser.h
	8	dpb.cc
	9	dpb.h
	10	image.cc
	11	intrapred.cc
	12	md5.cc
	13	nal.cc
	14	pps.cc
	15	transform.cc
	16	refpic.cc
	17	sao.cc
	18	scan.cc
	19	sei.cc
	20	slice.cc
	21	sps.cc
	22	util.cc
	23	vps.cc
	24	bitstream.h
	25	cabac.h
	26	deblock.h
	27	decctx.h
	28	image.h
	29	intrapred.h
	30	md5.h
	31	nal.h
	32	pps.h
	33	transform.h
	34	refpic.h
	35	sao.h
	36	scan.h
	37	sei.h
	38	slice.h
	39	sps.h
	40	util.h
	41	vps.h
	42	vui.h vui.cc
	43	motion.cc motion.h
	44	threads.cc threads.h
	45	visualize.cc visualize.h
	46	acceleration.h
	47	fallback.cc fallback.h fallback-motion.cc fallback-motion.h
	48	fallback-dct.h fallback-dct.cc
	49	quality.cc quality.h
	50	configparam.cc configparam.h
	51	image-io.h image-io.cc
	52	alloc_pool.h alloc_pool.cc
	53	en265.h en265.cc
	54	contextmodel.cc
	55	)
	56
	57	if(MSVC)
	58	set (libde265_sources
	59	${libde265_sources}
	60	../extra/win32cond.c
	61	../extra/win32cond.h
	62	)
	63	endif()
	64
	65	add_definitions(-DLIBDE265_EXPORTS)
	66
	67	add_library(${LIBDE265_LIBRARY_NAME} SHARED ${libde265_sources})
	68
	69	target_link_libraries(${LIBDE265_LIBRARY_NAME} ${CMAKE_THREAD_LIBS_INIT})
	70
	71	if(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
	72	SET_TARGET_PROPERTIES(${LIBDE265_LIBRARY_NAME} PROPERTIES COMPILE_FLAGS "-fPIC")
	73	endif(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
	74
	75	add_subdirectory (encoder)
	76
	77	target_link_libraries(${LIBDE265_LIBRARY_NAME} encoder)
	78
	79	if(SUPPORTS_SSE4_1)
	80	add_definitions(-DHAVE_SSE4_1)
	81	add_subdirectory (x86)
	82	target_link_libraries(${LIBDE265_LIBRARY_NAME} x86)
	83	endif()

+62

-32

libde265/Makefile.am less more

21	21	libde265_la_LIBADD = -lstdc++
22	22
23	23	libde265_la_SOURCES = \
	24	acceleration.h \
	25	alloc_pool.h \
	26	alloc_pool.cc \
24	27	bitstream.cc \
	28	bitstream.h \
25	29	cabac.cc \
	30	cabac.h \
	31	configparam.cc \
	32	configparam.h \
	33	contextmodel.cc \
	34	contextmodel.h \
26	35	de265.cc \
27	36	deblock.cc \
	37	deblock.h \
28	38	decctx.cc \
29		nal-parser.cc \
30		nal-parser.h \
	39	decctx.h \
	40	en265.h \
	41	en265.cc \
	42	fallback.cc \
	43	fallback.h \
	44	fallback-dct.h \
	45	fallback-dct.cc \
	46	fallback-motion.cc \
	47	fallback-motion.h \
31	48	dpb.cc \
32	49	dpb.h \
33	50	image.cc \
	51	image.h \
	52	image-io.h \
	53	image-io.cc \
34	54	intrapred.cc \
	55	intrapred.h \
35	56	md5.cc \
	57	md5.h \
	58	motion.cc \
	59	motion.h \
36	60	nal.cc \
	61	nal.h \
	62	nal-parser.cc \
	63	nal-parser.h \
37	64	pps.cc \
	65	pps.h \
	66	quality.cc \
	67	quality.h \
	68	refpic.cc \
	69	refpic.h \
	70	sao.cc \
	71	sao.h \
	72	scan.cc \
	73	scan.h \
	74	sei.cc \
	75	sei.h \
	76	slice.cc \
	77	slice.h \
	78	sps.cc \
	79	sps.h \
	80	threads.cc \
	81	threads.h \
38	82	transform.cc \
39		refpic.cc \
40		sao.cc \
41		scan.cc \
42		sei.cc \
43		slice.cc \
44		sps.cc \
	83	transform.h \
45	84	util.cc \
	85	util.h \
	86	visualize.cc \
	87	visualize.h \
46	88	vps.cc \
47		bitstream.h \
48		cabac.h \
49		deblock.h \
50		decctx.h \
51		image.h \
52		intrapred.h \
53		md5.h \
54		nal.h \
55		pps.h \
56		transform.h \
57		refpic.h \
58		sao.h \
59		scan.h \
60		sei.h \
61		slice.h \
62		sps.h \
63		util.h \
64	89	vps.h \
65		motion.cc motion.h \
66		threads.cc threads.h \
67		visualize.cc visualize.h \
68		acceleration.h \
69		fallback.cc fallback.h fallback-motion.cc fallback-motion.h \
70		fallback-dct.h fallback-dct.cc
	90	vui.cc \
	91	vui.h
	92
	93	SUBDIRS = encoder
	94	libde265_la_LIBADD += encoder/libde265_encoder.la
71	95
72	96	if ENABLE_SSE_OPT
73		SUBDIRS = x86
	97	SUBDIRS += x86
74	98	libde265_la_LIBADD += x86/libde265_x86.la
	99	endif
	100
	101	if ENABLE_ARM_OPT
	102	SUBDIRS += arm
	103	libde265_la_LIBADD += arm/libde265_arm.la
75	104	endif
76	105
77	106	if MINGW

80	109	endif
81	110
82	111	EXTRA_DIST = Makefile.vc7 \
	112	CMakeLists.txt \
83	113	../extra/stdbool.h \
84	114	../extra/stdint.h
85	115

+237

-160

libde265/Makefile.in less more

81	81	target_triplet = @target@
82	82	@HAVE_VISIBILITY_TRUE@am__append_1 = -DHAVE_VISIBILITY
83	83	@HAVE_VISIBILITY_TRUE@am__append_2 = -DHAVE_VISIBILITY
84		@ENABLE_SSE_OPT_TRUE@am__append_3 = x86/libde265_x86.la
85		@MINGW_TRUE@am__append_4 = ../extra/win32cond.c ../extra/win32cond.h
86		@MINGW_TRUE@am__append_5 = -no-undefined -static-libgcc -static-libstdc++
	84	@ENABLE_SSE_OPT_TRUE@am__append_3 = x86
	85	@ENABLE_SSE_OPT_TRUE@am__append_4 = x86/libde265_x86.la
	86	@ENABLE_ARM_OPT_TRUE@am__append_5 = arm
	87	@ENABLE_ARM_OPT_TRUE@am__append_6 = arm/libde265_arm.la
	88	@MINGW_TRUE@am__append_7 = ../extra/win32cond.c ../extra/win32cond.h
	89	@MINGW_TRUE@am__append_8 = -no-undefined -static-libgcc -static-libstdc++
87	90	subdir = libde265
88	91	DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/Makefile.am \
89	92	$(srcdir)/de265-version.h.in $(top_srcdir)/depcomp \
90	93	$(libde265_la_HEADERS) COPYING
91	94	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
92	95	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	96	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
93	97	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
94	98	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
95	99	$(top_srcdir)/m4/lt~obsolete.m4 \
96	100	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
97		$(top_srcdir)/m4/visibility.m4 $(top_srcdir)/configure.ac
	101	$(top_srcdir)/configure.ac
98	102	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
99	103	$(ACLOCAL_M4)
100	104	mkinstalldirs = $(install_sh) -d

130	134	}
131	135	am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(libde265_ladir)"
132	136	LTLIBRARIES = $(lib_LTLIBRARIES)
133		libde265_la_DEPENDENCIES = $(am__append_3)
134		am__libde265_la_SOURCES_DIST = bitstream.cc cabac.cc de265.cc \
135		deblock.cc decctx.cc nal-parser.cc nal-parser.h dpb.cc dpb.h \
136		image.cc intrapred.cc md5.cc nal.cc pps.cc transform.cc \
137		refpic.cc sao.cc scan.cc sei.cc slice.cc sps.cc util.cc vps.cc \
138		bitstream.h cabac.h deblock.h decctx.h image.h intrapred.h \
139		md5.h nal.h pps.h transform.h refpic.h sao.h scan.h sei.h \
140		slice.h sps.h util.h vps.h motion.cc motion.h threads.cc \
141		threads.h visualize.cc visualize.h acceleration.h fallback.cc \
142		fallback.h fallback-motion.cc fallback-motion.h fallback-dct.h \
143		fallback-dct.cc ../extra/win32cond.c ../extra/win32cond.h
	137	libde265_la_DEPENDENCIES = encoder/libde265_encoder.la $(am__append_4) \
	138	$(am__append_6)
	139	am__libde265_la_SOURCES_DIST = acceleration.h alloc_pool.h \
	140	alloc_pool.cc bitstream.cc bitstream.h cabac.cc cabac.h \
	141	configparam.cc configparam.h contextmodel.cc contextmodel.h \
	142	de265.cc deblock.cc deblock.h decctx.cc decctx.h en265.h \
	143	en265.cc fallback.cc fallback.h fallback-dct.h fallback-dct.cc \
	144	fallback-motion.cc fallback-motion.h dpb.cc dpb.h image.cc \
	145	image.h image-io.h image-io.cc intrapred.cc intrapred.h md5.cc \
	146	md5.h motion.cc motion.h nal.cc nal.h nal-parser.cc \
	147	nal-parser.h pps.cc pps.h quality.cc quality.h refpic.cc \
	148	refpic.h sao.cc sao.h scan.cc scan.h sei.cc sei.h slice.cc \
	149	slice.h sps.cc sps.h threads.cc threads.h transform.cc \
	150	transform.h util.cc util.h visualize.cc visualize.h vps.cc \
	151	vps.h vui.cc vui.h ../extra/win32cond.c ../extra/win32cond.h
144	152	am__dirstamp = $(am__leading_dot)dirstamp
145	153	@MINGW_TRUE@am__objects_1 = ../extra/libde265_la-win32cond.lo
146		am_libde265_la_OBJECTS = libde265_la-bitstream.lo libde265_la-cabac.lo \
	154	am_libde265_la_OBJECTS = libde265_la-alloc_pool.lo \
	155	libde265_la-bitstream.lo libde265_la-cabac.lo \
	156	libde265_la-configparam.lo libde265_la-contextmodel.lo \
147	157	libde265_la-de265.lo libde265_la-deblock.lo \
148		libde265_la-decctx.lo libde265_la-nal-parser.lo \
149		libde265_la-dpb.lo libde265_la-image.lo \
150		libde265_la-intrapred.lo libde265_la-md5.lo libde265_la-nal.lo \
151		libde265_la-pps.lo libde265_la-transform.lo \
152		libde265_la-refpic.lo libde265_la-sao.lo libde265_la-scan.lo \
153		libde265_la-sei.lo libde265_la-slice.lo libde265_la-sps.lo \
154		libde265_la-util.lo libde265_la-vps.lo libde265_la-motion.lo \
155		libde265_la-threads.lo libde265_la-visualize.lo \
156		libde265_la-fallback.lo libde265_la-fallback-motion.lo \
157		libde265_la-fallback-dct.lo $(am__objects_1)
	158	libde265_la-decctx.lo libde265_la-en265.lo \
	159	libde265_la-fallback.lo libde265_la-fallback-dct.lo \
	160	libde265_la-fallback-motion.lo libde265_la-dpb.lo \
	161	libde265_la-image.lo libde265_la-image-io.lo \
	162	libde265_la-intrapred.lo libde265_la-md5.lo \
	163	libde265_la-motion.lo libde265_la-nal.lo \
	164	libde265_la-nal-parser.lo libde265_la-pps.lo \
	165	libde265_la-quality.lo libde265_la-refpic.lo \
	166	libde265_la-sao.lo libde265_la-scan.lo libde265_la-sei.lo \
	167	libde265_la-slice.lo libde265_la-sps.lo libde265_la-threads.lo \
	168	libde265_la-transform.lo libde265_la-util.lo \
	169	libde265_la-visualize.lo libde265_la-vps.lo libde265_la-vui.lo \
	170	$(am__objects_1)
158	171	libde265_la_OBJECTS = $(am_libde265_la_OBJECTS)
159	172	AM_V_lt = $(am__v_lt_@AM_V@)
160	173	am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)

258	271	done \| $(am__uniquify_input)`
259	272	ETAGS = etags
260	273	CTAGS = ctags
261		DIST_SUBDIRS = x86
	274	DIST_SUBDIRS = encoder x86 arm
262	275	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
263	276	am__relativize = \
264	277	dir0=`pwd`; \

286	299	done; \
287	300	reldir="$$dir2"
288	301	ACLOCAL = @ACLOCAL@
	302	ALLOCA = @ALLOCA@
289	303	AMTAR = @AMTAR@
290	304	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
291	305	AR = @AR@

294	308	AUTOMAKE = @AUTOMAKE@
295	309	AWK = @AWK@
296	310	CC = @CC@
	311	CCAS = @CCAS@
	312	CCASDEPMODE = @CCASDEPMODE@
	313	CCASFLAGS = @CCASFLAGS@
297	314	CCDEPMODE = @CCDEPMODE@
298	315	CFLAGS = @CFLAGS@
299		CFLAG_VISIBILITY = @CFLAG_VISIBILITY@
300	316	CPP = @CPP@
301	317	CPPFLAGS = @CPPFLAGS@
302	318	CXX = @CXX@

316	332	EXEEXT = @EXEEXT@
317	333	FGREP = @FGREP@
318	334	GREP = @GREP@
319		HAVE_VISIBILITY = @HAVE_VISIBILITY@
	335	HAVE_CXX11 = @HAVE_CXX11@
320	336	INSTALL = @INSTALL@
321	337	INSTALL_DATA = @INSTALL_DATA@
322	338	INSTALL_PROGRAM = @INSTALL_PROGRAM@

439	455	$(am__append_2)
440	456	libde265_la_LDFLAGS = -version-info \
441	457	$(LIBDE265_CURRENT):$(LIBDE265_REVISION):$(LIBDE265_AGE) \
442		$(am__append_5)
443		libde265_la_LIBADD = -lstdc++ $(am__append_3)
444		libde265_la_SOURCES = bitstream.cc cabac.cc de265.cc deblock.cc \
445		decctx.cc nal-parser.cc nal-parser.h dpb.cc dpb.h image.cc \
446		intrapred.cc md5.cc nal.cc pps.cc transform.cc refpic.cc \
447		sao.cc scan.cc sei.cc slice.cc sps.cc util.cc vps.cc \
448		bitstream.h cabac.h deblock.h decctx.h image.h intrapred.h \
449		md5.h nal.h pps.h transform.h refpic.h sao.h scan.h sei.h \
450		slice.h sps.h util.h vps.h motion.cc motion.h threads.cc \
451		threads.h visualize.cc visualize.h acceleration.h fallback.cc \
452		fallback.h fallback-motion.cc fallback-motion.h fallback-dct.h \
453		fallback-dct.cc $(am__append_4)
454		@ENABLE_SSE_OPT_TRUE@SUBDIRS = x86
	458	$(am__append_8)
	459	libde265_la_LIBADD = -lstdc++ encoder/libde265_encoder.la \
	460	$(am__append_4) $(am__append_6)
	461	libde265_la_SOURCES = acceleration.h alloc_pool.h alloc_pool.cc \
	462	bitstream.cc bitstream.h cabac.cc cabac.h configparam.cc \
	463	configparam.h contextmodel.cc contextmodel.h de265.cc \
	464	deblock.cc deblock.h decctx.cc decctx.h en265.h en265.cc \
	465	fallback.cc fallback.h fallback-dct.h fallback-dct.cc \
	466	fallback-motion.cc fallback-motion.h dpb.cc dpb.h image.cc \
	467	image.h image-io.h image-io.cc intrapred.cc intrapred.h md5.cc \
	468	md5.h motion.cc motion.h nal.cc nal.h nal-parser.cc \
	469	nal-parser.h pps.cc pps.h quality.cc quality.h refpic.cc \
	470	refpic.h sao.cc sao.h scan.cc scan.h sei.cc sei.h slice.cc \
	471	slice.h sps.cc sps.h threads.cc threads.h transform.cc \
	472	transform.h util.cc util.h visualize.cc visualize.h vps.cc \
	473	vps.h vui.cc vui.h $(am__append_7)
	474	SUBDIRS = encoder $(am__append_3) $(am__append_5)
455	475	EXTRA_DIST = Makefile.vc7 \
	476	CMakeLists.txt \
456	477	../extra/stdbool.h \
457	478	../extra/stdint.h
458	479

552	573	-rm -f *.tab.c
553	574
554	575	@AMDEP_TRUE@@am__include@ @am__quote@../extra/$(DEPDIR)/libde265_la-win32cond.Plo@am__quote@
	576	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-alloc_pool.Plo@am__quote@
555	577	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-bitstream.Plo@am__quote@
556	578	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-cabac.Plo@am__quote@
	579	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-configparam.Plo@am__quote@
	580	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-contextmodel.Plo@am__quote@
557	581	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-de265.Plo@am__quote@
558	582	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-deblock.Plo@am__quote@
559	583	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-decctx.Plo@am__quote@
560	584	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-dpb.Plo@am__quote@
	585	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-en265.Plo@am__quote@
561	586	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-fallback-dct.Plo@am__quote@
562	587	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-fallback-motion.Plo@am__quote@
563	588	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-fallback.Plo@am__quote@
	589	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-image-io.Plo@am__quote@
564	590	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-image.Plo@am__quote@
565	591	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-intrapred.Plo@am__quote@
566	592	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-md5.Plo@am__quote@

568	594	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-nal-parser.Plo@am__quote@
569	595	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-nal.Plo@am__quote@
570	596	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-pps.Plo@am__quote@
	597	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-quality.Plo@am__quote@
571	598	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-refpic.Plo@am__quote@
572	599	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-sao.Plo@am__quote@
573	600	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-scan.Plo@am__quote@

579	606	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-util.Plo@am__quote@
580	607	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-visualize.Plo@am__quote@
581	608	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-vps.Plo@am__quote@
	609	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_la-vui.Plo@am__quote@
582	610
583	611	.c.o:
584	612	@am__fastdepCC_TRUE@ $(AM_V_CC)depbase=`echo $@ \| sed 's\|[^/]*$$\|$(DEPDIR)/&\|;s\|\.o$$\|\|'`;\

635	663	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
636	664	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
637	665
	666	libde265_la-alloc_pool.lo: alloc_pool.cc
	667	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-alloc_pool.lo -MD -MP -MF $(DEPDIR)/libde265_la-alloc_pool.Tpo -c -o libde265_la-alloc_pool.lo `test -f 'alloc_pool.cc' \|\| echo '$(srcdir)/'`alloc_pool.cc
	668	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-alloc_pool.Tpo $(DEPDIR)/libde265_la-alloc_pool.Plo
	669	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='alloc_pool.cc' object='libde265_la-alloc_pool.lo' libtool=yes @AMDEPBACKSLASH@
	670	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	671	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-alloc_pool.lo `test -f 'alloc_pool.cc' \|\| echo '$(srcdir)/'`alloc_pool.cc
	672
638	673	libde265_la-bitstream.lo: bitstream.cc
639	674	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-bitstream.lo -MD -MP -MF $(DEPDIR)/libde265_la-bitstream.Tpo -c -o libde265_la-bitstream.lo `test -f 'bitstream.cc' \|\| echo '$(srcdir)/'`bitstream.cc
640	675	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-bitstream.Tpo $(DEPDIR)/libde265_la-bitstream.Plo

649	684	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
650	685	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-cabac.lo `test -f 'cabac.cc' \|\| echo '$(srcdir)/'`cabac.cc
651	686
	687	libde265_la-configparam.lo: configparam.cc
	688	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-configparam.lo -MD -MP -MF $(DEPDIR)/libde265_la-configparam.Tpo -c -o libde265_la-configparam.lo `test -f 'configparam.cc' \|\| echo '$(srcdir)/'`configparam.cc
	689	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-configparam.Tpo $(DEPDIR)/libde265_la-configparam.Plo
	690	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='configparam.cc' object='libde265_la-configparam.lo' libtool=yes @AMDEPBACKSLASH@
	691	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	692	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-configparam.lo `test -f 'configparam.cc' \|\| echo '$(srcdir)/'`configparam.cc
	693
	694	libde265_la-contextmodel.lo: contextmodel.cc
	695	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-contextmodel.lo -MD -MP -MF $(DEPDIR)/libde265_la-contextmodel.Tpo -c -o libde265_la-contextmodel.lo `test -f 'contextmodel.cc' \|\| echo '$(srcdir)/'`contextmodel.cc
	696	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-contextmodel.Tpo $(DEPDIR)/libde265_la-contextmodel.Plo
	697	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='contextmodel.cc' object='libde265_la-contextmodel.lo' libtool=yes @AMDEPBACKSLASH@
	698	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	699	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-contextmodel.lo `test -f 'contextmodel.cc' \|\| echo '$(srcdir)/'`contextmodel.cc
	700
652	701	libde265_la-de265.lo: de265.cc
653	702	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-de265.lo -MD -MP -MF $(DEPDIR)/libde265_la-de265.Tpo -c -o libde265_la-de265.lo `test -f 'de265.cc' \|\| echo '$(srcdir)/'`de265.cc
654	703	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-de265.Tpo $(DEPDIR)/libde265_la-de265.Plo

670	719	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
671	720	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-decctx.lo `test -f 'decctx.cc' \|\| echo '$(srcdir)/'`decctx.cc
672	721
	722	libde265_la-en265.lo: en265.cc
	723	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-en265.lo -MD -MP -MF $(DEPDIR)/libde265_la-en265.Tpo -c -o libde265_la-en265.lo `test -f 'en265.cc' \|\| echo '$(srcdir)/'`en265.cc
	724	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-en265.Tpo $(DEPDIR)/libde265_la-en265.Plo
	725	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='en265.cc' object='libde265_la-en265.lo' libtool=yes @AMDEPBACKSLASH@
	726	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	727	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-en265.lo `test -f 'en265.cc' \|\| echo '$(srcdir)/'`en265.cc
	728
	729	libde265_la-fallback.lo: fallback.cc
	730	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-fallback.lo -MD -MP -MF $(DEPDIR)/libde265_la-fallback.Tpo -c -o libde265_la-fallback.lo `test -f 'fallback.cc' \|\| echo '$(srcdir)/'`fallback.cc
	731	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-fallback.Tpo $(DEPDIR)/libde265_la-fallback.Plo
	732	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='fallback.cc' object='libde265_la-fallback.lo' libtool=yes @AMDEPBACKSLASH@
	733	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	734	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-fallback.lo `test -f 'fallback.cc' \|\| echo '$(srcdir)/'`fallback.cc
	735
	736	libde265_la-fallback-dct.lo: fallback-dct.cc
	737	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-fallback-dct.lo -MD -MP -MF $(DEPDIR)/libde265_la-fallback-dct.Tpo -c -o libde265_la-fallback-dct.lo `test -f 'fallback-dct.cc' \|\| echo '$(srcdir)/'`fallback-dct.cc
	738	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-fallback-dct.Tpo $(DEPDIR)/libde265_la-fallback-dct.Plo
	739	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='fallback-dct.cc' object='libde265_la-fallback-dct.lo' libtool=yes @AMDEPBACKSLASH@
	740	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	741	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-fallback-dct.lo `test -f 'fallback-dct.cc' \|\| echo '$(srcdir)/'`fallback-dct.cc
	742
	743	libde265_la-fallback-motion.lo: fallback-motion.cc
	744	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-fallback-motion.lo -MD -MP -MF $(DEPDIR)/libde265_la-fallback-motion.Tpo -c -o libde265_la-fallback-motion.lo `test -f 'fallback-motion.cc' \|\| echo '$(srcdir)/'`fallback-motion.cc
	745	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-fallback-motion.Tpo $(DEPDIR)/libde265_la-fallback-motion.Plo
	746	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='fallback-motion.cc' object='libde265_la-fallback-motion.lo' libtool=yes @AMDEPBACKSLASH@
	747	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	748	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-fallback-motion.lo `test -f 'fallback-motion.cc' \|\| echo '$(srcdir)/'`fallback-motion.cc
	749
	750	libde265_la-dpb.lo: dpb.cc
	751	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-dpb.lo -MD -MP -MF $(DEPDIR)/libde265_la-dpb.Tpo -c -o libde265_la-dpb.lo `test -f 'dpb.cc' \|\| echo '$(srcdir)/'`dpb.cc
	752	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-dpb.Tpo $(DEPDIR)/libde265_la-dpb.Plo
	753	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='dpb.cc' object='libde265_la-dpb.lo' libtool=yes @AMDEPBACKSLASH@
	754	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	755	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-dpb.lo `test -f 'dpb.cc' \|\| echo '$(srcdir)/'`dpb.cc
	756
	757	libde265_la-image.lo: image.cc
	758	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-image.lo -MD -MP -MF $(DEPDIR)/libde265_la-image.Tpo -c -o libde265_la-image.lo `test -f 'image.cc' \|\| echo '$(srcdir)/'`image.cc
	759	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-image.Tpo $(DEPDIR)/libde265_la-image.Plo
	760	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='image.cc' object='libde265_la-image.lo' libtool=yes @AMDEPBACKSLASH@
	761	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	762	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-image.lo `test -f 'image.cc' \|\| echo '$(srcdir)/'`image.cc
	763
	764	libde265_la-image-io.lo: image-io.cc
	765	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-image-io.lo -MD -MP -MF $(DEPDIR)/libde265_la-image-io.Tpo -c -o libde265_la-image-io.lo `test -f 'image-io.cc' \|\| echo '$(srcdir)/'`image-io.cc
	766	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-image-io.Tpo $(DEPDIR)/libde265_la-image-io.Plo
	767	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='image-io.cc' object='libde265_la-image-io.lo' libtool=yes @AMDEPBACKSLASH@
	768	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	769	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-image-io.lo `test -f 'image-io.cc' \|\| echo '$(srcdir)/'`image-io.cc
	770
	771	libde265_la-intrapred.lo: intrapred.cc
	772	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-intrapred.lo -MD -MP -MF $(DEPDIR)/libde265_la-intrapred.Tpo -c -o libde265_la-intrapred.lo `test -f 'intrapred.cc' \|\| echo '$(srcdir)/'`intrapred.cc
	773	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-intrapred.Tpo $(DEPDIR)/libde265_la-intrapred.Plo
	774	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='intrapred.cc' object='libde265_la-intrapred.lo' libtool=yes @AMDEPBACKSLASH@
	775	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	776	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-intrapred.lo `test -f 'intrapred.cc' \|\| echo '$(srcdir)/'`intrapred.cc
	777
	778	libde265_la-md5.lo: md5.cc
	779	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-md5.lo -MD -MP -MF $(DEPDIR)/libde265_la-md5.Tpo -c -o libde265_la-md5.lo `test -f 'md5.cc' \|\| echo '$(srcdir)/'`md5.cc
	780	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-md5.Tpo $(DEPDIR)/libde265_la-md5.Plo
	781	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='md5.cc' object='libde265_la-md5.lo' libtool=yes @AMDEPBACKSLASH@
	782	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	783	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-md5.lo `test -f 'md5.cc' \|\| echo '$(srcdir)/'`md5.cc
	784
	785	libde265_la-motion.lo: motion.cc
	786	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-motion.lo -MD -MP -MF $(DEPDIR)/libde265_la-motion.Tpo -c -o libde265_la-motion.lo `test -f 'motion.cc' \|\| echo '$(srcdir)/'`motion.cc
	787	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-motion.Tpo $(DEPDIR)/libde265_la-motion.Plo
	788	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='motion.cc' object='libde265_la-motion.lo' libtool=yes @AMDEPBACKSLASH@
	789	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	790	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-motion.lo `test -f 'motion.cc' \|\| echo '$(srcdir)/'`motion.cc
	791
	792	libde265_la-nal.lo: nal.cc
	793	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-nal.lo -MD -MP -MF $(DEPDIR)/libde265_la-nal.Tpo -c -o libde265_la-nal.lo `test -f 'nal.cc' \|\| echo '$(srcdir)/'`nal.cc
	794	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-nal.Tpo $(DEPDIR)/libde265_la-nal.Plo
	795	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='nal.cc' object='libde265_la-nal.lo' libtool=yes @AMDEPBACKSLASH@
	796	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	797	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-nal.lo `test -f 'nal.cc' \|\| echo '$(srcdir)/'`nal.cc
	798
673	799	libde265_la-nal-parser.lo: nal-parser.cc
674	800	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-nal-parser.lo -MD -MP -MF $(DEPDIR)/libde265_la-nal-parser.Tpo -c -o libde265_la-nal-parser.lo `test -f 'nal-parser.cc' \|\| echo '$(srcdir)/'`nal-parser.cc
675	801	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-nal-parser.Tpo $(DEPDIR)/libde265_la-nal-parser.Plo

677	803	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
678	804	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-nal-parser.lo `test -f 'nal-parser.cc' \|\| echo '$(srcdir)/'`nal-parser.cc
679	805
680		libde265_la-dpb.lo: dpb.cc
681		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-dpb.lo -MD -MP -MF $(DEPDIR)/libde265_la-dpb.Tpo -c -o libde265_la-dpb.lo `test -f 'dpb.cc' \|\| echo '$(srcdir)/'`dpb.cc
682		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-dpb.Tpo $(DEPDIR)/libde265_la-dpb.Plo
683		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='dpb.cc' object='libde265_la-dpb.lo' libtool=yes @AMDEPBACKSLASH@
684		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
685		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-dpb.lo `test -f 'dpb.cc' \|\| echo '$(srcdir)/'`dpb.cc
686
687		libde265_la-image.lo: image.cc
688		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-image.lo -MD -MP -MF $(DEPDIR)/libde265_la-image.Tpo -c -o libde265_la-image.lo `test -f 'image.cc' \|\| echo '$(srcdir)/'`image.cc
689		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-image.Tpo $(DEPDIR)/libde265_la-image.Plo
690		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='image.cc' object='libde265_la-image.lo' libtool=yes @AMDEPBACKSLASH@
691		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
692		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-image.lo `test -f 'image.cc' \|\| echo '$(srcdir)/'`image.cc
693
694		libde265_la-intrapred.lo: intrapred.cc
695		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-intrapred.lo -MD -MP -MF $(DEPDIR)/libde265_la-intrapred.Tpo -c -o libde265_la-intrapred.lo `test -f 'intrapred.cc' \|\| echo '$(srcdir)/'`intrapred.cc
696		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-intrapred.Tpo $(DEPDIR)/libde265_la-intrapred.Plo
697		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='intrapred.cc' object='libde265_la-intrapred.lo' libtool=yes @AMDEPBACKSLASH@
698		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
699		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-intrapred.lo `test -f 'intrapred.cc' \|\| echo '$(srcdir)/'`intrapred.cc
700
701		libde265_la-md5.lo: md5.cc
702		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-md5.lo -MD -MP -MF $(DEPDIR)/libde265_la-md5.Tpo -c -o libde265_la-md5.lo `test -f 'md5.cc' \|\| echo '$(srcdir)/'`md5.cc
703		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-md5.Tpo $(DEPDIR)/libde265_la-md5.Plo
704		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='md5.cc' object='libde265_la-md5.lo' libtool=yes @AMDEPBACKSLASH@
705		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
706		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-md5.lo `test -f 'md5.cc' \|\| echo '$(srcdir)/'`md5.cc
707
708		libde265_la-nal.lo: nal.cc
709		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-nal.lo -MD -MP -MF $(DEPDIR)/libde265_la-nal.Tpo -c -o libde265_la-nal.lo `test -f 'nal.cc' \|\| echo '$(srcdir)/'`nal.cc
710		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-nal.Tpo $(DEPDIR)/libde265_la-nal.Plo
711		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='nal.cc' object='libde265_la-nal.lo' libtool=yes @AMDEPBACKSLASH@
712		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
713		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-nal.lo `test -f 'nal.cc' \|\| echo '$(srcdir)/'`nal.cc
714
715	806	libde265_la-pps.lo: pps.cc
716	807	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-pps.lo -MD -MP -MF $(DEPDIR)/libde265_la-pps.Tpo -c -o libde265_la-pps.lo `test -f 'pps.cc' \|\| echo '$(srcdir)/'`pps.cc
717	808	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-pps.Tpo $(DEPDIR)/libde265_la-pps.Plo

719	810	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
720	811	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-pps.lo `test -f 'pps.cc' \|\| echo '$(srcdir)/'`pps.cc
721	812
	813	libde265_la-quality.lo: quality.cc
	814	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-quality.lo -MD -MP -MF $(DEPDIR)/libde265_la-quality.Tpo -c -o libde265_la-quality.lo `test -f 'quality.cc' \|\| echo '$(srcdir)/'`quality.cc
	815	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-quality.Tpo $(DEPDIR)/libde265_la-quality.Plo
	816	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='quality.cc' object='libde265_la-quality.lo' libtool=yes @AMDEPBACKSLASH@
	817	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	818	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-quality.lo `test -f 'quality.cc' \|\| echo '$(srcdir)/'`quality.cc
	819
	820	libde265_la-refpic.lo: refpic.cc
	821	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-refpic.lo -MD -MP -MF $(DEPDIR)/libde265_la-refpic.Tpo -c -o libde265_la-refpic.lo `test -f 'refpic.cc' \|\| echo '$(srcdir)/'`refpic.cc
	822	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-refpic.Tpo $(DEPDIR)/libde265_la-refpic.Plo
	823	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='refpic.cc' object='libde265_la-refpic.lo' libtool=yes @AMDEPBACKSLASH@
	824	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	825	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-refpic.lo `test -f 'refpic.cc' \|\| echo '$(srcdir)/'`refpic.cc
	826
	827	libde265_la-sao.lo: sao.cc
	828	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-sao.lo -MD -MP -MF $(DEPDIR)/libde265_la-sao.Tpo -c -o libde265_la-sao.lo `test -f 'sao.cc' \|\| echo '$(srcdir)/'`sao.cc
	829	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-sao.Tpo $(DEPDIR)/libde265_la-sao.Plo
	830	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sao.cc' object='libde265_la-sao.lo' libtool=yes @AMDEPBACKSLASH@
	831	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	832	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-sao.lo `test -f 'sao.cc' \|\| echo '$(srcdir)/'`sao.cc
	833
	834	libde265_la-scan.lo: scan.cc
	835	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-scan.lo -MD -MP -MF $(DEPDIR)/libde265_la-scan.Tpo -c -o libde265_la-scan.lo `test -f 'scan.cc' \|\| echo '$(srcdir)/'`scan.cc
	836	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-scan.Tpo $(DEPDIR)/libde265_la-scan.Plo
	837	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='scan.cc' object='libde265_la-scan.lo' libtool=yes @AMDEPBACKSLASH@
	838	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	839	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-scan.lo `test -f 'scan.cc' \|\| echo '$(srcdir)/'`scan.cc
	840
	841	libde265_la-sei.lo: sei.cc
	842	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-sei.lo -MD -MP -MF $(DEPDIR)/libde265_la-sei.Tpo -c -o libde265_la-sei.lo `test -f 'sei.cc' \|\| echo '$(srcdir)/'`sei.cc
	843	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-sei.Tpo $(DEPDIR)/libde265_la-sei.Plo
	844	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sei.cc' object='libde265_la-sei.lo' libtool=yes @AMDEPBACKSLASH@
	845	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	846	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-sei.lo `test -f 'sei.cc' \|\| echo '$(srcdir)/'`sei.cc
	847
	848	libde265_la-slice.lo: slice.cc
	849	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-slice.lo -MD -MP -MF $(DEPDIR)/libde265_la-slice.Tpo -c -o libde265_la-slice.lo `test -f 'slice.cc' \|\| echo '$(srcdir)/'`slice.cc
	850	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-slice.Tpo $(DEPDIR)/libde265_la-slice.Plo
	851	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='slice.cc' object='libde265_la-slice.lo' libtool=yes @AMDEPBACKSLASH@
	852	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	853	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-slice.lo `test -f 'slice.cc' \|\| echo '$(srcdir)/'`slice.cc
	854
	855	libde265_la-sps.lo: sps.cc
	856	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-sps.lo -MD -MP -MF $(DEPDIR)/libde265_la-sps.Tpo -c -o libde265_la-sps.lo `test -f 'sps.cc' \|\| echo '$(srcdir)/'`sps.cc
	857	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-sps.Tpo $(DEPDIR)/libde265_la-sps.Plo
	858	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sps.cc' object='libde265_la-sps.lo' libtool=yes @AMDEPBACKSLASH@
	859	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	860	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-sps.lo `test -f 'sps.cc' \|\| echo '$(srcdir)/'`sps.cc
	861
	862	libde265_la-threads.lo: threads.cc
	863	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-threads.lo -MD -MP -MF $(DEPDIR)/libde265_la-threads.Tpo -c -o libde265_la-threads.lo `test -f 'threads.cc' \|\| echo '$(srcdir)/'`threads.cc
	864	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-threads.Tpo $(DEPDIR)/libde265_la-threads.Plo
	865	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='threads.cc' object='libde265_la-threads.lo' libtool=yes @AMDEPBACKSLASH@
	866	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	867	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-threads.lo `test -f 'threads.cc' \|\| echo '$(srcdir)/'`threads.cc
	868
722	869	libde265_la-transform.lo: transform.cc
723	870	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-transform.lo -MD -MP -MF $(DEPDIR)/libde265_la-transform.Tpo -c -o libde265_la-transform.lo `test -f 'transform.cc' \|\| echo '$(srcdir)/'`transform.cc
724	871	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-transform.Tpo $(DEPDIR)/libde265_la-transform.Plo

726	873	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
727	874	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-transform.lo `test -f 'transform.cc' \|\| echo '$(srcdir)/'`transform.cc
728	875
729		libde265_la-refpic.lo: refpic.cc
730		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-refpic.lo -MD -MP -MF $(DEPDIR)/libde265_la-refpic.Tpo -c -o libde265_la-refpic.lo `test -f 'refpic.cc' \|\| echo '$(srcdir)/'`refpic.cc
731		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-refpic.Tpo $(DEPDIR)/libde265_la-refpic.Plo
732		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='refpic.cc' object='libde265_la-refpic.lo' libtool=yes @AMDEPBACKSLASH@
733		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
734		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-refpic.lo `test -f 'refpic.cc' \|\| echo '$(srcdir)/'`refpic.cc
735
736		libde265_la-sao.lo: sao.cc
737		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-sao.lo -MD -MP -MF $(DEPDIR)/libde265_la-sao.Tpo -c -o libde265_la-sao.lo `test -f 'sao.cc' \|\| echo '$(srcdir)/'`sao.cc
738		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-sao.Tpo $(DEPDIR)/libde265_la-sao.Plo
739		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sao.cc' object='libde265_la-sao.lo' libtool=yes @AMDEPBACKSLASH@
740		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
741		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-sao.lo `test -f 'sao.cc' \|\| echo '$(srcdir)/'`sao.cc
742
743		libde265_la-scan.lo: scan.cc
744		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-scan.lo -MD -MP -MF $(DEPDIR)/libde265_la-scan.Tpo -c -o libde265_la-scan.lo `test -f 'scan.cc' \|\| echo '$(srcdir)/'`scan.cc
745		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-scan.Tpo $(DEPDIR)/libde265_la-scan.Plo
746		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='scan.cc' object='libde265_la-scan.lo' libtool=yes @AMDEPBACKSLASH@
747		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
748		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-scan.lo `test -f 'scan.cc' \|\| echo '$(srcdir)/'`scan.cc
749
750		libde265_la-sei.lo: sei.cc
751		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-sei.lo -MD -MP -MF $(DEPDIR)/libde265_la-sei.Tpo -c -o libde265_la-sei.lo `test -f 'sei.cc' \|\| echo '$(srcdir)/'`sei.cc
752		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-sei.Tpo $(DEPDIR)/libde265_la-sei.Plo
753		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sei.cc' object='libde265_la-sei.lo' libtool=yes @AMDEPBACKSLASH@
754		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
755		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-sei.lo `test -f 'sei.cc' \|\| echo '$(srcdir)/'`sei.cc
756
757		libde265_la-slice.lo: slice.cc
758		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-slice.lo -MD -MP -MF $(DEPDIR)/libde265_la-slice.Tpo -c -o libde265_la-slice.lo `test -f 'slice.cc' \|\| echo '$(srcdir)/'`slice.cc
759		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-slice.Tpo $(DEPDIR)/libde265_la-slice.Plo
760		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='slice.cc' object='libde265_la-slice.lo' libtool=yes @AMDEPBACKSLASH@
761		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
762		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-slice.lo `test -f 'slice.cc' \|\| echo '$(srcdir)/'`slice.cc
763
764		libde265_la-sps.lo: sps.cc
765		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-sps.lo -MD -MP -MF $(DEPDIR)/libde265_la-sps.Tpo -c -o libde265_la-sps.lo `test -f 'sps.cc' \|\| echo '$(srcdir)/'`sps.cc
766		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-sps.Tpo $(DEPDIR)/libde265_la-sps.Plo
767		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sps.cc' object='libde265_la-sps.lo' libtool=yes @AMDEPBACKSLASH@
768		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
769		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-sps.lo `test -f 'sps.cc' \|\| echo '$(srcdir)/'`sps.cc
770
771	876	libde265_la-util.lo: util.cc
772	877	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-util.lo -MD -MP -MF $(DEPDIR)/libde265_la-util.Tpo -c -o libde265_la-util.lo `test -f 'util.cc' \|\| echo '$(srcdir)/'`util.cc
773	878	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-util.Tpo $(DEPDIR)/libde265_la-util.Plo

775	880	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
776	881	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-util.lo `test -f 'util.cc' \|\| echo '$(srcdir)/'`util.cc
777	882
	883	libde265_la-visualize.lo: visualize.cc
	884	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-visualize.lo -MD -MP -MF $(DEPDIR)/libde265_la-visualize.Tpo -c -o libde265_la-visualize.lo `test -f 'visualize.cc' \|\| echo '$(srcdir)/'`visualize.cc
	885	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-visualize.Tpo $(DEPDIR)/libde265_la-visualize.Plo
	886	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='visualize.cc' object='libde265_la-visualize.lo' libtool=yes @AMDEPBACKSLASH@
	887	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	888	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-visualize.lo `test -f 'visualize.cc' \|\| echo '$(srcdir)/'`visualize.cc
	889
778	890	libde265_la-vps.lo: vps.cc
779	891	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-vps.lo -MD -MP -MF $(DEPDIR)/libde265_la-vps.Tpo -c -o libde265_la-vps.lo `test -f 'vps.cc' \|\| echo '$(srcdir)/'`vps.cc
780	892	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-vps.Tpo $(DEPDIR)/libde265_la-vps.Plo

782	894	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
783	895	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-vps.lo `test -f 'vps.cc' \|\| echo '$(srcdir)/'`vps.cc
784	896
785		libde265_la-motion.lo: motion.cc
786		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-motion.lo -MD -MP -MF $(DEPDIR)/libde265_la-motion.Tpo -c -o libde265_la-motion.lo `test -f 'motion.cc' \|\| echo '$(srcdir)/'`motion.cc
787		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-motion.Tpo $(DEPDIR)/libde265_la-motion.Plo
788		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='motion.cc' object='libde265_la-motion.lo' libtool=yes @AMDEPBACKSLASH@
789		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
790		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-motion.lo `test -f 'motion.cc' \|\| echo '$(srcdir)/'`motion.cc
791
792		libde265_la-threads.lo: threads.cc
793		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-threads.lo -MD -MP -MF $(DEPDIR)/libde265_la-threads.Tpo -c -o libde265_la-threads.lo `test -f 'threads.cc' \|\| echo '$(srcdir)/'`threads.cc
794		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-threads.Tpo $(DEPDIR)/libde265_la-threads.Plo
795		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='threads.cc' object='libde265_la-threads.lo' libtool=yes @AMDEPBACKSLASH@
796		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
797		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-threads.lo `test -f 'threads.cc' \|\| echo '$(srcdir)/'`threads.cc
798
799		libde265_la-visualize.lo: visualize.cc
800		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-visualize.lo -MD -MP -MF $(DEPDIR)/libde265_la-visualize.Tpo -c -o libde265_la-visualize.lo `test -f 'visualize.cc' \|\| echo '$(srcdir)/'`visualize.cc
801		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-visualize.Tpo $(DEPDIR)/libde265_la-visualize.Plo
802		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='visualize.cc' object='libde265_la-visualize.lo' libtool=yes @AMDEPBACKSLASH@
803		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
804		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-visualize.lo `test -f 'visualize.cc' \|\| echo '$(srcdir)/'`visualize.cc
805
806		libde265_la-fallback.lo: fallback.cc
807		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-fallback.lo -MD -MP -MF $(DEPDIR)/libde265_la-fallback.Tpo -c -o libde265_la-fallback.lo `test -f 'fallback.cc' \|\| echo '$(srcdir)/'`fallback.cc
808		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-fallback.Tpo $(DEPDIR)/libde265_la-fallback.Plo
809		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='fallback.cc' object='libde265_la-fallback.lo' libtool=yes @AMDEPBACKSLASH@
810		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
811		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-fallback.lo `test -f 'fallback.cc' \|\| echo '$(srcdir)/'`fallback.cc
812
813		libde265_la-fallback-motion.lo: fallback-motion.cc
814		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-fallback-motion.lo -MD -MP -MF $(DEPDIR)/libde265_la-fallback-motion.Tpo -c -o libde265_la-fallback-motion.lo `test -f 'fallback-motion.cc' \|\| echo '$(srcdir)/'`fallback-motion.cc
815		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-fallback-motion.Tpo $(DEPDIR)/libde265_la-fallback-motion.Plo
816		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='fallback-motion.cc' object='libde265_la-fallback-motion.lo' libtool=yes @AMDEPBACKSLASH@
817		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
818		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-fallback-motion.lo `test -f 'fallback-motion.cc' \|\| echo '$(srcdir)/'`fallback-motion.cc
819
820		libde265_la-fallback-dct.lo: fallback-dct.cc
821		@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-fallback-dct.lo -MD -MP -MF $(DEPDIR)/libde265_la-fallback-dct.Tpo -c -o libde265_la-fallback-dct.lo `test -f 'fallback-dct.cc' \|\| echo '$(srcdir)/'`fallback-dct.cc
822		@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-fallback-dct.Tpo $(DEPDIR)/libde265_la-fallback-dct.Plo
823		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='fallback-dct.cc' object='libde265_la-fallback-dct.lo' libtool=yes @AMDEPBACKSLASH@
824		@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
825		@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-fallback-dct.lo `test -f 'fallback-dct.cc' \|\| echo '$(srcdir)/'`fallback-dct.cc
	897	libde265_la-vui.lo: vui.cc
	898	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_la-vui.lo -MD -MP -MF $(DEPDIR)/libde265_la-vui.Tpo -c -o libde265_la-vui.lo `test -f 'vui.cc' \|\| echo '$(srcdir)/'`vui.cc
	899	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_la-vui.Tpo $(DEPDIR)/libde265_la-vui.Plo
	900	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='vui.cc' object='libde265_la-vui.lo' libtool=yes @AMDEPBACKSLASH@
	901	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	902	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libde265_la_CPPFLAGS) $(CPPFLAGS) $(libde265_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_la-vui.lo `test -f 'vui.cc' \|\| echo '$(srcdir)/'`vui.cc
826	903
827	904	mostlyclean-libtool:
828	905	-rm -f *.lo

+26

-1

libde265/Makefile.vc7 less more

1	1	# Makefile for Microsoft Visual Studio 2003
2	2	#
3	3	CFLAGS=/I..\extra /I.. /I.
4		CC=cl /nologo
	4	CC=cl /nologo
5	5	LINK=link /nologo /subsystem:console
6	6	DEFINES=/DWIN32 /D_WIN32_WINNT=0x0400 /DNDEBUG /DLIBDE265_EXPORTS /D_CRT_SECURE_NO_WARNINGS /DHAVE_SSE4_1 /DNOMINMAX
7	7

29	29	CFLAGS=$(CFLAGS) $(DEFINES)
30	30
31	31	OBJS=\
	32	alloc_pool.obj \
32	33	bitstream.obj \
33	34	cabac.obj \
	35	configparam.obj \
	36	contextmodel.obj \
34	37	de265.obj \
35	38	deblock.obj \
36	39	decctx.obj \
37	40	dpb.obj \
	41	en265.obj \
38	42	fallback-dct.obj \
39	43	fallback-motion.obj \
40	44	fallback.obj \
41	45	image.obj \
	46	image-io.obj \
42	47	intrapred.obj \
43	48	md5.obj \
44	49	motion.obj \
45	50	nal.obj \
46	51	nal-parser.obj \
47	52	pps.obj \
	53	quality.obj \
48	54	refpic.obj \
49	55	sao.obj \
50	56	scan.obj \

56	62	util.obj \
57	63	visualize.obj \
58	64	vps.obj \
	65	vui.obj \
	66	encoder\analyze.obj \
	67	encoder\encode.obj \
	68	encoder\encoder-context.obj \
	69	encoder\encoder-params.obj \
	70	encoder\encpicbuf.obj \
	71	encoder\sop.obj \
	72	encoder\algo\algo.obj \
	73	encoder\algo\cb-interpartmode.obj \
	74	encoder\algo\cb-intra-inter.obj \
	75	encoder\algo\cb-intrapartmode.obj \
	76	encoder\algo\cb-mergeindex.obj \
	77	encoder\algo\cb-skip.obj \
	78	encoder\algo\cb-split.obj \
	79	encoder\algo\coding-options.obj \
	80	encoder\algo\ctb-qscale.obj \
	81	encoder\algo\pb-mv.obj \
	82	encoder\algo\tb-intrapredmode.obj \
	83	encoder\algo\tb-split.obj \
59	84	x86\sse.obj \
60	85	x86\sse-dct.obj \
61	86	x86\sse-motion.obj \

+306

-20

libde265/acceleration.h less more

22	22
23	23	#include <stddef.h>
24	24	#include <stdint.h>
	25	#include <assert.h>
	26
25	27
26	28	struct acceleration_functions
27	29	{
28	30	void (put_weighted_pred_avg_8)(uint8_t _dst, ptrdiff_t dststride,
29		int16_t src1, int16_t src2, ptrdiff_t srcstride,
	31	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
30	32	int width, int height);
31	33
32	34	void (put_unweighted_pred_8)(uint8_t _dst, ptrdiff_t dststride,
33		int16_t *src, ptrdiff_t srcstride,
	35	const int16_t *src, ptrdiff_t srcstride,
34	36	int width, int height);
35	37
36	38	void (put_weighted_pred_8)(uint8_t _dst, ptrdiff_t dststride,
37		int16_t *src, ptrdiff_t srcstride,
	39	const int16_t *src, ptrdiff_t srcstride,
38	40	int width, int height,
39	41	int w,int o,int log2WD);
40	42	void (put_weighted_bipred_8)(uint8_t _dst, ptrdiff_t dststride,
41		int16_t src1, int16_t src2, ptrdiff_t srcstride,
	43	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
42	44	int width, int height,
43	45	int w1,int o1, int w2,int o2, int log2WD);
44	46
	47
	48	void (put_weighted_pred_avg_16)(uint16_t _dst, ptrdiff_t dststride,
	49	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	50	int width, int height, int bit_depth);
	51
	52	void (put_unweighted_pred_16)(uint16_t _dst, ptrdiff_t dststride,
	53	const int16_t *src, ptrdiff_t srcstride,
	54	int width, int height, int bit_depth);
	55
	56	void (put_weighted_pred_16)(uint16_t _dst, ptrdiff_t dststride,
	57	const int16_t *src, ptrdiff_t srcstride,
	58	int width, int height,
	59	int w,int o,int log2WD, int bit_depth);
	60	void (put_weighted_bipred_16)(uint16_t _dst, ptrdiff_t dststride,
	61	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	62	int width, int height,
	63	int w1,int o1, int w2,int o2, int log2WD, int bit_depth);
	64
	65
	66	void put_weighted_pred_avg(void *_dst, ptrdiff_t dststride,
	67	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	68	int width, int height, int bit_depth) const;
	69
	70	void put_unweighted_pred(void *_dst, ptrdiff_t dststride,
	71	const int16_t *src, ptrdiff_t srcstride,
	72	int width, int height, int bit_depth) const;
	73
	74	void put_weighted_pred(void *_dst, ptrdiff_t dststride,
	75	const int16_t *src, ptrdiff_t srcstride,
	76	int width, int height,
	77	int w,int o,int log2WD, int bit_depth) const;
	78	void put_weighted_bipred(void *_dst, ptrdiff_t dststride,
	79	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	80	int width, int height,
	81	int w1,int o1, int w2,int o2, int log2WD, int bit_depth) const;
	82
	83
	84
	85
45	86	void (put_hevc_epel_8)(int16_t dst, ptrdiff_t dststride,
46		uint8_t *src, ptrdiff_t srcstride, int width, int height,
	87	const uint8_t *src, ptrdiff_t srcstride, int width, int height,
47	88	int mx, int my, int16_t* mcbuffer);
48	89	void (put_hevc_epel_h_8)(int16_t dst, ptrdiff_t dststride,
49		uint8_t *src, ptrdiff_t srcstride, int width, int height,
50		int mx, int my, int16_t* mcbuffer);
	90	const uint8_t *src, ptrdiff_t srcstride, int width, int height,
	91	int mx, int my, int16_t* mcbuffer, int bit_depth);
51	92	void (put_hevc_epel_v_8)(int16_t dst, ptrdiff_t dststride,
52		uint8_t *src, ptrdiff_t srcstride, int width, int height,
53		int mx, int my, int16_t* mcbuffer);
	93	const uint8_t *src, ptrdiff_t srcstride, int width, int height,
	94	int mx, int my, int16_t* mcbuffer, int bit_depth);
54	95	void (put_hevc_epel_hv_8)(int16_t dst, ptrdiff_t dststride,
55		uint8_t *src, ptrdiff_t srcstride, int width, int height,
56		int mx, int my, int16_t* mcbuffer);
	96	const uint8_t *src, ptrdiff_t srcstride, int width, int height,
	97	int mx, int my, int16_t* mcbuffer, int bit_depth);
57	98
58	99	void (put_hevc_qpel_8[4][4])(int16_t dst, ptrdiff_t dststride,
59		uint8_t *src, ptrdiff_t srcstride, int width, int height,
	100	const uint8_t *src, ptrdiff_t srcstride, int width, int height,
60	101	int16_t* mcbuffer);
61	102
62		void (transform_skip_8)(uint8_t _dst, int16_t *coeffs, ptrdiff_t _stride); // no transform
63		void (transform_bypass_8)(uint8_t dst, int16_t *coeffs, int nT, ptrdiff_t stride);
64		void (transform_4x4_luma_add_8)(uint8_t dst, int16_t *coeffs, ptrdiff_t stride); // iDST
65
66		void (transform_4x4_add_8)(uint8_t dst, int16_t *coeffs, ptrdiff_t stride); // iDCT
67		void (transform_8x8_add_8)(uint8_t dst, int16_t *coeffs, ptrdiff_t stride); // iDCT
68		void (transform_16x16_add_8)(uint8_t dst, int16_t *coeffs, ptrdiff_t stride); // iDCT
69		void (transform_32x32_add_8)(uint8_t dst, int16_t *coeffs, ptrdiff_t stride); // iDCT
	103
	104	void (put_hevc_epel_16)(int16_t dst, ptrdiff_t dststride,
	105	const uint16_t *src, ptrdiff_t srcstride, int width, int height,
	106	int mx, int my, int16_t* mcbuffer, int bit_depth);
	107	void (put_hevc_epel_h_16)(int16_t dst, ptrdiff_t dststride,
	108	const uint16_t *src, ptrdiff_t srcstride, int width, int height,
	109	int mx, int my, int16_t* mcbuffer, int bit_depth);
	110	void (put_hevc_epel_v_16)(int16_t dst, ptrdiff_t dststride,
	111	const uint16_t *src, ptrdiff_t srcstride, int width, int height,
	112	int mx, int my, int16_t* mcbuffer, int bit_depth);
	113	void (put_hevc_epel_hv_16)(int16_t dst, ptrdiff_t dststride,
	114	const uint16_t *src, ptrdiff_t srcstride, int width, int height,
	115	int mx, int my, int16_t* mcbuffer, int bit_depth);
	116
	117	void (put_hevc_qpel_16[4][4])(int16_t dst, ptrdiff_t dststride,
	118	const uint16_t *src, ptrdiff_t srcstride, int width, int height,
	119	int16_t* mcbuffer, int bit_depth);
	120
	121
	122	void put_hevc_epel(int16_t *dst, ptrdiff_t dststride,
	123	const void *src, ptrdiff_t srcstride, int width, int height,
	124	int mx, int my, int16_t* mcbuffer, int bit_depth) const;
	125	void put_hevc_epel_h(int16_t *dst, ptrdiff_t dststride,
	126	const void *src, ptrdiff_t srcstride, int width, int height,
	127	int mx, int my, int16_t* mcbuffer, int bit_depth) const;
	128	void put_hevc_epel_v(int16_t *dst, ptrdiff_t dststride,
	129	const void *src, ptrdiff_t srcstride, int width, int height,
	130	int mx, int my, int16_t* mcbuffer, int bit_depth) const;
	131	void put_hevc_epel_hv(int16_t *dst, ptrdiff_t dststride,
	132	const void *src, ptrdiff_t srcstride, int width, int height,
	133	int mx, int my, int16_t* mcbuffer, int bit_depth) const;
	134
	135	void put_hevc_qpel(int16_t *dst, ptrdiff_t dststride,
	136	const void *src, ptrdiff_t srcstride, int width, int height,
	137	int16_t* mcbuffer, int dX,int dY, int bit_depth) const;
	138
	139
	140	// --- inverse transforms ---
	141
	142	void (transform_bypass)(int32_t residual, const int16_t *coeffs, int nT);
	143	void (transform_bypass_rdpcm_v)(int32_t r, const int16_t *coeffs, int nT);
	144	void (transform_bypass_rdpcm_h)(int32_t r, const int16_t *coeffs, int nT);
	145
	146	// 8 bit
	147
	148	void (transform_skip_8)(uint8_t _dst, const int16_t *coeffs, ptrdiff_t _stride); // no transform
	149	void (transform_skip_rdpcm_v_8)(uint8_t _dst, const int16_t *coeffs, int nT, ptrdiff_t _stride);
	150	void (transform_skip_rdpcm_h_8)(uint8_t _dst, const int16_t *coeffs, int nT, ptrdiff_t _stride);
	151	void (transform_4x4_dst_add_8)(uint8_t dst, const int16_t *coeffs, ptrdiff_t stride); // iDST
	152	void (transform_add_8[4])(uint8_t dst, const int16_t *coeffs, ptrdiff_t stride); // iDCT
	153
	154	// 9-16 bit
	155
	156	void (transform_skip_16)(uint16_t _dst, const int16_t *coeffs, ptrdiff_t _stride, int bit_depth); // no transform
	157	void (transform_4x4_dst_add_16)(uint16_t dst, const int16_t *coeffs, ptrdiff_t stride, int bit_depth); // iDST
	158	void (transform_add_16[4])(uint16_t dst, const int16_t *coeffs, ptrdiff_t stride, int bit_depth); // iDCT
	159
	160
	161	void (rotate_coefficients)(int16_t coeff, int nT);
	162
	163	void (transform_idst_4x4)(int32_t dst, const int16_t *coeffs, int bdShift, int max_coeff_bits);
	164	void (transform_idct_4x4)(int32_t dst, const int16_t *coeffs, int bdShift, int max_coeff_bits);
	165	void (transform_idct_8x8)(int32_t dst, const int16_t *coeffs, int bdShift, int max_coeff_bits);
	166	void (transform_idct_16x16)(int32_t dst,const int16_t *coeffs,int bdShift, int max_coeff_bits);
	167	void (transform_idct_32x32)(int32_t dst,const int16_t *coeffs,int bdShift, int max_coeff_bits);
	168	void (add_residual_8)(uint8_t dst, ptrdiff_t stride, const int32_t* r, int nT, int bit_depth);
	169	void (add_residual_16)(uint16_t dst,ptrdiff_t stride,const int32_t* r, int nT, int bit_depth);
	170
	171	template <class pixel_t>
	172	void add_residual(pixel_t dst, ptrdiff_t stride, const int32_t r, int nT, int bit_depth) const;
	173
	174	void (rdpcm_v)(int32_t residual, const int16_t* coeffs, int nT,int tsShift,int bdShift);
	175	void (rdpcm_h)(int32_t residual, const int16_t* coeffs, int nT,int tsShift,int bdShift);
	176
	177	void (transform_skip_residual)(int32_t residual, const int16_t *coeffs, int nT,
	178	int tsShift,int bdShift);
	179
	180
	181	template <class pixel_t> void transform_skip(pixel_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth) const;
	182	template <class pixel_t> void transform_skip_rdpcm_v(pixel_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth) const;
	183	template <class pixel_t> void transform_skip_rdpcm_h(pixel_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth) const;
	184	template <class pixel_t> void transform_4x4_dst_add(pixel_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth) const;
	185	template <class pixel_t> void transform_add(int sizeIdx, pixel_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth) const;
	186
	187
	188
	189	// --- forward transforms ---
	190
	191	void (fwd_transform_4x4_dst_8)(int16_t coeffs, const int16_t* src, ptrdiff_t stride); // fDST
	192
	193	// indexed with (log2TbSize-2)
	194	void (fwd_transform_8[4]) (int16_t coeffs, const int16_t *src, ptrdiff_t stride); // fDCT
	195
	196
	197	// forward Hadamard transform (without scaling factor)
	198	// (4x4,8x8,16x16,32x32) indexed with (log2TbSize-2)
	199	void (hadamard_transform_8[4]) (int16_t coeffs, const int16_t *src, ptrdiff_t stride);
70	200	};
71	201
	202
	203	/*
	204	template <> inline void acceleration_functions::put_weighted_pred_avg<uint8_t>(uint8_t *_dst, ptrdiff_t dststride,
	205	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	206	int width, int height, int bit_depth) { put_weighted_pred_avg_8(_dst,dststride,src1,src2,srcstride,width,height); }
	207	template <> inline void acceleration_functions::put_weighted_pred_avg<uint16_t>(uint16_t *_dst, ptrdiff_t dststride,
	208	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	209	int width, int height, int bit_depth) { put_weighted_pred_avg_16(_dst,dststride,src1,src2,
	210	srcstride,width,height,bit_depth); }
	211
	212	template <> inline void acceleration_functions::put_unweighted_pred<uint8_t>(uint8_t *_dst, ptrdiff_t dststride,
	213	const int16_t *src, ptrdiff_t srcstride,
	214	int width, int height, int bit_depth) { put_unweighted_pred_8(_dst,dststride,src,srcstride,width,height); }
	215	template <> inline void acceleration_functions::put_unweighted_pred<uint16_t>(uint16_t *_dst, ptrdiff_t dststride,
	216	const int16_t *src, ptrdiff_t srcstride,
	217	int width, int height, int bit_depth) { put_unweighted_pred_16(_dst,dststride,src,srcstride,width,height,bit_depth); }
	218
	219	template <> inline void acceleration_functions::put_weighted_pred<uint8_t>(uint8_t *_dst, ptrdiff_t dststride,
	220	const int16_t *src, ptrdiff_t srcstride,
	221	int width, int height,
	222	int w,int o,int log2WD, int bit_depth) { put_weighted_pred_8(_dst,dststride,src,srcstride,width,height,w,o,log2WD); }
	223	template <> inline void acceleration_functions::put_weighted_pred<uint16_t>(uint16_t *_dst, ptrdiff_t dststride,
	224	const int16_t *src, ptrdiff_t srcstride,
	225	int width, int height,
	226	int w,int o,int log2WD, int bit_depth) { put_weighted_pred_16(_dst,dststride,src,srcstride,width,height,w,o,log2WD,bit_depth); }
	227
	228	template <> inline void acceleration_functions::put_weighted_bipred<uint8_t>(uint8_t *_dst, ptrdiff_t dststride,
	229	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	230	int width, int height,
	231	int w1,int o1, int w2,int o2, int log2WD, int bit_depth) { put_weighted_bipred_8(_dst,dststride,src1,src2,srcstride,
	232	width,height,
	233	w1,o1,w2,o2,log2WD); }
	234	template <> inline void acceleration_functions::put_weighted_bipred<uint16_t>(uint16_t *_dst, ptrdiff_t dststride,
	235	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	236	int width, int height,
	237	int w1,int o1, int w2,int o2, int log2WD, int bit_depth) { put_weighted_bipred_16(_dst,dststride,src1,src2,srcstride,
	238	width,height,
	239	w1,o1,w2,o2,log2WD,bit_depth); }
	240	*/
	241
	242
	243	inline void acceleration_functions::put_weighted_pred_avg(void* _dst, ptrdiff_t dststride,
	244	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	245	int width, int height, int bit_depth) const
	246	{
	247	if (bit_depth <= 8)
	248	put_weighted_pred_avg_8((uint8_t*)_dst,dststride,src1,src2,srcstride,width,height);
	249	else
	250	put_weighted_pred_avg_16((uint16_t*)_dst,dststride,src1,src2,srcstride,width,height,bit_depth);
	251	}
	252
	253
	254	inline void acceleration_functions::put_unweighted_pred(void* _dst, ptrdiff_t dststride,
	255	const int16_t *src, ptrdiff_t srcstride,
	256	int width, int height, int bit_depth) const
	257	{
	258	if (bit_depth <= 8)
	259	put_unweighted_pred_8((uint8_t*)_dst,dststride,src,srcstride,width,height);
	260	else
	261	put_unweighted_pred_16((uint16_t*)_dst,dststride,src,srcstride,width,height,bit_depth);
	262	}
	263
	264
	265	inline void acceleration_functions::put_weighted_pred(void* _dst, ptrdiff_t dststride,
	266	const int16_t *src, ptrdiff_t srcstride,
	267	int width, int height,
	268	int w,int o,int log2WD, int bit_depth) const
	269	{
	270	if (bit_depth <= 8)
	271	put_weighted_pred_8((uint8_t*)_dst,dststride,src,srcstride,width,height,w,o,log2WD);
	272	else
	273	put_weighted_pred_16((uint16_t*)_dst,dststride,src,srcstride,width,height,w,o,log2WD,bit_depth);
	274	}
	275
	276
	277	inline void acceleration_functions::put_weighted_bipred(void* _dst, ptrdiff_t dststride,
	278	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	279	int width, int height,
	280	int w1,int o1, int w2,int o2, int log2WD, int bit_depth) const
	281	{
	282	if (bit_depth <= 8)
	283	put_weighted_bipred_8((uint8_t*)_dst,dststride,src1,src2,srcstride, width,height, w1,o1,w2,o2,log2WD);
	284	else
	285	put_weighted_bipred_16((uint16_t*)_dst,dststride,src1,src2,srcstride, width,height, w1,o1,w2,o2,log2WD,bit_depth);
	286	}
	287
	288
	289
	290	inline void acceleration_functions::put_hevc_epel(int16_t *dst, ptrdiff_t dststride,
	291	const void *src, ptrdiff_t srcstride, int width, int height,
	292	int mx, int my, int16_t* mcbuffer, int bit_depth) const
	293	{
	294	if (bit_depth <= 8)
	295	put_hevc_epel_8(dst,dststride,(const uint8_t*)src,srcstride,width,height,mx,my,mcbuffer);
	296	else
	297	put_hevc_epel_16(dst,dststride,(const uint16_t*)src,srcstride,width,height,mx,my,mcbuffer, bit_depth);
	298	}
	299
	300	inline void acceleration_functions::put_hevc_epel_h(int16_t *dst, ptrdiff_t dststride,
	301	const void *src, ptrdiff_t srcstride, int width, int height,
	302	int mx, int my, int16_t* mcbuffer, int bit_depth) const
	303	{
	304	if (bit_depth <= 8)
	305	put_hevc_epel_h_8(dst,dststride,(const uint8_t*)src,srcstride,width,height,mx,my,mcbuffer,bit_depth);
	306	else
	307	put_hevc_epel_h_16(dst,dststride,(const uint16_t*)src,srcstride,width,height,mx,my,mcbuffer,bit_depth);
	308	}
	309
	310	inline void acceleration_functions::put_hevc_epel_v(int16_t *dst, ptrdiff_t dststride,
	311	const void *src, ptrdiff_t srcstride, int width, int height,
	312	int mx, int my, int16_t* mcbuffer, int bit_depth) const
	313	{
	314	if (bit_depth <= 8)
	315	put_hevc_epel_v_8(dst,dststride,(const uint8_t*)src,srcstride,width,height,mx,my,mcbuffer,bit_depth);
	316	else
	317	put_hevc_epel_v_16(dst,dststride,(const uint16_t*)src,srcstride,width,height,mx,my,mcbuffer, bit_depth);
	318	}
	319
	320	inline void acceleration_functions::put_hevc_epel_hv(int16_t *dst, ptrdiff_t dststride,
	321	const void *src, ptrdiff_t srcstride, int width, int height,
	322	int mx, int my, int16_t* mcbuffer, int bit_depth) const
	323	{
	324	if (bit_depth <= 8)
	325	put_hevc_epel_hv_8(dst,dststride,(const uint8_t*)src,srcstride,width,height,mx,my,mcbuffer,bit_depth);
	326	else
	327	put_hevc_epel_hv_16(dst,dststride,(const uint16_t*)src,srcstride,width,height,mx,my,mcbuffer, bit_depth);
	328	}
	329
	330	inline void acceleration_functions::put_hevc_qpel(int16_t *dst, ptrdiff_t dststride,
	331	const void *src, ptrdiff_t srcstride, int width, int height,
	332	int16_t* mcbuffer, int dX,int dY, int bit_depth) const
	333	{
	334	if (bit_depth <= 8)
	335	put_hevc_qpel_8[dX][dY](dst,dststride,(const uint8_t*)src,srcstride,width,height,mcbuffer);
	336	else
	337	put_hevc_qpel_16[dX][dY](dst,dststride,(const uint16_t*)src,srcstride,width,height,mcbuffer, bit_depth);
	338	}
	339
	340	template <> inline void acceleration_functions::transform_skip<uint8_t>(uint8_t dst, const int16_t coeffs,ptrdiff_t stride, int bit_depth) const { transform_skip_8(dst,coeffs,stride); }
	341	template <> inline void acceleration_functions::transform_skip<uint16_t>(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth) const { transform_skip_16(dst,coeffs,stride, bit_depth); }
	342
	343	template <> inline void acceleration_functions::transform_skip_rdpcm_v<uint8_t>(uint8_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth) const { assert(bit_depth==8); transform_skip_rdpcm_v_8(dst,coeffs,nT,stride); }
	344	template <> inline void acceleration_functions::transform_skip_rdpcm_h<uint8_t>(uint8_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth) const { assert(bit_depth==8); transform_skip_rdpcm_h_8(dst,coeffs,nT,stride); }
	345	template <> inline void acceleration_functions::transform_skip_rdpcm_v<uint16_t>(uint16_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth) const { assert(false); /transform_skip_rdpcm_v_8(dst,coeffs,nT,stride);/ }
	346	template <> inline void acceleration_functions::transform_skip_rdpcm_h<uint16_t>(uint16_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth) const { assert(false); /transform_skip_rdpcm_h_8(dst,coeffs,nT,stride);/ }
	347
	348
	349	template <> inline void acceleration_functions::transform_4x4_dst_add<uint8_t>(uint8_t dst, const int16_t coeffs, ptrdiff_t stride,int bit_depth) const { transform_4x4_dst_add_8(dst,coeffs,stride); }
	350	template <> inline void acceleration_functions::transform_4x4_dst_add<uint16_t>(uint16_t dst, const int16_t coeffs, ptrdiff_t stride,int bit_depth) const { transform_4x4_dst_add_16(dst,coeffs,stride,bit_depth); }
	351
	352	template <> inline void acceleration_functions::transform_add<uint8_t>(int sizeIdx, uint8_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth) const { transform_add_8[sizeIdx](dst,coeffs,stride); }
	353	template <> inline void acceleration_functions::transform_add<uint16_t>(int sizeIdx, uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth) const { transform_add_16[sizeIdx](dst,coeffs,stride,bit_depth); }
	354
	355	template <> inline void acceleration_functions::add_residual(uint8_t dst, ptrdiff_t stride, const int32_t r, int nT, int bit_depth) const { add_residual_8(dst,stride,r,nT,bit_depth); }
	356	template <> inline void acceleration_functions::add_residual(uint16_t dst, ptrdiff_t stride, const int32_t r, int nT, int bit_depth) const { add_residual_16(dst,stride,r,nT,bit_depth); }
	357
72	358	#endif

+99

-0

libde265/alloc_pool.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "libde265/alloc_pool.h"
	23	#include "libde265/util.h"
	24	#include <assert.h>
	25	#include <stdio.h>
	26
	27	#define DEBUG_MEMORY 1
	28
	29
	30	alloc_pool::alloc_pool(size_t objSize, int poolSize, bool grow)
	31	: mObjSize(objSize),
	32	mPoolSize(poolSize),
	33	mGrow(grow)
	34	{
	35	m_freeList.reserve(poolSize);
	36	m_memBlocks.reserve(8);
	37
	38	add_memory_block();
	39	}
	40
	41
	42	void alloc_pool::add_memory_block()
	43	{
	44	uint8_t* p = new uint8_t[mObjSize * mPoolSize];
	45	m_memBlocks.push_back(p);
	46
	47	for (int i=0;i<mPoolSize;i++)
	48	{
	49	m_freeList.push_back(p + (mPoolSize-1-i) * mObjSize);
	50	}
	51	}
	52
	53	alloc_pool::~alloc_pool()
	54	{
	55	FOR_LOOP(uint8_t*, p, m_memBlocks) {
	56	delete[] p;
	57	}
	58	}
	59
	60
	61	void* alloc_pool::new_obj(const size_t size)
	62	{
	63	if (size != mObjSize) {
	64	return ::operator new(size);
	65	}
	66
	67	if (m_freeList.size()==0) {
	68	if (mGrow) {
	69	add_memory_block();
	70	if (DEBUG_MEMORY) { fprintf(stderr,"additional block allocated in memory pool\n"); }
	71	}
	72	else {
	73	return NULL;
	74	}
	75	}
	76
	77	assert(!m_freeList.empty());
	78
	79	void* p = m_freeList.back();
	80	m_freeList.pop_back();
	81
	82	return p;
	83	}
	84
	85
	86	void alloc_pool::delete_obj(void* obj)
	87	{
	88	int memBlockSize = mObjSize * mPoolSize;
	89
	90	FOR_LOOP(uint8_t*, memBlk, m_memBlocks) {
	91	if (memBlk <= obj && obj < memBlk + memBlockSize) {
	92	m_freeList.push_back(obj);
	93	return;
	94	}
	95	}
	96
	97	::operator delete(obj);
	98	}

+56

-0

libde265/alloc_pool.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef ALLOC_POOL_H
	23	#define ALLOC_POOL_H
	24
	25	#include <vector>
	26	#include <cstddef>
	27	#ifdef HAVE_STDINT_H
	28	#include <stdint.h>
	29	#else
	30	#include <cstdint>
	31	#endif
	32
	33
	34	class alloc_pool
	35	{
	36	public:
	37	alloc_pool(size_t objSize, int poolSize=1000, bool grow=true);
	38	~alloc_pool();
	39
	40	void* new_obj(const size_t size);
	41	void delete_obj(void*);
	42	void purge();
	43
	44	private:
	45	size_t mObjSize;
	46	int mPoolSize;
	47	bool mGrow;
	48
	49	std::vector<uint8_t*> m_memBlocks;
	50	std::vector<void*> m_freeList;
	51
	52	void add_memory_block();
	53	};
	54
	55	#endif

+38

-0

libde265/arm/Makefile.am less more

	0	noinst_LTLIBRARIES = libde265_arm.la
	1
	2	libde265_arm_la_CXXFLAGS = -I.. $(CFLAG_VISIBILITY)
	3	libde265_arm_la_SOURCES = arm.cc arm.h
	4	libde265_arm_la_LIBADD =
	5
	6	if HAVE_VISIBILITY
	7	libde265_arm_la_CXXFLAGS += -DHAVE_VISIBILITY
	8	endif
	9
	10
	11	if ENABLE_NEON_OPT
	12	# NEON specific functions
	13
	14	noinst_LTLIBRARIES += libde265_arm_neon.la
	15	libde265_arm_la_LIBADD += libde265_arm_neon.la
	16	libde265_arm_neon_la_CXXFLAGS = -mfpu=neon -I.. $(CFLAG_VISIBILITY)
	17	libde265_arm_neon_la_CCASFLAGS = -mfpu=neon -I.. \
	18	-DHAVE_NEON \
	19	-DEXTERN_ASM= \
	20	-DHAVE_AS_FUNC \
	21	-DHAVE_SECTION_DATA_REL_RO
	22
	23	if ENABLE_ARM_THUMB
	24	libde265_arm_neon_la_CCASFLAGS += -DCONFIG_THUMB
	25	endif
	26
	27	libde265_arm_neon_la_SOURCES = \
	28	asm.S \
	29	cpudetect.S \
	30	hevcdsp_qpel_neon.S \
	31	neon.S
	32
	33	if HAVE_VISIBILITY
	34	libde265_arm_neon_la_CXXFLAGS += -DHAVE_VISIBILITY
	35	endif
	36
	37	endif

+736

-0

libde265/arm/Makefile.in less more

	0	# Makefile.in generated by automake 1.14.1 from Makefile.am.
	1	# @configure_input@
	2
	3	# Copyright (C) 1994-2013 Free Software Foundation, Inc.
	4
	5	# This Makefile.in is free software; the Free Software Foundation
	6	# gives unlimited permission to copy and/or distribute it,
	7	# with or without modifications, as long as this notice is preserved.
	8
	9	# This program is distributed in the hope that it will be useful,
	10	# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
	11	# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
	12	# PARTICULAR PURPOSE.
	13
	14	@SET_MAKE@
	15
	16	VPATH = @srcdir@
	17	am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
	18	am__make_running_with_option = \
	19	case $${target_option-} in \
	20	?) ;; \
	21	*) echo "am__make_running_with_option: internal error: invalid" \
	22	"target option '$${target_option-}' specified" >&2; \
	23	exit 1;; \
	24	esac; \
	25	has_opt=no; \
	26	sane_makeflags=$$MAKEFLAGS; \
	27	if $(am__is_gnu_make); then \
	28	sane_makeflags=$$MFLAGS; \
	29	else \
	30	case $$MAKEFLAGS in \
	31	\\[\ \ ]) \
	32	bs=\\; \
	33	sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
	34	\| sed "s/$$bs$$bs[$$bs $$bs ]*//g"`;; \
	35	esac; \
	36	fi; \
	37	skip_next=no; \
	38	strip_trailopt () \
	39	{ \
	40	flg=`printf '%s\n' "$$flg" \| sed "s/$$1.*$$//"`; \
	41	}; \
	42	for flg in $$sane_makeflags; do \
	43	test $$skip_next = yes && { skip_next=no; continue; }; \
	44	case $$flg in \
	45	=\|--*) continue;; \
	46	-*I) strip_trailopt 'I'; skip_next=yes;; \
	47	-I?) strip_trailopt 'I';; \
	48	-*O) strip_trailopt 'O'; skip_next=yes;; \
	49	-O?) strip_trailopt 'O';; \
	50	-*l) strip_trailopt 'l'; skip_next=yes;; \
	51	-l?) strip_trailopt 'l';; \
	52	-[dEDm]) skip_next=yes;; \
	53	-[JT]) skip_next=yes;; \
	54	esac; \
	55	case $$flg in \
	56	$$target_option) has_opt=yes; break;; \
	57	esac; \
	58	done; \
	59	test $$has_opt = yes
	60	am__make_dryrun = (target_option=n; $(am__make_running_with_option))
	61	am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
	62	pkgdatadir = $(datadir)/@PACKAGE@
	63	pkgincludedir = $(includedir)/@PACKAGE@
	64	pkglibdir = $(libdir)/@PACKAGE@
	65	pkglibexecdir = $(libexecdir)/@PACKAGE@
	66	am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
	67	install_sh_DATA = $(install_sh) -c -m 644
	68	install_sh_PROGRAM = $(install_sh) -c
	69	install_sh_SCRIPT = $(install_sh) -c
	70	INSTALL_HEADER = $(INSTALL_DATA)
	71	transform = $(program_transform_name)
	72	NORMAL_INSTALL = :
	73	PRE_INSTALL = :
	74	POST_INSTALL = :
	75	NORMAL_UNINSTALL = :
	76	PRE_UNINSTALL = :
	77	POST_UNINSTALL = :
	78	build_triplet = @build@
	79	host_triplet = @host@
	80	target_triplet = @target@
	81
	82	# NEON specific functions
	83	@ENABLE_NEON_OPT_TRUE@am__append_1 = libde265_arm_neon.la
	84	@ENABLE_NEON_OPT_TRUE@am__append_2 = libde265_arm_neon.la
	85	subdir = libde265/arm
	86	DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/Makefile.am \
	87	$(top_srcdir)/depcomp
	88	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
	89	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	90	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
	91	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
	92	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
	93	$(top_srcdir)/m4/lt~obsolete.m4 \
	94	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
	95	$(top_srcdir)/configure.ac
	96	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
	97	$(ACLOCAL_M4)
	98	mkinstalldirs = $(install_sh) -d
	99	CONFIG_HEADER = $(top_builddir)/config.h
	100	CONFIG_CLEAN_FILES =
	101	CONFIG_CLEAN_VPATH_FILES =
	102	LTLIBRARIES = $(noinst_LTLIBRARIES)
	103	libde265_arm_la_DEPENDENCIES = $(am__append_2)
	104	am_libde265_arm_la_OBJECTS = libde265_arm_la-arm.lo
	105	libde265_arm_la_OBJECTS = $(am_libde265_arm_la_OBJECTS)
	106	AM_V_lt = $(am__v_lt_@AM_V@)
	107	am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
	108	am__v_lt_0 = --silent
	109	am__v_lt_1 =
	110	libde265_arm_la_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
	111	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
	112	$(libde265_arm_la_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) \
	113	$(LDFLAGS) -o $@
	114	libde265_arm_neon_la_LIBADD =
	115	am__libde265_arm_neon_la_SOURCES_DIST = asm.S cpudetect.S \
	116	hevcdsp_qpel_neon.S neon.S
	117	@ENABLE_NEON_OPT_TRUE@am_libde265_arm_neon_la_OBJECTS = \
	118	@ENABLE_NEON_OPT_TRUE@ libde265_arm_neon_la-asm.lo \
	119	@ENABLE_NEON_OPT_TRUE@ libde265_arm_neon_la-cpudetect.lo \
	120	@ENABLE_NEON_OPT_TRUE@ libde265_arm_neon_la-hevcdsp_qpel_neon.lo \
	121	@ENABLE_NEON_OPT_TRUE@ libde265_arm_neon_la-neon.lo
	122	libde265_arm_neon_la_OBJECTS = $(am_libde265_arm_neon_la_OBJECTS)
	123	@ENABLE_NEON_OPT_TRUE@am_libde265_arm_neon_la_rpath =
	124	AM_V_P = $(am__v_P_@AM_V@)
	125	am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
	126	am__v_P_0 = false
	127	am__v_P_1 = :
	128	AM_V_GEN = $(am__v_GEN_@AM_V@)
	129	am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
	130	am__v_GEN_0 = @echo " GEN " $@;
	131	am__v_GEN_1 =
	132	AM_V_at = $(am__v_at_@AM_V@)
	133	am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
	134	am__v_at_0 = @
	135	am__v_at_1 =
	136	DEFAULT_INCLUDES = -I.@am__isrc@ -I$(top_builddir)
	137	depcomp = $(SHELL) $(top_srcdir)/depcomp
	138	am__depfiles_maybe = depfiles
	139	am__mv = mv -f
	140	CPPASCOMPILE = $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
	141	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CCASFLAGS) $(CCASFLAGS)
	142	LTCPPASCOMPILE = $(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) \
	143	$(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) \
	144	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	145	$(AM_CCASFLAGS) $(CCASFLAGS)
	146	AM_V_CPPAS = $(am__v_CPPAS_@AM_V@)
	147	am__v_CPPAS_ = $(am__v_CPPAS_@AM_DEFAULT_V@)
	148	am__v_CPPAS_0 = @echo " CPPAS " $@;
	149	am__v_CPPAS_1 =
	150	CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
	151	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
	152	LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	153	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
	154	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	155	$(AM_CXXFLAGS) $(CXXFLAGS)
	156	AM_V_CXX = $(am__v_CXX_@AM_V@)
	157	am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
	158	am__v_CXX_0 = @echo " CXX " $@;
	159	am__v_CXX_1 =
	160	CXXLD = $(CXX)
	161	CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	162	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
	163	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
	164	AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
	165	am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
	166	am__v_CXXLD_0 = @echo " CXXLD " $@;
	167	am__v_CXXLD_1 =
	168	COMPILE = $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) \
	169	$(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS)
	170	LTCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CC $(AM_LIBTOOLFLAGS) \
	171	$(LIBTOOLFLAGS) --mode=compile $(CC) $(DEFS) \
	172	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	173	$(AM_CFLAGS) $(CFLAGS)
	174	AM_V_CC = $(am__v_CC_@AM_V@)
	175	am__v_CC_ = $(am__v_CC_@AM_DEFAULT_V@)
	176	am__v_CC_0 = @echo " CC " $@;
	177	am__v_CC_1 =
	178	CCLD = $(CC)
	179	LINK = $(LIBTOOL) $(AM_V_lt) --tag=CC $(AM_LIBTOOLFLAGS) \
	180	$(LIBTOOLFLAGS) --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
	181	$(AM_LDFLAGS) $(LDFLAGS) -o $@
	182	AM_V_CCLD = $(am__v_CCLD_@AM_V@)
	183	am__v_CCLD_ = $(am__v_CCLD_@AM_DEFAULT_V@)
	184	am__v_CCLD_0 = @echo " CCLD " $@;
	185	am__v_CCLD_1 =
	186	SOURCES = $(libde265_arm_la_SOURCES) $(libde265_arm_neon_la_SOURCES)
	187	DIST_SOURCES = $(libde265_arm_la_SOURCES) \
	188	$(am__libde265_arm_neon_la_SOURCES_DIST)
	189	am__can_run_installinfo = \
	190	case $$AM_UPDATE_INFO_DIR in \
	191	n\|no\|NO) false;; \
	192	*) (install-info --version) >/dev/null 2>&1;; \
	193	esac
	194	am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
	195	# Read a list of newline-separated strings from the standard input,
	196	# and print each of them once, without duplicates. Input order is
	197	# not preserved.
	198	am__uniquify_input = $(AWK) '\
	199	BEGIN { nonempty = 0; } \
	200	{ items[$$0] = 1; nonempty = 1; } \
	201	END { if (nonempty) { for (i in items) print i; }; } \
	202	'
	203	# Make sure the list of sources is unique. This is necessary because,
	204	# e.g., the same source file might be shared among _SOURCES variables
	205	# for different programs/libraries.
	206	am__define_uniq_tagged_files = \
	207	list='$(am__tagged_files)'; \
	208	unique=`for i in $$list; do \
	209	if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
	210	done \| $(am__uniquify_input)`
	211	ETAGS = etags
	212	CTAGS = ctags
	213	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
	214	ACLOCAL = @ACLOCAL@
	215	ALLOCA = @ALLOCA@
	216	AMTAR = @AMTAR@
	217	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
	218	AR = @AR@
	219	AUTOCONF = @AUTOCONF@
	220	AUTOHEADER = @AUTOHEADER@
	221	AUTOMAKE = @AUTOMAKE@
	222	AWK = @AWK@
	223	CC = @CC@
	224	CCAS = @CCAS@
	225	CCASDEPMODE = @CCASDEPMODE@
	226	CCASFLAGS = @CCASFLAGS@
	227	CCDEPMODE = @CCDEPMODE@
	228	CFLAGS = @CFLAGS@
	229	CPP = @CPP@
	230	CPPFLAGS = @CPPFLAGS@
	231	CXX = @CXX@
	232	CXXCPP = @CXXCPP@
	233	CXXDEPMODE = @CXXDEPMODE@
	234	CXXFLAGS = @CXXFLAGS@
	235	CYGPATH_W = @CYGPATH_W@
	236	DEFS = @DEFS@
	237	DEPDIR = @DEPDIR@
	238	DLLTOOL = @DLLTOOL@
	239	DSYMUTIL = @DSYMUTIL@
	240	DUMPBIN = @DUMPBIN@
	241	ECHO_C = @ECHO_C@
	242	ECHO_N = @ECHO_N@
	243	ECHO_T = @ECHO_T@
	244	EGREP = @EGREP@
	245	EXEEXT = @EXEEXT@
	246	FGREP = @FGREP@
	247	GREP = @GREP@
	248	HAVE_CXX11 = @HAVE_CXX11@
	249	INSTALL = @INSTALL@
	250	INSTALL_DATA = @INSTALL_DATA@
	251	INSTALL_PROGRAM = @INSTALL_PROGRAM@
	252	INSTALL_SCRIPT = @INSTALL_SCRIPT@
	253	INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
	254	LD = @LD@
	255	LDFLAGS = @LDFLAGS@
	256	LIBDE265_AGE = @LIBDE265_AGE@
	257	LIBDE265_CURRENT = @LIBDE265_CURRENT@
	258	LIBDE265_REVISION = @LIBDE265_REVISION@
	259	LIBOBJS = @LIBOBJS@
	260	LIBS = @LIBS@
	261	LIBTOOL = @LIBTOOL@
	262	LIPO = @LIPO@
	263	LN_S = @LN_S@
	264	LTLIBOBJS = @LTLIBOBJS@
	265	MAKEINFO = @MAKEINFO@
	266	MANIFEST_TOOL = @MANIFEST_TOOL@
	267	MKDIR_P = @MKDIR_P@
	268	NM = @NM@
	269	NMEDIT = @NMEDIT@
	270	NUMERIC_VERSION = @NUMERIC_VERSION@
	271	OBJDUMP = @OBJDUMP@
	272	OBJEXT = @OBJEXT@
	273	OTOOL = @OTOOL@
	274	OTOOL64 = @OTOOL64@
	275	PACKAGE = @PACKAGE@
	276	PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
	277	PACKAGE_NAME = @PACKAGE_NAME@
	278	PACKAGE_STRING = @PACKAGE_STRING@
	279	PACKAGE_TARNAME = @PACKAGE_TARNAME@
	280	PACKAGE_URL = @PACKAGE_URL@
	281	PACKAGE_VERSION = @PACKAGE_VERSION@
	282	PATH_SEPARATOR = @PATH_SEPARATOR@
	283	PKG_CONFIG = @PKG_CONFIG@
	284	PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
	285	PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
	286	QTCHOOSER = @QTCHOOSER@
	287	QTMOC = @QTMOC@
	288	QT_CFLAGS = @QT_CFLAGS@
	289	QT_LIBS = @QT_LIBS@
	290	RANLIB = @RANLIB@
	291	SDL_CFLAGS = @SDL_CFLAGS@
	292	SDL_LIBS = @SDL_LIBS@
	293	SED = @SED@
	294	SET_MAKE = @SET_MAKE@
	295	SHELL = @SHELL@
	296	STRIP = @STRIP@
	297	SWSCALE_CFLAGS = @SWSCALE_CFLAGS@
	298	SWSCALE_LIBS = @SWSCALE_LIBS@
	299	VERSION = @VERSION@
	300	VIDEOGFX_CFLAGS = @VIDEOGFX_CFLAGS@
	301	VIDEOGFX_LIBS = @VIDEOGFX_LIBS@
	302	abs_builddir = @abs_builddir@
	303	abs_srcdir = @abs_srcdir@
	304	abs_top_builddir = @abs_top_builddir@
	305	abs_top_srcdir = @abs_top_srcdir@
	306	ac_ct_AR = @ac_ct_AR@
	307	ac_ct_CC = @ac_ct_CC@
	308	ac_ct_CXX = @ac_ct_CXX@
	309	ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
	310	am__include = @am__include@
	311	am__leading_dot = @am__leading_dot@
	312	am__quote = @am__quote@
	313	am__tar = @am__tar@
	314	am__untar = @am__untar@
	315	bindir = @bindir@
	316	build = @build@
	317	build_alias = @build_alias@
	318	build_cpu = @build_cpu@
	319	build_os = @build_os@
	320	build_vendor = @build_vendor@
	321	builddir = @builddir@
	322	datadir = @datadir@
	323	datarootdir = @datarootdir@
	324	docdir = @docdir@
	325	dvidir = @dvidir@
	326	exec_prefix = @exec_prefix@
	327	host = @host@
	328	host_alias = @host_alias@
	329	host_cpu = @host_cpu@
	330	host_os = @host_os@
	331	host_vendor = @host_vendor@
	332	htmldir = @htmldir@
	333	includedir = @includedir@
	334	infodir = @infodir@
	335	install_sh = @install_sh@
	336	libdir = @libdir@
	337	libexecdir = @libexecdir@
	338	localedir = @localedir@
	339	localstatedir = @localstatedir@
	340	mandir = @mandir@
	341	mkdir_p = @mkdir_p@
	342	oldincludedir = @oldincludedir@
	343	pdfdir = @pdfdir@
	344	prefix = @prefix@
	345	program_transform_name = @program_transform_name@
	346	psdir = @psdir@
	347	sbindir = @sbindir@
	348	sharedstatedir = @sharedstatedir@
	349	srcdir = @srcdir@
	350	sysconfdir = @sysconfdir@
	351	target = @target@
	352	target_alias = @target_alias@
	353	target_cpu = @target_cpu@
	354	target_os = @target_os@
	355	target_vendor = @target_vendor@
	356	top_build_prefix = @top_build_prefix@
	357	top_builddir = @top_builddir@
	358	top_srcdir = @top_srcdir@
	359	noinst_LTLIBRARIES = libde265_arm.la $(am__append_1)
	360	libde265_arm_la_CXXFLAGS = -I.. $(CFLAG_VISIBILITY)
	361	libde265_arm_la_SOURCES = arm.cc arm.h
	362	libde265_arm_la_LIBADD = $(am__append_2)
	363	@ENABLE_NEON_OPT_TRUE@libde265_arm_neon_la_CXXFLAGS = -mfpu=neon -I.. $(CFLAG_VISIBILITY)
	364	@ENABLE_NEON_OPT_TRUE@libde265_arm_neon_la_CCASFLAGS = -mfpu=neon -I.. \
	365	@ENABLE_NEON_OPT_TRUE@ -DHAVE_NEON \
	366	@ENABLE_NEON_OPT_TRUE@ -DEXTERN_ASM= \
	367	@ENABLE_NEON_OPT_TRUE@ -DHAVE_AS_FUNC \
	368	@ENABLE_NEON_OPT_TRUE@ -DHAVE_SECTION_DATA_REL_RO
	369
	370	@ENABLE_NEON_OPT_TRUE@libde265_arm_neon_la_SOURCES = \
	371	@ENABLE_NEON_OPT_TRUE@ asm.S \
	372	@ENABLE_NEON_OPT_TRUE@ cpudetect.S \
	373	@ENABLE_NEON_OPT_TRUE@ hevcdsp_qpel_neon.S \
	374	@ENABLE_NEON_OPT_TRUE@ neon.S
	375
	376	all: all-am
	377
	378	.SUFFIXES:
	379	.SUFFIXES: .S .cc .lo .o .obj
	380	$(srcdir)/Makefile.in: $(srcdir)/Makefile.am $(am__configure_deps)
	381	@for dep in $?; do \
	382	case '$(am__configure_deps)' in \
	383	$$dep) \
	384	( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
	385	&& { if test -f $@; then exit 0; else break; fi; }; \
	386	exit 1;; \
	387	esac; \
	388	done; \
	389	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu libde265/arm/Makefile'; \
	390	$(am__cd) $(top_srcdir) && \
	391	$(AUTOMAKE) --gnu libde265/arm/Makefile
	392	.PRECIOUS: Makefile
	393	Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
	394	@case '$?' in \
	395	config.status) \
	396	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
	397	*) \
	398	echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
	399	cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
	400	esac;
	401
	402	$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
	403	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	404
	405	$(top_srcdir)/configure: $(am__configure_deps)
	406	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	407	$(ACLOCAL_M4): $(am__aclocal_m4_deps)
	408	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	409	$(am__aclocal_m4_deps):
	410
	411	clean-noinstLTLIBRARIES:
	412	-test -z "$(noinst_LTLIBRARIES)" \|\| rm -f $(noinst_LTLIBRARIES)
	413	@list='$(noinst_LTLIBRARIES)'; \
	414	locs=`for p in $$list; do echo $$p; done \| \
	415	sed 's\|^[^/]$$\|.\|; s\|/[^/]$$\|\|; s\|$$\|/so_locations\|' \| \
	416	sort -u`; \
	417	test -z "$$locs" \|\| { \
	418	echo rm -f $${locs}; \
	419	rm -f $${locs}; \
	420	}
	421
	422	libde265_arm.la: $(libde265_arm_la_OBJECTS) $(libde265_arm_la_DEPENDENCIES) $(EXTRA_libde265_arm_la_DEPENDENCIES)
	423	$(AM_V_CXXLD)$(libde265_arm_la_LINK) $(libde265_arm_la_OBJECTS) $(libde265_arm_la_LIBADD) $(LIBS)
	424
	425	libde265_arm_neon.la: $(libde265_arm_neon_la_OBJECTS) $(libde265_arm_neon_la_DEPENDENCIES) $(EXTRA_libde265_arm_neon_la_DEPENDENCIES)
	426	$(AM_V_CCLD)$(LINK) $(am_libde265_arm_neon_la_rpath) $(libde265_arm_neon_la_OBJECTS) $(libde265_arm_neon_la_LIBADD) $(LIBS)
	427
	428	mostlyclean-compile:
	429	-rm -f *.$(OBJEXT)
	430
	431	distclean-compile:
	432	-rm -f *.tab.c
	433
	434	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_arm_la-arm.Plo@am__quote@
	435	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_arm_neon_la-asm.Plo@am__quote@
	436	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_arm_neon_la-cpudetect.Plo@am__quote@
	437	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_arm_neon_la-hevcdsp_qpel_neon.Plo@am__quote@
	438	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_arm_neon_la-neon.Plo@am__quote@
	439
	440	.S.o:
	441	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(CPPASCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	442	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	443	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	444	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	445	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(CPPASCOMPILE) -c -o $@ $<
	446
	447	.S.obj:
	448	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(CPPASCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
	449	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	450	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	451	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	452	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(CPPASCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
	453
	454	.S.lo:
	455	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(LTCPPASCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	456	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Plo
	457	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
	458	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	459	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(LTCPPASCOMPILE) -c -o $@ $<
	460
	461	libde265_arm_neon_la-asm.lo: asm.S
	462	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -MT libde265_arm_neon_la-asm.lo -MD -MP -MF $(DEPDIR)/libde265_arm_neon_la-asm.Tpo -c -o libde265_arm_neon_la-asm.lo `test -f 'asm.S' \|\| echo '$(srcdir)/'`asm.S
	463	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_arm_neon_la-asm.Tpo $(DEPDIR)/libde265_arm_neon_la-asm.Plo
	464	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='asm.S' object='libde265_arm_neon_la-asm.lo' libtool=yes @AMDEPBACKSLASH@
	465	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	466	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -c -o libde265_arm_neon_la-asm.lo `test -f 'asm.S' \|\| echo '$(srcdir)/'`asm.S
	467
	468	libde265_arm_neon_la-cpudetect.lo: cpudetect.S
	469	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -MT libde265_arm_neon_la-cpudetect.lo -MD -MP -MF $(DEPDIR)/libde265_arm_neon_la-cpudetect.Tpo -c -o libde265_arm_neon_la-cpudetect.lo `test -f 'cpudetect.S' \|\| echo '$(srcdir)/'`cpudetect.S
	470	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_arm_neon_la-cpudetect.Tpo $(DEPDIR)/libde265_arm_neon_la-cpudetect.Plo
	471	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='cpudetect.S' object='libde265_arm_neon_la-cpudetect.lo' libtool=yes @AMDEPBACKSLASH@
	472	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	473	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -c -o libde265_arm_neon_la-cpudetect.lo `test -f 'cpudetect.S' \|\| echo '$(srcdir)/'`cpudetect.S
	474
	475	libde265_arm_neon_la-hevcdsp_qpel_neon.lo: hevcdsp_qpel_neon.S
	476	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -MT libde265_arm_neon_la-hevcdsp_qpel_neon.lo -MD -MP -MF $(DEPDIR)/libde265_arm_neon_la-hevcdsp_qpel_neon.Tpo -c -o libde265_arm_neon_la-hevcdsp_qpel_neon.lo `test -f 'hevcdsp_qpel_neon.S' \|\| echo '$(srcdir)/'`hevcdsp_qpel_neon.S
	477	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_arm_neon_la-hevcdsp_qpel_neon.Tpo $(DEPDIR)/libde265_arm_neon_la-hevcdsp_qpel_neon.Plo
	478	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='hevcdsp_qpel_neon.S' object='libde265_arm_neon_la-hevcdsp_qpel_neon.lo' libtool=yes @AMDEPBACKSLASH@
	479	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	480	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -c -o libde265_arm_neon_la-hevcdsp_qpel_neon.lo `test -f 'hevcdsp_qpel_neon.S' \|\| echo '$(srcdir)/'`hevcdsp_qpel_neon.S
	481
	482	libde265_arm_neon_la-neon.lo: neon.S
	483	@am__fastdepCCAS_TRUE@ $(AM_V_CPPAS)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -MT libde265_arm_neon_la-neon.lo -MD -MP -MF $(DEPDIR)/libde265_arm_neon_la-neon.Tpo -c -o libde265_arm_neon_la-neon.lo `test -f 'neon.S' \|\| echo '$(srcdir)/'`neon.S
	484	@am__fastdepCCAS_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_arm_neon_la-neon.Tpo $(DEPDIR)/libde265_arm_neon_la-neon.Plo
	485	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS)source='neon.S' object='libde265_arm_neon_la-neon.lo' libtool=yes @AMDEPBACKSLASH@
	486	@AMDEP_TRUE@@am__fastdepCCAS_FALSE@ DEPDIR=$(DEPDIR) $(CCASDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	487	@am__fastdepCCAS_FALSE@ $(AM_V_CPPAS@am__nodep@)$(LIBTOOL) $(AM_V_lt) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CCAS) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_neon_la_CCASFLAGS) $(CCASFLAGS) -c -o libde265_arm_neon_la-neon.lo `test -f 'neon.S' \|\| echo '$(srcdir)/'`neon.S
	488
	489	.cc.o:
	490	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	491	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	492	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	493	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	494	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
	495
	496	.cc.obj:
	497	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
	498	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	499	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	500	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	501	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
	502
	503	.cc.lo:
	504	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	505	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Plo
	506	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
	507	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	508	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
	509
	510	libde265_arm_la-arm.lo: arm.cc
	511	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_arm_la-arm.lo -MD -MP -MF $(DEPDIR)/libde265_arm_la-arm.Tpo -c -o libde265_arm_la-arm.lo `test -f 'arm.cc' \|\| echo '$(srcdir)/'`arm.cc
	512	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_arm_la-arm.Tpo $(DEPDIR)/libde265_arm_la-arm.Plo
	513	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='arm.cc' object='libde265_arm_la-arm.lo' libtool=yes @AMDEPBACKSLASH@
	514	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	515	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_arm_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_arm_la-arm.lo `test -f 'arm.cc' \|\| echo '$(srcdir)/'`arm.cc
	516
	517	mostlyclean-libtool:
	518	-rm -f *.lo
	519
	520	clean-libtool:
	521	-rm -rf .libs _libs
	522
	523	ID: $(am__tagged_files)
	524	$(am__define_uniq_tagged_files); mkid -fID $$unique
	525	tags: tags-am
	526	TAGS: tags
	527
	528	tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	529	set x; \
	530	here=`pwd`; \
	531	$(am__define_uniq_tagged_files); \
	532	shift; \
	533	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
	534	test -n "$$unique" \|\| unique=$$empty_fix; \
	535	if test $$# -gt 0; then \
	536	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	537	"$$@" $$unique; \
	538	else \
	539	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	540	$$unique; \
	541	fi; \
	542	fi
	543	ctags: ctags-am
	544
	545	CTAGS: ctags
	546	ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	547	$(am__define_uniq_tagged_files); \
	548	test -z "$(CTAGS_ARGS)$$unique" \
	549	\|\| $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
	550	$$unique
	551
	552	GTAGS:
	553	here=`$(am__cd) $(top_builddir) && pwd` \
	554	&& $(am__cd) $(top_srcdir) \
	555	&& gtags -i $(GTAGS_ARGS) "$$here"
	556	cscopelist: cscopelist-am
	557
	558	cscopelist-am: $(am__tagged_files)
	559	list='$(am__tagged_files)'; \
	560	case "$(srcdir)" in \
	561	[\\/]* \| ?:[\\/]*) sdir="$(srcdir)" ;; \
	562	*) sdir=$(subdir)/$(srcdir) ;; \
	563	esac; \
	564	for i in $$list; do \
	565	if test -f "$$i"; then \
	566	echo "$(subdir)/$$i"; \
	567	else \
	568	echo "$$sdir/$$i"; \
	569	fi; \
	570	done >> $(top_builddir)/cscope.files
	571
	572	distclean-tags:
	573	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
	574
	575	distdir: $(DISTFILES)
	576	@srcdirstrip=`echo "$(srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	577	topsrcdirstrip=`echo "$(top_srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	578	list='$(DISTFILES)'; \
	579	dist_files=`for file in $$list; do echo $$file; done \| \
	580	sed -e "s\|^$$srcdirstrip/\|\|;t" \
	581	-e "s\|^$$topsrcdirstrip/\|$(top_builddir)/\|;t"`; \
	582	case $$dist_files in \
	583	/) $(MKDIR_P) `echo "$$dist_files" \| \
	584	sed '/\//!d;s\|^\|$(distdir)/\|;s,/[^/]*$$,,' \| \
	585	sort -u` ;; \
	586	esac; \
	587	for file in $$dist_files; do \
	588	if test -f $$file \|\| test -d $$file; then d=.; else d=$(srcdir); fi; \
	589	if test -d $$d/$$file; then \
	590	dir=`echo "/$$file" \| sed -e 's,/[^/]*$$,,'`; \
	591	if test -d "$(distdir)/$$file"; then \
	592	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	593	fi; \
	594	if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
	595	cp -fpR $(srcdir)/$$file "$(distdir)$$dir" \|\| exit 1; \
	596	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	597	fi; \
	598	cp -fpR $$d/$$file "$(distdir)$$dir" \|\| exit 1; \
	599	else \
	600	test -f "$(distdir)/$$file" \
	601	\|\| cp -p $$d/$$file "$(distdir)/$$file" \
	602	\|\| exit 1; \
	603	fi; \
	604	done
	605	check-am: all-am
	606	check: check-am
	607	all-am: Makefile $(LTLIBRARIES)
	608	installdirs:
	609	install: install-am
	610	install-exec: install-exec-am
	611	install-data: install-data-am
	612	uninstall: uninstall-am
	613
	614	install-am: all-am
	615	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
	616
	617	installcheck: installcheck-am
	618	install-strip:
	619	if test -z '$(STRIP)'; then \
	620	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	621	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	622	install; \
	623	else \
	624	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	625	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	626	"INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
	627	fi
	628	mostlyclean-generic:
	629
	630	clean-generic:
	631
	632	distclean-generic:
	633	-test -z "$(CONFIG_CLEAN_FILES)" \|\| rm -f $(CONFIG_CLEAN_FILES)
	634	-test . = "$(srcdir)" \|\| test -z "$(CONFIG_CLEAN_VPATH_FILES)" \|\| rm -f $(CONFIG_CLEAN_VPATH_FILES)
	635
	636	maintainer-clean-generic:
	637	@echo "This command is intended for maintainers to use"
	638	@echo "it deletes files that may require special tools to rebuild."
	639	clean: clean-am
	640
	641	clean-am: clean-generic clean-libtool clean-noinstLTLIBRARIES \
	642	mostlyclean-am
	643
	644	distclean: distclean-am
	645	-rm -rf ./$(DEPDIR)
	646	-rm -f Makefile
	647	distclean-am: clean-am distclean-compile distclean-generic \
	648	distclean-tags
	649
	650	dvi: dvi-am
	651
	652	dvi-am:
	653
	654	html: html-am
	655
	656	html-am:
	657
	658	info: info-am
	659
	660	info-am:
	661
	662	install-data-am:
	663
	664	install-dvi: install-dvi-am
	665
	666	install-dvi-am:
	667
	668	install-exec-am:
	669
	670	install-html: install-html-am
	671
	672	install-html-am:
	673
	674	install-info: install-info-am
	675
	676	install-info-am:
	677
	678	install-man:
	679
	680	install-pdf: install-pdf-am
	681
	682	install-pdf-am:
	683
	684	install-ps: install-ps-am
	685
	686	install-ps-am:
	687
	688	installcheck-am:
	689
	690	maintainer-clean: maintainer-clean-am
	691	-rm -rf ./$(DEPDIR)
	692	-rm -f Makefile
	693	maintainer-clean-am: distclean-am maintainer-clean-generic
	694
	695	mostlyclean: mostlyclean-am
	696
	697	mostlyclean-am: mostlyclean-compile mostlyclean-generic \
	698	mostlyclean-libtool
	699
	700	pdf: pdf-am
	701
	702	pdf-am:
	703
	704	ps: ps-am
	705
	706	ps-am:
	707
	708	uninstall-am:
	709
	710	.MAKE: install-am install-strip
	711
	712	.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean clean-generic \
	713	clean-libtool clean-noinstLTLIBRARIES cscopelist-am ctags \
	714	ctags-am distclean distclean-compile distclean-generic \
	715	distclean-libtool distclean-tags distdir dvi dvi-am html \
	716	html-am info info-am install install-am install-data \
	717	install-data-am install-dvi install-dvi-am install-exec \
	718	install-exec-am install-html install-html-am install-info \
	719	install-info-am install-man install-pdf install-pdf-am \
	720	install-ps install-ps-am install-strip installcheck \
	721	installcheck-am installdirs maintainer-clean \
	722	maintainer-clean-generic mostlyclean mostlyclean-compile \
	723	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
	724	tags tags-am uninstall uninstall-am
	725
	726
	727	@HAVE_VISIBILITY_TRUE@ libde265_arm_la_CXXFLAGS += -DHAVE_VISIBILITY
	728
	729	@ENABLE_ARM_THUMB_TRUE@@ENABLE_NEON_OPT_TRUE@ libde265_arm_neon_la_CCASFLAGS += -DCONFIG_THUMB
	730
	731	@ENABLE_NEON_OPT_TRUE@@HAVE_VISIBILITY_TRUE@ libde265_arm_neon_la_CXXFLAGS += -DHAVE_VISIBILITY
	732
	733	# Tell versions [3.59,3.63) of GNU make to not export all variables.
	734	# Otherwise a system limit (for SysV at least) may be exceeded.
	735	.NOEXPORT:

+123

-0

libde265/arm/arm.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2015 struktur AG, Joachim Bauch <bauch@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifdef HAVE_CONFIG_H
	21	#include "config.h"
	22	#endif
	23
	24	#include "arm.h"
	25
	26	#ifdef HAVE_NEON
	27
	28	#define QPEL_FUNC(name) \
	29	extern "C" void ff_##name(int16_t dst, ptrdiff_t dststride, const uint8_t src, ptrdiff_t srcstride, \
	30	int height, int width); \
	31	void libde265_##name(int16_t dst, ptrdiff_t dststride, const uint8_t src, ptrdiff_t srcstride, \
	32	int width, int height, int16_t* mcbuffer) { \
	33	ff_##name(dst, dststride, src, srcstride, height, width); \
	34	}
	35
	36	QPEL_FUNC(hevc_put_qpel_v1_neon_8);
	37	QPEL_FUNC(hevc_put_qpel_v2_neon_8);
	38	QPEL_FUNC(hevc_put_qpel_v3_neon_8);
	39	QPEL_FUNC(hevc_put_qpel_h1_neon_8);
	40	QPEL_FUNC(hevc_put_qpel_h2_neon_8);
	41	QPEL_FUNC(hevc_put_qpel_h3_neon_8);
	42	QPEL_FUNC(hevc_put_qpel_h1v1_neon_8);
	43	QPEL_FUNC(hevc_put_qpel_h1v2_neon_8);
	44	QPEL_FUNC(hevc_put_qpel_h1v3_neon_8);
	45	QPEL_FUNC(hevc_put_qpel_h2v1_neon_8);
	46	QPEL_FUNC(hevc_put_qpel_h2v2_neon_8);
	47	QPEL_FUNC(hevc_put_qpel_h2v3_neon_8);
	48	QPEL_FUNC(hevc_put_qpel_h3v1_neon_8);
	49	QPEL_FUNC(hevc_put_qpel_h3v2_neon_8);
	50	QPEL_FUNC(hevc_put_qpel_h3v3_neon_8);
	51	#undef QPEL_FUNC
	52
	53	#if defined(HAVE_SIGNAL_H) && defined(HAVE_SETJMP_H)
	54
	55	#include <signal.h>
	56	#include <setjmp.h>
	57
	58	extern "C" void libde265_detect_neon(void);
	59
	60	static jmp_buf jump_env;
	61
	62	static void sighandler(int sig) {
	63	(void)sig;
	64	longjmp(jump_env, 1);
	65	}
	66
	67	static bool has_NEON() {
	68	static bool checked_NEON = false;
	69	static bool have_NEON = false;
	70
	71	if (!checked_NEON) {
	72	void (*oldsignal)(int);
	73
	74	checked_NEON = true;
	75	oldsignal = signal(SIGILL, sighandler);
	76	if (setjmp(jump_env)) {
	77	signal(SIGILL, oldsignal);
	78	have_NEON = false;
	79	return false;
	80	}
	81	libde265_detect_neon();
	82	signal(SIGILL, oldsignal);
	83	have_NEON = true;
	84	}
	85
	86	return have_NEON;
	87	}
	88
	89	#else // #if defined(HAVE_SIGNAL_H) && defined(HAVE_SETJMP_H)
	90
	91	#warning "Don't know how to detect NEON support at runtime- will be disabled"
	92
	93	static bool has_NEON() {
	94	return false;
	95	}
	96
	97	#endif
	98
	99	#endif // #ifdef HAVE_NEON
	100
	101	void init_acceleration_functions_arm(struct acceleration_functions* accel)
	102	{
	103	#ifdef HAVE_NEON
	104	if (has_NEON()) {
	105	accel->put_hevc_qpel_8[0][1] = libde265_hevc_put_qpel_v1_neon_8;
	106	accel->put_hevc_qpel_8[0][2] = libde265_hevc_put_qpel_v2_neon_8;
	107	accel->put_hevc_qpel_8[0][3] = libde265_hevc_put_qpel_v3_neon_8;
	108	accel->put_hevc_qpel_8[1][0] = libde265_hevc_put_qpel_h1_neon_8;
	109	accel->put_hevc_qpel_8[1][1] = libde265_hevc_put_qpel_h1v1_neon_8;
	110	accel->put_hevc_qpel_8[1][2] = libde265_hevc_put_qpel_h1v2_neon_8;
	111	accel->put_hevc_qpel_8[1][3] = libde265_hevc_put_qpel_h1v3_neon_8;
	112	accel->put_hevc_qpel_8[2][0] = libde265_hevc_put_qpel_h2_neon_8;
	113	accel->put_hevc_qpel_8[2][1] = libde265_hevc_put_qpel_h2v1_neon_8;
	114	accel->put_hevc_qpel_8[2][2] = libde265_hevc_put_qpel_h2v2_neon_8;
	115	accel->put_hevc_qpel_8[2][3] = libde265_hevc_put_qpel_h2v3_neon_8;
	116	accel->put_hevc_qpel_8[3][0] = libde265_hevc_put_qpel_h3_neon_8;
	117	accel->put_hevc_qpel_8[3][1] = libde265_hevc_put_qpel_h3v1_neon_8;
	118	accel->put_hevc_qpel_8[3][2] = libde265_hevc_put_qpel_h3v2_neon_8;
	119	accel->put_hevc_qpel_8[3][3] = libde265_hevc_put_qpel_h3v3_neon_8;
	120	}
	121	#endif // #ifdef HAVE_NEON
	122	}

+28

-0

libde265/arm/arm.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2015 struktur AG, Joachim Bauch <bauch@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifndef LIBDE265_ARM_H
	21	#define LIBDE265_ARM_H
	22
	23	#include "acceleration.h"
	24
	25	void init_acceleration_functions_arm(struct acceleration_functions* accel);
	26
	27	#endif // LIBDE265_ARM_H

+325

-0

libde265/arm/asm.S less more

	0	/*
	1	* Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
	2	*
	3	* This file is part of FFmpeg.
	4	*
	5	* FFmpeg is free software; you can redistribute it and/or
	6	* modify it under the terms of the GNU Lesser General Public
	7	* License as published by the Free Software Foundation; either
	8	* version 2.1 of the License, or (at your option) any later version.
	9	*
	10	* FFmpeg is distributed in the hope that it will be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	13	* Lesser General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU Lesser General Public
	16	* License along with FFmpeg; if not, write to the Free Software
	17	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
	18	*/
	19
	20	#include "config.h"
	21
	22	#ifdef __ELF__
	23	# define ELF
	24	#else
	25	# define ELF @
	26	#endif
	27
	28	#if CONFIG_THUMB
	29	# define A @
	30	# define T
	31	#else
	32	# define A
	33	# define T @
	34	#endif
	35
	36	#if HAVE_AS_FUNC
	37	# define FUNC
	38	#else
	39	# define FUNC @
	40	#endif
	41
	42	#if HAVE_NEON
	43	.arch armv7-a
	44	#elif HAVE_ARMV6T2
	45	.arch armv6t2
	46	#elif HAVE_ARMV6
	47	.arch armv6
	48	#elif HAVE_ARMV5TE
	49	.arch armv5te
	50	#endif
	51
	52	#if HAVE_NEON
	53	.fpu neon
	54	#elif HAVE_VFP
	55	.fpu vfp
	56	#endif
	57
	58	.syntax unified
	59	T .thumb
	60	ELF .eabi_attribute 25, 1 @ Tag_ABI_align_preserved
	61	ELF .section .note.GNU-stack,"",%progbits @ Mark stack as non-executable
	62
	63	.macro function name, export=0, align=2
	64	.set .Lpic_idx, 0
	65	.set .Lpic_gp, 0
	66	.macro endfunc
	67	.if .Lpic_idx
	68	.align 2
	69	.altmacro
	70	put_pic %(.Lpic_idx - 1)
	71	.noaltmacro
	72	.endif
	73	ELF .size \name, . - \name
	74	FUNC .endfunc
	75	.purgem endfunc
	76	.endm
	77	.text
	78	.align \align
	79	.if \export
	80	.global EXTERN_ASM\name
	81	ELF .type EXTERN_ASM\name, %function
	82	FUNC .func EXTERN_ASM\name
	83	EXTERN_ASM\name:
	84	.else
	85	ELF .type \name, %function
	86	FUNC .func \name
	87	\name:
	88	.endif
	89	.endm
	90
	91	.macro const name, align=2, relocate=0
	92	.macro endconst
	93	ELF .size \name, . - \name
	94	.purgem endconst
	95	.endm
	96	.if HAVE_SECTION_DATA_REL_RO && \relocate
	97	.section .data.rel.ro
	98	.else
	99	.section .rodata
	100	.endif
	101	.align \align
	102	\name:
	103	.endm
	104
	105	#if !HAVE_ARMV6T2_EXTERNAL
	106	.macro movw rd, val
	107	mov \rd, \val & 255
	108	orr \rd, \val & ~255
	109	.endm
	110	#endif
	111
	112	.macro mov32 rd, val
	113	#if HAVE_ARMV6T2_EXTERNAL
	114	movw \rd, #(\val) & 0xffff
	115	.if (\val) >> 16
	116	movt \rd, #(\val) >> 16
	117	.endif
	118	#else
	119	ldr \rd, =\val
	120	#endif
	121	.endm
	122
	123	.macro put_pic num
	124	put_pic_\num
	125	.endm
	126
	127	.macro do_def_pic num, val, label
	128	.macro put_pic_\num
	129	.if \num
	130	.altmacro
	131	put_pic %(\num - 1)
	132	.noaltmacro
	133	.endif
	134	\label: .word \val
	135	.purgem put_pic_\num
	136	.endm
	137	.endm
	138
	139	.macro def_pic val, label
	140	.altmacro
	141	do_def_pic %.Lpic_idx, \val, \label
	142	.noaltmacro
	143	.set .Lpic_idx, .Lpic_idx + 1
	144	.endm
	145
	146	.macro ldpic rd, val, indir=0
	147	ldr \rd, .Lpicoff\@
	148	.Lpic\@:
	149	.if \indir
	150	A ldr \rd, [pc, \rd]
	151	T add \rd, pc
	152	T ldr \rd, [\rd]
	153	.else
	154	add \rd, pc
	155	.endif
	156	def_pic \val - (.Lpic\@ + (8 >> CONFIG_THUMB)), .Lpicoff\@
	157	.endm
	158
	159	.macro movrel rd, val
	160	#if CONFIG_PIC
	161	ldpic \rd, \val
	162	#elif HAVE_ARMV6T2_EXTERNAL && !defined(__APPLE__)
	163	movw \rd, #:lower16:\val
	164	movt \rd, #:upper16:\val
	165	#else
	166	ldr \rd, =\val
	167	#endif
	168	.endm
	169
	170	.macro movrelx rd, val, gp
	171	#if CONFIG_PIC && defined(__ELF__)
	172	.ifnb \gp
	173	.if .Lpic_gp
	174	.unreq gp
	175	.endif
	176	gp .req \gp
	177	ldpic gp, _GLOBAL_OFFSET_TABLE_
	178	.elseif !.Lpic_gp
	179	gp .req r12
	180	ldpic gp, _GLOBAL_OFFSET_TABLE_
	181	.endif
	182	.set .Lpic_gp, 1
	183	ldr \rd, .Lpicoff\@
	184	ldr \rd, [gp, \rd]
	185	def_pic \val(GOT), .Lpicoff\@
	186	#elif CONFIG_PIC && defined(__APPLE__)
	187	ldpic \rd, .Lpic\@, indir=1
	188	.non_lazy_symbol_pointer
	189	.Lpic\@:
	190	.indirect_symbol \val
	191	.word 0
	192	.text
	193	#else
	194	movrel \rd, \val
	195	#endif
	196	.endm
	197
	198	.macro add_sh rd, rn, rm, sh:vararg
	199	A add \rd, \rn, \rm, \sh
	200	T mov \rm, \rm, \sh
	201	T add \rd, \rn, \rm
	202	.endm
	203
	204	.macro ldr_pre rt, rn, rm:vararg
	205	A ldr \rt, [\rn, \rm]!
	206	T add \rn, \rn, \rm
	207	T ldr \rt, [\rn]
	208	.endm
	209
	210	.macro ldr_dpre rt, rn, rm:vararg
	211	A ldr \rt, [\rn, -\rm]!
	212	T sub \rn, \rn, \rm
	213	T ldr \rt, [\rn]
	214	.endm
	215
	216	.macro ldr_nreg rt, rn, rm:vararg
	217	A ldr \rt, [\rn, -\rm]
	218	T sub \rt, \rn, \rm
	219	T ldr \rt, [\rt]
	220	.endm
	221
	222	.macro ldr_post rt, rn, rm:vararg
	223	A ldr \rt, [\rn], \rm
	224	T ldr \rt, [\rn]
	225	T add \rn, \rn, \rm
	226	.endm
	227
	228	.macro ldrc_pre cc, rt, rn, rm:vararg
	229	A ldr\cc \rt, [\rn, \rm]!
	230	T itt \cc
	231	T add\cc \rn, \rn, \rm
	232	T ldr\cc \rt, [\rn]
	233	.endm
	234
	235	.macro ldrd_reg rt, rt2, rn, rm
	236	A ldrd \rt, \rt2, [\rn, \rm]
	237	T add \rt, \rn, \rm
	238	T ldrd \rt, \rt2, [\rt]
	239	.endm
	240
	241	.macro ldrd_post rt, rt2, rn, rm
	242	A ldrd \rt, \rt2, [\rn], \rm
	243	T ldrd \rt, \rt2, [\rn]
	244	T add \rn, \rn, \rm
	245	.endm
	246
	247	.macro ldrh_pre rt, rn, rm
	248	A ldrh \rt, [\rn, \rm]!
	249	T add \rn, \rn, \rm
	250	T ldrh \rt, [\rn]
	251	.endm
	252
	253	.macro ldrh_dpre rt, rn, rm
	254	A ldrh \rt, [\rn, -\rm]!
	255	T sub \rn, \rn, \rm
	256	T ldrh \rt, [\rn]
	257	.endm
	258
	259	.macro ldrh_post rt, rn, rm
	260	A ldrh \rt, [\rn], \rm
	261	T ldrh \rt, [\rn]
	262	T add \rn, \rn, \rm
	263	.endm
	264
	265	.macro ldrb_post rt, rn, rm
	266	A ldrb \rt, [\rn], \rm
	267	T ldrb \rt, [\rn]
	268	T add \rn, \rn, \rm
	269	.endm
	270
	271	.macro str_post rt, rn, rm:vararg
	272	A str \rt, [\rn], \rm
	273	T str \rt, [\rn]
	274	T add \rn, \rn, \rm
	275	.endm
	276
	277	.macro strb_post rt, rn, rm:vararg
	278	A strb \rt, [\rn], \rm
	279	T strb \rt, [\rn]
	280	T add \rn, \rn, \rm
	281	.endm
	282
	283	.macro strd_post rt, rt2, rn, rm
	284	A strd \rt, \rt2, [\rn], \rm
	285	T strd \rt, \rt2, [\rn]
	286	T add \rn, \rn, \rm
	287	.endm
	288
	289	.macro strh_pre rt, rn, rm
	290	A strh \rt, [\rn, \rm]!
	291	T add \rn, \rn, \rm
	292	T strh \rt, [\rn]
	293	.endm
	294
	295	.macro strh_dpre rt, rn, rm
	296	A strh \rt, [\rn, -\rm]!
	297	T sub \rn, \rn, \rm
	298	T strh \rt, [\rn]
	299	.endm
	300
	301	.macro strh_post rt, rn, rm
	302	A strh \rt, [\rn], \rm
	303	T strh \rt, [\rn]
	304	T add \rn, \rn, \rm
	305	.endm
	306
	307	.macro strh_dpost rt, rn, rm
	308	A strh \rt, [\rn], -\rm
	309	T strh \rt, [\rn]
	310	T sub \rn, \rn, \rm
	311	.endm
	312
	313	#if HAVE_VFP_ARGS
	314	ELF .eabi_attribute 28, 1
	315	# define VFP
	316	# define NOVFP @
	317	#else
	318	# define VFP @
	319	# define NOVFP
	320	#endif
	321
	322	#define GLUE(a, b) a ## b
	323	#define JOIN(a, b) GLUE(a, b)
	324	#define X(s) JOIN(EXTERN_ASM, s)

+29

-0

libde265/arm/cpudetect.S less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2015 struktur AG, Joachim Bauch <bauch@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "asm.S"
	21	#include "neon.S"
	22
	23	// we execute a simple NEON instruction and check if SIGILL is triggered to
	24	// detect if the CPU support NEON code
	25	function libde265_detect_neon, export=1
	26	vand q0, q0, q0
	27	bx lr
	28	endfunc

+1004

-0

libde265/arm/hevcdsp_qpel_neon.S less more

	0	/*
	1	* Copyright (c) 2014 - 2015 Seppo Tomperi <seppo.tomperi@vtt.fi>
	2	*
	3	* This file is part of FFmpeg.
	4	*
	5	* FFmpeg is free software; you can redistribute it and/or
	6	* modify it under the terms of the GNU Lesser General Public
	7	* License as published by the Free Software Foundation; either
	8	* version 2.1 of the License, or (at your option) any later version.
	9	*
	10	* FFmpeg is distributed in the hope that it will be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	13	* Lesser General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU Lesser General Public
	16	* License along with FFmpeg; if not, write to the Free Software
	17	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
	18	*/
	19
	20	/*
	21	* This is commit 63ca0fe8288dbd300c9bb814cb671e5d889f691c from
	22	* https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/arm/hevcdsp_qpel_neon.S
	23	*/
	24
	25	#include "asm.S"
	26	#include "neon.S"
	27
	28	#define MAX_PB_SIZE #64
	29
	30	.macro regshuffle_d8
	31	vmov d16, d17
	32	vmov d17, d18
	33	vmov d18, d19
	34	vmov d19, d20
	35	vmov d20, d21
	36	vmov d21, d22
	37	vmov d22, d23
	38	.endm
	39
	40	.macro regshuffle_q8
	41	vmov q0, q1
	42	vmov q1, q2
	43	vmov q2, q3
	44	vmov q3, q4
	45	vmov q4, q5
	46	vmov q5, q6
	47	vmov q6, q7
	48	.endm
	49
	50	.macro vextin8
	51	pld [r2]
	52	vld1.8 {q11}, [r2], r3
	53	vext.8 d16, d22, d23, #1
	54	vext.8 d17, d22, d23, #2
	55	vext.8 d18, d22, d23, #3
	56	vext.8 d19, d22, d23, #4
	57	vext.8 d20, d22, d23, #5
	58	vext.8 d21, d22, d23, #6
	59	vext.8 d22, d22, d23, #7
	60	.endm
	61
	62	.macro loadin8
	63	pld [r2]
	64	vld1.8 {d16}, [r2], r3
	65	pld [r2]
	66	vld1.8 {d17}, [r2], r3
	67	pld [r2]
	68	vld1.8 {d18}, [r2], r3
	69	pld [r2]
	70	vld1.8 {d19}, [r2], r3
	71	pld [r2]
	72	vld1.8 {d20}, [r2], r3
	73	pld [r2]
	74	vld1.8 {d21}, [r2], r3
	75	pld [r2]
	76	vld1.8 {d22}, [r2], r3
	77	pld [r2]
	78	vld1.8 {d23}, [r2], r3
	79	.endm
	80
	81	.macro qpel_filter_1_32b
	82	vmov.i16 d16, #58
	83	vmov.i16 d17, #10
	84	vmull.s16 q9, d6, d16 // 58 * d0
	85	vmull.s16 q10, d7, d16 // 58 * d1
	86	vmov.i16 d16, #17
	87	vmull.s16 q11, d4, d17 // 10 * c0
	88	vmull.s16 q12, d5, d17 // 10 * c1
	89	vmov.i16 d17, #5
	90	vmull.s16 q13, d8, d16 // 17 * e0
	91	vmull.s16 q14, d9, d16 // 17 * e1
	92	vmull.s16 q15, d10, d17 // 5 * f0
	93	vmull.s16 q8, d11, d17 // 5 * f1
	94	vsub.s32 q9, q11 // 58 * d0 - 10 * c0
	95	vsub.s32 q10, q12 // 58 * d1 - 10 * c1
	96	vshll.s16 q11, d2, #2 // 4 * b0
	97	vshll.s16 q12, d3, #2 // 4 * b1
	98	vadd.s32 q9, q13 // 58 * d0 - 10 * c0 + 17 * e0
	99	vadd.s32 q10, q14 // 58 * d1 - 10 * c1 + 17 * e1
	100	vsubl.s16 q13, d12, d0 // g0 - a0
	101	vsubl.s16 q14, d13, d1 // g1 - a1
	102	vadd.s32 q9, q11 // 58 * d0 - 10 * c0 + 17 * e0 + 4 * b0
	103	vadd.s32 q10, q12 // 58 * d1 - 10 * c1 + 17 * e1 + 4 * b1
	104	vsub.s32 q13, q15 // g0 - a0 - 5 * f0
	105	vsub.s32 q14, q8 // g1 - a1 - 5 * f1
	106	vadd.s32 q9, q13 // 58 * d0 - 10 * c0 + 17 * e0 + 4 * b0 + g0 - a0 - 5 * f0
	107	vadd.s32 q10, q14 // 58 * d1 - 10 * c1 + 17 * e1 + 4 * b1 + g1 - a1 - 5 * f1
	108	vqshrn.s32 d16, q9, #6
	109	vqshrn.s32 d17, q10, #6
	110	.endm
	111
	112	// input q0 - q7
	113	// output q8
	114	.macro qpel_filter_2_32b
	115	vmov.i32 q8, #11
	116	vaddl.s16 q9, d6, d8 // d0 + e0
	117	vaddl.s16 q10, d7, d9 // d1 + e1
	118	vaddl.s16 q11, d4, d10 // c0 + f0
	119	vaddl.s16 q12, d5, d11 // c1 + f1
	120	vmul.s32 q11, q8 // 11 * (c0 + f0)
	121	vmul.s32 q12, q8 // 11 * (c1 + f1)
	122	vmov.i32 q8, #40
	123	vaddl.s16 q15, d2, d12 // b0 + g0
	124	vmul.s32 q9, q8 // 40 * (d0 + e0)
	125	vmul.s32 q10, q8 // 40 * (d1 + e1)
	126	vaddl.s16 q8, d3, d13 // b1 + g1
	127	vaddl.s16 q13, d0, d14 // a0 + h0
	128	vaddl.s16 q14, d1, d15 // a1 + h1
	129	vshl.s32 q15, #2 // 4*(b0+g0)
	130	vshl.s32 q8, #2 // 4*(b1+g1)
	131	vadd.s32 q11, q13 // 11 * (c0 + f0) + a0 + h0
	132	vadd.s32 q12, q14 // 11 * (c1 + f1) + a1 + h1
	133	vadd.s32 q9, q15 // 40 * (d0 + e0) + 4*(b0+g0)
	134	vadd.s32 q10, q8 // 40 * (d1 + e1) + 4*(b1+g1)
	135	vsub.s32 q9, q11 // 40 * (d0 + e0) + 4(b0+g0) - (11 (c0 + f0) + a0 + h0)
	136	vsub.s32 q10, q12 // 40 * (d1 + e1) + 4(b1+g1) - (11 (c1 + f1) + a1 + h1)
	137	vqshrn.s32 d16, q9, #6
	138	vqshrn.s32 d17, q10, #6
	139	.endm
	140
	141	.macro qpel_filter_3_32b
	142	vmov.i16 d16, #58
	143	vmov.i16 d17, #10
	144	vmull.s16 q9, d8, d16 // 58 * d0
	145	vmull.s16 q10, d9, d16 // 58 * d1
	146	vmov.i16 d16, #17
	147	vmull.s16 q11, d10, d17 // 10 * c0
	148	vmull.s16 q12, d11, d17 // 10 * c1
	149	vmov.i16 d17, #5
	150	vmull.s16 q13, d6, d16 // 17 * e0
	151	vmull.s16 q14, d7, d16 // 17 * e1
	152	vmull.s16 q15, d4, d17 // 5 * f0
	153	vmull.s16 q8, d5, d17 // 5 * f1
	154	vsub.s32 q9, q11 // 58 * d0 - 10 * c0
	155	vsub.s32 q10, q12 // 58 * d1 - 10 * c1
	156	vshll.s16 q11, d12, #2 // 4 * b0
	157	vshll.s16 q12, d13, #2 // 4 * b1
	158	vadd.s32 q9, q13 // 58 * d0 - 10 * c0 + 17 * e0
	159	vadd.s32 q10, q14 // 58 * d1 - 10 * c1 + 17 * e1
	160	vsubl.s16 q13, d2, d14 // g0 - a0
	161	vsubl.s16 q14, d3, d15 // g1 - a1
	162	vadd.s32 q9, q11 // 58 * d0 - 10 * c0 + 17 * e0 + 4 * b0
	163	vadd.s32 q10, q12 // 58 * d1 - 10 * c1 + 17 * e1 + 4 * b1
	164	vsub.s32 q13, q15 // g0 - a0 - 5 * f0
	165	vsub.s32 q14, q8 // g1 - a1 - 5 * f1
	166	vadd.s32 q9, q13 // 58 * d0 - 10 * c0 + 17 * e0 + 4 * b0 + g0 - a0 - 5 * f0
	167	vadd.s32 q10, q14 // 58 * d1 - 10 * c1 + 17 * e1 + 4 * b1 + g1 - a1 - 5 * f1
	168	vqshrn.s32 d16, q9, #6
	169	vqshrn.s32 d17, q10, #6
	170	.endm
	171
	172	.macro qpel_filter_1 out=q7
	173	vmov.u8 d24, #58
	174	vmov.u8 d25, #10
	175	vshll.u8 q13, d20, #4 // 16*e
	176	vshll.u8 q14, d21, #2 // 4*f
	177	vmull.u8 \out, d19, d24 // 58*d
	178	vaddw.u8 q13, q13, d20 // 17*e
	179	vmull.u8 q15, d18, d25 // 10*c
	180	vaddw.u8 q14, q14, d21 // 5*f
	181	vsubl.u8 q12, d22, d16 // g - a
	182	vadd.u16 \out, q13 // 58d + 17e
	183	vshll.u8 q13, d17, #2 // 4*b
	184	vadd.u16 q15, q14 // 10c + 5f
	185	vadd.s16 q13, q12 // - a + 4*b + g
	186	vsub.s16 \out, q15 // -10c + 58d + 17e -5f
	187	vadd.s16 \out, q13 // -a + 4b -10c + 58d + 17e -5*f
	188	.endm
	189
	190	.macro qpel_filter_2 out=q7
	191	vmov.i16 q12, #10
	192	vmov.i16 q14, #11
	193	vaddl.u8 q13, d19, d20 // d + e
	194	vaddl.u8 q15, d18, d21 // c + f
	195	vmul.u16 q13, q12 // 10 * (d+e)
	196	vmul.u16 q15, q14 // 11 * ( c + f)
	197	vaddl.u8 \out, d17, d22 // b + g
	198	vaddl.u8 q12, d16, d23 // a + h
	199	vadd.u16 \out, q13 // b + 10 * (d + e) + g
	200	vadd.s16 q12, q15
	201	vshl.u16 \out, #2 // 4 * (b + 10 * (d + e) + g)
	202	vsub.s16 \out, q12
	203	.endm
	204
	205	.macro qpel_filter_3 out=q7
	206	vmov.u8 d24, #58
	207	vmov.u8 d25, #10
	208	vshll.u8 q13, d19, #4 // 16*e
	209	vshll.u8 q14, d18, #2 // 4*f
	210	vmull.u8 \out, d20, d24 // 58*d
	211	vaddw.u8 q13, q13, d19 // 17*e
	212	vmull.u8 q15, d21, d25 // 10*c
	213	vaddw.u8 q14, q14, d18 // 5*f
	214	vsubl.u8 q12, d17, d23 // g - a
	215	vadd.u16 \out, q13 // 58d + 17e
	216	vshll.u8 q13, d22, #2 // 4*b
	217	vadd.u16 q15, q14 // 10c + 5f
	218	vadd.s16 q13, q12 // - a + 4*b + g
	219	vsub.s16 \out, q15 // -10c + 58d + 17e -5f
	220	vadd.s16 \out, q13 // -a + 4b -10c + 58d + 17e -5*f
	221	.endm
	222
	223	.macro hevc_put_qpel_vX_neon_8 filter
	224	push {r4, r5, r6, r7}
	225	ldr r4, [sp, #16] // height
	226	ldr r5, [sp, #20] // width
	227	vpush {d8-d15}
	228	sub r2, r2, r3, lsl #1
	229	sub r2, r3
	230	mov r12, r4
	231	mov r6, r0
	232	mov r7, r2
	233	lsl r1, #1
	234	0: loadin8
	235	cmp r5, #4
	236	beq 4f
	237	8: subs r4, #1
	238	\filter
	239	vst1.16 {q7}, [r0], r1
	240	regshuffle_d8
	241	vld1.8 {d23}, [r2], r3
	242	bne 8b
	243	subs r5, #8
	244	beq 99f
	245	mov r4, r12
	246	add r6, #16
	247	mov r0, r6
	248	add r7, #8
	249	mov r2, r7
	250	b 0b
	251	4: subs r4, #1
	252	\filter
	253	vst1.16 d14, [r0], r1
	254	regshuffle_d8
	255	vld1.32 {d23[0]}, [r2], r3
	256	bne 4b
	257	99: vpop {d8-d15}
	258	pop {r4, r5, r6, r7}
	259	bx lr
	260	.endm
	261
	262	.macro hevc_put_qpel_uw_vX_neon_8 filter
	263	push {r4-r10}
	264	ldr r5, [sp, #28] // width
	265	ldr r4, [sp, #32] // height
	266	ldr r8, [sp, #36] // src2
	267	ldr r9, [sp, #40] // src2stride
	268	vpush {d8-d15}
	269	sub r2, r2, r3, lsl #1
	270	sub r2, r3
	271	mov r12, r4
	272	mov r6, r0
	273	mov r7, r2
	274	cmp r8, #0
	275	bne .Lbi\@
	276	0: loadin8
	277	cmp r5, #4
	278	beq 4f
	279	8: subs r4, #1
	280	\filter
	281	vqrshrun.s16 d0, q7, #6
	282	vst1.8 d0, [r0], r1
	283	regshuffle_d8
	284	vld1.8 {d23}, [r2], r3
	285	bne 8b
	286	subs r5, #8
	287	beq 99f
	288	mov r4, r12
	289	add r6, #8
	290	mov r0, r6
	291	add r7, #8
	292	mov r2, r7
	293	b 0b
	294	4: subs r4, #1
	295	\filter
	296	vqrshrun.s16 d0, q7, #6
	297	vst1.32 d0[0], [r0], r1
	298	regshuffle_d8
	299	vld1.32 {d23[0]}, [r2], r3
	300	bne 4b
	301	b 99f
	302	.Lbi\@: lsl r9, #1
	303	mov r10, r8
	304	0: loadin8
	305	cmp r5, #4
	306	beq 4f
	307	8: subs r4, #1
	308	\filter
	309	vld1.16 {q0}, [r8], r9
	310	vqadd.s16 q0, q7
	311	vqrshrun.s16 d0, q0, #7
	312	vst1.8 d0, [r0], r1
	313	regshuffle_d8
	314	vld1.8 {d23}, [r2], r3
	315	bne 8b
	316	subs r5, #8
	317	beq 99f
	318	mov r4, r12
	319	add r6, #8
	320	mov r0, r6
	321	add r10, #16
	322	mov r8, r10
	323	add r7, #8
	324	mov r2, r7
	325	b 0b
	326	4: subs r4, #1
	327	\filter
	328	vld1.16 d0, [r8], r9
	329	vqadd.s16 d0, d14
	330	vqrshrun.s16 d0, q0, #7
	331	vst1.32 d0[0], [r0], r1
	332	regshuffle_d8
	333	vld1.32 {d23[0]}, [r2], r3
	334	bne 4b
	335	99: vpop {d8-d15}
	336	pop {r4-r10}
	337	bx lr
	338	.endm
	339
	340	function ff_hevc_put_qpel_v1_neon_8, export=1
	341	hevc_put_qpel_vX_neon_8 qpel_filter_1
	342	endfunc
	343
	344	function ff_hevc_put_qpel_v2_neon_8, export=1
	345	hevc_put_qpel_vX_neon_8 qpel_filter_2
	346	endfunc
	347
	348	function ff_hevc_put_qpel_v3_neon_8, export=1
	349	hevc_put_qpel_vX_neon_8 qpel_filter_3
	350	endfunc
	351
	352
	353	function ff_hevc_put_qpel_uw_v1_neon_8, export=1
	354	hevc_put_qpel_uw_vX_neon_8 qpel_filter_1
	355	endfunc
	356
	357	function ff_hevc_put_qpel_uw_v2_neon_8, export=1
	358	hevc_put_qpel_uw_vX_neon_8 qpel_filter_2
	359	endfunc
	360
	361	function ff_hevc_put_qpel_uw_v3_neon_8, export=1
	362	hevc_put_qpel_uw_vX_neon_8 qpel_filter_3
	363	endfunc
	364
	365	.macro hevc_put_qpel_hX_neon_8 filter
	366	push {r4, r5, r6, r7}
	367	ldr r4, [sp, #16] // height
	368	ldr r5, [sp, #20] // width
	369
	370	vpush {d8-d15}
	371	sub r2, #4
	372	lsl r1, #1
	373	mov r12, r4
	374	mov r6, r0
	375	mov r7, r2
	376	cmp r5, #4
	377	beq 4f
	378	8: subs r4, #1
	379	vextin8
	380	\filter
	381	vst1.16 {q7}, [r0], r1
	382	bne 8b
	383	subs r5, #8
	384	beq 99f
	385	mov r4, r12
	386	add r6, #16
	387	mov r0, r6
	388	add r7, #8
	389	mov r2, r7
	390	cmp r5, #4
	391	bne 8b
	392	4: subs r4, #1
	393	vextin8
	394	\filter
	395	vst1.16 d14, [r0], r1
	396	bne 4b
	397	99: vpop {d8-d15}
	398	pop {r4, r5, r6, r7}
	399	bx lr
	400	.endm
	401
	402	.macro hevc_put_qpel_uw_hX_neon_8 filter
	403	push {r4-r10}
	404	ldr r5, [sp, #28] // width
	405	ldr r4, [sp, #32] // height
	406	ldr r8, [sp, #36] // src2
	407	ldr r9, [sp, #40] // src2stride
	408	vpush {d8-d15}
	409	sub r2, #4
	410	mov r12, r4
	411	mov r6, r0
	412	mov r7, r2
	413	cmp r8, #0
	414	bne .Lbi\@
	415	cmp r5, #4
	416	beq 4f
	417	8: subs r4, #1
	418	vextin8
	419	\filter
	420	vqrshrun.s16 d0, q7, #6
	421	vst1.8 d0, [r0], r1
	422	bne 8b
	423	subs r5, #8
	424	beq 99f
	425	mov r4, r12
	426	add r6, #8
	427	mov r0, r6
	428	add r7, #8
	429	mov r2, r7
	430	cmp r5, #4
	431	bne 8b
	432	4: subs r4, #1
	433	vextin8
	434	\filter
	435	vqrshrun.s16 d0, q7, #6
	436	vst1.32 d0[0], [r0], r1
	437	bne 4b
	438	b 99f
	439	.Lbi\@:
	440	lsl r9, #1
	441	cmp r5, #4
	442	beq 4f
	443	mov r10, r8
	444	8: subs r4, #1
	445	vextin8
	446	\filter
	447	vld1.16 {q0}, [r8], r9
	448	vqadd.s16 q0, q7
	449	vqrshrun.s16 d0, q0, #7
	450	vst1.8 d0, [r0], r1
	451	bne 8b
	452	subs r5, #8
	453	beq 99f
	454	mov r4, r12
	455	add r6, #8
	456	add r10, #16
	457	mov r8, r10
	458	mov r0, r6
	459	add r7, #8
	460	mov r2, r7
	461	cmp r5, #4
	462	bne 8b
	463	4: subs r4, #1
	464	vextin8
	465	\filter
	466	vld1.16 d0, [r8], r9
	467	vqadd.s16 d0, d14
	468	vqrshrun.s16 d0, q0, #7
	469	vst1.32 d0[0], [r0], r1
	470	bne 4b
	471	99: vpop {d8-d15}
	472	pop {r4-r10}
	473	bx lr
	474	.endm
	475
	476	function ff_hevc_put_qpel_h1_neon_8, export=1
	477	hevc_put_qpel_hX_neon_8 qpel_filter_1
	478	endfunc
	479
	480	function ff_hevc_put_qpel_h2_neon_8, export=1
	481	hevc_put_qpel_hX_neon_8 qpel_filter_2
	482	endfunc
	483
	484	function ff_hevc_put_qpel_h3_neon_8, export=1
	485	hevc_put_qpel_hX_neon_8 qpel_filter_3
	486	endfunc
	487
	488
	489	function ff_hevc_put_qpel_uw_h1_neon_8, export=1
	490	hevc_put_qpel_uw_hX_neon_8 qpel_filter_1
	491	endfunc
	492
	493	function ff_hevc_put_qpel_uw_h2_neon_8, export=1
	494	hevc_put_qpel_uw_hX_neon_8 qpel_filter_2
	495	endfunc
	496
	497	function ff_hevc_put_qpel_uw_h3_neon_8, export=1
	498	hevc_put_qpel_uw_hX_neon_8 qpel_filter_3
	499	endfunc
	500
	501	.macro hevc_put_qpel_hXvY_neon_8 filterh filterv
	502	push {r4, r5, r6, r7}
	503	ldr r4, [sp, #16] // height
	504	ldr r5, [sp, #20] // width
	505
	506	vpush {d8-d15}
	507	sub r2, #4
	508	sub r2, r2, r3, lsl #1
	509	sub r2, r3 // extra_before 3
	510	lsl r1, #1
	511	mov r12, r4
	512	mov r6, r0
	513	mov r7, r2
	514	0: vextin8
	515	\filterh q0
	516	vextin8
	517	\filterh q1
	518	vextin8
	519	\filterh q2
	520	vextin8
	521	\filterh q3
	522	vextin8
	523	\filterh q4
	524	vextin8
	525	\filterh q5
	526	vextin8
	527	\filterh q6
	528	vextin8
	529	\filterh q7
	530	cmp r5, #4
	531	beq 4f
	532	8: subs r4, #1
	533	\filterv
	534	vst1.16 {q8}, [r0], r1
	535	regshuffle_q8
	536	vextin8
	537	\filterh q7
	538	bne 8b
	539	subs r5, #8
	540	beq 99f
	541	mov r4, r12
	542	add r6, #16
	543	mov r0, r6
	544	add r7, #8
	545	mov r2, r7
	546	b 0b
	547	4: subs r4, #1
	548	\filterv
	549	vst1.16 d16, [r0], r1
	550	regshuffle_q8
	551	vextin8
	552	\filterh q7
	553	bne 4b
	554	99: vpop {d8-d15}
	555	pop {r4, r5, r6, r7}
	556	bx lr
	557	.endm
	558
	559	.macro hevc_put_qpel_uw_hXvY_neon_8 filterh filterv
	560	push {r4-r10}
	561	ldr r5, [sp, #28] // width
	562	ldr r4, [sp, #32] // height
	563	ldr r8, [sp, #36] // src2
	564	ldr r9, [sp, #40] // src2stride
	565	vpush {d8-d15}
	566	sub r2, #4
	567	sub r2, r2, r3, lsl #1
	568	sub r2, r3 // extra_before 3
	569	mov r12, r4
	570	mov r6, r0
	571	mov r7, r2
	572	cmp r8, #0
	573	bne .Lbi\@
	574	0: vextin8
	575	\filterh q0
	576	vextin8
	577	\filterh q1
	578	vextin8
	579	\filterh q2
	580	vextin8
	581	\filterh q3
	582	vextin8
	583	\filterh q4
	584	vextin8
	585	\filterh q5
	586	vextin8
	587	\filterh q6
	588	vextin8
	589	\filterh q7
	590	cmp r5, #4
	591	beq 4f
	592	8: subs r4, #1
	593	\filterv
	594	vqrshrun.s16 d0, q8, #6
	595	vst1.8 d0, [r0], r1
	596	regshuffle_q8
	597	vextin8
	598	\filterh q7
	599	bne 8b
	600	subs r5, #8
	601	beq 99f
	602	mov r4, r12
	603	add r6, #8
	604	mov r0, r6
	605	add r7, #8
	606	mov r2, r7
	607	b 0b
	608	4: subs r4, #1
	609	\filterv
	610	vqrshrun.s16 d0, q8, #6
	611	vst1.32 d0[0], [r0], r1
	612	regshuffle_q8
	613	vextin8
	614	\filterh q7
	615	bne 4b
	616	b 99f
	617	.Lbi\@: lsl r9, #1
	618	mov r10, r8
	619	0: vextin8
	620	\filterh q0
	621	vextin8
	622	\filterh q1
	623	vextin8
	624	\filterh q2
	625	vextin8
	626	\filterh q3
	627	vextin8
	628	\filterh q4
	629	vextin8
	630	\filterh q5
	631	vextin8
	632	\filterh q6
	633	vextin8
	634	\filterh q7
	635	cmp r5, #4
	636	beq 4f
	637	8: subs r4, #1
	638	\filterv
	639	vld1.16 {q0}, [r8], r9
	640	vqadd.s16 q0, q8
	641	vqrshrun.s16 d0, q0, #7
	642	vst1.8 d0, [r0], r1
	643	regshuffle_q8
	644	vextin8
	645	\filterh q7
	646	bne 8b
	647	subs r5, #8
	648	beq 99f
	649	mov r4, r12
	650	add r6, #8
	651	mov r0, r6
	652	add r10, #16
	653	mov r8, r10
	654	add r7, #8
	655	mov r2, r7
	656	b 0b
	657	4: subs r4, #1
	658	\filterv
	659	vld1.16 d0, [r8], r9
	660	vqadd.s16 d0, d16
	661	vqrshrun.s16 d0, q0, #7
	662	vst1.32 d0[0], [r0], r1
	663	regshuffle_q8
	664	vextin8
	665	\filterh q7
	666	bne 4b
	667	99: vpop {d8-d15}
	668	pop {r4-r10}
	669	bx lr
	670	.endm
	671
	672
	673	function ff_hevc_put_qpel_h1v1_neon_8, export=1
	674	hevc_put_qpel_hXvY_neon_8 qpel_filter_1 qpel_filter_1_32b
	675	endfunc
	676
	677	function ff_hevc_put_qpel_h2v1_neon_8, export=1
	678	hevc_put_qpel_hXvY_neon_8 qpel_filter_2 qpel_filter_1_32b
	679	endfunc
	680
	681	function ff_hevc_put_qpel_h3v1_neon_8, export=1
	682	hevc_put_qpel_hXvY_neon_8 qpel_filter_3 qpel_filter_1_32b
	683	endfunc
	684
	685	function ff_hevc_put_qpel_h1v2_neon_8, export=1
	686	hevc_put_qpel_hXvY_neon_8 qpel_filter_1 qpel_filter_2_32b
	687	endfunc
	688
	689	function ff_hevc_put_qpel_h2v2_neon_8, export=1
	690	hevc_put_qpel_hXvY_neon_8 qpel_filter_2 qpel_filter_2_32b
	691	endfunc
	692
	693	function ff_hevc_put_qpel_h3v2_neon_8, export=1
	694	hevc_put_qpel_hXvY_neon_8 qpel_filter_3 qpel_filter_2_32b
	695	endfunc
	696
	697	function ff_hevc_put_qpel_h1v3_neon_8, export=1
	698	hevc_put_qpel_hXvY_neon_8 qpel_filter_1 qpel_filter_3_32b
	699	endfunc
	700
	701	function ff_hevc_put_qpel_h2v3_neon_8, export=1
	702	hevc_put_qpel_hXvY_neon_8 qpel_filter_2 qpel_filter_3_32b
	703	endfunc
	704
	705	function ff_hevc_put_qpel_h3v3_neon_8, export=1
	706	hevc_put_qpel_hXvY_neon_8 qpel_filter_3 qpel_filter_3_32b
	707	endfunc
	708
	709
	710	function ff_hevc_put_qpel_uw_h1v1_neon_8, export=1
	711	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_1 qpel_filter_1_32b
	712	endfunc
	713
	714	function ff_hevc_put_qpel_uw_h2v1_neon_8, export=1
	715	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_2 qpel_filter_1_32b
	716	endfunc
	717
	718	function ff_hevc_put_qpel_uw_h3v1_neon_8, export=1
	719	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_3 qpel_filter_1_32b
	720	endfunc
	721
	722	function ff_hevc_put_qpel_uw_h1v2_neon_8, export=1
	723	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_1 qpel_filter_2_32b
	724	endfunc
	725
	726	function ff_hevc_put_qpel_uw_h2v2_neon_8, export=1
	727	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_2 qpel_filter_2_32b
	728	endfunc
	729
	730	function ff_hevc_put_qpel_uw_h3v2_neon_8, export=1
	731	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_3 qpel_filter_2_32b
	732	endfunc
	733
	734	function ff_hevc_put_qpel_uw_h1v3_neon_8, export=1
	735	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_1 qpel_filter_3_32b
	736	endfunc
	737
	738	function ff_hevc_put_qpel_uw_h2v3_neon_8, export=1
	739	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_2 qpel_filter_3_32b
	740	endfunc
	741
	742	function ff_hevc_put_qpel_uw_h3v3_neon_8, export=1
	743	hevc_put_qpel_uw_hXvY_neon_8 qpel_filter_3 qpel_filter_3_32b
	744	endfunc
	745
	746	.macro init_put_pixels
	747	pld [r1]
	748	pld [r1, r2]
	749	mov r12, MAX_PB_SIZE
	750	lsl r12, #1
	751	.endm
	752
	753	function ff_hevc_put_pixels_w2_neon_8, export=1
	754	init_put_pixels
	755	vmov.u8 d5, #255
	756	vshr.u64 d5, #32
	757	0: subs r3, #1
	758	vld1.32 {d0[0]}, [r1], r2
	759	pld [r1]
	760	vld1.32 d6, [r0]
	761	vshll.u8 q0, d0, #6
	762	vbit d6, d0, d5
	763	vst1.32 d6, [r0], r12
	764	bne 0b
	765	bx lr
	766	endfunc
	767
	768	function ff_hevc_put_pixels_w4_neon_8, export=1
	769	init_put_pixels
	770	0: subs r3, #2
	771	vld1.32 {d0[0]}, [r1], r2
	772	vld1.32 {d0[1]}, [r1], r2
	773	pld [r1]
	774	pld [r1, r2]
	775	vshll.u8 q0, d0, #6
	776	vst1.64 {d0}, [r0], r12
	777	vst1.64 {d1}, [r0], r12
	778	bne 0b
	779	bx lr
	780	endfunc
	781
	782	function ff_hevc_put_pixels_w6_neon_8, export=1
	783	init_put_pixels
	784	vmov.u8 q10, #255
	785	vshr.u64 d21, #32
	786	0: subs r3, #1
	787	vld1.16 {d0}, [r1], r2
	788	pld [r1]
	789	vshll.u8 q0, d0, #6
	790	vld1.8 {q12}, [r0]
	791	vbit q12, q0, q10
	792	vst1.8 {q12}, [r0], r12
	793	bne 0b
	794	bx lr
	795	endfunc
	796
	797	function ff_hevc_put_pixels_w8_neon_8, export=1
	798	init_put_pixels
	799	0: subs r3, #2
	800	vld1.8 {d0}, [r1], r2
	801	vld1.8 {d2}, [r1], r2
	802	pld [r1]
	803	pld [r1, r2]
	804	vshll.u8 q0, d0, #6
	805	vshll.u8 q1, d2, #6
	806	vst1.16 {q0}, [r0], r12
	807	vst1.16 {q1}, [r0], r12
	808	bne 0b
	809	bx lr
	810	endfunc
	811
	812	function ff_hevc_put_pixels_w12_neon_8, export=1
	813	init_put_pixels
	814	0: subs r3, #2
	815	vld1.64 {d0}, [r1]
	816	add r1, #8
	817	vld1.32 {d1[0]}, [r1], r2
	818	sub r1, #8
	819	vld1.64 {d2}, [r1]
	820	add r1, #8
	821	vld1.32 {d1[1]}, [r1], r2
	822	sub r1, #8
	823	pld [r1]
	824	pld [r1, r2]
	825	vshll.u8 q8, d0, #6
	826	vshll.u8 q9, d1, #6
	827	vshll.u8 q10, d2, #6
	828	vmov d22, d19
	829	vst1.64 {d16, d17, d18}, [r0], r12
	830	vst1.64 {d20, d21, d22}, [r0], r12
	831	bne 0b
	832	bx lr
	833	endfunc
	834
	835	function ff_hevc_put_pixels_w16_neon_8, export=1
	836	init_put_pixels
	837	0: subs r3, #2
	838	vld1.8 {q0}, [r1], r2
	839	vld1.8 {q1}, [r1], r2
	840	pld [r1]
	841	pld [r1, r2]
	842	vshll.u8 q8, d0, #6
	843	vshll.u8 q9, d1, #6
	844	vshll.u8 q10, d2, #6
	845	vshll.u8 q11, d3, #6
	846	vst1.8 {q8, q9}, [r0], r12
	847	vst1.8 {q10, q11}, [r0], r12
	848	bne 0b
	849	bx lr
	850	endfunc
	851
	852	function ff_hevc_put_pixels_w24_neon_8, export=1
	853	init_put_pixels
	854	0: subs r3, #1
	855	vld1.8 {d0, d1, d2}, [r1], r2
	856	pld [r1]
	857	vshll.u8 q10, d0, #6
	858	vshll.u8 q11, d1, #6
	859	vshll.u8 q12, d2, #6
	860	vstm r0, {q10, q11, q12}
	861	add r0, r12
	862	bne 0b
	863	bx lr
	864	endfunc
	865
	866	function ff_hevc_put_pixels_w32_neon_8, export=1
	867	init_put_pixels
	868	0: subs r3, #1
	869	vld1.8 {q0, q1}, [r1], r2
	870	pld [r1]
	871	vshll.u8 q8, d0, #6
	872	vshll.u8 q9, d1, #6
	873	vshll.u8 q10, d2, #6
	874	vshll.u8 q11, d3, #6
	875	vstm r0, {q8, q9, q10, q11}
	876	add r0, r12
	877	bne 0b
	878	bx lr
	879	endfunc
	880
	881	function ff_hevc_put_pixels_w48_neon_8, export=1
	882	init_put_pixels
	883	0: subs r3, #1
	884	vld1.8 {q0, q1}, [r1]
	885	add r1, #32
	886	vld1.8 {q2}, [r1], r2
	887	sub r1, #32
	888	pld [r1]
	889	vshll.u8 q8, d0, #6
	890	vshll.u8 q9, d1, #6
	891	vshll.u8 q10, d2, #6
	892	vshll.u8 q11, d3, #6
	893	vshll.u8 q12, d4, #6
	894	vshll.u8 q13, d5, #6
	895	vstm r0, {q8, q9, q10, q11, q12, q13}
	896	add r0, r12
	897	bne 0b
	898	bx lr
	899	endfunc
	900
	901	function ff_hevc_put_pixels_w64_neon_8, export=1
	902	init_put_pixels
	903	0: subs r3, #1
	904	vld1.8 {q0, q1}, [r1]
	905	add r1, #32
	906	vld1.8 {q2, q3}, [r1], r2
	907	sub r1, #32
	908	pld [r1]
	909	vshll.u8 q8, d0, #6
	910	vshll.u8 q9, d1, #6
	911	vshll.u8 q10, d2, #6
	912	vshll.u8 q11, d3, #6
	913	vshll.u8 q12, d4, #6
	914	vshll.u8 q13, d5, #6
	915	vshll.u8 q14, d6, #6
	916	vshll.u8 q15, d7, #6
	917	vstm r0, {q8, q9, q10, q11, q12, q13, q14, q15}
	918	add r0, r12
	919	bne 0b
	920	bx lr
	921	endfunc
	922
	923	function ff_hevc_put_qpel_uw_pixels_neon_8, export=1
	924	push {r4-r9}
	925	ldr r5, [sp, #24] // width
	926	ldr r4, [sp, #28] // height
	927	ldr r8, [sp, #32] // src2
	928	ldr r9, [sp, #36] // src2stride
	929	vpush {d8-d15}
	930	cmp r8, #0
	931	bne 2f
	932	1: subs r4, #1
	933	vld1.8 {d0}, [r2], r3
	934	vst1.8 d0, [r0], r1
	935	bne 1b
	936	vpop {d8-d15}
	937	pop {r4-r9}
	938	bx lr
	939	2: subs r4, #1
	940	vld1.8 {d0}, [r2], r3
	941	vld1.16 {q1}, [r8], r9
	942	vshll.u8 q0, d0, #6
	943	vqadd.s16 q0, q1
	944	vqrshrun.s16 d0, q0, #7
	945	vst1.8 d0, [r0], r1
	946	bne 2b
	947	vpop {d8-d15}
	948	pop {r4-r9}
	949	bx lr
	950	endfunc
	951
	952	.macro put_qpel_uw_pixels width, regs, regs2, regs3, regs4
	953	function ff_hevc_put_qpel_uw_pixels_w\width\()_neon_8, export=1
	954	ldr r12, [sp] // height
	955	1: subs r12, #4
	956	vld1.32 {\regs} , [r2], r3
	957	vld1.32 {\regs2} , [r2], r3
	958	vld1.32 {\regs3} , [r2], r3
	959	vld1.32 {\regs4} , [r2], r3
	960	vst1.32 {\regs} , [r0], r1
	961	vst1.32 {\regs2} , [r0], r1
	962	vst1.32 {\regs3} , [r0], r1
	963	vst1.32 {\regs4} , [r0], r1
	964	bne 1b
	965	bx lr
	966	endfunc
	967	.endm
	968
	969	.macro put_qpel_uw_pixels_m width, regs, regs2, regs3, regs4
	970	function ff_hevc_put_qpel_uw_pixels_w\width\()_neon_8, export=1
	971	push {r4-r5}
	972	ldr r12, [sp, #8] // height
	973	1: subs r12, #2
	974	mov r4, r2
	975	vld1.32 {\regs} , [r2]!
	976	vld1.32 {\regs2} , [r2]
	977	add r2, r4, r3
	978	mov r4, r2
	979	vld1.32 {\regs3} , [r2]!
	980	vld1.32 {\regs4} , [r2]
	981	add r2, r4, r3
	982	mov r5, r0
	983	vst1.32 {\regs} , [r0]!
	984	vst1.32 {\regs2} , [r0]
	985	add r0, r5, r1
	986	mov r5, r0
	987	vst1.32 {\regs3} , [r0]!
	988	vst1.32 {\regs4} , [r0]
	989	add r0, r5, r1
	990	bne 1b
	991	pop {r4-r5}
	992	bx lr
	993	endfunc
	994	.endm
	995
	996	put_qpel_uw_pixels 4, d0[0], d0[1], d1[0], d1[1]
	997	put_qpel_uw_pixels 8, d0, d1, d2, d3
	998	put_qpel_uw_pixels_m 12, d0, d1[0], d2, d3[0]
	999	put_qpel_uw_pixels 16, q0, q1, q2, q3
	1000	put_qpel_uw_pixels 24, d0-d2, d3-d5, d16-d18, d19-d21
	1001	put_qpel_uw_pixels 32, q0-q1, q2-q3, q8-q9, q10-q11
	1002	put_qpel_uw_pixels_m 48, q0-q1, q2, q8-q9, q10
	1003	put_qpel_uw_pixels_m 64, q0-q1, q2-q3, q8-q9, q10-q11

+59

-0

libde265/arm/neon.S less more

	0	/*
	1	* Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
	2	*
	3	* This file is part of FFmpeg.
	4	*
	5	* FFmpeg is free software; you can redistribute it and/or
	6	* modify it under the terms of the GNU Lesser General Public
	7	* License as published by the Free Software Foundation; either
	8	* version 2.1 of the License, or (at your option) any later version.
	9	*
	10	* FFmpeg is distributed in the hope that it will be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	13	* Lesser General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU Lesser General Public
	16	* License along with FFmpeg; if not, write to the Free Software
	17	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
	18	*/
	19
	20	.macro transpose_8x8 r0, r1, r2, r3, r4, r5, r6, r7
	21	vtrn.32 \r0, \r4
	22	vtrn.32 \r1, \r5
	23	vtrn.32 \r2, \r6
	24	vtrn.32 \r3, \r7
	25	vtrn.16 \r0, \r2
	26	vtrn.16 \r1, \r3
	27	vtrn.16 \r4, \r6
	28	vtrn.16 \r5, \r7
	29	vtrn.8 \r0, \r1
	30	vtrn.8 \r2, \r3
	31	vtrn.8 \r4, \r5
	32	vtrn.8 \r6, \r7
	33	.endm
	34
	35	.macro transpose_4x4 r0, r1, r2, r3
	36	vtrn.16 \r0, \r2
	37	vtrn.16 \r1, \r3
	38	vtrn.8 \r0, \r1
	39	vtrn.8 \r2, \r3
	40	.endm
	41
	42	.macro swap4 r0, r1, r2, r3, r4, r5, r6, r7
	43	vswp \r0, \r4
	44	vswp \r1, \r5
	45	vswp \r2, \r6
	46	vswp \r3, \r7
	47	.endm
	48
	49	.macro transpose16_4x4 r0, r1, r2, r3, r4, r5, r6, r7
	50	vtrn.32 \r0, \r2
	51	vtrn.32 \r1, \r3
	52	vtrn.32 \r4, \r6
	53	vtrn.32 \r5, \r7
	54	vtrn.16 \r0, \r1
	55	vtrn.16 \r2, \r3
	56	vtrn.16 \r4, \r5
	57	vtrn.16 \r6, \r7
	58	.endm

-1

libde265/bitstream.cc less more

141	141	int offset = 0;
142	142	if (num_zeros != 0) {
143	143	offset = get_bits(br, num_zeros);
144		return offset + (1<<num_zeros)-1;
	144	int value = offset + (1<<num_zeros)-1;
	145	assert(value>0);
	146	return value;
145	147	} else {
146	148	return 0;
147	149	}

+631

-37

libde265/cabac.cc less more

22	22
23	23	#include <stdint.h>
24	24	#include <stdio.h>
	25	#include <stdlib.h>
25	26	#include <assert.h>
	27
	28	#define INITIAL_CABAC_BUFFER_CAPACITY 4096
26	29
27	30
28	31	static const uint8_t LPS_table[64][4] =

131	134
132	135	void init_CABAC_decoder(CABAC_decoder* decoder, uint8_t* bitstream, int length)
133	136	{
	137	assert(length >= 0);
	138
134	139	decoder->bitstream_start = bitstream;
135	140	decoder->bitstream_curr = bitstream;
136	141	decoder->bitstream_end = bitstream+length;

152	157	}
153	158
154	159
155		//#include <sys/types.h>
156		//#include <signal.h>
157
158	160	int decode_CABAC_bit(CABAC_decoder* decoder, context_model* model)
159	161	{
160		//if (logcnt >= 1100000) { enablelog(); }
161
162		// if (logcnt==400068770) { raise(SIGINT); }
163
164	162	logtrace(LogCABAC,"[%3d] decodeBin r:%x v:%x state:%d\n",logcnt,decoder->range, decoder->value, model->state);
165
166		//assert(decoder->range>=0x100);
167	163
168	164	int decoded_bit;
169	165	int LPS = LPS_table[model->state][ ( decoder->range >> 6 ) - 4 ];

177	173	{
178	174	logtrace(LogCABAC,"[%3d] MPS\n",logcnt);
179	175
180		// MPS path
	176	// MPS path
181	177
182	178	decoded_bit = model->MPSbit;
183	179	model->state = next_state_MPS[model->state];

193	189	if (decoder->bits_needed == 0)
194	190	{
195	191	decoder->bits_needed = -8;
196		if (decoder->bitstream_curr != decoder->bitstream_end)
	192	if (decoder->bitstream_curr < decoder->bitstream_end)
197	193	{ decoder->value \|= *decoder->bitstream_curr++; }
198	194	}
199	195	}

201	197	else
202	198	{
203	199	logtrace(LogCABAC,"[%3d] LPS\n",logcnt);
204
205		// LPS path
	200	//printf("%d %d\n", model->state, 0);
	201
	202	// LPS path
	203
	204	decoder->value = (decoder->value - scaled_range);
206	205
207	206	int num_bits = renorm_table[ LPS >> 3 ];
208		decoder->value = (decoder->value - scaled_range);
209
210	207	decoder->value <<= num_bits;
211	208	decoder->range = LPS << num_bits; /* this is always >= 0x100 except for state 63,
212	209	but state 63 is never used */
	210
	211	int num_bitsTab = renorm_table[ LPS >> 3 ];
	212
	213	assert(num_bits == num_bitsTab);
	214
213	215	decoded_bit = 1 - model->MPSbit;
214	216
215	217	if (model->state==0) { model->MPSbit = 1-model->MPSbit; }

220	222	if (decoder->bits_needed >= 0)
221	223	{
222	224	logtrace(LogCABAC,"bits_needed: %d\n", decoder->bits_needed);
223		if (decoder->bitstream_curr != decoder->bitstream_end)
	225	if (decoder->bitstream_curr < decoder->bitstream_end)
224	226	{ decoder->value \|= (*decoder->bitstream_curr++) << decoder->bits_needed; }
225	227
226	228	decoder->bits_needed -= 8;

232	234	logcnt++;
233	235	#endif
234	236
235		//assert(decoder->range>=0x100);
236
237	237	return decoded_bit;
238	238	}
239	239
240	240	int decode_CABAC_term_bit(CABAC_decoder* decoder)
241	241	{
	242	logtrace(LogCABAC,"CABAC term: range=%x\n", decoder->range);
	243
242	244	decoder->range -= 2;
243	245	uint32_t scaledRange = decoder->range << 7;
244	246

260	262	{
261	263	decoder->bits_needed = -8;
262	264
263		if (decoder->bitstream_curr != decoder->bitstream_end) {
	265	if (decoder->bitstream_curr < decoder->bitstream_end) {
264	266	decoder->value += (*decoder->bitstream_curr++);
265	267	}
266	268	}

276	278	{
277	279	logtrace(LogCABAC,"[%3d] bypass r:%x v:%x\n",logcnt,decoder->range, decoder->value);
278	280
279		//assert(decoder->range>=0x100);
280
281	281	decoder->value <<= 1;
282	282	decoder->bits_needed++;
283	283
284	284	if (decoder->bits_needed >= 0)
285	285	{
286		//assert(decoder->bits_needed==0);
287
288		decoder->bits_needed = -8;
289		decoder->value \|= *decoder->bitstream_curr++;
	286	if (decoder->bitstream_end > decoder->bitstream_curr) {
	287	decoder->bits_needed = -8;
	288	decoder->value \|= *decoder->bitstream_curr++;
	289	}
290	290	}
291	291
292	292	int bit;

306	306	logcnt++;
307	307	#endif
308	308
309		//assert(decoder->range>=0x100);
310
311	309	return bit;
312	310	}
313	311

339	337
340	338	int decode_CABAC_FL_bypass_parallel(CABAC_decoder* decoder, int nBits)
341	339	{
342		logtrace(LogCABAC,"[%3d] bypass group r:%x v:%x\n",logcnt,decoder->range, decoder->value);
	340	logtrace(LogCABAC,"[%3d] bypass group r:%x v:%x (nBits=%d)\n",logcnt,
	341	decoder->range, decoder->value, nBits);
343	342
344	343	decoder->value <<= nBits;
345	344	decoder->bits_needed+=nBits;
346	345
347	346	if (decoder->bits_needed >= 0)
348	347	{
349		int input = *decoder->bitstream_curr++;
350		input <<= decoder->bits_needed;
351
352		decoder->bits_needed -= 8;
353		decoder->value \|= input;
	348	if (decoder->bitstream_end > decoder->bitstream_curr) {
	349	int input = *decoder->bitstream_curr++;
	350	input <<= decoder->bits_needed;
	351
	352	decoder->bits_needed -= 8;
	353	decoder->value \|= input;
	354	}
354	355	}
355	356
356	357	uint32_t scaled_range = decoder->range << 7;

360	361
361	362	logtrace(LogCABAC,"[%3d] -> value %d r:%x v:%x\n", logcnt+nBits-1,
362	363	value, decoder->range, decoder->value);
	364
363	365	#ifdef DE265_LOG_TRACE
364	366	logcnt+=nBits;
365	367	#endif
366	368
367		//assert(decoder->range>=0x100);
368
369	369	return value;
370	370	}
371	371

373	373	int decode_CABAC_FL_bypass(CABAC_decoder* decoder, int nBits)
374	374	{
375	375	int value=0;
376
377	376
378	377	if (likely(nBits<=8)) {
379	378	if (nBits==0) {

398	397	value \|= decode_CABAC_bypass(decoder);
399	398	}
400	399	}
401
402	400	logtrace(LogCABAC," -> FL: %d\n", value);
403	401
404	402	return value;

415	413
416	414	return (prefix << cRiceParam) \| suffix;
417	415	}
	416
	417
	418	#define MAX_PREFIX 32
418	419
419	420	int decode_CABAC_EGk_bypass(CABAC_decoder* decoder, int k)
420	421	{

430	431	base += 1<<n;
431	432	n++;
432	433	}
	434
	435	if (n == k+MAX_PREFIX) {
	436	return 0; // TODO: error
	437	}
433	438	}
434	439
435	440	int suffix = decode_CABAC_FL_bypass(decoder, n);
436	441	return base + suffix;
437	442	}
438	443
	444
	445	// ---------------------------------------------------------------------------
	446
	447	void CABAC_encoder::add_trailing_bits()
	448	{
	449	write_bit(1);
	450	int nZeros = number_free_bits_in_byte();
	451	write_bits(0, nZeros);
	452	}
	453
	454
	455
	456	CABAC_encoder_bitstream::CABAC_encoder_bitstream()
	457	{
	458	data_mem = NULL;
	459	data_capacity = 0;
	460	data_size = 0;
	461	state = 0;
	462
	463	vlc_buffer_len = 0;
	464
	465	init_CABAC();
	466	}
	467
	468	CABAC_encoder_bitstream::~CABAC_encoder_bitstream()
	469	{
	470	free(data_mem);
	471	}
	472
	473	void CABAC_encoder_bitstream::reset()
	474	{
	475	data_size = 0;
	476	state = 0;
	477
	478	vlc_buffer_len = 0;
	479
	480	init_CABAC();
	481	}
	482
	483	void CABAC_encoder_bitstream::write_bits(uint32_t bits,int n)
	484	{
	485	vlc_buffer <<= n;
	486	vlc_buffer \|= bits;
	487	vlc_buffer_len += n;
	488
	489	while (vlc_buffer_len>=8) {
	490	append_byte((vlc_buffer >> (vlc_buffer_len-8)) & 0xFF);
	491	vlc_buffer_len -= 8;
	492	}
	493	}
	494
	495	void CABAC_encoder::write_uvlc(int value)
	496	{
	497	assert(value>=0);
	498
	499	int nLeadingZeros=0;
	500	int base=0;
	501	int range=1;
	502
	503	while (value>=base+range) {
	504	base += range;
	505	range <<= 1;
	506	nLeadingZeros++;
	507	}
	508
	509	write_bits((1<<nLeadingZeros) \| (value-base),2*nLeadingZeros+1);
	510	}
	511
	512	void CABAC_encoder::write_svlc(int value)
	513	{
	514	if (value==0) write_bits(1,1);
	515	else if (value>0) write_uvlc(2*value-1);
	516	else write_uvlc(-2*value);
	517	}
	518
	519	void CABAC_encoder_bitstream::flush_VLC()
	520	{
	521	while (vlc_buffer_len>=8) {
	522	append_byte((vlc_buffer >> (vlc_buffer_len-8)) & 0xFF);
	523	vlc_buffer_len -= 8;
	524	}
	525
	526	if (vlc_buffer_len>0) {
	527	append_byte(vlc_buffer << (8-vlc_buffer_len));
	528	vlc_buffer_len = 0;
	529	}
	530
	531	vlc_buffer = 0;
	532	}
	533
	534	void CABAC_encoder_bitstream::skip_bits(int nBits)
	535	{
	536	while (nBits>=8) {
	537	write_bits(0,8);
	538	nBits-=8;
	539	}
	540
	541	if (nBits>0) {
	542	write_bits(0,nBits);
	543	}
	544	}
	545
	546
	547	int CABAC_encoder_bitstream::number_free_bits_in_byte() const
	548	{
	549	if ((vlc_buffer_len % 8)==0) return 0;
	550	return 8- (vlc_buffer_len % 8);
	551	}
	552
	553
	554	void CABAC_encoder_bitstream::check_size_and_resize(int nBytes)
	555	{
	556	if (data_size+nBytes > data_capacity) { // 1 extra byte for stuffing
	557	if (data_capacity==0) {
	558	data_capacity = INITIAL_CABAC_BUFFER_CAPACITY;
	559	} else {
	560	data_capacity *= 2;
	561	}
	562
	563	data_mem = (uint8_t*)realloc(data_mem,data_capacity);
	564	}
	565	}
	566
	567
	568	void CABAC_encoder_bitstream::append_byte(int byte)
	569	{
	570	check_size_and_resize(2);
	571
	572	// --- emulation prevention ---
	573
	574	/* These byte sequences may never occur in the bitstream:
	575	0x000000 / 0x000001 / 0x000002
	576
	577	Hence, we have to add a 0x03 before the third byte.
	578	We also have to add a 0x03 for this sequence: 0x000003, because
	579	the escape byte itself also has to be escaped.
	580	*/
	581
	582	// S0 --(0)--> S1 --(0)--> S2 --(0,1,2,3)--> add stuffing
	583
	584	if (byte<=3) {
	585	/**/ if (state< 2 && byte==0) { state++; }
	586	else if (state==2 && byte<=3) {
	587	data_mem[ data_size++ ] = 3;
	588
	589	if (byte==0) state=1;
	590	else state=0;
	591	}
	592	else { state=0; }
	593	}
	594	else { state=0; }
	595
	596
	597	// write actual data byte
	598
	599	data_mem[ data_size++ ] = byte;
	600	}
	601
	602
	603	void CABAC_encoder_bitstream::write_startcode()
	604	{
	605	check_size_and_resize(3);
	606
	607	data_mem[ data_size+0 ] = 0;
	608	data_mem[ data_size+1 ] = 0;
	609	data_mem[ data_size+2 ] = 1;
	610	data_size+=3;
	611	}
	612
	613	void CABAC_encoder_bitstream::init_CABAC()
	614	{
	615	range = 510;
	616	low = 0;
	617
	618	bits_left = 23;
	619	buffered_byte = 0xFF;
	620	num_buffered_bytes = 0;
	621	}
	622
	623	void CABAC_encoder_bitstream::flush_CABAC()
	624	{
	625	if (low >> (32 - bits_left))
	626	{
	627	append_byte(buffered_byte + 1);
	628	while (num_buffered_bytes > 1)
	629	{
	630	append_byte(0x00);
	631	num_buffered_bytes--;
	632	}
	633
	634	low -= 1 << (32 - bits_left);
	635	}
	636	else
	637	{
	638	if (num_buffered_bytes > 0)
	639	{
	640	append_byte(buffered_byte);
	641	}
	642
	643	while (num_buffered_bytes > 1)
	644	{
	645	append_byte(0xff);
	646	num_buffered_bytes--;
	647	}
	648	}
	649
	650	// printf("low: %08x nbits left:%d filled:%d\n",low,bits_left,32-bits_left);
	651
	652	write_bits(low >> 8, 24-bits_left);
	653	}
	654
	655
	656	void CABAC_encoder_bitstream::write_out()
	657	{
	658	//logtrace(LogCABAC,"low = %08x (bits_left=%d)\n",low,bits_left);
	659	int leadByte = low >> (24 - bits_left);
	660	bits_left += 8;
	661	low &= 0xffffffffu >> bits_left;
	662
	663	//logtrace(LogCABAC,"write byte %02x\n",leadByte);
	664	//logtrace(LogCABAC,"-> low = %08x\n",low);
	665
	666	if (leadByte == 0xff)
	667	{
	668	num_buffered_bytes++;
	669	}
	670	else
	671	{
	672	if (num_buffered_bytes > 0)
	673	{
	674	int carry = leadByte >> 8;
	675	int byte = buffered_byte + carry;
	676	buffered_byte = leadByte & 0xff;
	677	append_byte(byte);
	678
	679	byte = ( 0xff + carry ) & 0xff;
	680	while ( num_buffered_bytes > 1 )
	681	{
	682	append_byte(byte);
	683	num_buffered_bytes--;
	684	}
	685	}
	686	else
	687	{
	688	num_buffered_bytes = 1;
	689	buffered_byte = leadByte;
	690	}
	691	}
	692	}
	693
	694	void CABAC_encoder_bitstream::testAndWriteOut()
	695	{
	696	// logtrace(LogCABAC,"bits_left = %d\n",bits_left);
	697
	698	if (bits_left < 12)
	699	{
	700	write_out();
	701	}
	702	}
	703
	704
	705	#ifdef DE265_LOG_TRACE
	706	int encBinCnt=1;
	707	#endif
	708
	709	void CABAC_encoder_bitstream::write_CABAC_bit(int modelIdx, int bin)
	710	{
	711	context_model* model = &(*mCtxModels)[modelIdx];
	712	//m_uiBinsCoded += m_binCountIncrement;
	713	//rcCtxModel.setBinsCoded( 1 );
	714
	715	logtrace(LogCABAC,"[%d] range=%x low=%x state=%d, bin=%d\n",
	716	encBinCnt, range,low, model->state,bin);
	717
	718	/*
	719	printf("[%d] range=%x low=%x state=%d, bin=%d\n",
	720	encBinCnt, range,low, model->state,bin);
	721
	722	printf("%d %d X\n",model->state,bin != model->MPSbit);
	723	*/
	724
	725	#ifdef DE265_LOG_TRACE
	726	encBinCnt++;
	727	#endif
	728
	729	uint32_t LPS = LPS_table[model->state][ ( range >> 6 ) - 4 ];
	730	range -= LPS;
	731
	732	if (bin != model->MPSbit)
	733	{
	734	//logtrace(LogCABAC,"LPS\n");
	735
	736	int num_bits = renorm_table[ LPS >> 3 ];
	737	low = (low + range) << num_bits;
	738	range = LPS << num_bits;
	739
	740	if (model->state==0) { model->MPSbit = 1-model->MPSbit; }
	741
	742	model->state = next_state_LPS[model->state];
	743
	744	bits_left -= num_bits;
	745	}
	746	else
	747	{
	748	//logtrace(LogCABAC,"MPS\n");
	749
	750	model->state = next_state_MPS[model->state];
	751
	752
	753	// renorm
	754
	755	if (range >= 256) { return; }
	756
	757	low <<= 1;
	758	range <<= 1;
	759	bits_left--;
	760	}
	761
	762	testAndWriteOut();
	763	}
	764
	765	void CABAC_encoder_bitstream::write_CABAC_bypass(int bin)
	766	{
	767	logtrace(LogCABAC,"[%d] bypass = %d, range=%x\n",encBinCnt,bin,range);
	768	/*
	769	printf("[%d] bypass = %d, range=%x\n",encBinCnt,bin,range);
	770	printf("%d %d X\n",64, -1);
	771	*/
	772
	773	#ifdef DE265_LOG_TRACE
	774	encBinCnt++;
	775	#endif
	776
	777	// BinsCoded += m_binCountIncrement;
	778	low <<= 1;
	779
	780	if (bin)
	781	{
	782	low += range;
	783	}
	784	bits_left--;
	785
	786	testAndWriteOut();
	787	}
	788
	789	void CABAC_encoder::write_CABAC_TU_bypass(int value, int cMax)
	790	{
	791	for (int i=0;i<value;i++) {
	792	write_CABAC_bypass(1);
	793	}
	794
	795	if (value<cMax) {
	796	write_CABAC_bypass(0);
	797	}
	798	}
	799
	800	void CABAC_encoder::write_CABAC_FL_bypass(int value, int n)
	801	{
	802	while (n>0) {
	803	n--;
	804	write_CABAC_bypass(value & (1<<n));
	805	}
	806	}
	807
	808	void CABAC_encoder_bitstream::write_CABAC_term_bit(int bit)
	809	{
	810	logtrace(LogCABAC,"CABAC term: range=%x\n", range);
	811
	812	range -= 2;
	813
	814	if (bit) {
	815	low += range;
	816
	817	low <<= 7;
	818	range = 2 << 7;
	819	bits_left -= 7;
	820	}
	821	else if (range >= 256)
	822	{
	823	return;
	824	}
	825	else
	826	{
	827	low <<= 1;
	828	range <<= 1;
	829	bits_left--;
	830	}
	831
	832	testAndWriteOut();
	833	}
	834
	835
	836
	837
	838	static const uint32_t entropy_table[128] = {
	839	// -------------------- 200 --------------------
	840	/* state= 0 / 0x07d13 / 0.977164 /, 0x08255 / 1.018237 */,
	841	/* state= 1 / 0x07738 / 0.931417 /, 0x086ef / 1.054179 */,
	842	/* state= 2 / 0x0702b / 0.876323 /, 0x0935a / 1.151195 */,
	843	/* state= 3 / 0x069e6 / 0.827333 /, 0x09c7f / 1.222650 */,
	844	/* state= 4 / 0x062e8 / 0.772716 /, 0x0a2c7 / 1.271708 */,
	845	/* state= 5 / 0x05c18 / 0.719488 /, 0x0ae25 / 1.360532 */,
	846	/* state= 6 / 0x05632 / 0.673414 /, 0x0b724 / 1.430793 */,
	847	/* state= 7 / 0x05144 / 0.634904 /, 0x0c05d / 1.502850 */,
	848	/* state= 8 / 0x04bdf / 0.592754 /, 0x0ccf2 / 1.601145 */,
	849	/* state= 9 / 0x0478d / 0.559012 /, 0x0d57b / 1.667843 */,
	850	/* state=10 / 0x042ad / 0.520924 /, 0x0de81 / 1.738336 */,
	851	/* state=11 / 0x03f4d / 0.494564 /, 0x0e4b8 / 1.786871 */,
	852	/* state=12 / 0x03a9d / 0.457945 /, 0x0f471 / 1.909721 */,
	853	/* state=13 / 0x037d5 / 0.436201 /, 0x0fc56 / 1.971385 */,
	854	/* state=14 / 0x034c2 / 0.412177 /, 0x10236 / 2.017284 */,
	855	/* state=15 / 0x031a6 / 0.387895 /, 0x10d5c / 2.104394 */,
	856	/* state=16 / 0x02e62 / 0.362383 /, 0x11b34 / 2.212552 */,
	857	/* state=17 / 0x02c20 / 0.344752 /, 0x120b4 / 2.255512 */,
	858	/* state=18 / 0x029b8 / 0.325943 /, 0x1294d / 2.322672 */,
	859	/* state=19 / 0x02791 / 0.309143 /, 0x135e1 / 2.420959 */,
	860	/* state=20 / 0x02562 / 0.292057 /, 0x13e37 / 2.486077 */,
	861	/* state=21 / 0x0230d / 0.273846 /, 0x144fd / 2.539000 */,
	862	/* state=22 / 0x02193 / 0.262308 /, 0x150c9 / 2.631150 */,
	863	/* state=23 / 0x01f5d / 0.245026 /, 0x15ca0 / 2.723641 */,
	864	/* state=24 / 0x01de7 / 0.233617 /, 0x162f9 / 2.773246 */,
	865	/* state=25 / 0x01c2f / 0.220208 /, 0x16d99 / 2.856259 */,
	866	/* state=26 / 0x01a8e / 0.207459 /, 0x17a93 / 2.957634 */,
	867	/* state=27 / 0x0195a / 0.198065 /, 0x18051 / 3.002477 */,
	868	/* state=28 / 0x01809 / 0.187778 /, 0x18764 / 3.057759 */,
	869	/* state=29 / 0x0164a / 0.174144 /, 0x19460 / 3.159206 */,
	870	/* state=30 / 0x01539 / 0.165824 /, 0x19f20 / 3.243181 */,
	871	/* state=31 / 0x01452 / 0.158756 /, 0x1a465 / 3.284334 */,
	872	/* state=32 / 0x0133b / 0.150261 /, 0x1b422 / 3.407303 */,
	873	/* state=33 / 0x0120c / 0.140995 /, 0x1bce5 / 3.475767 */,
	874	/* state=34 / 0x01110 / 0.133315 /, 0x1c394 / 3.527962 */,
	875	/* state=35 / 0x0104d / 0.127371 /, 0x1d059 / 3.627736 */,
	876	/* state=36 / 0x00f8b / 0.121451 /, 0x1d74b / 3.681983 */,
	877	/* state=37 / 0x00ef4 / 0.116829 /, 0x1dfd0 / 3.748540 */,
	878	/* state=38 / 0x00e10 / 0.109864 /, 0x1e6d3 / 3.803335 */,
	879	/* state=39 / 0x00d3f / 0.103507 /, 0x1f925 / 3.946462 */,
	880	/* state=40 / 0x00cc4 / 0.099758 /, 0x1fda7 / 3.981667 */,
	881	/* state=41 / 0x00c42 / 0.095792 /, 0x203f8 / 4.031012 */,
	882	/* state=42 / 0x00b78 / 0.089610 /, 0x20f7d / 4.121014 */,
	883	/* state=43 / 0x00afc / 0.085830 /, 0x21dd6 / 4.233102 */,
	884	/* state=44 / 0x00a5e / 0.081009 /, 0x22419 / 4.282016 */,
	885	/* state=45 / 0x00a1b / 0.078950 /, 0x22a5e / 4.331015 */,
	886	/* state=46 / 0x00989 / 0.074514 /, 0x23756 / 4.432323 */,
	887	/* state=47 / 0x0091b / 0.071166 /, 0x24225 / 4.516775 */,
	888	/* state=48 / 0x008cf / 0.068837 /, 0x2471a / 4.555487 */,
	889	/* state=49 / 0x00859 / 0.065234 /, 0x25313 / 4.649048 */,
	890	/* state=50 / 0x00814 / 0.063140 /, 0x25d67 / 4.729721 */,
	891	/* state=51 / 0x007b6 / 0.060272 /, 0x2651f / 4.790028 */,
	892	/* state=52 / 0x0076e / 0.058057 /, 0x2687c / 4.816294 */,
	893	/* state=53 / 0x00707 / 0.054924 /, 0x27da7 / 4.981661 */,
	894	/* state=54 / 0x006d5 / 0.053378 /, 0x28172 / 5.011294 */,
	895	/* state=55 / 0x00659 / 0.049617 /, 0x28948 / 5.072512 */,
	896	/* state=56 / 0x00617 / 0.047598 /, 0x297c5 / 5.185722 */,
	897	/* state=57 / 0x005dd / 0.045814 /, 0x2a2df / 5.272434 */,
	898	/* state=58 / 0x005c1 / 0.044965 /, 0x2a581 / 5.293019 */,
	899	/* state=59 / 0x00574 / 0.042619 /, 0x2ad59 / 5.354304 */,
	900	/* state=60 / 0x0053b / 0.040882 /, 0x2bba5 / 5.465973 */,
	901	/* state=61 / 0x0050c / 0.039448 /, 0x2c596 / 5.543651 */,
	902	/* state=62 / 0x004e9 / 0.038377 /, 0x2cd88 / 5.605741 */,
	903	0x00400 , 0x2d000 /* dummy, should never be used */
	904	};
	905
	906
	907	static const uint32_t entropy_table_orig[128] = {
	908	0x07b23, 0x085f9, 0x074a0, 0x08cbc, 0x06ee4, 0x09354, 0x067f4, 0x09c1b,
	909	0x060b0, 0x0a62a, 0x05a9c, 0x0af5b, 0x0548d, 0x0b955, 0x04f56, 0x0c2a9,
	910	0x04a87, 0x0cbf7, 0x045d6, 0x0d5c3, 0x04144, 0x0e01b, 0x03d88, 0x0e937,
	911	0x039e0, 0x0f2cd, 0x03663, 0x0fc9e, 0x03347, 0x10600, 0x03050, 0x10f95,
	912	0x02d4d, 0x11a02, 0x02ad3, 0x12333, 0x0286e, 0x12cad, 0x02604, 0x136df,
	913	0x02425, 0x13f48, 0x021f4, 0x149c4, 0x0203e, 0x1527b, 0x01e4d, 0x15d00,
	914	0x01c99, 0x166de, 0x01b18, 0x17017, 0x019a5, 0x17988, 0x01841, 0x18327,
	915	0x016df, 0x18d50, 0x015d9, 0x19547, 0x0147c, 0x1a083, 0x0138e, 0x1a8a3,
	916	0x01251, 0x1b418, 0x01166, 0x1bd27, 0x01068, 0x1c77b, 0x00f7f, 0x1d18e,
	917	0x00eda, 0x1d91a, 0x00e19, 0x1e254, 0x00d4f, 0x1ec9a, 0x00c90, 0x1f6e0,
	918	0x00c01, 0x1fef8, 0x00b5f, 0x208b1, 0x00ab6, 0x21362, 0x00a15, 0x21e46,
	919	0x00988, 0x2285d, 0x00934, 0x22ea8, 0x008a8, 0x239b2, 0x0081d, 0x24577,
	920	0x007c9, 0x24ce6, 0x00763, 0x25663, 0x00710, 0x25e8f, 0x006a0, 0x26a26,
	921	0x00672, 0x26f23, 0x005e8, 0x27ef8, 0x005ba, 0x284b5, 0x0055e, 0x29057,
	922	0x0050c, 0x29bab, 0x004c1, 0x2a674, 0x004a7, 0x2aa5e, 0x0046f, 0x2b32f,
	923	0x0041f, 0x2c0ad, 0x003e7, 0x2ca8d, 0x003ba, 0x2d323, 0x0010c, 0x3bfbb
	924	};
	925
	926
	927	const uint32_t entropy_table_theory[128] =
	928	{
	929	0x08000, 0x08000, 0x076da, 0x089a0, 0x06e92, 0x09340, 0x0670a, 0x09cdf, 0x06029, 0x0a67f, 0x059dd, 0x0b01f, 0x05413, 0x0b9bf, 0x04ebf, 0x0c35f,
	930	0x049d3, 0x0ccff, 0x04546, 0x0d69e, 0x0410d, 0x0e03e, 0x03d22, 0x0e9de, 0x0397d, 0x0f37e, 0x03619, 0x0fd1e, 0x032ee, 0x106be, 0x02ffa, 0x1105d,
	931	0x02d37, 0x119fd, 0x02aa2, 0x1239d, 0x02836, 0x12d3d, 0x025f2, 0x136dd, 0x023d1, 0x1407c, 0x021d2, 0x14a1c, 0x01ff2, 0x153bc, 0x01e2f, 0x15d5c,
	932	0x01c87, 0x166fc, 0x01af7, 0x1709b, 0x0197f, 0x17a3b, 0x0181d, 0x183db, 0x016d0, 0x18d7b, 0x01595, 0x1971b, 0x0146c, 0x1a0bb, 0x01354, 0x1aa5a,
	933	0x0124c, 0x1b3fa, 0x01153, 0x1bd9a, 0x01067, 0x1c73a, 0x00f89, 0x1d0da, 0x00eb7, 0x1da79, 0x00df0, 0x1e419, 0x00d34, 0x1edb9, 0x00c82, 0x1f759,
	934	0x00bda, 0x200f9, 0x00b3c, 0x20a99, 0x00aa5, 0x21438, 0x00a17, 0x21dd8, 0x00990, 0x22778, 0x00911, 0x23118, 0x00898, 0x23ab8, 0x00826, 0x24458,
	935	0x007ba, 0x24df7, 0x00753, 0x25797, 0x006f2, 0x26137, 0x00696, 0x26ad7, 0x0063f, 0x27477, 0x005ed, 0x27e17, 0x0059f, 0x287b6, 0x00554, 0x29156,
	936	0x0050e, 0x29af6, 0x004cc, 0x2a497, 0x0048d, 0x2ae35, 0x00451, 0x2b7d6, 0x00418, 0x2c176, 0x003e2, 0x2cb15, 0x003af, 0x2d4b5, 0x0037f, 0x2de55
	937	};
	938
	939
	940	void CABAC_encoder_estim::write_CABAC_bit(int modelIdx, int bit)
	941	{
	942	context_model* model = &(*mCtxModels)[modelIdx];
	943	//printf("[%d] state=%d, bin=%d\n", encBinCnt, model->state,bit);
	944	//encBinCnt++;
	945
	946	int idx = model->state<<1;
	947
	948	if (bit==model->MPSbit) {
	949	model->state = next_state_MPS[model->state];
	950	}
	951	else {
	952	idx++;
	953	if (model->state==0) { model->MPSbit = 1-model->MPSbit; }
	954	model->state = next_state_LPS[model->state];
	955	}
	956
	957	mFracBits += entropy_table[idx];
	958
	959	//printf("-> %08lx %f\n",entropy_table[idx], entropy_table[idx] / float(1<<15));
	960	}
	961
	962
	963	float CABAC_encoder::RDBits_for_CABAC_bin(int modelIdx, int bit)
	964	{
	965	context_model* model = &(*mCtxModels)[modelIdx];
	966	int idx = model->state<<1;
	967
	968	if (bit!=model->MPSbit) {
	969	idx++;
	970	}
	971
	972	return entropy_table[idx] / float(1<<15);
	973	}
	974
	975
	976	void CABAC_encoder::write_CABAC_EGk(int val, int k)
	977	{
	978	while (val >= ( 1 << k ) ) {
	979	write_CABAC_bypass(1);
	980	val = val - ( 1 << k );
	981	k++;
	982	}
	983
	984	write_CABAC_bypass(0);
	985
	986	while (k) {
	987	k--;
	988	write_CABAC_bypass((val >> k) & 1);
	989	}
	990	}
	991
	992
	993
	994	void CABAC_encoder_estim_constant::write_CABAC_bit(int modelIdx, int bit)
	995	{
	996	context_model* model = &(*mCtxModels)[modelIdx];
	997	int idx = model->state<<1;
	998
	999	if (bit!=model->MPSbit) {
	1000	idx++;
	1001	}
	1002
	1003	mFracBits += entropy_table[idx];
	1004	}
	1005
	1006
	1007
	1008	#if 0
	1009	void printtab(int idx,int s)
	1010	{
	1011	printf("%d %f %f %f\n", s,
	1012	double(entropy_table[idx])/0x8000,
	1013	double(entropy_table_orig[idx])/0x8000,
	1014	double(entropy_table_f265[idx])/0x8000);
	1015	}
	1016
	1017
	1018	void plot_tables()
	1019	{
	1020	for (int i=-62;i<=0;i++) {
	1021	int idx = -i *2;
	1022	int s = i;
	1023	printtab(idx,s);
	1024	}
	1025
	1026	for (int i=0;i<=62;i++) {
	1027	int idx = 2*i +1;
	1028	int s = i;
	1029	printtab(idx,s);
	1030	}
	1031	}
	1032	#endif

+161

-6

libde265/cabac.h less more

21	21	#define DE265_CABAC_H
22	22
23	23	#include <stdint.h>
	24	#include "contextmodel.h"
24	25
25	26
26	27	typedef struct {

32	33	uint32_t value;
33	34	int16_t bits_needed;
34	35	} CABAC_decoder;
35
36
37		typedef struct {
38		uint8_t MPSbit : 1;
39		uint8_t state : 7;
40		} context_model;
41	36
42	37
43	38	void init_CABAC_decoder(CABAC_decoder* decoder, uint8_t* bitstream, int length);

52	47	int decode_CABAC_TR_bypass(CABAC_decoder* decoder, int cRiceParam, int cTRMax);
53	48	int decode_CABAC_EGk_bypass(CABAC_decoder* decoder, int k);
54	49
	50
	51	// ---------------------------------------------------------------------------
	52
	53	class CABAC_encoder
	54	{
	55	public:
	56	CABAC_encoder() : mCtxModels(NULL) { }
	57	virtual ~CABAC_encoder() { }
	58
	59	virtual int size() const = 0;
	60	virtual void reset() = 0;
	61
	62	// --- VLC ---
	63
	64	virtual void write_bits(uint32_t bits,int n) = 0;
	65	virtual void write_bit(int bit) { write_bits(bit,1); }
	66	virtual void write_uvlc(int value);
	67	virtual void write_svlc(int value);
	68	virtual void write_startcode() = 0;
	69	virtual void skip_bits(int nBits) = 0;
	70
	71	virtual void add_trailing_bits();
	72	virtual int number_free_bits_in_byte() const = 0;
	73
	74	// output all remaining bits and fill with zeros to next byte boundary
	75	virtual void flush_VLC() { }
	76
	77
	78	// --- CABAC ---
	79
	80	void set_context_models(context_model_table* models) { mCtxModels=models; }
	81
	82	virtual void init_CABAC() { }
	83	virtual void write_CABAC_bit(int modelIdx, int bit) = 0;
	84	virtual void write_CABAC_bypass(int bit) = 0;
	85	virtual void write_CABAC_TU_bypass(int value, int cMax);
	86	virtual void write_CABAC_FL_bypass(int value, int nBits);
	87	virtual void write_CABAC_term_bit(int bit) = 0;
	88	virtual void flush_CABAC() { }
	89
	90	void write_CABAC_EGk(int absolute_symbol, int k); // absolute_symbol >= 0
	91
	92	virtual bool modifies_context() const = 0;
	93
	94	float RDBits_for_CABAC_bin(int modelIdx, int bit);
	95
	96	protected:
	97	context_model_table* mCtxModels;
	98	};
	99
	100
	101	class CABAC_encoder_bitstream : public CABAC_encoder
	102	{
	103	public:
	104	CABAC_encoder_bitstream();
	105	~CABAC_encoder_bitstream();
	106
	107	virtual void reset();
	108
	109	virtual int size() const { return data_size; }
	110	uint8_t* data() const { return data_mem; }
	111
	112	// --- VLC ---
	113
	114	virtual void write_bits(uint32_t bits,int n);
	115	virtual void write_startcode();
	116	virtual void skip_bits(int nBits);
	117
	118	virtual int number_free_bits_in_byte() const;
	119
	120	// output all remaining bits and fill with zeros to next byte boundary
	121	virtual void flush_VLC();
	122
	123
	124	// --- CABAC ---
	125
	126	virtual void init_CABAC();
	127	virtual void write_CABAC_bit(int modelIdx, int bit);
	128	virtual void write_CABAC_bypass(int bit);
	129	virtual void write_CABAC_term_bit(int bit);
	130	virtual void flush_CABAC();
	131
	132	virtual bool modifies_context() const { return true; }
	133
	134	private:
	135	// data buffer
	136
	137	uint8_t* data_mem;
	138	uint32_t data_capacity;
	139	uint32_t data_size;
	140	char state; // for inserting emulation-prevention bytes
	141
	142	// VLC
	143
	144	uint32_t vlc_buffer;
	145	uint32_t vlc_buffer_len;
	146
	147
	148	// CABAC
	149
	150	uint32_t range;
	151	uint32_t low;
	152	int8_t bits_left;
	153	uint8_t buffered_byte;
	154	uint16_t num_buffered_bytes;
	155
	156
	157	void check_size_and_resize(int nBytes);
	158	void testAndWriteOut();
	159	void write_out();
	160	void append_byte(int byte);
	161	};
	162
	163
	164	class CABAC_encoder_estim : public CABAC_encoder
	165	{
	166	public:
	167	CABAC_encoder_estim() : mFracBits(0) { }
	168
	169	virtual void reset() { mFracBits=0; }
	170
	171	virtual int size() const { return mFracBits>>(15+3); }
	172
	173	uint64_t getFracBits() const { return mFracBits; }
	174	float getRDBits() const { return mFracBits / float(1<<15); }
	175
	176	// --- VLC ---
	177
	178	virtual void write_bits(uint32_t bits,int n) { mFracBits += n<<15; }
	179	virtual void write_bit(int bit) { mFracBits+=1<<15; }
	180	virtual void write_startcode() { mFracBits += (1<<15)83; }
	181	virtual void skip_bits(int nBits) { mFracBits += nBits<<15; }
	182	virtual int number_free_bits_in_byte() const { return 0; } // TODO, good enough for now
	183
	184	// --- CABAC ---
	185
	186	virtual void write_CABAC_bit(int modelIdx, int bit);
	187	virtual void write_CABAC_bypass(int bit) {
	188	mFracBits += 0x8000;
	189	}
	190	virtual void write_CABAC_FL_bypass(int value, int nBits) {
	191	mFracBits += nBits<<15;
	192	}
	193	virtual void write_CABAC_term_bit(int bit) { /* not implemented (not needed) */ }
	194
	195	virtual bool modifies_context() const { return true; }
	196
	197	protected:
	198	uint64_t mFracBits;
	199	};
	200
	201
	202	class CABAC_encoder_estim_constant : public CABAC_encoder_estim
	203	{
	204	public:
	205	void write_CABAC_bit(int modelIdx, int bit);
	206
	207	virtual bool modifies_context() const { return false; }
	208	};
	209
55	210	#endif

+485

-0

libde265/configparam.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "configparam.h"
	23
	24	#include <string.h>
	25	#include <ctype.h>
	26	#include <sstream>
	27	#include <iomanip>
	28	#include <iostream>
	29	#include <algorithm>
	30	#include <typeinfo>
	31
	32	#ifndef RTTI_ENABLED
	33	#error "Need to compile with RTTI enabled."
	34	#endif
	35
	36	static void remove_option(int* argc,char** argv,int idx, int n=1)
	37	{
	38	for (int i=idx+n;i<*argc;i++) {
	39	argv[i-n] = argv[i];
	40	}
	41
	42	*argc-=n;
	43	}
	44
	45
	46	bool option_string::processCmdLineArguments(char** argv, int* argc, int idx)
	47	{
	48	if (argv==NULL) { return false; }
	49	if (idx >= *argc) { return false; }
	50
	51	value = argv[idx];
	52	value_set = true;
	53
	54	remove_option(argc,argv,idx,1);
	55
	56	return true;
	57	}
	58
	59
	60	void option_int::set_range(int mini,int maxi)
	61	{
	62	have_low_limit =true;
	63	have_high_limit=true;
	64	low_limit =mini;
	65	high_limit=maxi;
	66	}
	67
	68	std::string option_int::getTypeDescr() const
	69	{
	70	std::stringstream sstr;
	71	sstr << "(int)";
	72
	73	if (have_low_limit \|\| have_high_limit) { sstr << " "; }
	74	if (have_low_limit) { sstr << low_limit << " <= "; }
	75	if (have_low_limit \|\| have_high_limit) { sstr << "x"; }
	76	if (have_high_limit) { sstr << " <= " << high_limit; }
	77
	78	if (!valid_values_set.empty()) {
	79	sstr << " {";
	80	bool first=true;
	81	FOR_LOOP(int, v, valid_values_set) {
	82	if (!first) sstr << ","; else first=false;
	83	sstr << v;
	84	}
	85	sstr << "}";
	86	}
	87
	88	return sstr.str();
	89	}
	90
	91	bool option_int::processCmdLineArguments(char** argv, int* argc, int idx)
	92	{
	93	if (argv==NULL) { return false; }
	94	if (idx >= *argc) { return false; }
	95
	96	int v = atoi(argv[idx]);
	97	if (!is_valid(v)) { return false; }
	98
	99	value = v;
	100	value_set = true;
	101
	102	remove_option(argc,argv,idx,1);
	103
	104	return true;
	105	}
	106
	107	bool option_int::is_valid(int v) const
	108	{
	109	if (have_low_limit && v<low_limit) { return false; }
	110	if (have_high_limit && v>high_limit) { return false; }
	111
	112	if (!valid_values_set.empty()) {
	113	auto iter = std::find(valid_values_set.begin(), valid_values_set.end(), v);
	114	if (iter==valid_values_set.end()) { return false; }
	115	}
	116
	117	return true;
	118	}
	119
	120	std::string option_int::get_default_string() const
	121	{
	122	std::stringstream sstr;
	123	sstr << default_value;
	124	return sstr.str();
	125	}
	126
	127
	128	std::string choice_option_base::getTypeDescr() const
	129	{
	130	std::vector<std::string> choices = get_choice_names();
	131
	132	std::stringstream sstr;
	133	sstr << "{";
	134
	135	bool first=true;
	136	#ifdef FOR_LOOP_AUTO_SUPPORT
	137	FOR_LOOP(auto, c, choices) {
	138	#else
	139	FOR_LOOP(std::string, c, choices) {
	140	#endif
	141	if (first) { first=false; }
	142	else { sstr << ","; }
	143
	144	sstr << c;
	145	}
	146
	147	sstr << "}";
	148	return sstr.str();
	149	}
	150
	151
	152	bool choice_option_base::processCmdLineArguments(char** argv, int* argc, int idx)
	153	{
	154	if (argv==NULL) { return false; }
	155	if (idx >= *argc) { return false; }
	156
	157	std::string value = argv[idx];
	158
	159	std::cout << "set " << value << "\n";
	160	bool success = set_value(value);
	161	std::cout << "success " << success << "\n";
	162
	163	remove_option(argc,argv,idx,1);
	164
	165	return success;
	166	}
	167
	168
	169	static char* fill_strings_into_memory(const std::vector<std::string>& strings_list)
	170	{
	171	// calculate memory requirement
	172
	173	int totalStringLengths = 0;
	174	#ifdef FOR_LOOP_AUTO_SUPPORT
	175	FOR_LOOP(auto, str, strings_list) {
	176	#else
	177	FOR_LOOP(std::string, str, strings_list) {
	178	#endif
	179	totalStringLengths += str.length() +1; // +1 for null termination
	180	}
	181
	182	int numStrings = strings_list.size();
	183
	184	int pointersSize = (numStrings+1) * sizeof(const char*);
	185
	186	char* memory = new char[pointersSize + totalStringLengths];
	187
	188
	189	// copy strings to memory area
	190
	191	char* stringPtr = memory + (numStrings+1) * sizeof(const char*);
	192	const char tablePtr = (const char)memory;
	193
	194	#ifdef FOR_LOOP_AUTO_SUPPORT
	195	FOR_LOOP(auto, str, strings_list) {
	196	#else
	197	FOR_LOOP(std::string, str, strings_list) {
	198	#endif
	199	*tablePtr++ = stringPtr;
	200
	201	strcpy(stringPtr, str.c_str());
	202	stringPtr += str.length()+1;
	203	}
	204
	205	*tablePtr = NULL;
	206
	207	return memory;
	208	}
	209
	210
	211	const char** choice_option_base::get_choices_string_table() const
	212	{
	213	if (choice_string_table==NULL) {
	214	choice_string_table = fill_strings_into_memory(get_choice_names());
	215	}
	216
	217	return (const char**)choice_string_table;
	218	}
	219
	220
	221
	222	bool config_parameters::parse_command_line_params(int* argc, char** argv, int* first_idx_ptr,
	223	bool ignore_unknown_options)
	224	{
	225	int first_idx=1;
	226	if (first_idx_ptr) { first_idx = *first_idx_ptr; }
	227
	228	for (int i=first_idx;i < *argc;i++) {
	229
	230	if (argv[i][0]=='-') {
	231	// option
	232
	233	if (argv[i][1]=='-') {
	234	// long option
	235
	236	bool option_found=false;
	237
	238	for (int o=0;o<mOptions.size();o++) {
	239	if (mOptions[o]->hasLongOption() && strcmp(mOptions[o]->getLongOption().c_str(),
	240	argv[i]+2)==0) {
	241	option_found=true;
	242
	243	bool success = mOptions[o]->processCmdLineArguments(argv,argc, i+1);
	244	if (!success) {
	245	if (first_idx_ptr) { *first_idx_ptr = i; }
	246	return false;
	247	}
	248
	249	remove_option(argc,argv,i);
	250	i--;
	251
	252	break;
	253	}
	254	}
	255
	256	if (option_found == false && !ignore_unknown_options) {
	257	return false;
	258	}
	259	}
	260	else {
	261	// short option
	262
	263	bool is_single_option = (argv[i][1] != 0 && argv[i][2]==0);
	264	bool do_remove_option = true;
	265
	266	for (int n=1; argv[i][n]; n++) {
	267	char option = argv[i][n];
	268
	269	bool option_found=false;
	270
	271	for (int o=0;o<mOptions.size();o++) {
	272	if (mOptions[o]->getShortOption() == option) {
	273	option_found=true;
	274
	275	bool success;
	276	if (is_single_option) {
	277	success = mOptions[o]->processCmdLineArguments(argv,argc, i+1);
	278	}
	279	else {
	280	success = mOptions[o]->processCmdLineArguments(NULL,NULL, 0);
	281	}
	282
	283	if (!success) {
	284	if (first_idx_ptr) { *first_idx_ptr = i; }
	285	return false;
	286	}
	287
	288	break;
	289	}
	290	}
	291
	292	if (!option_found) {
	293	if (!ignore_unknown_options) {
	294	fprintf(stderr, "unknown option -%c\n",option);
	295	return false;
	296	}
	297	else {
	298	do_remove_option=false;
	299	}
	300	}
	301
	302	} // all short options
	303
	304	if (do_remove_option) {
	305	remove_option(argc,argv,i);
	306	i--;
	307	}
	308	} // is short option
	309	} // is option
	310	} // all command line arguments
	311
	312	return true;
	313	}
	314
	315
	316	void config_parameters::print_params() const
	317	{
	318	for (int i=0;i<mOptions.size();i++) {
	319	const option_base* o = mOptions[i];
	320
	321	std::stringstream sstr;
	322	sstr << " ";
	323	if (o->hasShortOption()) {
	324	sstr << '-' << o->getShortOption();
	325	} else {
	326	sstr << " ";
	327	}
	328
	329	if (o->hasShortOption() && o->hasLongOption()) {
	330	sstr << ", ";
	331	} else {
	332	sstr << " ";
	333	}
	334
	335	if (o->hasLongOption()) {
	336	sstr << "--" << std::setw(12) << std::left << o->getLongOption();
	337	} else {
	338	sstr << " ";
	339	}
	340
	341	sstr << " ";
	342	sstr << o->getTypeDescr();
	343
	344	if (o->has_default()) {
	345	sstr << ", default=" << o->get_default_string();
	346	}
	347
	348	sstr << "\n";
	349
	350	std::cerr << sstr.str();
	351	}
	352	}
	353
	354
	355	void config_parameters::add_option(option_base* o)
	356	{
	357	mOptions.push_back(o);
	358	delete[] param_string_table; // delete old table, since we got a new parameter
	359	param_string_table = NULL;
	360	}
	361
	362
	363	std::vector<std::string> config_parameters::get_parameter_IDs() const
	364	{
	365	std::vector<std::string> ids;
	366
	367	#ifdef FOR_LOOP_AUTO_SUPPORT
	368	FOR_LOOP(auto, option, mOptions) {
	369	#else
	370	FOR_LOOP(option_base*, option, mOptions) {
	371	#endif
	372	ids.push_back(option->get_name());
	373	}
	374
	375	return ids;
	376	}
	377
	378
	379	enum en265_parameter_type config_parameters::get_parameter_type(const char* param) const
	380	{
	381	option_base* option = find_option(param);
	382	assert(option);
	383
	384	if (dynamic_cast<option_int*> (option)) { return en265_parameter_int; }
	385	if (dynamic_cast<option_bool*> (option)) { return en265_parameter_bool; }
	386	if (dynamic_cast<option_string*>(option)) { return en265_parameter_string; }
	387	if (dynamic_cast<choice_option_base*>(option)) { return en265_parameter_choice; }
	388
	389	assert(false);
	390	return en265_parameter_bool;
	391	}
	392
	393
	394	std::vector<std::string> config_parameters::get_parameter_choices(const char* param) const
	395	{
	396	option_base* option = find_option(param);
	397	assert(option);
	398
	399	choice_option_base* o = dynamic_cast<choice_option_base*>(option);
	400	assert(o);
	401
	402	return o->get_choice_names();
	403	}
	404
	405
	406	option_base* config_parameters::find_option(const char* param) const
	407	{
	408	#ifdef FOR_LOOP_AUTO_SUPPORT
	409	FOR_LOOP(auto, o, mOptions) {
	410	#else
	411	FOR_LOOP(option_base*, o, mOptions) {
	412	#endif
	413	if (strcmp(o->get_name().c_str(), param)==0) { return o; }
	414	}
	415
	416	return NULL;
	417	}
	418
	419
	420	bool config_parameters::set_bool(const char* param, bool value)
	421	{
	422	option_base* option = find_option(param);
	423	assert(option);
	424
	425	option_bool* o = dynamic_cast<option_bool*>(option);
	426	assert(o);
	427
	428	return o->set(value);
	429	}
	430
	431	bool config_parameters::set_int(const char* param, int value)
	432	{
	433	option_base* option = find_option(param);
	434	assert(option);
	435
	436	option_int* o = dynamic_cast<option_int*>(option);
	437	assert(o);
	438
	439	return o->set(value);
	440	}
	441
	442	bool config_parameters::set_string(const char* param, const char* value)
	443	{
	444	option_base* option = find_option(param);
	445	assert(option);
	446
	447	option_string* o = dynamic_cast<option_string*>(option);
	448	assert(o);
	449
	450	return o->set(value);
	451	}
	452
	453	bool config_parameters::set_choice(const char* param, const char* value)
	454	{
	455	option_base* option = find_option(param);
	456	assert(option);
	457
	458	choice_option_base* o = dynamic_cast<choice_option_base*>(option);
	459	assert(o);
	460
	461	return o->set(value);
	462	}
	463
	464
	465
	466	const char** config_parameters::get_parameter_choices_table(const char* param) const
	467	{
	468	option_base* option = find_option(param);
	469	assert(option);
	470
	471	choice_option_base* o = dynamic_cast<choice_option_base*>(option);
	472	assert(o);
	473
	474	return o->get_choices_string_table();
	475	}
	476
	477	const char** config_parameters::get_parameter_string_table() const
	478	{
	479	if (param_string_table==NULL) {
	480	param_string_table = fill_strings_into_memory(get_parameter_IDs());
	481	}
	482
	483	return (const char**)param_string_table;
	484	}

+400

-0

libde265/configparam.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CONFIG_PARAM_H
	23	#define CONFIG_PARAM_H
	24
	25	#include "en265.h"
	26	#include "util.h"
	27
	28	#include <climits>
	29	#include <vector>
	30	#include <string>
	31	#include <stddef.h>
	32	#include <assert.h>
	33
	34
	35	/* Notes: probably best to keep cmd-line-options here. So it will be:
	36	- automatically consistent even when having different combinations of algorithms
	37	- no other place to edit
	38	- if needed, one can still override it at another place
	39	*/
	40
	41	// TODO: set a stack of default prefixes in config_parameters, such that all options added
	42	// will receive this namespace prefix.
	43
	44	// TODO: add the possibility to remove long options again, i.e., not use the default id name
	45	class option_base
	46	{
	47	public:
	48	option_base() : mShortOption(0), mLongOption(NULL) { }
	49	option_base(const char* name) : mIDName(name), mShortOption(0), mLongOption(NULL) { }
	50	virtual ~option_base() { }
	51
	52
	53	// --- option identifier ---
	54
	55	void set_ID(const char* name) { mIDName=name; }
	56	void add_namespace_prefix(std::string prefix) { mPrefix = prefix + ":" + mPrefix; }
	57
	58	std::string get_name() const { return mPrefix + mIDName; }
	59
	60
	61	// --- description ---
	62
	63	void set_description(std::string descr) { mDescription = descr; }
	64	std::string get_description() const { return mDescription; }
	65
	66
	67	// --- value ---
	68
	69	virtual bool is_defined() const = 0;
	70	bool is_undefined() const { return !is_defined(); }
	71
	72	virtual bool has_default() const = 0;
	73
	74
	75	// --- command line options ----
	76
	77	void set_cmd_line_options(const char* long_option, char short_option = 0)
	78	{
	79	mShortOption = short_option;
	80	mLongOption = long_option;
	81	}
	82
	83	void set_short_option(char short_option) { mShortOption=short_option; }
	84
	85	void unsetCmdLineOption()
	86	{
	87	mShortOption = 0;
	88	mLongOption = NULL;
	89	}
	90
	91	bool hasShortOption() const { return mShortOption!=0; }
	92	char getShortOption() const { return mShortOption; }
	93	bool hasLongOption() const { return true; } //mLongOption!=NULL; }
	94	std::string getLongOption() const { return mLongOption ? std::string(mLongOption) : get_name(); }
	95
	96	virtual LIBDE265_API bool processCmdLineArguments(char** argv, int* argc, int idx) { return false; }
	97
	98
	99
	100	virtual std::string getTypeDescr() const = 0;
	101
	102	virtual std::string get_default_string() const { return "N/A"; }
	103
	104	private:
	105	std::string mPrefix;
	106	std::string mIDName;
	107
	108	std::string mDescription;
	109
	110	char mShortOption;
	111	const char* mLongOption;
	112	};
	113
	114
	115
	116	class option_bool : public option_base
	117	{
	118	public:
	119	option_bool() : value_set(false), default_set(false) { }
	120
	121	operator bool() const {
	122	assert(value_set \|\| default_set);
	123	return value_set ? value : default_value;
	124	}
	125
	126	virtual bool is_defined() const { return value_set \|\| default_set; }
	127	virtual bool has_default() const { return default_set; }
	128
	129	void set_default(bool v) { default_value=v; default_set=true; }
	130	virtual std::string get_default_string() const { return default_value ? "true":"false"; }
	131
	132	virtual std::string getTypeDescr() const { return "boolean"; }
	133	virtual LIBDE265_API bool processCmdLineArguments(char** argv, int* argc, int idx) { value=true; return true; }
	134
	135	bool set(bool v) { value_set=true; value=v; return true; }
	136
	137	private:
	138	bool value_set;
	139	bool value;
	140
	141	bool default_set;
	142	bool default_value;
	143	};
	144
	145
	146	class option_string : public option_base
	147	{
	148	public:
	149	option_string() : value_set(false), default_set(false) { }
	150
	151	const option_string& operator=(std::string v) { value=v; value_set=true; return *this; }
	152
	153	operator std::string() const { return get(); }
	154	std::string get() const {
	155	assert(value_set \|\| default_set);
	156	return value_set ? value : default_value;
	157	}
	158
	159	virtual bool is_defined() const { return value_set \|\| default_set; }
	160	virtual bool has_default() const { return default_set; }
	161
	162	void set_default(std::string v) { default_value=v; default_set=true; }
	163	virtual LIBDE265_API std::string get_default_string() const { return default_value; }
	164
	165	virtual LIBDE265_API std::string getTypeDescr() const { return "(string)"; }
	166	virtual LIBDE265_API bool processCmdLineArguments(char** argv, int* argc, int idx);
	167
	168	bool set(std::string v) { value_set=true; value=v; return true; }
	169
	170	private:
	171	bool value_set;
	172	std::string value;
	173
	174	bool default_set;
	175	std::string default_value;
	176	};
	177
	178
	179	class option_int : public option_base
	180	{
	181	public:
	182	option_int() : value_set(false), default_set(false),
	183	have_low_limit(false), have_high_limit(false) { }
	184
	185	void set_minimum(int mini) { have_low_limit =true; low_limit =mini; }
	186	void set_maximum(int maxi) { have_high_limit=true; high_limit=maxi; }
	187	void set_range(int mini,int maxi);
	188	void set_valid_values(const std::vector<int>& v) { valid_values_set = v; }
	189
	190	const option_int& operator=(int v) { value=v; value_set=true; return *this; }
	191
	192	int operator() () const {
	193	assert(value_set \|\| default_set);
	194	return value_set ? value : default_value;
	195	}
	196	operator int() const { return operator()(); }
	197
	198	virtual bool is_defined() const { return value_set \|\| default_set; }
	199	virtual bool has_default() const { return default_set; }
	200
	201	void set_default(int v) { default_value=v; default_set=true; }
	202	virtual LIBDE265_API std::string get_default_string() const;
	203
	204	virtual LIBDE265_API std::string getTypeDescr() const;
	205	virtual LIBDE265_API bool processCmdLineArguments(char** argv, int* argc, int idx);
	206
	207	bool set(int v) {
	208	if (is_valid(v)) { value_set=true; value=v; return true; }
	209	else { return false; }
	210	}
	211
	212	private:
	213	bool value_set;
	214	int value;
	215
	216	bool default_set;
	217	int default_value;
	218
	219	bool have_low_limit, have_high_limit;
	220	int low_limit, high_limit;
	221
	222	std::vector<int> valid_values_set;
	223
	224	bool is_valid(int v) const;
	225	};
	226
	227
	228
	229	class choice_option_base : public option_base
	230	{
	231	public:
	232	choice_option_base() : choice_string_table(NULL) { }
	233	~choice_option_base() { delete[] choice_string_table; }
	234
	235	bool set(std::string v) { return set_value(v); }
	236	virtual bool set_value(const std::string& val) = 0;
	237	virtual std::vector<std::string> get_choice_names() const = 0;
	238
	239	virtual std::string getTypeDescr() const;
	240	virtual LIBDE265_API bool processCmdLineArguments(char** argv, int* argc, int idx);
	241
	242	const char** get_choices_string_table() const;
	243
	244	protected:
	245	void invalidate_choices_string_table() {
	246	delete[] choice_string_table;
	247	choice_string_table = NULL;
	248	}
	249
	250	private:
	251	mutable char* choice_string_table;
	252	};
	253
	254
	255	template <class T> class choice_option : public choice_option_base
	256	{
	257	public:
	258	choice_option() : default_set(false), value_set(false) { }
	259
	260	// --- initialization ---
	261
	262	void add_choice(const std::string& s, T id, bool default_value=false) {
	263	choices.push_back( std::make_pair(s,id) );
	264	if (default_value) {
	265	defaultID = id;
	266	defaultValue = s;
	267	default_set = true;
	268	}
	269
	270	invalidate_choices_string_table();
	271	}
	272
	273	void set_default(T val) {
	274	#ifdef FOR_LOOP_AUTO_SUPPORT
	275	FOR_LOOP(auto, c, choices) {
	276	#else
	277	for (typename std::vector< std::pair<std::string,T> >::const_iterator it=choices.begin(); it!=choices.end(); ++it) {
	278	const std::pair<std::string,T> & c = *it;
	279	#endif
	280	if (c.second == val) {
	281	defaultID = val;
	282	defaultValue = c.first;
	283	default_set = true;
	284	return;
	285	}
	286	}
	287
	288	assert(false); // value does not exist
	289	}
	290
	291
	292	// --- usage ---
	293
	294	bool set_value(const std::string& val) // returns false if it is not a valid option
	295	{
	296	value_set = true;
	297	selectedValue=val;
	298
	299	validValue = false;
	300
	301	#ifdef FOR_LOOP_AUTO_SUPPORT
	302	FOR_LOOP(auto, c, choices) {
	303	#else
	304	for (typename std::vector< std::pair<std::string,T> >::const_iterator it=choices.begin(); it!=choices.end(); ++it) {
	305	const std::pair<std::string,T> & c = *it;
	306	#endif
	307	if (val == c.first) {
	308	selectedID = c.second;
	309	validValue = true;
	310	}
	311	}
	312
	313	return validValue;
	314	}
	315
	316	bool isValidValue() const { return validValue; }
	317
	318	const std::string& getValue() const {
	319	assert(value_set \|\| default_set);
	320	return value_set ? selectedValue : defaultValue;
	321	}
	322	void setID(T id) { selectedID=id; validValue=true; }
	323	const T getID() const { return value_set ? selectedID : defaultID; }
	324
	325	virtual bool is_defined() const { return value_set \|\| default_set; }
	326	virtual bool has_default() const { return default_set; }
	327
	328	std::vector<std::string> get_choice_names() const
	329	{
	330	std::vector<std::string> names;
	331	#ifdef FOR_LOOP_AUTO_SUPPORT
	332	FOR_LOOP(auto, p, choices) {
	333	#else
	334	for (typename std::vector< std::pair<std::string,T> >::const_iterator it=choices.begin(); it!=choices.end(); ++it) {
	335	const std::pair<std::string,T> & p = *it;
	336	#endif
	337	names.push_back(p.first);
	338	}
	339	return names;
	340	}
	341
	342	std::string get_default_string() const { return defaultValue; }
	343
	344	T operator() () const { return (T)getID(); }
	345
	346	private:
	347	std::vector< std::pair<std::string,T> > choices;
	348
	349	bool default_set;
	350	std::string defaultValue;
	351	T defaultID;
	352
	353	bool value_set;
	354	std::string selectedValue;
	355	T selectedID;
	356
	357	bool validValue;
	358	};
	359
	360
	361
	362
	363	class config_parameters
	364	{
	365	public:
	366	config_parameters() : param_string_table(NULL) { }
	367	~config_parameters() { delete[] param_string_table; }
	368
	369	void LIBDE265_API add_option(option_base* o);
	370
	371	void LIBDE265_API print_params() const;
	372	bool LIBDE265_API parse_command_line_params(int* argc, char** argv, int* first_idx=NULL,
	373	bool ignore_unknown_options=false);
	374
	375
	376	// --- connection to C API ---
	377
	378	std::vector<std::string> get_parameter_IDs() const;
	379	enum en265_parameter_type get_parameter_type(const char* param) const;
	380
	381	std::vector<std::string> get_parameter_choices(const char* param) const;
	382
	383	bool set_bool(const char* param, bool value);
	384	bool set_int(const char* param, int value);
	385	bool set_string(const char* param, const char* value);
	386	bool set_choice(const char* param, const char* value);
	387
	388	const char** get_parameter_string_table() const;
	389	const char** get_parameter_choices_table(const char* param) const;
	390
	391	private:
	392	std::vector<option_base*> mOptions;
	393
	394	option_base* find_option(const char* param) const;
	395
	396	mutable char* param_string_table;
	397	};
	398
	399	#endif

+347

-0

libde265/contextmodel.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "slice.h"
	21	#include <assert.h>
	22	#include <iomanip>
	23	#include <sstream>
	24
	25	bool D = false;
	26
	27	context_model_table::context_model_table()
	28	: model(NULL), refcnt(NULL)
	29	{
	30	}
	31
	32
	33	context_model_table::context_model_table(const context_model_table& src)
	34	{
	35	if (D) printf("%p c'tor = %p\n",this,&src);
	36
	37	if (src.refcnt) {
	38	(*(src.refcnt))++;
	39	}
	40
	41	refcnt = src.refcnt;
	42	model = src.model;
	43	}
	44
	45
	46	context_model_table::~context_model_table()
	47	{
	48	if (D) printf("%p destructor\n",this);
	49
	50	if (refcnt) {
	51	(*refcnt)--;
	52	if (*refcnt==0) {
	53	if (D) printf("mfree %p\n",model);
	54	delete[] model;
	55	delete refcnt;
	56	}
	57	}
	58	}
	59
	60
	61	void context_model_table::init(int initType, int QPY)
	62	{
	63	if (D) printf("%p init\n",this);
	64
	65	decouple_or_alloc_with_empty_data();
	66
	67	initialize_CABAC_models(model, initType, QPY);
	68	}
	69
	70
	71	void context_model_table::release()
	72	{
	73	if (D) printf("%p release %p\n",this,refcnt);
	74
	75	if (!refcnt) { return; }
	76
	77	// if (*refcnt == 1) { return; } <- keep memory for later, but does not work when we believe that we freed the memory and nulled all references
	78
	79	(*refcnt)--;
	80	if (refcnt==0) {
	81	delete[] model;
	82	delete refcnt;
	83	}
	84
	85	model = nullptr;
	86	refcnt= nullptr;
	87	}
	88
	89
	90	void context_model_table::decouple()
	91	{
	92	if (D) printf("%p decouple (%p)\n",this,refcnt);
	93
	94	assert(refcnt); // not necessarily so, but we never use it on an unitialized object
	95
	96	if (*refcnt > 1) {
	97	(*refcnt)--;
	98
	99	context_model* oldModel = model;
	100
	101	model = new context_model[CONTEXT_MODEL_TABLE_LENGTH];
	102	refcnt= new int;
	103	*refcnt=1;
	104
	105	memcpy(model,oldModel,sizeof(context_model)*CONTEXT_MODEL_TABLE_LENGTH);
	106	}
	107	}
	108
	109
	110	context_model_table context_model_table::transfer()
	111	{
	112	context_model_table newtable;
	113	newtable.model = model;
	114	newtable.refcnt= refcnt;
	115
	116	model =nullptr;
	117	refcnt=nullptr;
	118
	119	return newtable;
	120	}
	121
	122
	123	context_model_table& context_model_table::operator=(const context_model_table& src)
	124	{
	125	if (D) printf("%p assign = %p\n",this,&src);
	126
	127	// assert(src.refcnt); // not necessarily so, but we never use it on an unitialized object
	128
	129	if (!src.refcnt) {
	130	release();
	131	return *this;
	132	}
	133
	134	(*(src.refcnt))++;
	135
	136	release();
	137
	138	model = src.model;
	139	refcnt= src.refcnt;
	140
	141	return *this;
	142	}
	143
	144
	145	bool context_model_table::operator==(const context_model_table& b) const
	146	{
	147	if (b.model == model) return true;
	148	if (b.model == nullptr \|\| model == nullptr) return false;
	149
	150	for (int i=0;i<CONTEXT_MODEL_TABLE_LENGTH;i++) {
	151	if (!(b.model[i] == model[i])) return false;
	152	}
	153
	154	return true;
	155	}
	156
	157
	158	std::string context_model_table::debug_dump() const
	159	{
	160	int hash = 0;
	161	for (int i=0;i<CONTEXT_MODEL_TABLE_LENGTH;i++) {
	162	hash ^= ((i+7)*model[i].state) & 0xFFFF;
	163	}
	164
	165	std::stringstream sstr;
	166	sstr << std::hex << hash;
	167	return sstr.str();
	168	}
	169
	170
	171	void context_model_table::decouple_or_alloc_with_empty_data()
	172	{
	173	if (refcnt && *refcnt==1) { return; }
	174
	175	if (refcnt) {
	176	assert(*refcnt>1);
	177	(*refcnt)--;
	178	}
	179
	180	if (D) printf("%p (alloc)\n",this);
	181
	182	model = new context_model[CONTEXT_MODEL_TABLE_LENGTH];
	183	refcnt= new int;
	184	*refcnt=1;
	185	}
	186
	187
	188
	189
	190
	191
	192	static void set_initValue(int SliceQPY,
	193	context_model* model, int initValue, int nContexts)
	194	{
	195	int slopeIdx = initValue >> 4;
	196	int intersecIdx = initValue & 0xF;
	197	int m = slopeIdx*5 - 45;
	198	int n = (intersecIdx<<3) - 16;
	199	int preCtxState = Clip3(1,126, ((m*Clip3(0,51, SliceQPY))>>4)+n);
	200
	201	// logtrace(LogSlice,"QP=%d slopeIdx=%d intersecIdx=%d m=%d n=%d\n",SliceQPY,slopeIdx,intersecIdx,m,n);
	202
	203	for (int i=0;i<nContexts;i++) {
	204	model[i].MPSbit=(preCtxState<=63) ? 0 : 1;
	205	model[i].state = model[i].MPSbit ? (preCtxState-64) : (63-preCtxState);
	206
	207	// model state will always be between [0;62]
	208
	209	assert(model[i].state <= 62);
	210	}
	211	}
	212
	213
	214	static const int initValue_split_cu_flag[3][3] = {
	215	{ 139,141,157 },
	216	{ 107,139,126 },
	217	{ 107,139,126 },
	218	};
	219	static const int initValue_cu_skip_flag[2][3] = {
	220	{ 197,185,201 },
	221	{ 197,185,201 },
	222	};
	223	static const int initValue_part_mode[9] = { 184,154,139, 154,154,154, 139,154,154 };
	224	static const int initValue_prev_intra_luma_pred_flag[3] = { 184,154,183 };
	225	static const int initValue_intra_chroma_pred_mode[3] = { 63,152,152 };
	226	static const int initValue_cbf_luma[4] = { 111,141,153,111 };
	227	static const int initValue_cbf_chroma[12] = { 94,138,182,154,149,107,167,154,149,92,167,154 };
	228	static const int initValue_split_transform_flag[9] = { 153,138,138, 124,138,94, 224,167,122 }; // FIX712
	229	static const int initValue_last_significant_coefficient_prefix[54] = {
	230	110,110,124,125,140,153,125,127,140,109,111,143,127,111, 79,108,123, 63,
	231	125,110, 94,110, 95, 79,125,111,110, 78,110,111,111, 95, 94,108,123,108,
	232	125,110,124,110, 95, 94,125,111,111, 79,125,126,111,111, 79,108,123, 93
	233	};
	234	static const int initValue_coded_sub_block_flag[12] = { 91,171,134,141,121,140,61,154,121,140,61,154 };
	235	static const int initValue_significant_coeff_flag[3][42] = {
	236	{
	237	111, 111, 125, 110, 110, 94, 124, 108, 124, 107, 125, 141, 179, 153, 125, 107,
	238	125, 141, 179, 153, 125, 107, 125, 141, 179, 153, 125, 140, 139, 182, 182, 152,
	239	136, 152, 136, 153, 136, 139, 111, 136, 139, 111
	240	},
	241	{
	242	155, 154, 139, 153, 139, 123, 123, 63, 153, 166, 183, 140, 136, 153, 154, 166,
	243	183, 140, 136, 153, 154, 166, 183, 140, 136, 153, 154, 170, 153, 123, 123, 107,
	244	121, 107, 121, 167, 151, 183, 140, 151, 183, 140,
	245	},
	246	{
	247	170, 154, 139, 153, 139, 123, 123, 63, 124, 166, 183, 140, 136, 153, 154, 166,
	248	183, 140, 136, 153, 154, 166, 183, 140, 136, 153, 154, 170, 153, 138, 138, 122,
	249	121, 122, 121, 167, 151, 183, 140, 151, 183, 140
	250	},
	251	};
	252	static const int initValue_significant_coeff_flag_skipmode[3][2] = {
	253	{ 141,111 }, { 140,140 }, { 140,140 }
	254	};
	255
	256	static const int initValue_coeff_abs_level_greater1_flag[72] = {
	257	140, 92,137,138,140,152,138,139,153, 74,149, 92,139,107,122,152,
	258	140,179,166,182,140,227,122,197,154,196,196,167,154,152,167,182,
	259	182,134,149,136,153,121,136,137,169,194,166,167,154,167,137,182,
	260	154,196,167,167,154,152,167,182,182,134,149,136,153,121,136,122,
	261	169,208,166,167,154,152,167,182
	262	};
	263	static const int initValue_coeff_abs_level_greater2_flag[18] = {
	264	138,153,136,167,152,152,107,167, 91,122,107,167,
	265	107,167, 91,107,107,167
	266	};
	267	static const int initValue_sao_merge_leftUp_flag[3] = { 153,153,153 };
	268	static const int initValue_sao_type_idx_lumaChroma_flag[3] = { 200,185,160 };
	269	static const int initValue_cu_qp_delta_abs[2] = { 154,154 };
	270	static const int initValue_transform_skip_flag[2] = { 139,139 };
	271	static const int initValue_merge_flag[2] = { 110,154 };
	272	static const int initValue_merge_idx[2] = { 122,137 };
	273	static const int initValue_pred_mode_flag[2] = { 149,134 };
	274	static const int initValue_abs_mvd_greater01_flag[4] = { 140,198,169,198 };
	275	static const int initValue_mvp_lx_flag[1] = { 168 };
	276	static const int initValue_rqt_root_cbf[1] = { 79 };
	277	static const int initValue_ref_idx_lX[2] = { 153,153 };
	278	static const int initValue_inter_pred_idc[5] = { 95,79,63,31,31 };
	279	static const int initValue_cu_transquant_bypass_flag[3] = { 154,154,154 };
	280
	281
	282	static void init_context(int SliceQPY,
	283	context_model* model,
	284	const int* initValues, int len)
	285	{
	286	for (int i=0;i<len;i++)
	287	{
	288	set_initValue(SliceQPY, &model[i], initValues[i], 1);
	289	}
	290	}
	291
	292
	293	static void init_context_const(int SliceQPY,
	294	context_model* model,
	295	int initValue, int len)
	296	{
	297	set_initValue(SliceQPY, model, initValue, len);
	298	}
	299
	300	void initialize_CABAC_models(context_model context_model_table[CONTEXT_MODEL_TABLE_LENGTH],
	301	int initType,
	302	int QPY)
	303	{
	304	context_model* cm = context_model_table; // just an abbreviation
	305
	306	if (initType > 0) {
	307	init_context(QPY, cm+CONTEXT_MODEL_CU_SKIP_FLAG, initValue_cu_skip_flag[initType-1], 3);
	308	init_context(QPY, cm+CONTEXT_MODEL_PRED_MODE_FLAG, &initValue_pred_mode_flag[initType-1], 1);
	309	init_context(QPY, cm+CONTEXT_MODEL_MERGE_FLAG, &initValue_merge_flag[initType-1],1);
	310	init_context(QPY, cm+CONTEXT_MODEL_MERGE_IDX, &initValue_merge_idx[initType-1], 1);
	311	init_context(QPY, cm+CONTEXT_MODEL_INTER_PRED_IDC, initValue_inter_pred_idc, 5);
	312	init_context(QPY, cm+CONTEXT_MODEL_REF_IDX_LX, initValue_ref_idx_lX, 2);
	313	init_context(QPY, cm+CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG, &initValue_abs_mvd_greater01_flag[initType == 1 ? 0 : 2], 2);
	314	init_context(QPY, cm+CONTEXT_MODEL_MVP_LX_FLAG, initValue_mvp_lx_flag, 1);
	315	init_context(QPY, cm+CONTEXT_MODEL_RQT_ROOT_CBF, initValue_rqt_root_cbf, 1);
	316
	317	init_context_const(QPY, cm+CONTEXT_MODEL_RDPCM_FLAG, 139, 2);
	318	init_context_const(QPY, cm+CONTEXT_MODEL_RDPCM_DIR, 139, 2);
	319	}
	320
	321	init_context(QPY, cm+CONTEXT_MODEL_SPLIT_CU_FLAG, initValue_split_cu_flag[initType], 3);
	322	init_context(QPY, cm+CONTEXT_MODEL_PART_MODE, &initValue_part_mode[(initType!=2 ? initType : 5)], 4);
	323	init_context(QPY, cm+CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG, &initValue_prev_intra_luma_pred_flag[initType], 1);
	324	init_context(QPY, cm+CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE, &initValue_intra_chroma_pred_mode[initType], 1);
	325	init_context(QPY, cm+CONTEXT_MODEL_CBF_LUMA, &initValue_cbf_luma[initType == 0 ? 0 : 2], 2);
	326	init_context(QPY, cm+CONTEXT_MODEL_CBF_CHROMA, &initValue_cbf_chroma[initType * 4], 4);
	327	init_context(QPY, cm+CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG, &initValue_split_transform_flag[initType * 3], 3);
	328	init_context(QPY, cm+CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX, &initValue_last_significant_coefficient_prefix[initType * 18], 18);
	329	init_context(QPY, cm+CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX, &initValue_last_significant_coefficient_prefix[initType * 18], 18);
	330	init_context(QPY, cm+CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG, &initValue_coded_sub_block_flag[initType * 4], 4);
	331	init_context(QPY, cm+CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG, initValue_significant_coeff_flag[initType], 42);
	332	init_context(QPY, cm+CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG+42, initValue_significant_coeff_flag_skipmode[initType], 2);
	333
	334	init_context(QPY, cm+CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG, &initValue_coeff_abs_level_greater1_flag[initType * 24], 24);
	335	init_context(QPY, cm+CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG, &initValue_coeff_abs_level_greater2_flag[initType * 6], 6);
	336	init_context(QPY, cm+CONTEXT_MODEL_SAO_MERGE_FLAG, &initValue_sao_merge_leftUp_flag[initType], 1);
	337	init_context(QPY, cm+CONTEXT_MODEL_SAO_TYPE_IDX, &initValue_sao_type_idx_lumaChroma_flag[initType], 1);
	338	init_context(QPY, cm+CONTEXT_MODEL_CU_QP_DELTA_ABS, initValue_cu_qp_delta_abs, 2);
	339	init_context(QPY, cm+CONTEXT_MODEL_TRANSFORM_SKIP_FLAG, initValue_transform_skip_flag, 2);
	340	init_context(QPY, cm+CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG, &initValue_cu_transquant_bypass_flag[initType], 1);
	341
	342	init_context_const(QPY, cm+CONTEXT_MODEL_LOG2_RES_SCALE_ABS_PLUS1, 154, 8);
	343	init_context_const(QPY, cm+CONTEXT_MODEL_RES_SCALE_SIGN_FLAG, 154, 2);
	344	init_context_const(QPY, cm+CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_FLAG, 154, 1);
	345	init_context_const(QPY, cm+CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_IDX, 154, 1);
	346	}

+130

-0

libde265/contextmodel.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	* Min Chen <chenm003@163.com>
	6	*
	7	* This file is part of libde265.
	8	*
	9	* libde265 is free software: you can redistribute it and/or modify
	10	* it under the terms of the GNU Lesser General Public License as
	11	* published by the Free Software Foundation, either version 3 of
	12	* the License, or (at your option) any later version.
	13	*
	14	* libde265 is distributed in the hope that it will be useful,
	15	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	16	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	17	* GNU Lesser General Public License for more details.
	18	*
	19	* You should have received a copy of the GNU Lesser General Public License
	20	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	21	*/
	22
	23	#ifndef DE265_CONTEXTMODEL_H
	24	#define DE265_CONTEXTMODEL_H
	25
	26	#include "libde265/cabac.h"
	27	#include "libde265/de265.h"
	28
	29	#include <string.h>
	30	#include <string>
	31
	32
	33	struct context_model {
	34	uint8_t MPSbit : 1;
	35	uint8_t state : 7;
	36
	37	bool operator==(context_model b) const { return state==b.state && MPSbit==b.MPSbit; }
	38	bool operator!=(context_model b) const { return state!=b.state \|\| MPSbit!=b.MPSbit; }
	39	};
	40
	41
	42	enum context_model_index {
	43	// SAO
	44	CONTEXT_MODEL_SAO_MERGE_FLAG = 0,
	45	CONTEXT_MODEL_SAO_TYPE_IDX = CONTEXT_MODEL_SAO_MERGE_FLAG +1,
	46
	47	// CB-tree
	48	CONTEXT_MODEL_SPLIT_CU_FLAG = CONTEXT_MODEL_SAO_TYPE_IDX + 1,
	49	CONTEXT_MODEL_CU_SKIP_FLAG = CONTEXT_MODEL_SPLIT_CU_FLAG + 3,
	50
	51	// intra-prediction
	52	CONTEXT_MODEL_PART_MODE = CONTEXT_MODEL_CU_SKIP_FLAG + 3,
	53	CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG = CONTEXT_MODEL_PART_MODE + 4,
	54	CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE = CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG + 1,
	55
	56	// transform-tree
	57	CONTEXT_MODEL_CBF_LUMA = CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE + 1,
	58	CONTEXT_MODEL_CBF_CHROMA = CONTEXT_MODEL_CBF_LUMA + 2,
	59	CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG = CONTEXT_MODEL_CBF_CHROMA + 4,
	60	CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_FLAG = CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG + 3,
	61	CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_IDX = CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_FLAG + 1,
	62
	63	// residual
	64	CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX = CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_IDX + 1,
	65	CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX = CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX + 18,
	66	CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG = CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX + 18,
	67	CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG = CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG + 4,
	68	CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG = CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG + 42+2,
	69	CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG = CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG + 24,
	70
	71	CONTEXT_MODEL_CU_QP_DELTA_ABS = CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG + 6,
	72	CONTEXT_MODEL_TRANSFORM_SKIP_FLAG = CONTEXT_MODEL_CU_QP_DELTA_ABS + 2,
	73	CONTEXT_MODEL_RDPCM_FLAG = CONTEXT_MODEL_TRANSFORM_SKIP_FLAG + 2,
	74	CONTEXT_MODEL_RDPCM_DIR = CONTEXT_MODEL_RDPCM_FLAG + 2,
	75
	76	// motion
	77	CONTEXT_MODEL_MERGE_FLAG = CONTEXT_MODEL_RDPCM_DIR + 2,
	78	CONTEXT_MODEL_MERGE_IDX = CONTEXT_MODEL_MERGE_FLAG + 1,
	79	CONTEXT_MODEL_PRED_MODE_FLAG = CONTEXT_MODEL_MERGE_IDX + 1,
	80	CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG = CONTEXT_MODEL_PRED_MODE_FLAG + 1,
	81	CONTEXT_MODEL_MVP_LX_FLAG = CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG + 2,
	82	CONTEXT_MODEL_RQT_ROOT_CBF = CONTEXT_MODEL_MVP_LX_FLAG + 1,
	83	CONTEXT_MODEL_REF_IDX_LX = CONTEXT_MODEL_RQT_ROOT_CBF + 1,
	84	CONTEXT_MODEL_INTER_PRED_IDC = CONTEXT_MODEL_REF_IDX_LX + 2,
	85	CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG = CONTEXT_MODEL_INTER_PRED_IDC + 5,
	86	CONTEXT_MODEL_LOG2_RES_SCALE_ABS_PLUS1 = CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG + 1,
	87	CONTEXT_MODEL_RES_SCALE_SIGN_FLAG = CONTEXT_MODEL_LOG2_RES_SCALE_ABS_PLUS1 + 8,
	88	CONTEXT_MODEL_TABLE_LENGTH = CONTEXT_MODEL_RES_SCALE_SIGN_FLAG + 2
	89	};
	90
	91
	92
	93	void initialize_CABAC_models(context_model context_model_table[CONTEXT_MODEL_TABLE_LENGTH],
	94	int initType,
	95	int QPY);
	96
	97
	98	class context_model_table
	99	{
	100	public:
	101	context_model_table();
	102	context_model_table(const context_model_table&);
	103	~context_model_table();
	104
	105	void init(int initType, int QPY);
	106	void release();
	107	void decouple();
	108	context_model_table transfer();
	109	context_model_table copy() const { context_model_table t=*this; t.decouple(); return t; }
	110
	111	bool empty() const { return refcnt != NULL; }
	112
	113	context_model& operator[](int i) { return model[i]; }
	114
	115	context_model_table& operator=(const context_model_table&);
	116
	117	bool operator==(const context_model_table&) const;
	118
	119	std::string debug_dump() const;
	120
	121	private:
	122	void decouple_or_alloc_with_empty_data();
	123
	124	context_model* model; // [CONTEXT_MODEL_TABLE_LENGTH]
	125	int* refcnt;
	126	};
	127
	128
	129	#endif

-2

libde265/de265-version.h less more

27	27	#define LIBDE265_VERSION_H
28	28
29	29	/* Numeric representation of the version */
30		#define LIBDE265_NUMERIC_VERSION 0x00090000
	30	#define LIBDE265_NUMERIC_VERSION 0x01000200
31	31
32	32	/* Version string */
33		#define LIBDE265_VERSION "0.9"
	33	#define LIBDE265_VERSION "1.0.2"
34	34
35	35	#endif

+50

-7

libde265/de265.cc less more

46	46	return (LIBDE265_NUMERIC_VERSION);
47	47	}
48	48
	49	LIBDE265_API int de265_get_version_number_major(void)
	50	{
	51	return ((LIBDE265_NUMERIC_VERSION)>>24) & 0xFF;
	52	}
	53
	54	LIBDE265_API int de265_get_version_number_minor(void)
	55	{
	56	return ((LIBDE265_NUMERIC_VERSION)>>16) & 0xFF;
	57	}
	58
	59	LIBDE265_API int de265_get_version_number_maintenance(void)
	60	{
	61	return ((LIBDE265_NUMERIC_VERSION)>>8) & 0xFF;
	62	}
	63
	64
49	65	LIBDE265_API const char* de265_get_error_text(de265_error err)
50	66	{
51	67	switch (err) {
52	68	case DE265_OK: return "no error";
53	69	case DE265_ERROR_NO_SUCH_FILE: return "no such file";
54	70	//case DE265_ERROR_NO_STARTCODE: return "no startcode found";
55		case DE265_ERROR_EOF: return "end of file";
	71	//case DE265_ERROR_EOF: return "end of file";
56	72	case DE265_ERROR_COEFFICIENT_OUT_OF_IMAGE_BOUNDS: return "coefficient out of image bounds";
57	73	case DE265_ERROR_CHECKSUM_MISMATCH: return "image checksum mismatch";
58	74	case DE265_ERROR_CTB_OUTSIDE_IMAGE_AREA: return "CTB outside of image area";

63	79	case DE265_ERROR_LIBRARY_INITIALIZATION_FAILED: return "global library initialization failed";
64	80	case DE265_ERROR_LIBRARY_NOT_INITIALIZED: return "cannot free library data (not initialized";
65	81
66		case DE265_ERROR_MAX_THREAD_CONTEXTS_EXCEEDED:
67		return "internal error: maximum number of thread contexts exceeded";
68		case DE265_ERROR_MAX_NUMBER_OF_SLICES_EXCEEDED:
69		return "internal error: maximum number of slices exceeded";
	82	//case DE265_ERROR_MAX_THREAD_CONTEXTS_EXCEEDED:
	83	// return "internal error: maximum number of thread contexts exceeded";
	84	//case DE265_ERROR_MAX_NUMBER_OF_SLICES_EXCEEDED:
	85	// return "internal error: maximum number of slices exceeded";
	86	case DE265_ERROR_NOT_IMPLEMENTED_YET:
	87	return "unimplemented decoder feature";
70	88	//case DE265_ERROR_SCALING_LIST_NOT_IMPLEMENTED:
71	89	//return "scaling list not implemented";
	90
72	91	case DE265_ERROR_WAITING_FOR_INPUT_DATA:
73	92	return "no more input data, decoder stalled";
74	93	case DE265_ERROR_CANNOT_PROCESS_SEI:
75	94	return "SEI data cannot be processed";
	95	case DE265_ERROR_PARAMETER_PARSING:
	96	return "command-line parameter error";
	97	case DE265_ERROR_NO_INITIAL_SLICE_HEADER:
	98	return "first slice missing, cannot decode dependent slice";
	99	case DE265_ERROR_PREMATURE_END_OF_SLICE:
	100	return "premature end of slice data";
	101	case DE265_ERROR_UNSPECIFIED_DECODING_ERROR:
	102	return "unspecified decoding error";
76	103
77	104	case DE265_WARNING_NO_WPP_CANNOT_USE_MULTITHREADING:
78	105	return "Cannot run decoder multi-threaded because stream does not support WPP";

126	153	return "cannot apply SAO because we ran out of memory";
127	154	case DE265_WARNING_SPS_MISSING_CANNOT_DECODE_SEI:
128	155	return "SPS header missing, cannot decode SEI";
	156	case DE265_WARNING_COLLOCATED_MOTION_VECTOR_OUTSIDE_IMAGE_AREA:
	157	return "collocated motion-vector is outside image area";
129	158
130	159	default: return "unknown error";
131	160	}

586	615	}
587	616	}
588	617
	618	LIBDE265_API int de265_get_bits_per_pixel(const struct de265_image* img,int channel)
	619	{
	620	switch (channel) {
	621	case 0:
	622	return img->sps.BitDepth_Y;
	623	case 1:
	624	case 2:
	625	return img->sps.BitDepth_C;
	626	default:
	627	return 0;
	628	}
	629	}
	630
589	631	LIBDE265_API enum de265_chroma de265_get_chroma_format(const struct de265_image* img)
590	632	{
591	633	return img->get_chroma_format();

597	639
598	640	uint8_t* data = img->pixels_confwin[channel];
599	641
600		if (stride) *stride = img->get_image_stride(channel);
	642	if (stride) stride = img->get_image_stride(channel) ((de265_get_bits_per_pixel(img, channel)+7) / 8);
601	643
602	644	return data;
603	645	}

611	653
612	654	LIBDE265_API void de265_set_image_plane(de265_image* img, int cIdx, void* mem, int stride, void *userdata)
613	655	{
	656	// The internal "stride" is the number of pixels per line.
	657	stride = stride / ((de265_get_bits_per_pixel(img, cIdx)+7) / 8);
614	658	img->set_image_plane(cIdx, (uint8_t*)mem, stride, userdata);
615	659	}
616	660

655	699	if (nuh_temporal_id) *nuh_temporal_id = img->nal_hdr.nuh_temporal_id;
656	700	}
657	701	}
658

+21

-7

libde265/de265.h less more

71	71	LIBDE265_API const char *de265_get_version(void);
72	72	LIBDE265_API uint32_t de265_get_version_number(void);
73	73
	74	LIBDE265_API int de265_get_version_number_major(void);
	75	LIBDE265_API int de265_get_version_number_minor(void);
	76	LIBDE265_API int de265_get_version_number_maintenance(void);
	77
74	78
75	79	/* === error codes === */
76	80

78	82	DE265_OK = 0,
79	83	DE265_ERROR_NO_SUCH_FILE=1,
80	84	//DE265_ERROR_NO_STARTCODE=2, obsolet
81		DE265_ERROR_EOF=3,
	85	//DE265_ERROR_EOF=3,
82	86	DE265_ERROR_COEFFICIENT_OUT_OF_IMAGE_BOUNDS=4,
83	87	DE265_ERROR_CHECKSUM_MISMATCH=5,
84	88	DE265_ERROR_CTB_OUTSIDE_IMAGE_AREA=6,

90	94	DE265_ERROR_LIBRARY_NOT_INITIALIZED=12,
91	95	DE265_ERROR_WAITING_FOR_INPUT_DATA=13,
92	96	DE265_ERROR_CANNOT_PROCESS_SEI=14,
	97	DE265_ERROR_PARAMETER_PARSING=15,
	98	DE265_ERROR_NO_INITIAL_SLICE_HEADER=16,
	99	DE265_ERROR_PREMATURE_END_OF_SLICE=17,
	100	DE265_ERROR_UNSPECIFIED_DECODING_ERROR=18,
93	101
94	102	// --- errors that should become obsolete in later libde265 versions ---
95	103
96		DE265_ERROR_MAX_THREAD_CONTEXTS_EXCEEDED = 500,
97		DE265_ERROR_MAX_NUMBER_OF_SLICES_EXCEEDED = 501,
	104	//DE265_ERROR_MAX_THREAD_CONTEXTS_EXCEEDED = 500, obsolet
	105	//DE265_ERROR_MAX_NUMBER_OF_SLICES_EXCEEDED = 501, obsolet
	106	DE265_ERROR_NOT_IMPLEMENTED_YET = 502,
98	107	//DE265_ERROR_SCALING_LIST_NOT_IMPLEMENTED = 502, obsolet
99	108
100	109	// --- warnings ---

124	133	DE265_WARNING_NUMBER_OF_THREADS_LIMITED_TO_MAXIMUM=1022,
125	134	DE265_NON_EXISTING_LT_REFERENCE_CANDIDATE_IN_SLICE_HEADER=1023,
126	135	DE265_WARNING_CANNOT_APPLY_SAO_OUT_OF_MEMORY=1024,
127		DE265_WARNING_SPS_MISSING_CANNOT_DECODE_SEI=1025
	136	DE265_WARNING_SPS_MISSING_CANNOT_DECODE_SEI=1025,
	137	DE265_WARNING_COLLOCATED_MOTION_VECTOR_OUTSIDE_IMAGE_AREA=1026
128	138	} de265_error;
129	139
130	140	LIBDE265_API const char* de265_get_error_text(de265_error err);

147	157
148	158	enum de265_chroma {
149	159	de265_chroma_mono=0,
150		de265_chroma_420=1, // currently the only used format
	160	de265_chroma_420=1,
151	161	de265_chroma_422=2,
152	162	de265_chroma_444=3
153	163	};

158	168	LIBDE265_API int de265_get_image_width(const struct de265_image*,int channel);
159	169	LIBDE265_API int de265_get_image_height(const struct de265_image*,int channel);
160	170	LIBDE265_API enum de265_chroma de265_get_chroma_format(const struct de265_image*);
	171	LIBDE265_API int de265_get_bits_per_pixel(const struct de265_image*,int channel);
	172	/* The \|out_stride\| is returned as "bytes per line" if a non-NULL parameter is given. */
161	173	LIBDE265_API const uint8_t* de265_get_image_plane(const struct de265_image, int channel, int out_stride);
162	174	LIBDE265_API void* de265_get_image_plane_user_data(const struct de265_image*, int channel);
163	175	LIBDE265_API de265_PTS de265_get_image_PTS(const struct de265_image*);

311	323
312	324	struct de265_image_allocation
313	325	{
314		int (get_buffer)(de265_decoder_context ctx,
	326	int (get_buffer)(de265_decoder_context ctx, // first parameter deprecated
315	327	struct de265_image_spec* spec,
316	328	struct de265_image* img,
317	329	void* userdata);
318		void (release_buffer)(de265_decoder_context ctx,
	330	void (release_buffer)(de265_decoder_context ctx, // first parameter deprecated
319	331	struct de265_image* img,
320	332	void* userdata);
321	333	};

383	395	de265_acceleration_SSE4 = 40,
384	396	de265_acceleration_AVX = 50, // not implemented yet
385	397	de265_acceleration_AVX2 = 60, // not implemented yet
	398	de265_acceleration_ARM = 70,
	399	de265_acceleration_NEON = 80,
386	400	de265_acceleration_AUTO = 10000
387	401	};
388	402

+143

-68

libde265/deblock.cc less more

159	159	int x0ctb = x0 >> ctbshift;
160	160	int y0ctb = y0 >> ctbshift;
161	161
	162	// check for corrupted streams
	163	if (img->is_SliceHeader_available(x0,y0)==false) {
	164	return false;
	165	}
	166
162	167	// check whether we should filter this slice
163	168
164	169	slice_segment_header* shdr = img->get_SliceHeader(x0,y0);

174	179
175	180	if (x0 && ((x0 & ctb_mask) == 0)) { // left edge at CTB boundary
176	181	if (shdr->slice_loop_filter_across_slices_enabled_flag == 0 &&
	182	img->is_SliceHeader_available(x0-1,y0) && // for corrupted streams
177	183	shdr->SliceAddrRS != img->get_SliceHeader(x0-1,y0)->SliceAddrRS)
178	184	{
179	185	filterLeftCbEdge = 0;

187	193
188	194	if (y0 && ((y0 & ctb_mask) == 0)) { // top edge at CTB boundary
189	195	if (shdr->slice_loop_filter_across_slices_enabled_flag == 0 &&
	196	img->is_SliceHeader_available(x0,y0-1) && // for corrupted streams
190	197	shdr->SliceAddrRS != img->get_SliceHeader(x0,y0-1)->SliceAddrRS)
191	198	{
192	199	filterTopCbEdge = 0;

280	287
281	288	bS = 0;
282	289
283		const PredVectorInfo* mviP = img->get_mv_info(xDiOpp,yDiOpp);
284		const PredVectorInfo* mviQ = img->get_mv_info(xDi ,yDi);
	290	const MotionVectorSpec* mviP = img->get_mv_info(xDiOpp,yDiOpp);
	291	const MotionVectorSpec* mviQ = img->get_mv_info(xDi ,yDi);
285	292
286	293	slice_segment_header* shdrP = img->get_SliceHeader(xDiOpp,yDiOpp);
287	294	slice_segment_header* shdrQ = img->get_SliceHeader(xDi ,yDi);

392	399
393	400
394	401	// 8.7.2.4
395		void edge_filtering_luma(de265_image* img, bool vertical,
396		int yStart,int yEnd, int xStart,int xEnd)
397		{
	402	template <class pixel_t>
	403	void edge_filtering_luma_internal(de265_image* img, bool vertical,
	404	int yStart,int yEnd, int xStart,int xEnd)
	405	{
	406	//printf("luma %d-%d %d-%d\n",xStart,xEnd,yStart,yEnd);
	407
398	408	int xIncr = vertical ? 2 : 1;
399	409	int yIncr = vertical ? 1 : 2;
400	410

407	417
408	418	for (int y=yStart;y<yEnd;y+=yIncr)
409	419	for (int x=xStart;x<xEnd;x+=xIncr) {
410		int xDi = x<<2;
411		int yDi = y<<2;
	420	// x;y in deblocking units (4x4 pixels)
	421
	422	int xDi = x<<2; // *4 -> pixel resolution
	423	int yDi = y<<2; // *4 -> pixel resolution
412	424	int bS = img->get_deblk_bS(xDi,yDi);
	425
	426	//printf("x,y:%d,%d xDi,yDi:%d,%d\n",x,y,xDi,yDi);
413	427
414	428	logtrace(LogDeblock,"deblock POC=%d %c --- x:%d y:%d bS:%d---\n",
415	429	img->PicOrderCntVal,vertical ? 'V':'H',xDi,yDi,bS);

449	463
450	464	// 8.7.2.4.3
451	465
452		uint8_t* ptr = img->get_image_plane_at_pos(0, xDi,yDi);
453
454		uint8_t q[4][4], p[4][4];
	466	pixel_t* ptr = img->get_image_plane_at_pos_NEW<pixel_t>(0, xDi,yDi);
	467
	468	pixel_t q[4][4], p[4][4];
455	469	for (int k=0;k<4;k++)
456	470	for (int i=0;i<4;i++)
457	471	{

576	590
577	591	logtrace(LogDeblock,"line:%d\n",k);
578	592
579		const uint8_t p0 = p[k][0];
580		const uint8_t p1 = p[k][1];
581		const uint8_t p2 = p[k][2];
582		const uint8_t p3 = p[k][3];
583		const uint8_t q0 = q[k][0];
584		const uint8_t q1 = q[k][1];
585		const uint8_t q2 = q[k][2];
586		const uint8_t q3 = q[k][3];
	593	const pixel_t p0 = p[k][0];
	594	const pixel_t p1 = p[k][1];
	595	const pixel_t p2 = p[k][2];
	596	const pixel_t p3 = p[k][3];
	597	const pixel_t q0 = q[k][0];
	598	const pixel_t q1 = q[k][1];
	599	const pixel_t q2 = q[k][2];
	600	const pixel_t q3 = q[k][3];
587	601
588	602	if (dE==2) {
589	603	// strong filtering
590	604
591	605	//nDp=nDq=3;
592	606
593		uint8_t pnew[3],qnew[3];
	607	pixel_t pnew[3],qnew[3];
594	608	pnew[0] = Clip3(p0-2tc,p0+2tc, (p2 + 2p1 + 2p0 + 2*q0 + q1 +4)>>3);
595	609	pnew[1] = Clip3(p1-2tc,p1+2tc, (p2 + p1 + p0 + q0+2)>>2);
596	610	pnew[2] = Clip3(p2-2tc,p2+2tc, (2p3 + 3p2 + p1 + p0 + q0 + 4)>>3);

628	642	delta = Clip3(-tc,tc,delta);
629	643	logtrace(LogDeblock," deblk + %d;%d [%02x->%02x] - %d;%d [%02x->%02x] delta:%d\n",
630	644	vertical ? xDi-1 : xDi+k,
631		vertical ? yDi+k : yDi-1, p0,Clip1_8bit(p0+delta),
	645	vertical ? yDi+k : yDi-1, p0,Clip_BitDepth(p0+delta, bitDepth_Y),
632	646	vertical ? xDi : xDi+k,
633		vertical ? yDi+k : yDi, q0,Clip1_8bit(q0-delta),
	647	vertical ? yDi+k : yDi, q0,Clip_BitDepth(q0-delta, bitDepth_Y),
634	648	delta);
635	649
636	650	if (vertical) {
637		if (filterP) { ptr[-0-1+k*stride] = Clip1_8bit(p0+delta); }
638		if (filterQ) { ptr[ 0 +k*stride] = Clip1_8bit(q0-delta); }
	651	if (filterP) { ptr[-0-1+k*stride] = Clip_BitDepth(p0+delta, bitDepth_Y); }
	652	if (filterQ) { ptr[ 0 +k*stride] = Clip_BitDepth(q0-delta, bitDepth_Y); }
639	653	}
640	654	else {
641		if (filterP) { ptr[ k -1*stride] = Clip1_8bit(p0+delta); }
642		if (filterQ) { ptr[ k +0*stride] = Clip1_8bit(q0-delta); }
	655	if (filterP) { ptr[ k -1*stride] = Clip_BitDepth(p0+delta, bitDepth_Y); }
	656	if (filterQ) { ptr[ k +0*stride] = Clip_BitDepth(q0-delta, bitDepth_Y); }
643	657	}
644	658
645	659	//ptr[ 0+k*stride] = 200;

652	666	vertical ? yDi+k : yDi-2,
653	667	delta_p);
654	668
655		if (vertical) { ptr[-1-1+k*stride] = Clip1_8bit(p1+delta_p); }
656		else { ptr[ k -2*stride] = Clip1_8bit(p1+delta_p); }
	669	if (vertical) { ptr[-1-1+k*stride] = Clip_BitDepth(p1+delta_p, bitDepth_Y); }
	670	else { ptr[ k -2*stride] = Clip_BitDepth(p1+delta_p, bitDepth_Y); }
657	671	}
658	672
659	673	if (dEq==1 && filterQ) {

664	678	vertical ? yDi+k : yDi+1,
665	679	delta_q);
666	680
667		if (vertical) { ptr[ 1 +k*stride] = Clip1_8bit(q1+delta_q); }
668		else { ptr[ k +1*stride] = Clip1_8bit(q1+delta_q); }
	681	if (vertical) { ptr[ 1 +k*stride] = Clip_BitDepth(q1+delta_q, bitDepth_Y); }
	682	else { ptr[ k +1*stride] = Clip_BitDepth(q1+delta_q, bitDepth_Y); }
669	683	}
670	684
671	685	//nDp = dEp+1;

680	694	}
681	695	}
682	696
	697
	698	void edge_filtering_luma(de265_image* img, bool vertical,
	699	int yStart,int yEnd, int xStart,int xEnd)
	700	{
	701	if (img->high_bit_depth(0)) {
	702	edge_filtering_luma_internal<uint16_t>(img,vertical,yStart,yEnd,xStart,xEnd);
	703	}
	704	else {
	705	edge_filtering_luma_internal<uint8_t>(img,vertical,yStart,yEnd,xStart,xEnd);
	706	}
	707	}
683	708
684	709	void edge_filtering_luma_CTB(de265_image* img, bool vertical, int xCtb,int yCtb)
685	710	{

695	720
696	721
697	722	// 8.7.2.4
698		void edge_filtering_chroma(de265_image* img, bool vertical, int yStart,int yEnd,
699		int xStart,int xEnd)
700		{
701		int xIncr = vertical ? 4 : 2;
702		int yIncr = vertical ? 2 : 4;
	723	/** ?Start and ?End values in 4-luma pixels resolution.
	724	*/
	725	template <class pixel_t>
	726	void edge_filtering_chroma_internal(de265_image* img, bool vertical,
	727	int yStart,int yEnd,
	728	int xStart,int xEnd)
	729	{
	730	//printf("chroma %d-%d %d-%d\n",xStart,xEnd,yStart,yEnd);
	731
	732	const int SubWidthC = img->sps.SubWidthC;
	733	const int SubHeightC = img->sps.SubHeightC;
	734
	735	int xIncr = vertical ? 2 : 1;
	736	int yIncr = vertical ? 1 : 2;
	737
	738	xIncr *= SubWidthC;
	739	yIncr *= SubHeightC;
703	740
704	741	const int stride = img->get_image_stride(1);
705	742
706	743	xEnd = libde265_min(xEnd,img->get_deblk_width());
707	744	yEnd = libde265_min(yEnd,img->get_deblk_height());
708	745
	746	int bitDepth_C = img->sps.BitDepth_C;
	747
709	748	for (int y=yStart;y<yEnd;y+=yIncr)
710	749	for (int x=xStart;x<xEnd;x+=xIncr) {
711		int xDi = x*2;
712		int yDi = y*2;
713		int bS = img->get_deblk_bS(2xDi,2yDi);
	750	int xDi = x << (3-SubWidthC);
	751	int yDi = y << (3-SubHeightC);
	752
	753	//printf("x,y:%d,%d xDi,yDi:%d,%d\n",x,y,xDi,yDi);
	754
	755	int bS = img->get_deblk_bS(xDiSubWidthC,yDiSubHeightC);
714	756
715	757	if (bS>1) {
716	758	// 8.7.2.4.5

720	762	img->pps.pic_cb_qp_offset :
721	763	img->pps.pic_cr_qp_offset);
722	764
723		uint8_t* ptr = img->get_image_plane_at_pos(cplane+1, xDi,yDi);
724
725		uint8_t p[2][4];
726		uint8_t q[2][4];
	765	pixel_t* ptr = img->get_image_plane_at_pos_NEW<pixel_t>(cplane+1, xDi,yDi);
	766
	767	pixel_t p[2][4];
	768	pixel_t q[2][4];
727	769
728	770	logtrace(LogDeblock,"-%s- %d %d\n",cplane==0 ? "Cb" : "Cr",xDi,yDi);
729	771

758	800	}
759	801	#endif
760	802
761		int QP_Q = img->get_QPY(2xDi,2yDi);
	803	int QP_Q = img->get_QPY(SubWidthCxDi,SubHeightCyDi);
762	804	int QP_P = (vertical ?
763		img->get_QPY(2xDi-1,2yDi) :
764		img->get_QPY(2xDi,2yDi-1));
	805	img->get_QPY(SubWidthCxDi-1,SubHeightCyDi) :
	806	img->get_QPY(SubWidthCxDi,SubHeightCyDi-1));
765	807	int qP_i = ((QP_Q+QP_P+1)>>1) + cQpPicOffset;
766		int QP_C = table8_22(qP_i);
	808	int QP_C;
	809	if (img->sps.ChromaArrayType == CHROMA_420) {
	810	QP_C = table8_22(qP_i);
	811	} else {
	812	QP_C = libde265_min(qP_i, 51);
	813	}
	814
767	815
768	816	//printf("POC=%d\n",ctx->img->PicOrderCntVal);
769	817	logtrace(LogDeblock,"%d %d: ((%d+%d+1)>>1) + %d = qP_i=%d (QP_C=%d)\n",
770		2xDi,2yDi, QP_Q,QP_P,cQpPicOffset,qP_i,QP_C);
771
772		int sliceIndexQ00 = img->get_SliceHeaderIndex(2xDi,2yDi);
	818	SubWidthCxDi,SubHeightCyDi, QP_Q,QP_P,cQpPicOffset,qP_i,QP_C);
	819
	820	int sliceIndexQ00 = img->get_SliceHeaderIndex(SubWidthCxDi,SubHeightCyDi);
773	821	int tc_offset = img->slices[sliceIndexQ00]->slice_tc_offset;
774	822
775	823	int Q = Clip3(0,53, QP_C + 2*(bS-1) + tc_offset);

781	829
782	830	if (vertical) {
783	831	bool filterP = true;
784		if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(2xDi-1,2yDi)) filterP=false;
785		if (img->get_cu_transquant_bypass(2xDi-1,2yDi)) filterP=false;
	832	if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(SubWidthCxDi-1,SubHeightCyDi)) filterP=false;
	833	if (img->get_cu_transquant_bypass(SubWidthCxDi-1,SubHeightCyDi)) filterP=false;
786	834
787	835	bool filterQ = true;
788		if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(2xDi,2yDi)) filterQ=false;
789		if (img->get_cu_transquant_bypass(2xDi,2yDi)) filterQ=false;
	836	if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(SubWidthCxDi,SubHeightCyDi)) filterQ=false;
	837	if (img->get_cu_transquant_bypass(SubWidthCxDi,SubHeightCyDi)) filterQ=false;
790	838
791	839
792	840	for (int k=0;k<4;k++) {
793	841	int delta = Clip3(-tc,tc, ((((q[0][k]-p[0][k])<<2)+p[1][k]-q[1][k]+4)>>3));
794	842	logtrace(LogDeblock,"delta=%d\n",delta);
795		if (filterP) { ptr[-1+k*stride] = Clip1_8bit(p[0][k]+delta); }
796		if (filterQ) { ptr[ 0+k*stride] = Clip1_8bit(q[0][k]-delta); }
	843	if (filterP) { ptr[-1+k*stride] = Clip_BitDepth(p[0][k]+delta, bitDepth_C); }
	844	if (filterQ) { ptr[ 0+k*stride] = Clip_BitDepth(q[0][k]-delta, bitDepth_C); }
797	845	}
798	846	}
799	847	else {
800	848	bool filterP = true;
801		if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(2xDi,2yDi-1)) filterP=false;
802		if (img->get_cu_transquant_bypass(2xDi,2yDi-1)) filterP=false;
	849	if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(SubWidthCxDi,SubHeightCyDi-1)) filterP=false;
	850	if (img->get_cu_transquant_bypass(SubWidthCxDi,SubHeightCyDi-1)) filterP=false;
803	851
804	852	bool filterQ = true;
805		if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(2xDi,2yDi)) filterQ=false;
806		if (img->get_cu_transquant_bypass(2xDi,2yDi)) filterQ=false;
	853	if (img->sps.pcm_loop_filter_disable_flag && img->get_pcm_flag(SubWidthCxDi,SubHeightCyDi)) filterQ=false;
	854	if (img->get_cu_transquant_bypass(SubWidthCxDi,SubHeightCyDi)) filterQ=false;
807	855
808	856	for (int k=0;k<4;k++) {
809	857	int delta = Clip3(-tc,tc, ((((q[0][k]-p[0][k])<<2)+p[1][k]-q[1][k]+4)>>3));
810		if (filterP) { ptr[ k-1*stride] = Clip1_8bit(p[0][k]+delta); }
811		if (filterQ) { ptr[ k+0*stride] = Clip1_8bit(q[0][k]-delta); }
	858	if (filterP) { ptr[ k-1*stride] = Clip_BitDepth(p[0][k]+delta, bitDepth_C); }
	859	if (filterQ) { ptr[ k+0*stride] = Clip_BitDepth(q[0][k]-delta, bitDepth_C); }
812	860	}
813	861	}
814	862	}
815	863	}
816	864	}
817	865	}
	866
	867
	868	void edge_filtering_chroma(de265_image* img, bool vertical, int yStart,int yEnd,
	869	int xStart,int xEnd)
	870	{
	871	if (img->high_bit_depth(1)) {
	872	edge_filtering_chroma_internal<uint16_t>(img,vertical,yStart,yEnd,xStart,xEnd);
	873	}
	874	else {
	875	edge_filtering_chroma_internal<uint8_t>(img,vertical,yStart,yEnd,xStart,xEnd);
	876	}
	877	}
	878
818	879
819	880	void edge_filtering_chroma_CTB(de265_image* img, bool vertical, int xCtb,int yCtb)
820	881	{

836	897	bool vertical;
837	898
838	899	virtual void work();
	900	virtual std::string name() const {
	901	char buf[100];
	902	sprintf(buf,"deblock-%d",ctb_y);
	903	return buf;
	904	}
839	905	};
840	906
841	907
842	908	void thread_task_deblock_CTBRow::work()
843	909	{
844	910	state = Running;
845		img->thread_run();
	911	img->thread_run(this);
846	912
847	913	int xStart=0;
848	914	int xEnd = img->get_deblk_width();

899	965
900	966	if (deblocking_enabled) {
901	967	derive_boundaryStrength(img, vertical, first,last, xStart,xEnd);
902		edge_filtering_luma (img, vertical, first,last, xStart,xEnd);
903		edge_filtering_chroma (img, vertical, first,last, xStart,xEnd);
	968
	969	edge_filtering_luma(img, vertical, first,last, xStart,xEnd);
	970
	971	if (img->sps.ChromaArrayType != CHROMA_MONO) {
	972	edge_filtering_chroma(img, vertical, first,last, xStart,xEnd);
	973	}
904	974	}
905	975
906	976	for (int x=0;x<=rightCtb;x++) {

909	979	}
910	980
911	981	state = Finished;
912		img->thread_finishes();
	982	img->thread_finishes(this);
913	983	}
914	984
915	985

934	1004	task->vertical = (pass==0);
935	1005
936	1006	imgunit->tasks.push_back(task);
937		add_task(&ctx->thread_pool, task);
	1007	add_task(&ctx->thread_pool_, task);
938	1008	n++;
939	1009	}
940	1010	}

954	1024	logtrace(LogDeblock,"VERTICAL\n");
955	1025	derive_boundaryStrength(img, true ,0,img->get_deblk_height(),0,img->get_deblk_width());
956	1026	edge_filtering_luma (img, true ,0,img->get_deblk_height(),0,img->get_deblk_width());
957		edge_filtering_chroma (img, true ,0,img->get_deblk_height(),0,img->get_deblk_width());
958
	1027
	1028	if (img->sps.ChromaArrayType != CHROMA_MONO) {
	1029	edge_filtering_chroma (img, true ,0,img->get_deblk_height(),0,img->get_deblk_width());
	1030	}
959	1031	#if 0
960	1032	char buf[1000];
961	1033	sprintf(buf,"lf-after-V-%05d.yuv", ctx->img->PicOrderCntVal);

967	1039	logtrace(LogDeblock,"HORIZONTAL\n");
968	1040	derive_boundaryStrength(img, false ,0,img->get_deblk_height(),0,img->get_deblk_width());
969	1041	edge_filtering_luma (img, false ,0,img->get_deblk_height(),0,img->get_deblk_width());
970		edge_filtering_chroma (img, false ,0,img->get_deblk_height(),0,img->get_deblk_width());
	1042
	1043	if (img->sps.ChromaArrayType != CHROMA_MONO) {
	1044	edge_filtering_chroma (img, false ,0,img->get_deblk_height(),0,img->get_deblk_width());
	1045	}
971	1046
972	1047	#if 0
973	1048	sprintf(buf,"lf-after-H-%05d.yuv", ctx->img->PicOrderCntVal);

+282

-90

libde265/decctx.cc less more

39	39	#include "x86/sse.h"
40	40	#endif
41	41
	42	#ifdef HAVE_ARM
	43	#include "arm/arm.h"
	44	#endif
	45
42	46	#define SAVE_INTERMEDIATE_IMAGES 0
43	47
44	48	#if SAVE_INTERMEDIATE_IMAGES

87	91	IsCuQpDeltaCoded = false;
88	92	CuQpDelta = 0;
89	93
	94	IsCuChromaQpOffsetCoded = false;
	95	CuQpOffsetCb = 0;
	96	CuQpOffsetCr = 0;
	97
90	98	/*
91	99	currentQPY = 0;
92	100	currentQG_x = 0;

109	117	img = NULL;
110	118	shdr = NULL;
111	119
	120	imgunit = NULL;
	121	sliceunit = NULL;
	122
112	123
113	124	//memset(this,0,sizeof(thread_context));
114	125

117	128	int offset = (uintptr_t)_coeffBuf & 0x0f;
118	129
119	130	if (offset == 0) {
120		coeffBuf = (int16_t *) &_coeffBuf; // correctly aligned already
	131	coeffBuf = _coeffBuf; // correctly aligned already
121	132	}
122	133	else {
123	134	coeffBuf = (int16_t ) (((uint8_t )_coeffBuf) + (16-offset));

128	139
129	140
130	141	slice_unit::slice_unit(decoder_context* decctx)
131		: ctx(decctx),
132		nal(NULL),
	142	: nal(NULL),
133	143	shdr(NULL),
	144	imgunit(NULL),
134	145	flush_reorder_buffer(false),
	146	nThreads(0),
	147	first_decoded_CTB_RS(-1),
	148	last_decoded_CTB_RS(-1),
135	149	thread_contexts(NULL),
136		imgunit(NULL)
	150	ctx(decctx)
137	151	{
138	152	state = Unprocessed;
	153	nThreadContexts = 0;
139	154	}
140	155
141	156	slice_unit::~slice_unit()

153	168	assert(thread_contexts==NULL);
154	169
155	170	thread_contexts = new thread_context[n];
	171	nThreadContexts = n;
156	172	}
157	173
158	174

173	189	for (int i=0;i<tasks.size();i++) {
174	190	delete tasks[i];
175	191	}
	192	}
	193
	194
	195	base_context::base_context()
	196	{
	197	set_acceleration_functions(de265_acceleration_AUTO);
176	198	}
177	199
178	200

197	219	param_vps_headers_fd = -1;
198	220	param_pps_headers_fd = -1;
199	221	param_slice_headers_fd = -1;
200
201		set_acceleration_functions(de265_acceleration_AUTO);
202	222
203	223	param_image_allocation_functions = de265_image::default_image_allocation;
204	224	param_image_allocation_userdata = NULL;

319	339
320	340	de265_error decoder_context::start_thread_pool(int nThreads)
321	341	{
322		::start_thread_pool(&thread_pool, nThreads);
	342	::start_thread_pool(&thread_pool_, nThreads);
323	343
324	344	num_worker_threads = nThreads;
325	345

331	351	{
332	352	if (get_num_worker_threads()>0) {
333	353	//flush_thread_pool(&ctx->thread_pool);
334		::stop_thread_pool(&thread_pool);
	354	::stop_thread_pool(&thread_pool_);
335	355	}
336	356	}
337	357

340	360	{
341	361	if (num_worker_threads>0) {
342	362	//flush_thread_pool(&ctx->thread_pool);
343		::stop_thread_pool(&thread_pool);
	363	::stop_thread_pool(&thread_pool_);
344	364	}
345	365
346	366	// --------------------------------------------------

406	426	}
407	427	}
408	428
409		void decoder_context::set_acceleration_functions(enum de265_acceleration l)
	429	void base_context::set_acceleration_functions(enum de265_acceleration l)
410	430	{
411	431	// fill scalar functions first (so that function table is completely filled)
412	432

420	440	init_acceleration_functions_sse(&acceleration);
421	441	}
422	442	#endif
	443	#ifdef HAVE_ARM
	444	if (l>=de265_acceleration_ARM) {
	445	init_acceleration_functions_arm(&acceleration);
	446	}
	447	#endif
423	448	}
424	449
425	450

465	490	}
466	491
467	492
468		void decoder_context::add_task_decode_CTB_row(thread_context* tctx, bool firstSliceSubstream)
	493	void decoder_context::add_task_decode_CTB_row(thread_context* tctx,
	494	bool firstSliceSubstream,
	495	int ctbRow)
469	496	{
470	497	thread_task_ctb_row* task = new thread_task_ctb_row;
471	498	task->firstSliceSubstream = firstSliceSubstream;
472	499	task->tctx = tctx;
	500	task->debug_startCtbRow = ctbRow;
473	501	tctx->task = task;
474	502
475		add_task(&thread_pool, task);
	503	add_task(&thread_pool_, task);
476	504
477	505	tctx->imgunit->tasks.push_back(task);
478	506	}
479	507
480	508
481		void decoder_context::add_task_decode_slice_segment(thread_context* tctx, bool firstSliceSubstream)
	509	void decoder_context::add_task_decode_slice_segment(thread_context* tctx, bool firstSliceSubstream,
	510	int ctbx,int ctby)
482	511	{
483	512	thread_task_slice_segment* task = new thread_task_slice_segment;
484	513	task->firstSliceSubstream = firstSliceSubstream;
485	514	task->tctx = tctx;
	515	task->debug_startCtbX = ctbx;
	516	task->debug_startCtbY = ctby;
486	517	tctx->task = task;
487	518
488		add_task(&thread_pool, task);
	519	add_task(&thread_pool_, task);
489	520
490	521	tctx->imgunit->tasks.push_back(task);
491	522	}

495	526	{
496	527	logdebug(LogHeaders,"---> read VPS\n");
497	528
498		video_parameter_set vps = { 0 };
499		de265_error err = ::read_vps(this,&reader,&vps);
	529	video_parameter_set vps;
	530	de265_error err = vps.read(this,&reader);
500	531	if (err != DE265_OK) {
501	532	return err;
502	533	}
503	534
504	535	if (param_vps_headers_fd>=0) {
505		dump_vps(&vps, param_vps_headers_fd);
	536	vps.dump(param_vps_headers_fd);
506	537	}
507	538
508	539	process_vps(&vps);

522	553	}
523	554
524	555	if (param_sps_headers_fd>=0) {
525		sps.dump_sps(param_sps_headers_fd);
	556	sps.dump(param_sps_headers_fd);
526	557	}
527	558
528	559	process_sps(&sps);

539	570	bool success = pps.read(&reader,this);
540	571
541	572	if (param_pps_headers_fd>=0) {
542		pps.dump_pps(param_pps_headers_fd);
	573	pps.dump(param_pps_headers_fd);
543	574	}
544	575
545	576	if (success) {

603	634
604	635	if (process_slice_segment_header(this, shdr, &err, nal->pts, &nal_hdr, nal->user_data) == false)
605	636	{
606		img->integrity = INTEGRITY_NOT_DECODED;
	637	if (img!=NULL) img->integrity = INTEGRITY_NOT_DECODED;
607	638	nal_parser.free_NAL_unit(nal);
608	639	delete shdr;
609	640	return err;

649	680	image_units.back()->slice_units.push_back(sliceunit);
650	681	}
651	682
652		decode_some();
	683	bool did_work;
	684	err = decode_some(&did_work);
653	685
654	686	return DE265_OK;
655	687	}

664	696	}
665	697
666	698
667		de265_error decoder_context::decode_some()
	699	de265_error decoder_context::decode_some(bool* did_work)
668	700	{
669	701	de265_error err = DE265_OK;
670	702
671		if (0) {
672		static int cnt=0;
673		cnt++;
674		if (cnt<5) return DE265_OK;
675		}
	703	*did_work = false;
676	704
677	705	if (image_units.empty()) { return DE265_OK; } // nothing to do
678	706
679	707
680	708	// decode something if there is work to do
681	709
682		if ( ! image_units.empty() && ! image_units[0]->slice_units.empty() ) {
	710	if ( ! image_units.empty() ) { // && ! image_units[0]->slice_units.empty() ) {
683	711
684	712	image_unit* imgunit = image_units[0];
685		slice_unit* sliceunit = imgunit->slice_units[0];
686
687		pop_front(imgunit->slice_units);
688
689		if (sliceunit->flush_reorder_buffer) {
690		dpb.flush_reorder_buffer();
691		}
692
693		//err = decode_slice_unit_sequential(imgunit, sliceunit);
694		err = decode_slice_unit_parallel(imgunit, sliceunit);
695		if (err) {
696		return err;
697		}
698
699		delete sliceunit;
	713	slice_unit* sliceunit = imgunit->get_next_unprocessed_slice_segment();
	714
	715	if (sliceunit != NULL) {
	716
	717	//pop_front(imgunit->slice_units);
	718
	719	if (sliceunit->flush_reorder_buffer) {
	720	dpb.flush_reorder_buffer();
	721	}
	722
	723	*did_work = true;
	724
	725	//err = decode_slice_unit_sequential(imgunit, sliceunit);
	726	err = decode_slice_unit_parallel(imgunit, sliceunit);
	727	if (err) {
	728	return err;
	729	}
	730
	731	//delete sliceunit;
	732	}
700	733	}
701	734
702	735

704	737	// if we decoded all slices of the current image and there will not
705	738	// be added any more slices to the image, output the image
706	739
707		if ( ( image_units.size()>=2 && image_units[0]->slice_units.empty() ) \|\|
708		( image_units.size()>=1 && image_units[0]->slice_units.empty() &&
	740	if ( ( image_units.size()>=2 && image_units[0]->all_slice_segments_processed()) \|\|
	741	( image_units.size()>=1 && image_units[0]->all_slice_segments_processed() &&
709	742	nal_parser.number_of_NAL_units_pending()==0 &&
710	743	(nal_parser.is_end_of_stream() \|\| nal_parser.is_end_of_frame()) )) {
711	744
712	745	image_unit* imgunit = image_units[0];
	746
	747	*did_work=true;
713	748
714	749
715	750	// mark all CTBs as decoded even if they are not, because faulty input

767	802
768	803	remove_images_from_dpb(sliceunit->shdr->RemoveReferencesList);
769	804
	805	if (sliceunit->shdr->slice_segment_address >= imgunit->img->pps.CtbAddrRStoTS.size()) {
	806	return DE265_ERROR_CTB_OUTSIDE_IMAGE_AREA;
	807	}
	808
770	809
771	810	struct thread_context tctx;
772	811

774	813	tctx.img = imgunit->img;
775	814	tctx.decctx = this;
776	815	tctx.imgunit = imgunit;
	816	tctx.sliceunit= sliceunit;
777	817	tctx.CtbAddrInTS = imgunit->img->pps.CtbAddrRStoTS[tctx.shdr->slice_segment_address];
778	818	tctx.task = NULL;
779	819
780	820	init_thread_context(&tctx);
	821
	822	if (sliceunit->reader.bytes_remaining <= 0) {
	823	return DE265_ERROR_PREMATURE_END_OF_SLICE;
	824	}
781	825
782	826	init_CABAC_decoder(&tctx.cabac_decoder,
783	827	sliceunit->reader.data,

787	831
788	832	if (pps->entropy_coding_sync_enabled_flag &&
789	833	sliceunit->shdr->first_slice_segment_in_pic_flag) {
790		imgunit->ctx_models.resize( (img->sps.PicHeightInCtbsY-1) * CONTEXT_MODEL_TABLE_LENGTH );
791		}
792
793		if ((err=read_slice_segment_data(&tctx)) != DE265_OK)
794		{ return err; }
	834	imgunit->ctx_models.resize( (img->sps.PicHeightInCtbsY-1) ); //* CONTEXT_MODEL_TABLE_LENGTH );
	835	}
	836
	837	sliceunit->nThreads=1;
	838
	839	err=read_slice_segment_data(&tctx);
	840
	841	sliceunit->finished_threads.set_progress(1);
795	842
796	843	return err;
	844	}
	845
	846
	847	void decoder_context::mark_whole_slice_as_processed(image_unit* imgunit,
	848	slice_unit* sliceunit,
	849	int progress)
	850	{
	851	//printf("mark whole slice\n");
	852
	853
	854	// mark all CTBs upto the next slice segment as processed
	855
	856	slice_unit* nextSegment = imgunit->get_next_slice_segment(sliceunit);
	857	if (nextSegment) {
	858	/*
	859	printf("mark whole slice between %d and %d\n",
	860	sliceunit->shdr->slice_segment_address,
	861	nextSegment->shdr->slice_segment_address);
	862	*/
	863
	864	for (int ctb=sliceunit->shdr->slice_segment_address;
	865	ctb < nextSegment->shdr->slice_segment_address;
	866	ctb++)
	867	{
	868	if (ctb >= imgunit->img->number_of_ctbs())
	869	break;
	870
	871	imgunit->img->ctb_progress[ctb].set_progress(progress);
	872	}
	873	}
797	874	}
798	875
799	876

804	881
805	882	remove_images_from_dpb(sliceunit->shdr->RemoveReferencesList);
806	883
807
	884	/*
	885	printf("-------- decode --------\n");
	886	printf("IMAGE UNIT %p\n",imgunit);
	887	sliceunit->shdr->dump_slice_segment_header(sliceunit->ctx, 1);
	888	imgunit->dump_slices();
	889	*/
808	890
809	891	de265_image* img = imgunit->img;
810	892	const pic_parameter_set* pps = &img->pps;
	893
	894	sliceunit->state = slice_unit::InProgress;
811	895
812	896	bool use_WPP = (img->decctx->num_worker_threads > 0 &&
813	897	pps->entropy_coding_sync_enabled_flag);

825	909	}
826	910
827	911
	912	// If this is the first slice segment, mark all CTBs before this as processed
	913	// (the real first slice segment could be missing).
	914
	915	if (imgunit->is_first_slice_segment(sliceunit)) {
	916	slice_segment_header* shdr = sliceunit->shdr;
	917	int firstCTB = shdr->slice_segment_address;
	918
	919	for (int ctb=0;ctb<firstCTB;ctb++) {
	920	//printf("mark pre progress %d\n",ctb);
	921	img->ctb_progress[ctb].set_progress(CTB_PROGRESS_PREFILTER);
	922	}
	923	}
	924
	925
	926	// if there is a previous slice that has been completely decoded,
	927	// mark all CTBs until the start of this slice as completed
	928
	929	//printf("this slice: %p\n",sliceunit);
	930	slice_unit* prevSlice = imgunit->get_prev_slice_segment(sliceunit);
	931	//if (prevSlice) printf("prev slice state: %d\n",prevSlice->state);
	932	if (prevSlice && prevSlice->state == slice_unit::Decoded) {
	933	mark_whole_slice_as_processed(imgunit,prevSlice,CTB_PROGRESS_PREFILTER);
	934	}
	935
	936
828	937	// TODO: even though we cannot split this into several tasks, we should run it
829	938	// as a background thread
830	939	if (!use_WPP && !use_tiles) {
831		return decode_slice_unit_sequential(imgunit, sliceunit);
	940	//printf("SEQ\n");
	941	err = decode_slice_unit_sequential(imgunit, sliceunit);
	942	sliceunit->state = slice_unit::Decoded;
	943	mark_whole_slice_as_processed(imgunit,sliceunit,CTB_PROGRESS_PREFILTER);
	944	return err;
832	945	}
833	946
834	947
835	948	if (use_WPP && use_tiles) {
836	949	// TODO: this is not allowed ... output some warning or error
	950
	951	return DE265_WARNING_PPS_HEADER_INVALID;
837	952	}
838	953
839	954
840	955	if (use_WPP) {
841		return decode_slice_unit_WPP(imgunit, sliceunit);
	956	//printf("WPP\n");
	957	err = decode_slice_unit_WPP(imgunit, sliceunit);
	958	sliceunit->state = slice_unit::Decoded;
	959	mark_whole_slice_as_processed(imgunit,sliceunit,CTB_PROGRESS_PREFILTER);
	960	return err;
842	961	}
843	962	else if (use_tiles) {
844		return decode_slice_unit_tiles(imgunit, sliceunit);
	963	//printf("TILE\n");
	964	err = decode_slice_unit_tiles(imgunit, sliceunit);
	965	sliceunit->state = slice_unit::Decoded;
	966	mark_whole_slice_as_processed(imgunit,sliceunit,CTB_PROGRESS_PREFILTER);
	967	return err;
845	968	}
846	969
847	970	assert(false);
848		return DE265_OK;
	971	return err;
849	972	}
850	973
851	974

863	986
864	987
865	988	assert(img->num_threads_active() == 0);
866		img->thread_start(nRows);
867
868		//printf("-------- decode --------\n");
869	989
870	990
871	991	// reserve space to store entropy coding context models for each CTB row
872	992
873	993	if (shdr->first_slice_segment_in_pic_flag) {
874	994	// reserve space for nRows-1 because we don't need to save the CABAC model in the last CTB row
875		imgunit->ctx_models.resize( (img->sps.PicHeightInCtbsY-1) * CONTEXT_MODEL_TABLE_LENGTH );
	995	imgunit->ctx_models.resize( (img->sps.PicHeightInCtbsY-1) ); //* CONTEXT_MODEL_TABLE_LENGTH );
876	996	}
877	997
878	998

889	1009	ctbRow++;
890	1010	ctbAddrRS = ctbRow * ctbsWidth;
891	1011	}
	1012	else if (nRows>1 && (ctbAddrRS % ctbsWidth) != 0) {
	1013	// If slice segment consists of several WPP rows, each of them
	1014	// has to start at a row.
	1015
	1016	//printf("does not start at start\n");
	1017
	1018	err = DE265_WARNING_SLICEHEADER_INVALID;
	1019	break;
	1020	}
892	1021
893	1022
894	1023	// prepare thread context

899	1028	tctx->decctx = img->decctx;
900	1029	tctx->img = img;
901	1030	tctx->imgunit = imgunit;
	1031	tctx->sliceunit= sliceunit;
902	1032	tctx->CtbAddrInTS = pps->CtbAddrRStoTS[ctbAddrRS];
903	1033
904	1034	init_thread_context(tctx);

914	1044	if (entryPt==nRows-1) dataEnd = sliceunit->reader.bytes_remaining;
915	1045	else dataEnd = shdr->entry_point_offset[entryPt];
916	1046
	1047	if (dataStartIndex<0 \|\| dataEnd>sliceunit->reader.bytes_remaining \|\|
	1048	dataEnd <= dataStartIndex) {
	1049	//printf("WPP premature end\n");
	1050	err = DE265_ERROR_PREMATURE_END_OF_SLICE;
	1051	break;
	1052	}
	1053
917	1054	init_CABAC_decoder(&tctx->cabac_decoder,
918	1055	&sliceunit->reader.data[dataStartIndex],
919	1056	dataEnd-dataStartIndex);
920	1057
921	1058	// add task
922	1059
923		add_task_decode_CTB_row(tctx, entryPt==0);
	1060	//printf("start task for ctb-row: %d\n",ctbRow);
	1061	img->thread_start(1);
	1062	sliceunit->nThreads++;
	1063	add_task_decode_CTB_row(tctx, entryPt==0, ctbRow);
924	1064	}
925	1065
926	1066	#if 0

960	1100
961	1101
962	1102	assert(img->num_threads_active() == 0);
963		img->thread_start(nTiles);
964	1103
965	1104	sliceunit->allocate_thread_contexts(nTiles);
966	1105

973	1112	// entry points other than the first start at tile beginnings
974	1113	if (entryPt>0) {
975	1114	tileID++;
	1115
	1116	if (tileID >= pps->num_tile_columns * pps->num_tile_rows) {
	1117	err = DE265_WARNING_SLICEHEADER_INVALID;
	1118	break;
	1119	}
	1120
976	1121	int ctbX = pps->colBd[tileID % pps->num_tile_columns];
977	1122	int ctbY = pps->rowBd[tileID / pps->num_tile_columns];
978	1123	ctbAddrRS = ctbY * ctbsWidth + ctbX;

986	1131	tctx->decctx = img->decctx;
987	1132	tctx->img = img;
988	1133	tctx->imgunit = imgunit;
	1134	tctx->sliceunit= sliceunit;
989	1135	tctx->CtbAddrInTS = pps->CtbAddrRStoTS[ctbAddrRS];
990	1136
991	1137	init_thread_context(tctx);

1001	1147	if (entryPt==nTiles-1) dataEnd = sliceunit->reader.bytes_remaining;
1002	1148	else dataEnd = shdr->entry_point_offset[entryPt];
1003	1149
	1150	if (dataStartIndex<0 \|\| dataEnd>sliceunit->reader.bytes_remaining \|\|
	1151	dataEnd <= dataStartIndex) {
	1152	err = DE265_ERROR_PREMATURE_END_OF_SLICE;
	1153	break;
	1154	}
	1155
1004	1156	init_CABAC_decoder(&tctx->cabac_decoder,
1005	1157	&sliceunit->reader.data[dataStartIndex],
1006	1158	dataEnd-dataStartIndex);
1007	1159
1008	1160	// add task
1009	1161
1010		add_task_decode_slice_segment(tctx, entryPt==0);
	1162	//printf("add tiles thread\n");
	1163	img->thread_start(1);
	1164	sliceunit->nThreads++;
	1165	add_task_decode_slice_segment(tctx, entryPt==0,
	1166	ctbAddrRS % ctbsWidth,
	1167	ctbAddrRS / ctbsWidth);
1011	1168	}
1012	1169
1013	1170	img->wait_for_completion();

1016	1173	delete imgunit->tasks[i];
1017	1174	imgunit->tasks.clear();
1018	1175
1019		return DE265_OK;
	1176	return err;
1020	1177	}
1021	1178
1022	1179

1032	1189	bitreader_init(&reader, nal->data(), nal->size());
1033	1190
1034	1191	nal_header nal_hdr;
1035		nal_read_header(&reader, &nal_hdr);
	1192	nal_hdr.read(&reader);
1036	1193	ctx->process_nal_hdr(&nal_hdr);
	1194
	1195	if (nal_hdr.nuh_layer_id > 0) {
	1196	// Discard all NAL units with nuh_layer_id > 0
	1197	// These will have to be handeled by an SHVC decoder.
	1198	nal_parser.free_NAL_unit(nal);
	1199	return DE265_OK;
	1200	}
1037	1201
1038	1202	loginfo(LogHighlevel,"NAL: 0x%x 0x%x - unit type:%s temporal id:%d\n",
1039	1203	nal->data()[0], nal->data()[1],

1104	1268	// if the stream has ended, and no more NALs are to be decoded, flush all pictures
1105	1269
1106	1270	if (ctx->nal_parser.get_NAL_queue_length() == 0 &&
1107		ctx->nal_parser.is_end_of_stream() &&
	1271	(ctx->nal_parser.is_end_of_stream() \|\| ctx->nal_parser.is_end_of_frame()) &&
1108	1272	ctx->image_units.empty()) {
1109	1273
1110	1274	// flush all pending pictures into output queue

1142	1306	// decode one NAL from the queue
1143	1307
1144	1308	de265_error err = DE265_OK;
1145
1146		if (ctx->nal_parser.number_of_NAL_units_pending()) {
	1309	bool did_work = false;
	1310
	1311	if (ctx->nal_parser.get_NAL_queue_length()) { // number_of_NAL_units_pending()) {
1147	1312	NAL_unit* nal = ctx->nal_parser.pop_from_NAL_queue();
1148	1313	assert(nal);
1149	1314	err = ctx->decode_NAL(nal);
1150	1315	// ctx->nal_parser.free_NAL_unit(nal); TODO: do not free NAL with new loop
	1316	did_work=true;
1151	1317	}
1152	1318	else if (ctx->nal_parser.is_end_of_frame() == true &&
1153	1319	ctx->image_units.empty()) {

1156	1322	return DE265_ERROR_WAITING_FOR_INPUT_DATA;
1157	1323	}
1158	1324	else {
1159		err = decode_some();
	1325	err = decode_some(&did_work);
1160	1326	}
1161	1327
1162	1328	if (more) {
1163	1329	// decoding error is assumed to be unrecoverable
1164		*more = (err==DE265_OK);
	1330	*more = (err==DE265_OK && did_work);
1165	1331	}
1166	1332
1167	1333	return err;

1172	1338	{
1173	1339	nal_unit_type = nal->nal_unit_type;
1174	1340
1175		IdrPicFlag = (nal->nal_unit_type == NAL_UNIT_IDR_W_RADL \|\|
1176		nal->nal_unit_type == NAL_UNIT_IDR_N_LP);
1177
1178		RapPicFlag = (nal->nal_unit_type >= 16 &&
1179		nal->nal_unit_type <= 23);
	1341	IdrPicFlag = isIdrPic(nal->nal_unit_type);
	1342	RapPicFlag = isRapPic(nal->nal_unit_type);
1180	1343	}
1181	1344
1182	1345

1247	1410	ctx->img->PicOrderCntVal);
1248	1411
1249	1412	if (ctx->img->nal_hdr.nuh_temporal_id==0 &&
1250		(isReferenceNALU(ctx->nal_unit_type) &&
1251		(!isRASL(ctx->nal_unit_type) && !isRADL(ctx->nal_unit_type))) &&
1252		1 /* sub-layer non-reference picture */) // TODO
	1413	!isSublayerNonReference(ctx->nal_unit_type) &&
	1414	!isRASL(ctx->nal_unit_type) &&
	1415	!isRADL(ctx->nal_unit_type))
1253	1416	{
1254	1417	loginfo(LogHeaders,"set prevPicOrderCntLsb/Msb\n");
1255	1418

1607	1770	}
1608	1771	}
1609	1772
1610		if (hdr->num_ref_idx_l0_active > 15) {
	1773	/*
	1774	if (hdr->num_ref_idx_l0_active > 16) {
1611	1775	ctx->add_warning(DE265_WARNING_NONEXISTING_REFERENCE_PICTURE_ACCESSED, false);
1612	1776	return false;
1613	1777	}
1614
	1778	*/
	1779
	1780	assert(hdr->num_ref_idx_l0_active <= 16);
1615	1781	for (rIdx=0; rIdx<hdr->num_ref_idx_l0_active; rIdx++) {
1616	1782	int idx = hdr->ref_pic_list_modification_flag_l0 ? hdr->list_entry_l0[rIdx] : rIdx;
1617	1783

1619	1785	hdr->LongTermRefPic[0][rIdx] = isLongTerm[0][idx];
1620	1786
1621	1787	// remember POC of referenced image (needed in motion.c, derive_collocated_motion_vector)
1622		hdr->RefPicList_POC[0][rIdx] = ctx->dpb.get_image(hdr->RefPicList[0][rIdx])->PicOrderCntVal;
1623		hdr->RefPicList_PicState[0][rIdx] = ctx->dpb.get_image(hdr->RefPicList[0][rIdx])->PicState;
	1788	de265_image* img_0_rIdx = ctx->dpb.get_image(hdr->RefPicList[0][rIdx]);
	1789	if (img_0_rIdx==NULL) {
	1790	return false;
	1791	}
	1792	hdr->RefPicList_POC[0][rIdx] = img_0_rIdx->PicOrderCntVal;
	1793	hdr->RefPicList_PicState[0][rIdx] = img_0_rIdx->PicState;
1624	1794	}
1625	1795
1626	1796

1635	1805
1636	1806	int rIdx=0;
1637	1807	while (rIdx < NumRpsCurrTempList1) {
1638		for (int i=0;i<ctx->NumPocStCurrAfter && rIdx<NumRpsCurrTempList1; rIdx++,i++)
	1808	for (int i=0;i<ctx->NumPocStCurrAfter && rIdx<NumRpsCurrTempList1; rIdx++,i++) {
1639	1809	RefPicListTemp1[rIdx] = ctx->RefPicSetStCurrAfter[i];
1640
1641		for (int i=0;i<ctx->NumPocStCurrBefore && rIdx<NumRpsCurrTempList1; rIdx++,i++)
	1810	}
	1811
	1812	for (int i=0;i<ctx->NumPocStCurrBefore && rIdx<NumRpsCurrTempList1; rIdx++,i++) {
1642	1813	RefPicListTemp1[rIdx] = ctx->RefPicSetStCurrBefore[i];
	1814	}
1643	1815
1644	1816	for (int i=0;i<ctx->NumPocLtCurr && rIdx<NumRpsCurrTempList1; rIdx++,i++) {
1645	1817	RefPicListTemp1[rIdx] = ctx->RefPicSetLtCurr[i];
1646	1818	isLongTerm[1][rIdx] = true;
1647	1819	}
1648		}
1649
1650		assert(hdr->num_ref_idx_l1_active <= 15);
	1820
	1821	// This check is to prevent an endless loop when no images are added above.
	1822	if (rIdx==0) {
	1823	ctx->add_warning(DE265_WARNING_FAULTY_REFERENCE_PICTURE_LIST, false);
	1824	return false;
	1825	}
	1826	}
	1827
	1828	if (hdr->num_ref_idx_l0_active > 16) {
	1829	ctx->add_warning(DE265_WARNING_NONEXISTING_REFERENCE_PICTURE_ACCESSED, false);
	1830	return false;
	1831	}
	1832
	1833	assert(hdr->num_ref_idx_l1_active <= 16);
1651	1834	for (rIdx=0; rIdx<hdr->num_ref_idx_l1_active; rIdx++) {
1652	1835	int idx = hdr->ref_pic_list_modification_flag_l1 ? hdr->list_entry_l1[rIdx] : rIdx;
1653	1836

1655	1838	hdr->LongTermRefPic[1][rIdx] = isLongTerm[1][idx];
1656	1839
1657	1840	// remember POC of referenced imaged (needed in motion.c, derive_collocated_motion_vector)
1658		hdr->RefPicList_POC[1][rIdx] = ctx->dpb.get_image(hdr->RefPicList[1][rIdx])->PicOrderCntVal;
1659		hdr->RefPicList_PicState[1][rIdx] = ctx->dpb.get_image(hdr->RefPicList[1][rIdx])->PicState;
	1841	de265_image* img_1_rIdx = ctx->dpb.get_image(hdr->RefPicList[1][rIdx]);
	1842	if (img_1_rIdx == NULL) { return false; }
	1843	hdr->RefPicList_POC[1][rIdx] = img_1_rIdx->PicOrderCntVal;
	1844	hdr->RefPicList_PicState[1][rIdx] = img_1_rIdx->PicState;
1660	1845	}
1661	1846	}
1662	1847

1837	2022	ctx->img = img;
1838	2023
1839	2024	img->vps = *ctx->current_vps;
1840		img->sps = *ctx->current_sps;
	2025	//img->sps = *ctx->current_sps; // already set in new_image()
1841	2026	img->pps = *ctx->current_pps;
1842	2027	img->decctx = ctx;
1843	2028

1892	2077
1893	2078	first_decoded_picture = false;
1894	2079	}
	2080	else {
	2081	// claims to be not the first slice, but there is no active image available
	2082
	2083	if (ctx->img == NULL) {
	2084	return false;
	2085	}
	2086	}
1895	2087
1896	2088	if (hdr->slice_type == SLICE_TYPE_B \|\|
1897	2089	hdr->slice_type == SLICE_TYPE_P)

+131

-46

libde265/decctx.h less more

37	37	#define DE265_MAX_VPS_SETS 16 // this is the maximum as defined in the standard
38	38	#define DE265_MAX_SPS_SETS 16 // this is the maximum as defined in the standard
39	39	#define DE265_MAX_PPS_SETS 64 // this is the maximum as defined in the standard
40		#define MAX_THREAD_CONTEXTS 68 // enough for 4K @ 32 pixel CTBs, but TODO: make this dynamic
41	40
42	41	#define MAX_WARNINGS 20
43	42
44	43
45		struct slice_segment_header;
46		struct image_unit;
47
48
49		struct thread_context
	44	class slice_segment_header;
	45	class image_unit;
	46	class slice_unit;
	47	class decoder_context;
	48
	49
	50	class thread_context
50	51	{
	52	public:
51	53	thread_context();
52	54
53	55	int CtbAddrInRS;

58	60
59	61	// motion vectors
60	62
61		int8_t refIdx[2];
62		int16_t mvd[2][2]; // only in top left position
63		uint8_t merge_flag;
64		uint8_t merge_idx;
65		uint8_t mvp_lX_flag[2];
66		uint8_t inter_pred_idc; // enum InterPredIdc
	63	motion_spec motion;
67	64
68	65
69	66	// prediction
70	67
71		enum IntraPredMode IntraPredModeC; // chroma intra-prediction mode for current CB
	68	// enum IntraPredMode IntraPredModeC[4]; // chroma intra-prediction mode for current CB
	69	int ResScaleVal;
72	70
73	71
74	72	// residual data
75	73
76	74	uint8_t cu_transquant_bypass_flag;
77	75	uint8_t transform_skip_flag[3];
	76	uint8_t explicit_rdpcm_flag;
	77	uint8_t explicit_rdpcm_dir;
78	78
79	79	ALIGNED_16(int16_t) _coeffBuf[(32*32)+8]; // alignment required for SSE code !
80	80	int16_t *coeffBuf;

83	83	int16_t coeffPos[3][32*32];
84	84	int16_t nCoeff[3];
85	85
	86	int32_t residual_luma[32*32]; // only used when cross-comp-prediction is enabled
	87
86	88
87	89	// quantization
88	90
89	91	int IsCuQpDeltaCoded;
90	92	int CuQpDelta;
	93	int IsCuChromaQpOffsetCoded;
	94	int CuQpOffsetCb, CuQpOffsetCr;
91	95
92	96	int currentQPY;
93	97	int currentQG_x, currentQG_y;

97	101
98	102	CABAC_decoder cabac_decoder;
99	103
100		context_model ctx_model[CONTEXT_MODEL_TABLE_LENGTH];
101
102		struct decoder_context* decctx;
	104	context_model_table ctx_model;
	105	uint8_t StatCoeff[4];
	106
	107	decoder_context* decctx;
103	108	struct de265_image *img;
104		struct slice_segment_header* shdr;
105
106		struct image_unit* imgunit;
107		struct thread_task* task; // executing thread_task or NULL if not multi-threaded
	109	slice_segment_header* shdr;
	110
	111	image_unit* imgunit;
	112	slice_unit* sliceunit;
	113	thread_task* task; // executing thread_task or NULL if not multi-threaded
108	114
109	115	private:
110	116	thread_context(const thread_context&); // not allowed

130	136
131	137
132	138
133		struct slice_unit
	139	class slice_unit
134	140	{
	141	public:
135	142	slice_unit(decoder_context* decctx);
136	143	~slice_unit();
137	144

139	146	slice_segment_header* shdr; // not the owner (de265_image is owner)
140	147	bitreader reader;
141	148
142		struct image_unit* imgunit;
	149	image_unit* imgunit;
143	150
144	151	bool flush_reorder_buffer;
145	152
146		enum { Unprocessed,
147		Inprogress,
148		Decoded
	153
	154	// decoding status
	155
	156	enum SliceDecodingProgress { Unprocessed,
	157	InProgress,
	158	Decoded
149	159	} state;
150	160
	161	de265_progress_lock finished_threads;
	162	int nThreads;
	163
	164	int first_decoded_CTB_RS; // TODO
	165	int last_decoded_CTB_RS; // TODO
	166
151	167	void allocate_thread_contexts(int n);
152		thread_context* get_thread_context(int n) { return &thread_contexts[n]; }
	168	thread_context* get_thread_context(int n) {
	169	assert(n < nThreadContexts);
	170	return &thread_contexts[n];
	171	}
	172	int num_thread_contexts() const { return nThreadContexts; }
153	173
154	174	private:
155	175	thread_context* thread_contexts; /* NOTE: cannot use std::vector, because thread_context has
156	176	no copy constructor. */
157
	177	int nThreadContexts;
	178
	179	public:
158	180	decoder_context* ctx;
159	181
160
	182	private:
161	183	slice_unit(const slice_unit&); // not allowed
162	184	const slice_unit& operator=(const slice_unit&); // not allowed
163	185	};
164	186
165	187
166		struct image_unit
	188	class image_unit
167	189	{
	190	public:
168	191	image_unit();
169	192	~image_unit();
170	193

173	196
174	197	std::vector<slice_unit*> slice_units;
175	198	std::vector<sei_message> suffix_SEIs;
	199
	200	slice_unit* get_next_unprocessed_slice_segment() const {
	201	for (int i=0;i<slice_units.size();i++) {
	202	if (slice_units[i]->state == slice_unit::Unprocessed) {
	203	return slice_units[i];
	204	}
	205	}
	206
	207	return NULL;
	208	}
	209
	210	slice_unit* get_prev_slice_segment(slice_unit* s) const {
	211	for (int i=1; i<slice_units.size(); i++) {
	212	if (slice_units[i]==s) {
	213	return slice_units[i-1];
	214	}
	215	}
	216
	217	return NULL;
	218	}
	219
	220	slice_unit* get_next_slice_segment(slice_unit* s) const {
	221	for (int i=0; i<slice_units.size()-1; i++) {
	222	if (slice_units[i]==s) {
	223	return slice_units[i+1];
	224	}
	225	}
	226
	227	return NULL;
	228	}
	229
	230	void dump_slices() const {
	231	for (int i=0; i<slice_units.size(); i++) {
	232	printf("[%d] = %p\n",i,slice_units[i]);
	233	}
	234	}
	235
	236	bool all_slice_segments_processed() const {
	237	if (slice_units.size()==0) return true;
	238	if (slice_units.back()->state != slice_unit::Unprocessed) return true;
	239	return false;
	240	}
	241
	242	bool is_first_slice_segment(const slice_unit* s) const {
	243	if (slice_units.size()==0) return false;
	244	return (slice_units[0] == s);
	245	}
176	246
177	247	enum { Invalid, // headers not read yet
178	248	Unknown, // SPS/PPS available

191	261	/* Saved context models for WPP.
192	262	There is one saved model for the initialization of each CTB row.
193	263	The array is unused for non-WPP streams. */
194		std::vector<context_model> ctx_models; // TODO: move this into image ?
	264	std::vector<context_model_table> ctx_models; // TODO: move this into image ?
195	265	};
196	266
197	267
198
199		class decoder_context : public error_queue {
	268	class base_context : public error_queue
	269	{
	270	public:
	271	base_context();
	272	virtual ~base_context() { }
	273
	274	// --- accelerated DSP functions ---
	275
	276	void set_acceleration_functions(enum de265_acceleration);
	277
	278	struct acceleration_functions acceleration; // CPU optimized functions
	279
	280	//virtual /* / de265_image get_image(int dpb_index) { return dpb.get_image(dpb_index); }
	281	virtual const de265_image* get_image(int frame_id) const = 0;
	282	virtual bool has_image(int frame_id) const = 0;
	283	};
	284
	285
	286	class decoder_context : public base_context {
200	287	public:
201	288	decoder_context();
202	289	~decoder_context();

223	310	de265_error decode_NAL(NAL_unit* nal);
224	311
225	312	de265_error decode(int* more);
226		de265_error decode_some();
	313	de265_error decode_some(bool* did_work);
227	314
228	315	de265_error decode_slice_unit_sequential(image_unit* imgunit, slice_unit* sliceunit);
229	316	de265_error decode_slice_unit_parallel(image_unit* imgunit, slice_unit* sliceunit);
230	317	de265_error decode_slice_unit_WPP(image_unit* imgunit, slice_unit* sliceunit);
231	318	de265_error decode_slice_unit_tiles(image_unit* imgunit, slice_unit* sliceunit);
232	319
	320
233	321	void process_nal_hdr(nal_header*);
234	322	void process_vps(video_parameter_set*);
235	323	void process_sps(seq_parameter_set*);

263	351
264	352	de265_image_allocation param_image_allocation_functions;
265	353	void* param_image_allocation_userdata;
266
267
268		// --- accelerated DSP functions ---
269
270		void set_acceleration_functions(enum de265_acceleration);
271
272		struct acceleration_functions acceleration; // CPU optimized functions
273	354
274	355
275	356	// --- input stream data ---

308	389	pic_parameter_set* current_pps;
309	390
310	391	public:
311		struct thread_pool thread_pool;
	392	thread_pool thread_pool_;
312	393
313	394	private:
314	395	int num_worker_threads;

414	495	bool flush_reorder_buffer_at_this_frame;
415	496
416	497	private:
417		void init_thread_context(class thread_context* tctx);
418		void add_task_decode_CTB_row(thread_context* tctx, bool firstSliceSubstream);
419		void add_task_decode_slice_segment(thread_context* tctx, bool firstSliceSubstream);
420
	498	void init_thread_context(thread_context* tctx);
	499	void add_task_decode_CTB_row(thread_context* tctx, bool firstSliceSubstream, int ctbRow);
	500	void add_task_decode_slice_segment(thread_context* tctx, bool firstSliceSubstream,
	501	int ctbX,int ctbY);
	502
	503	void mark_whole_slice_as_processed(image_unit* imgunit,
	504	slice_unit* sliceunit,
	505	int progress);
421	506
422	507	void process_picture_order_count(decoder_context* ctx, slice_segment_header* hdr);
423	508	int generate_unavailable_reference_picture(decoder_context* ctx, const seq_parameter_set* sps,

427	512
428	513
429	514	void remove_images_from_dpb(const std::vector<int>& removeImageList);
430		void run_postprocessing_filters_sequential(de265_image* img);
	515	void run_postprocessing_filters_sequential(struct de265_image* img);
431	516	void run_postprocessing_filters_parallel(image_unit* img);
432	517	};
433	518

-1

libde265/dpb.cc less more

258	258	default: chroma = de265_chroma_420; assert(0); break; // should never happen
259	259	}
260	260
261		img->alloc_image(w,h, chroma, sps, true, decctx, pts, user_data, isOutputImage);
	261	img->alloc_image(w,h, chroma, sps, true, decctx, NULL, pts, user_data, isOutputImage);
262	262
263	263	img->integrity = INTEGRITY_CORRECT;
264	264

+23

-14

libde265/dpb.h less more

26	26	#include <deque>
27	27	#include <vector>
28	28
	29	class decoder_context;
29	30
30
31		struct decoded_picture_buffer {
	31	class decoded_picture_buffer {
	32	public:
32	33	decoded_picture_buffer();
33	34	~decoded_picture_buffer();
34	35

39	40	If there is no space for a new image, return -1. */
40	41	int new_image(const seq_parameter_set* sps, decoder_context* decctx,
41	42	de265_PTS pts, void* user_data, bool isOutputImage);
42
	43
43	44	/* Check for a free slot in the DPB. There are some slots reserved for
44	45	unavailable reference frames. If high_priority==true, these reserved slots
45	46	are included in the check. */

51	52	int size() const { return dpb.size(); }
52	53
53	54	/* Raw access to the images. */
54		/* / de265_image get_image(int index) { return dpb[index]; }
55		const de265_image* get_image(int index) const { return dpb[index]; }
	55
	56	/* / de265_image get_image(int index) {
	57	if (index>=dpb.size()) return NULL;
	58	return dpb[index];
	59	}
	60
	61	const de265_image* get_image(int index) const {
	62	if (index>=dpb.size()) return NULL;
	63	return dpb[index];
	64	}
56	65
57	66	/* Search DPB for the slot index of a specific picture. */
58	67	int DPB_index_of_picture_with_POC(int poc, int currentID, bool preferLongTerm=false) const;
59	68	int DPB_index_of_picture_with_LSB(int lsb, int currentID, bool preferLongTerm=false) const;
60	69	int DPB_index_of_picture_with_ID (int id) const;
61
	70
62	71
63	72	// --- reorder buffer ---
64	73
65		void insert_image_into_reorder_buffer(de265_image* img) {
	74	void insert_image_into_reorder_buffer(struct de265_image* img) {
66	75	reorder_output_queue.push_back(img);
67	76	}
68	77

70	79
71	80	// move next picture in reorder buffer to output queue
72	81	void output_next_picture_in_reorder_buffer();
73
	82
74	83	// Move all pictures in reorder buffer to output buffer. Return true if there were any pictures.
75	84	bool flush_reorder_buffer();
76
	85
77	86
78	87	// --- output buffer ---
79	88
80	89	int num_pictures_in_output_queue() const { return image_output_queue.size(); }
81	90
82	91	/* Get the next picture in the output queue, but do not remove it from the queue. */
83		de265_image* get_next_picture_in_output_queue() const { return image_output_queue.front(); }
	92	struct de265_image* get_next_picture_in_output_queue() const { return image_output_queue.front(); }
84	93
85	94	/* Remove the next picture in the output queue. */
86	95	void pop_next_picture_in_output_queue();

90	99
91	100	void log_dpb_content() const;
92	101	void log_dpb_queues() const;
93
	102
94	103	private:
95	104	int max_images_in_DPB;
96	105	int norm_images_in_DPB;
97	106
98		std::vector<de265_image*> dpb; // decoded picture buffer
	107	std::vector<struct de265_image*> dpb; // decoded picture buffer
99	108
100		std::vector<de265_image*> reorder_output_queue;
101		std::deque<de265_image*> image_output_queue;
	109	std::vector<struct de265_image*> reorder_output_queue;
	110	std::deque<struct de265_image*> image_output_queue;
102	111
103	112	private:
104	113	decoded_picture_buffer(const decoded_picture_buffer&); // no copy

+320

-0

libde265/en265.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "libde265/en265.h"
	23	#include "libde265/encoder/encoder-context.h"
	24
	25
	26	LIBDE265_API en265_encoder_context* en265_new_encoder(void)
	27	{
	28	de265_error init_err = de265_init();
	29	if (init_err != DE265_OK) {
	30	return NULL;
	31	}
	32
	33	encoder_context* ectx = new encoder_context();
	34	return (en265_encoder_context*)ectx;
	35	}
	36
	37
	38	LIBDE265_API de265_error en265_free_encoder(en265_encoder_context* e)
	39	{
	40	assert(e);
	41	encoder_context* ectx = (encoder_context*)e;
	42	delete ectx;
	43
	44	return de265_free();
	45	}
	46
	47
	48	LIBDE265_API void en265_set_image_release_function(en265_encoder_context* e,
	49	void (release_func)(en265_encoder_context,
	50	de265_image*,
	51	void* userdata),
	52	void* alloc_userdata)
	53	{
	54	assert(e);
	55	encoder_context* ectx = (encoder_context*)e;
	56
	57	ectx->param_image_allocation_userdata = alloc_userdata;
	58	ectx->release_func = release_func;
	59	}
	60
	61
	62	// ========== encoder parameters ==========
	63
	64	LIBDE265_API de265_error en265_parse_command_line_parameters(en265_encoder_context* e,
	65	int* argc, char** argv)
	66	{
	67	assert(e);
	68	encoder_context* ectx = (encoder_context*)e;
	69
	70	//if (!ectx->params_config.parse_command_line_params(argc,argv, &ectx->params, true)) {
	71	int first_idx=1;
	72	if (!ectx->params_config.parse_command_line_params(argc,argv, &first_idx, true)) {
	73	return DE265_ERROR_PARAMETER_PARSING;
	74	}
	75	else {
	76	return DE265_OK;
	77	}
	78	}
	79
	80	LIBDE265_API void en265_show_parameters(en265_encoder_context* e)
	81	{
	82	assert(e);
	83	encoder_context* ectx = (encoder_context*)e;
	84
	85	//ectx->params_config.show_params(&ectx->params);
	86
	87	ectx->params_config.print_params();
	88	}
	89
	90
	91	LIBDE265_API const char** en265_list_parameters(en265_encoder_context* e)
	92	{
	93	assert(e);
	94	encoder_context* ectx = (encoder_context*)e;
	95
	96	return ectx->params_config.get_parameter_string_table();
	97	}
	98
	99
	100	LIBDE265_API enum en265_parameter_type en265_get_parameter_type(en265_encoder_context* e,
	101	const char* parametername)
	102	{
	103	assert(e);
	104	encoder_context* ectx = (encoder_context*)e;
	105
	106	return ectx->params_config.get_parameter_type(parametername);
	107	}
	108
	109
	110	LIBDE265_API de265_error en265_set_parameter_bool(en265_encoder_context* e,
	111	const char* param,int value)
	112	{
	113	assert(e);
	114	encoder_context* ectx = (encoder_context*)e;
	115
	116	return ectx->params_config.set_bool(param,value) ? DE265_OK : DE265_ERROR_PARAMETER_PARSING;
	117	}
	118
	119
	120	LIBDE265_API de265_error en265_set_parameter_int(en265_encoder_context* e,
	121	const char* param,int value)
	122	{
	123	assert(e);
	124	encoder_context* ectx = (encoder_context*)e;
	125
	126	return ectx->params_config.set_int(param,value) ? DE265_OK : DE265_ERROR_PARAMETER_PARSING;
	127	}
	128
	129	LIBDE265_API de265_error en265_set_parameter_string(en265_encoder_context* e,
	130	const char* param,const char* value)
	131	{
	132	assert(e);
	133	encoder_context* ectx = (encoder_context*)e;
	134
	135	return ectx->params_config.set_string(param,value) ? DE265_OK : DE265_ERROR_PARAMETER_PARSING;
	136	}
	137
	138	LIBDE265_API de265_error en265_set_parameter_choice(en265_encoder_context* e,
	139	const char* param,const char* value)
	140	{
	141	assert(e);
	142	encoder_context* ectx = (encoder_context*)e;
	143
	144	return ectx->params_config.set_choice(param,value) ? DE265_OK : DE265_ERROR_PARAMETER_PARSING;
	145	}
	146
	147
	148	LIBDE265_API const char** en265_list_parameter_choices(en265_encoder_context* e,
	149	const char* parametername)
	150	{
	151	assert(e);
	152	encoder_context* ectx = (encoder_context*)e;
	153
	154	return ectx->params_config.get_parameter_choices_table(parametername);
	155	}
	156
	157
	158
	159	// ========== encoding loop ==========
	160
	161
	162	LIBDE265_API de265_error en265_start_encoder(en265_encoder_context* e, int number_of_threads)
	163	{
	164	assert(e);
	165	encoder_context* ectx = (encoder_context*)e;
	166
	167	ectx->start_encoder();
	168
	169	return DE265_OK;
	170	}
	171
	172
	173	LIBDE265_API struct de265_image* en265_allocate_image(en265_encoder_context* e,
	174	int width, int height, de265_chroma chroma,
	175	de265_PTS pts, void* image_userdata)
	176	{
	177	assert(e);
	178	encoder_context* ectx = (encoder_context*)e;
	179
	180	de265_image* img = new de265_image;
	181	if (img->alloc_image(width,height,de265_chroma_420, NULL, false,
	182	NULL,ectx, pts, image_userdata, true) != DE265_OK) {
	183	delete img;
	184	return NULL;
	185	}
	186
	187	return img;
	188	}
	189
	190	// Request a specification of the image memory layout for an image of the specified dimensions.
	191	LIBDE265_API void en265_get_image_spec(en265_encoder_context* e,
	192	int width, int height, de265_chroma chroma,
	193	struct de265_image_spec* out_spec)
	194	{
	195	out_spec->format = de265_image_format_YUV420P8;
	196	out_spec->width = width;
	197	out_spec->height= height;
	198	out_spec->alignment = 1;
	199
	200	out_spec->crop_left =0;
	201	out_spec->crop_right =0;
	202	out_spec->crop_top =0;
	203	out_spec->crop_bottom=0;
	204
	205	out_spec->visible_width = out_spec->width - out_spec->crop_left - out_spec->crop_right;
	206	out_spec->visible_height = out_spec->height - out_spec->crop_top - out_spec->crop_bottom;
	207	}
	208
	209	// Image memory layout specification for an image returned by en265_allocate_image().
	210	//LIBDE265_API void de265_get_image_spec_from_image(de265_image* img, struct de265_image_spec* spec);
	211
	212
	213
	214	LIBDE265_API de265_error en265_push_image(en265_encoder_context* e,
	215	struct de265_image* img)
	216	{
	217	assert(e);
	218	encoder_context* ectx = (encoder_context*)e;
	219
	220	ectx->sop->insert_new_input_image(img);
	221	return DE265_OK;
	222	}
	223
	224
	225	LIBDE265_API de265_error en265_push_eof(en265_encoder_context* e)
	226	{
	227	assert(e);
	228	encoder_context* ectx = (encoder_context*)e;
	229
	230	ectx->sop->insert_end_of_stream();
	231	return DE265_OK;
	232	}
	233
	234
	235	LIBDE265_API de265_error en265_block_on_input_queue_length(en265_encoder_context*,
	236	int max_pending_images,
	237	int timeout_ms)
	238	{
	239	// TODO
	240	return DE265_OK;
	241	}
	242
	243	LIBDE265_API de265_error en265_trim_input_queue(en265_encoder_context*, int max_pending_images)
	244	{
	245	// TODO
	246	return DE265_OK;
	247	}
	248
	249	LIBDE265_API int en265_current_input_queue_length(en265_encoder_context*)
	250	{
	251	// TODO
	252	return DE265_OK;
	253	}
	254
	255	LIBDE265_API de265_error en265_encode(en265_encoder_context* e)
	256	{
	257	assert(e);
	258	encoder_context* ectx = (encoder_context*)e;
	259
	260	while (ectx->picbuf.have_more_frames_to_encode())
	261	{
	262	de265_error result = ectx->encode_picture_from_input_buffer();
	263	if (result != DE265_OK) return result;
	264	}
	265
	266	return DE265_OK;
	267	}
	268
	269	LIBDE265_API enum en265_encoder_state en265_get_encoder_state(en265_encoder_context* e)
	270	{
	271	// TODO
	272	return EN265_STATE_IDLE;
	273	}
	274
	275	LIBDE265_API struct en265_packet* en265_get_packet(en265_encoder_context* e, int timeout_ms)
	276	{
	277	assert(e);
	278	encoder_context* ectx = (encoder_context*)e;
	279
	280	assert(timeout_ms==0); // TODO: blocking not implemented yet
	281
	282	if (ectx->output_packets.size()>0) {
	283	en265_packet* pck = ectx->output_packets.front();
	284	ectx->output_packets.pop_front();
	285
	286	return pck;
	287	}
	288	else {
	289	return NULL;
	290	}
	291	}
	292
	293	LIBDE265_API void en265_free_packet(en265_encoder_context* e, struct en265_packet* pck)
	294	{
	295	assert(e);
	296	encoder_context* ectx = (encoder_context*)e;
	297
	298	// Do not delete images here. They are owned by the EncPicBuf.
	299	//delete pck->input_image;
	300	//delete pck->reconstruction;
	301
	302	if (pck->frame_number >= 0) {
	303	ectx->mark_image_is_outputted(pck->frame_number);
	304
	305	ectx->release_input_image(pck->frame_number);
	306	}
	307
	308	delete[] pck->data;
	309	delete pck;
	310	}
	311
	312	LIBDE265_API int en265_number_of_queued_packets(en265_encoder_context* e)
	313	{
	314	assert(e);
	315	encoder_context* ectx = (encoder_context*)e;
	316
	317	return ectx->output_packets.size();
	318	}
	319

+216

-0

libde265/en265.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef EN265_H
	23	#define EN265_H
	24
	25	#ifdef __cplusplus
	26	extern "C" {
	27	#endif
	28
	29	#include <libde265/de265.h>
	30
	31
	32	// ========== encoder context ==========
	33
	34	typedef void en265_encoder_context; // private structure
	35
	36	/* Get a new encoder context. Must be freed with en265_free_encoder(). */
	37	LIBDE265_API en265_encoder_context* en265_new_encoder(void);
	38
	39	/* Free encoder context. May only be called once on a context. */
	40	LIBDE265_API de265_error en265_free_encoder(en265_encoder_context*);
	41
	42	/* The alloc_userdata pointer will be given to the release_func(). */
	43	LIBDE265_API void en265_set_image_release_function(en265_encoder_context*,
	44	void (release_func)(en265_encoder_context,
	45	struct de265_image*,
	46	void* userdata),
	47	void* alloc_userdata);
	48
	49	// ========== encoder parameters ==========
	50
	51	LIBDE265_API de265_error en265_set_parameter_bool(en265_encoder_context*,
	52	const char* parametername,int value);
	53	LIBDE265_API de265_error en265_set_parameter_int(en265_encoder_context*,
	54	const char* parametername,int value);
	55	LIBDE265_API de265_error en265_set_parameter_string(en265_encoder_context*,
	56	const char* parametername,const char* value);
	57	LIBDE265_API de265_error en265_set_parameter_choice(en265_encoder_context*,
	58	const char* parametername,const char* value);
	59
	60
	61	LIBDE265_API const char** en265_list_parameters(en265_encoder_context*);
	62
	63	enum en265_parameter_type {
	64	en265_parameter_bool,
	65	en265_parameter_int,
	66	en265_parameter_string,
	67	en265_parameter_choice
	68	};
	69
	70	LIBDE265_API enum en265_parameter_type en265_get_parameter_type(en265_encoder_context*,
	71	const char* parametername);
	72
	73	LIBDE265_API const char** en265_list_parameter_choices(en265_encoder_context*,
	74	const char* parametername);
	75
	76
	77	// --- convenience functions for command-line parameters ---
	78
	79	LIBDE265_API de265_error en265_parse_command_line_parameters(en265_encoder_context*,
	80	int* argc, char** argv);
	81	LIBDE265_API void en265_show_parameters(en265_encoder_context*);
	82
	83
	84
	85	// ========== encoding loop ==========
	86
	87	LIBDE265_API de265_error en265_start_encoder(en265_encoder_context*, int number_of_threads);
	88
	89	// If we have provided our own memory release function, no image memory will be allocated.
	90	LIBDE265_API struct de265_image* en265_allocate_image(en265_encoder_context*,
	91	int width, int height,
	92	enum de265_chroma chroma,
	93	de265_PTS pts, void* image_userdata);
	94
	95	LIBDE265_API void* de265_alloc_image_plane(struct de265_image* img, int cIdx,
	96	void* inputdata, int inputstride, void *userdata);
	97	LIBDE265_API void de265_free_image_plane(struct de265_image* img, int cIdx);
	98
	99
	100	// Request a specification of the image memory layout for an image of the specified dimensions.
	101	LIBDE265_API void en265_get_image_spec(en265_encoder_context*,
	102	int width, int height, enum de265_chroma chroma,
	103	struct de265_image_spec* out_spec);
	104
	105	// Image memory layout specification for an image returned by en265_allocate_image().
	106	/* TODO: do we need this?
	107	LIBDE265_API void de265_get_image_spec_from_image(de265_image* img, struct de265_image_spec* spec);
	108	*/
	109
	110
	111	LIBDE265_API de265_error en265_push_image(en265_encoder_context*,
	112	struct de265_image*); // non-blocking
	113
	114	LIBDE265_API de265_error en265_push_eof(en265_encoder_context*);
	115
	116	// block when there are more than max_input_images in the input queue
	117	LIBDE265_API de265_error en265_block_on_input_queue_length(en265_encoder_context*,
	118	int max_pending_images,
	119	int timeout_ms);
	120
	121	LIBDE265_API de265_error en265_trim_input_queue(en265_encoder_context*, int max_pending_images);
	122
	123	LIBDE265_API int en265_current_input_queue_length(en265_encoder_context*);
	124
	125	// Run encoder in main thread. Only use this when not using background threads.
	126	LIBDE265_API de265_error en265_encode(en265_encoder_context*);
	127
	128	enum en265_encoder_state
	129	{
	130	EN265_STATE_IDLE,
	131	EN265_STATE_WAITING_FOR_INPUT,
	132	EN265_STATE_WORKING,
	133	EN265_STATE_OUTPUT_QUEUE_FULL,
	134	EN265_STATE_EOS
	135	};
	136
	137
	138	LIBDE265_API enum en265_encoder_state en265_get_encoder_state(en265_encoder_context*);
	139
	140
	141	enum en265_packet_content_type {
	142	EN265_PACKET_VPS,
	143	EN265_PACKET_SPS,
	144	EN265_PACKET_PPS,
	145	EN265_PACKET_SEI,
	146	EN265_PACKET_SLICE,
	147	EN265_PACKET_SKIPPED_IMAGE
	148	};
	149
	150
	151	enum en265_nal_unit_type {
	152	EN265_NUT_TRAIL_N = 0,
	153	EN265_NUT_TRAIL_R = 1,
	154	EN265_NUT_TSA_N = 2,
	155	EN265_NUT_TSA_R = 3,
	156	EN265_NUT_STSA_N = 4,
	157	EN265_NUT_STSA_R = 5,
	158	EN265_NUT_RADL_N = 6,
	159	EN265_NUT_RADL_R = 7,
	160	EN265_NUT_RASL_N = 8,
	161	EN265_NUT_RASL_R = 9,
	162	EN265_NUT_BLA_W_LP = 16,
	163	EN265_NUT_BLA_W_RADL= 17,
	164	EN265_NUT_BLA_N_LP = 18,
	165	EN265_NUT_IDR_W_RADL= 19,
	166	EN265_NUT_IDR_N_LP = 20,
	167	EN265_NUT_CRA = 21,
	168	EN265_NUT_VPS = 32,
	169	EN265_NUT_SPS = 33,
	170	EN265_NUT_PPS = 34,
	171	EN265_NUT_AUD = 35,
	172	EN265_NUT_EOS = 36,
	173	EN265_NUT_EOB = 37,
	174	EN265_NUT_FD = 38,
	175	EN265_NUT_PREFIX_SEI = 39,
	176	EN265_NUT_SUFFIX_SEI = 40
	177	};
	178
	179
	180	struct en265_packet
	181	{
	182	int version; // currently: 1
	183
	184	const uint8_t* data;
	185	int length;
	186
	187	int frame_number;
	188
	189	enum en265_packet_content_type content_type;
	190	char complete_picture : 1;
	191	char final_slice : 1;
	192	char dependent_slice : 1;
	193
	194	enum en265_nal_unit_type nal_unit_type;
	195	unsigned char nuh_layer_id;
	196	unsigned char nuh_temporal_id;
	197
	198	en265_encoder_context* encoder_context;
	199
	200	const struct de265_image* input_image;
	201	const struct de265_image* reconstruction;
	202	};
	203
	204	// timeout_ms - timeout in milliseconds. 0 - no timeout, -1 - block forever
	205	LIBDE265_API struct en265_packet* en265_get_packet(en265_encoder_context*, int timeout_ms);
	206	LIBDE265_API void en265_free_packet(en265_encoder_context, struct en265_packet);
	207
	208	LIBDE265_API int en265_number_of_queued_packets(en265_encoder_context*);
	209
	210	#ifdef __cplusplus
	211	}
	212	#endif
	213
	214
	215	#endif

+18

-0

libde265/encoder/CMakeLists.txt less more

	0	set (encoder_sources
	1	analyze.cc analyze.h
	2	encode.h encode.cc
	3	encoder-params.h encoder-params.cc
	4	encoder-context.h encoder-context.cc
	5	encpicbuf.h encpicbuf.cc
	6	sop.h sop.cc
	7	)
	8
	9	add_library(encoder STATIC ${encoder_sources})
	10
	11	if(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
	12	SET_TARGET_PROPERTIES(encoder PROPERTIES COMPILE_FLAGS "-fPIC")
	13	endif(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
	14
	15	add_subdirectory (algo)
	16
	17	target_link_libraries(encoder algo)

+16

-0

libde265/encoder/Makefile.am less more

	0	noinst_LTLIBRARIES = libde265_encoder.la
	1
	2	libde265_encoder_la_CXXFLAGS = -I..
	3	libde265_encoder_la_SOURCES = \
	4	analyze.cc analyze.h \
	5	encode.h encode.cc \
	6	encoder-params.h encoder-params.cc \
	7	encoder-context.h encoder-context.cc \
	8	encpicbuf.h encpicbuf.cc \
	9	sop.h sop.cc
	10
	11	SUBDIRS=algo
	12	libde265_encoder_la_LIBADD = algo/libde265_encoder_algo.la
	13
	14	EXTRA_DIST = \
	15	CMakeLists.txt

+807

-0

libde265/encoder/Makefile.in less more

	0	# Makefile.in generated by automake 1.14.1 from Makefile.am.
	1	# @configure_input@
	2
	3	# Copyright (C) 1994-2013 Free Software Foundation, Inc.
	4
	5	# This Makefile.in is free software; the Free Software Foundation
	6	# gives unlimited permission to copy and/or distribute it,
	7	# with or without modifications, as long as this notice is preserved.
	8
	9	# This program is distributed in the hope that it will be useful,
	10	# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
	11	# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
	12	# PARTICULAR PURPOSE.
	13
	14	@SET_MAKE@
	15
	16	VPATH = @srcdir@
	17	am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
	18	am__make_running_with_option = \
	19	case $${target_option-} in \
	20	?) ;; \
	21	*) echo "am__make_running_with_option: internal error: invalid" \
	22	"target option '$${target_option-}' specified" >&2; \
	23	exit 1;; \
	24	esac; \
	25	has_opt=no; \
	26	sane_makeflags=$$MAKEFLAGS; \
	27	if $(am__is_gnu_make); then \
	28	sane_makeflags=$$MFLAGS; \
	29	else \
	30	case $$MAKEFLAGS in \
	31	\\[\ \ ]) \
	32	bs=\\; \
	33	sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
	34	\| sed "s/$$bs$$bs[$$bs $$bs ]*//g"`;; \
	35	esac; \
	36	fi; \
	37	skip_next=no; \
	38	strip_trailopt () \
	39	{ \
	40	flg=`printf '%s\n' "$$flg" \| sed "s/$$1.*$$//"`; \
	41	}; \
	42	for flg in $$sane_makeflags; do \
	43	test $$skip_next = yes && { skip_next=no; continue; }; \
	44	case $$flg in \
	45	=\|--*) continue;; \
	46	-*I) strip_trailopt 'I'; skip_next=yes;; \
	47	-I?) strip_trailopt 'I';; \
	48	-*O) strip_trailopt 'O'; skip_next=yes;; \
	49	-O?) strip_trailopt 'O';; \
	50	-*l) strip_trailopt 'l'; skip_next=yes;; \
	51	-l?) strip_trailopt 'l';; \
	52	-[dEDm]) skip_next=yes;; \
	53	-[JT]) skip_next=yes;; \
	54	esac; \
	55	case $$flg in \
	56	$$target_option) has_opt=yes; break;; \
	57	esac; \
	58	done; \
	59	test $$has_opt = yes
	60	am__make_dryrun = (target_option=n; $(am__make_running_with_option))
	61	am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
	62	pkgdatadir = $(datadir)/@PACKAGE@
	63	pkgincludedir = $(includedir)/@PACKAGE@
	64	pkglibdir = $(libdir)/@PACKAGE@
	65	pkglibexecdir = $(libexecdir)/@PACKAGE@
	66	am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
	67	install_sh_DATA = $(install_sh) -c -m 644
	68	install_sh_PROGRAM = $(install_sh) -c
	69	install_sh_SCRIPT = $(install_sh) -c
	70	INSTALL_HEADER = $(INSTALL_DATA)
	71	transform = $(program_transform_name)
	72	NORMAL_INSTALL = :
	73	PRE_INSTALL = :
	74	POST_INSTALL = :
	75	NORMAL_UNINSTALL = :
	76	PRE_UNINSTALL = :
	77	POST_UNINSTALL = :
	78	build_triplet = @build@
	79	host_triplet = @host@
	80	target_triplet = @target@
	81	subdir = libde265/encoder
	82	DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/Makefile.am \
	83	$(top_srcdir)/depcomp
	84	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
	85	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	86	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
	87	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
	88	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
	89	$(top_srcdir)/m4/lt~obsolete.m4 \
	90	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
	91	$(top_srcdir)/configure.ac
	92	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
	93	$(ACLOCAL_M4)
	94	mkinstalldirs = $(install_sh) -d
	95	CONFIG_HEADER = $(top_builddir)/config.h
	96	CONFIG_CLEAN_FILES =
	97	CONFIG_CLEAN_VPATH_FILES =
	98	LTLIBRARIES = $(noinst_LTLIBRARIES)
	99	libde265_encoder_la_DEPENDENCIES = algo/libde265_encoder_algo.la
	100	am_libde265_encoder_la_OBJECTS = libde265_encoder_la-analyze.lo \
	101	libde265_encoder_la-encode.lo \
	102	libde265_encoder_la-encoder-params.lo \
	103	libde265_encoder_la-encoder-context.lo \
	104	libde265_encoder_la-encpicbuf.lo libde265_encoder_la-sop.lo
	105	libde265_encoder_la_OBJECTS = $(am_libde265_encoder_la_OBJECTS)
	106	AM_V_lt = $(am__v_lt_@AM_V@)
	107	am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
	108	am__v_lt_0 = --silent
	109	am__v_lt_1 =
	110	libde265_encoder_la_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
	111	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
	112	$(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) \
	113	$(LDFLAGS) -o $@
	114	AM_V_P = $(am__v_P_@AM_V@)
	115	am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
	116	am__v_P_0 = false
	117	am__v_P_1 = :
	118	AM_V_GEN = $(am__v_GEN_@AM_V@)
	119	am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
	120	am__v_GEN_0 = @echo " GEN " $@;
	121	am__v_GEN_1 =
	122	AM_V_at = $(am__v_at_@AM_V@)
	123	am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
	124	am__v_at_0 = @
	125	am__v_at_1 =
	126	DEFAULT_INCLUDES = -I.@am__isrc@ -I$(top_builddir)
	127	depcomp = $(SHELL) $(top_srcdir)/depcomp
	128	am__depfiles_maybe = depfiles
	129	am__mv = mv -f
	130	CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
	131	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
	132	LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	133	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
	134	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	135	$(AM_CXXFLAGS) $(CXXFLAGS)
	136	AM_V_CXX = $(am__v_CXX_@AM_V@)
	137	am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
	138	am__v_CXX_0 = @echo " CXX " $@;
	139	am__v_CXX_1 =
	140	CXXLD = $(CXX)
	141	CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	142	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
	143	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
	144	AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
	145	am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
	146	am__v_CXXLD_0 = @echo " CXXLD " $@;
	147	am__v_CXXLD_1 =
	148	COMPILE = $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) \
	149	$(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS)
	150	LTCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CC $(AM_LIBTOOLFLAGS) \
	151	$(LIBTOOLFLAGS) --mode=compile $(CC) $(DEFS) \
	152	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	153	$(AM_CFLAGS) $(CFLAGS)
	154	AM_V_CC = $(am__v_CC_@AM_V@)
	155	am__v_CC_ = $(am__v_CC_@AM_DEFAULT_V@)
	156	am__v_CC_0 = @echo " CC " $@;
	157	am__v_CC_1 =
	158	CCLD = $(CC)
	159	LINK = $(LIBTOOL) $(AM_V_lt) --tag=CC $(AM_LIBTOOLFLAGS) \
	160	$(LIBTOOLFLAGS) --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
	161	$(AM_LDFLAGS) $(LDFLAGS) -o $@
	162	AM_V_CCLD = $(am__v_CCLD_@AM_V@)
	163	am__v_CCLD_ = $(am__v_CCLD_@AM_DEFAULT_V@)
	164	am__v_CCLD_0 = @echo " CCLD " $@;
	165	am__v_CCLD_1 =
	166	SOURCES = $(libde265_encoder_la_SOURCES)
	167	DIST_SOURCES = $(libde265_encoder_la_SOURCES)
	168	RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
	169	ctags-recursive dvi-recursive html-recursive info-recursive \
	170	install-data-recursive install-dvi-recursive \
	171	install-exec-recursive install-html-recursive \
	172	install-info-recursive install-pdf-recursive \
	173	install-ps-recursive install-recursive installcheck-recursive \
	174	installdirs-recursive pdf-recursive ps-recursive \
	175	tags-recursive uninstall-recursive
	176	am__can_run_installinfo = \
	177	case $$AM_UPDATE_INFO_DIR in \
	178	n\|no\|NO) false;; \
	179	*) (install-info --version) >/dev/null 2>&1;; \
	180	esac
	181	RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive \
	182	distclean-recursive maintainer-clean-recursive
	183	am__recursive_targets = \
	184	$(RECURSIVE_TARGETS) \
	185	$(RECURSIVE_CLEAN_TARGETS) \
	186	$(am__extra_recursive_targets)
	187	AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
	188	distdir
	189	am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
	190	# Read a list of newline-separated strings from the standard input,
	191	# and print each of them once, without duplicates. Input order is
	192	# not preserved.
	193	am__uniquify_input = $(AWK) '\
	194	BEGIN { nonempty = 0; } \
	195	{ items[$$0] = 1; nonempty = 1; } \
	196	END { if (nonempty) { for (i in items) print i; }; } \
	197	'
	198	# Make sure the list of sources is unique. This is necessary because,
	199	# e.g., the same source file might be shared among _SOURCES variables
	200	# for different programs/libraries.
	201	am__define_uniq_tagged_files = \
	202	list='$(am__tagged_files)'; \
	203	unique=`for i in $$list; do \
	204	if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
	205	done \| $(am__uniquify_input)`
	206	ETAGS = etags
	207	CTAGS = ctags
	208	DIST_SUBDIRS = $(SUBDIRS)
	209	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
	210	am__relativize = \
	211	dir0=`pwd`; \
	212	sed_first='s,^$[^/]$/.$$,\1,'; \
	213	sed_rest='s,^[^/]/,,'; \
	214	sed_last='s,^./$[^/]$$$,\1,'; \
	215	sed_butlast='s,/[^/]$$,,'; \
	216	while test -n "$$dir1"; do \
	217	first=`echo "$$dir1" \| sed -e "$$sed_first"`; \
	218	if test "$$first" != "."; then \
	219	if test "$$first" = ".."; then \
	220	dir2=`echo "$$dir0" \| sed -e "$$sed_last"`/"$$dir2"; \
	221	dir0=`echo "$$dir0" \| sed -e "$$sed_butlast"`; \
	222	else \
	223	first2=`echo "$$dir2" \| sed -e "$$sed_first"`; \
	224	if test "$$first2" = "$$first"; then \
	225	dir2=`echo "$$dir2" \| sed -e "$$sed_rest"`; \
	226	else \
	227	dir2="../$$dir2"; \
	228	fi; \
	229	dir0="$$dir0"/"$$first"; \
	230	fi; \
	231	fi; \
	232	dir1=`echo "$$dir1" \| sed -e "$$sed_rest"`; \
	233	done; \
	234	reldir="$$dir2"
	235	ACLOCAL = @ACLOCAL@
	236	ALLOCA = @ALLOCA@
	237	AMTAR = @AMTAR@
	238	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
	239	AR = @AR@
	240	AUTOCONF = @AUTOCONF@
	241	AUTOHEADER = @AUTOHEADER@
	242	AUTOMAKE = @AUTOMAKE@
	243	AWK = @AWK@
	244	CC = @CC@
	245	CCAS = @CCAS@
	246	CCASDEPMODE = @CCASDEPMODE@
	247	CCASFLAGS = @CCASFLAGS@
	248	CCDEPMODE = @CCDEPMODE@
	249	CFLAGS = @CFLAGS@
	250	CPP = @CPP@
	251	CPPFLAGS = @CPPFLAGS@
	252	CXX = @CXX@
	253	CXXCPP = @CXXCPP@
	254	CXXDEPMODE = @CXXDEPMODE@
	255	CXXFLAGS = @CXXFLAGS@
	256	CYGPATH_W = @CYGPATH_W@
	257	DEFS = @DEFS@
	258	DEPDIR = @DEPDIR@
	259	DLLTOOL = @DLLTOOL@
	260	DSYMUTIL = @DSYMUTIL@
	261	DUMPBIN = @DUMPBIN@
	262	ECHO_C = @ECHO_C@
	263	ECHO_N = @ECHO_N@
	264	ECHO_T = @ECHO_T@
	265	EGREP = @EGREP@
	266	EXEEXT = @EXEEXT@
	267	FGREP = @FGREP@
	268	GREP = @GREP@
	269	HAVE_CXX11 = @HAVE_CXX11@
	270	INSTALL = @INSTALL@
	271	INSTALL_DATA = @INSTALL_DATA@
	272	INSTALL_PROGRAM = @INSTALL_PROGRAM@
	273	INSTALL_SCRIPT = @INSTALL_SCRIPT@
	274	INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
	275	LD = @LD@
	276	LDFLAGS = @LDFLAGS@
	277	LIBDE265_AGE = @LIBDE265_AGE@
	278	LIBDE265_CURRENT = @LIBDE265_CURRENT@
	279	LIBDE265_REVISION = @LIBDE265_REVISION@
	280	LIBOBJS = @LIBOBJS@
	281	LIBS = @LIBS@
	282	LIBTOOL = @LIBTOOL@
	283	LIPO = @LIPO@
	284	LN_S = @LN_S@
	285	LTLIBOBJS = @LTLIBOBJS@
	286	MAKEINFO = @MAKEINFO@
	287	MANIFEST_TOOL = @MANIFEST_TOOL@
	288	MKDIR_P = @MKDIR_P@
	289	NM = @NM@
	290	NMEDIT = @NMEDIT@
	291	NUMERIC_VERSION = @NUMERIC_VERSION@
	292	OBJDUMP = @OBJDUMP@
	293	OBJEXT = @OBJEXT@
	294	OTOOL = @OTOOL@
	295	OTOOL64 = @OTOOL64@
	296	PACKAGE = @PACKAGE@
	297	PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
	298	PACKAGE_NAME = @PACKAGE_NAME@
	299	PACKAGE_STRING = @PACKAGE_STRING@
	300	PACKAGE_TARNAME = @PACKAGE_TARNAME@
	301	PACKAGE_URL = @PACKAGE_URL@
	302	PACKAGE_VERSION = @PACKAGE_VERSION@
	303	PATH_SEPARATOR = @PATH_SEPARATOR@
	304	PKG_CONFIG = @PKG_CONFIG@
	305	PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
	306	PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
	307	QTCHOOSER = @QTCHOOSER@
	308	QTMOC = @QTMOC@
	309	QT_CFLAGS = @QT_CFLAGS@
	310	QT_LIBS = @QT_LIBS@
	311	RANLIB = @RANLIB@
	312	SDL_CFLAGS = @SDL_CFLAGS@
	313	SDL_LIBS = @SDL_LIBS@
	314	SED = @SED@
	315	SET_MAKE = @SET_MAKE@
	316	SHELL = @SHELL@
	317	STRIP = @STRIP@
	318	SWSCALE_CFLAGS = @SWSCALE_CFLAGS@
	319	SWSCALE_LIBS = @SWSCALE_LIBS@
	320	VERSION = @VERSION@
	321	VIDEOGFX_CFLAGS = @VIDEOGFX_CFLAGS@
	322	VIDEOGFX_LIBS = @VIDEOGFX_LIBS@
	323	abs_builddir = @abs_builddir@
	324	abs_srcdir = @abs_srcdir@
	325	abs_top_builddir = @abs_top_builddir@
	326	abs_top_srcdir = @abs_top_srcdir@
	327	ac_ct_AR = @ac_ct_AR@
	328	ac_ct_CC = @ac_ct_CC@
	329	ac_ct_CXX = @ac_ct_CXX@
	330	ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
	331	am__include = @am__include@
	332	am__leading_dot = @am__leading_dot@
	333	am__quote = @am__quote@
	334	am__tar = @am__tar@
	335	am__untar = @am__untar@
	336	bindir = @bindir@
	337	build = @build@
	338	build_alias = @build_alias@
	339	build_cpu = @build_cpu@
	340	build_os = @build_os@
	341	build_vendor = @build_vendor@
	342	builddir = @builddir@
	343	datadir = @datadir@
	344	datarootdir = @datarootdir@
	345	docdir = @docdir@
	346	dvidir = @dvidir@
	347	exec_prefix = @exec_prefix@
	348	host = @host@
	349	host_alias = @host_alias@
	350	host_cpu = @host_cpu@
	351	host_os = @host_os@
	352	host_vendor = @host_vendor@
	353	htmldir = @htmldir@
	354	includedir = @includedir@
	355	infodir = @infodir@
	356	install_sh = @install_sh@
	357	libdir = @libdir@
	358	libexecdir = @libexecdir@
	359	localedir = @localedir@
	360	localstatedir = @localstatedir@
	361	mandir = @mandir@
	362	mkdir_p = @mkdir_p@
	363	oldincludedir = @oldincludedir@
	364	pdfdir = @pdfdir@
	365	prefix = @prefix@
	366	program_transform_name = @program_transform_name@
	367	psdir = @psdir@
	368	sbindir = @sbindir@
	369	sharedstatedir = @sharedstatedir@
	370	srcdir = @srcdir@
	371	sysconfdir = @sysconfdir@
	372	target = @target@
	373	target_alias = @target_alias@
	374	target_cpu = @target_cpu@
	375	target_os = @target_os@
	376	target_vendor = @target_vendor@
	377	top_build_prefix = @top_build_prefix@
	378	top_builddir = @top_builddir@
	379	top_srcdir = @top_srcdir@
	380	noinst_LTLIBRARIES = libde265_encoder.la
	381	libde265_encoder_la_CXXFLAGS = -I..
	382	libde265_encoder_la_SOURCES = \
	383	analyze.cc analyze.h \
	384	encode.h encode.cc \
	385	encoder-params.h encoder-params.cc \
	386	encoder-context.h encoder-context.cc \
	387	encpicbuf.h encpicbuf.cc \
	388	sop.h sop.cc
	389
	390	SUBDIRS = algo
	391	libde265_encoder_la_LIBADD = algo/libde265_encoder_algo.la
	392	EXTRA_DIST = \
	393	CMakeLists.txt
	394
	395	all: all-recursive
	396
	397	.SUFFIXES:
	398	.SUFFIXES: .cc .lo .o .obj
	399	$(srcdir)/Makefile.in: $(srcdir)/Makefile.am $(am__configure_deps)
	400	@for dep in $?; do \
	401	case '$(am__configure_deps)' in \
	402	$$dep) \
	403	( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
	404	&& { if test -f $@; then exit 0; else break; fi; }; \
	405	exit 1;; \
	406	esac; \
	407	done; \
	408	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu libde265/encoder/Makefile'; \
	409	$(am__cd) $(top_srcdir) && \
	410	$(AUTOMAKE) --gnu libde265/encoder/Makefile
	411	.PRECIOUS: Makefile
	412	Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
	413	@case '$?' in \
	414	config.status) \
	415	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
	416	*) \
	417	echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
	418	cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
	419	esac;
	420
	421	$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
	422	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	423
	424	$(top_srcdir)/configure: $(am__configure_deps)
	425	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	426	$(ACLOCAL_M4): $(am__aclocal_m4_deps)
	427	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	428	$(am__aclocal_m4_deps):
	429
	430	clean-noinstLTLIBRARIES:
	431	-test -z "$(noinst_LTLIBRARIES)" \|\| rm -f $(noinst_LTLIBRARIES)
	432	@list='$(noinst_LTLIBRARIES)'; \
	433	locs=`for p in $$list; do echo $$p; done \| \
	434	sed 's\|^[^/]$$\|.\|; s\|/[^/]$$\|\|; s\|$$\|/so_locations\|' \| \
	435	sort -u`; \
	436	test -z "$$locs" \|\| { \
	437	echo rm -f $${locs}; \
	438	rm -f $${locs}; \
	439	}
	440
	441	libde265_encoder.la: $(libde265_encoder_la_OBJECTS) $(libde265_encoder_la_DEPENDENCIES) $(EXTRA_libde265_encoder_la_DEPENDENCIES)
	442	$(AM_V_CXXLD)$(libde265_encoder_la_LINK) $(libde265_encoder_la_OBJECTS) $(libde265_encoder_la_LIBADD) $(LIBS)
	443
	444	mostlyclean-compile:
	445	-rm -f *.$(OBJEXT)
	446
	447	distclean-compile:
	448	-rm -f *.tab.c
	449
	450	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_la-analyze.Plo@am__quote@
	451	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_la-encode.Plo@am__quote@
	452	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_la-encoder-context.Plo@am__quote@
	453	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_la-encoder-params.Plo@am__quote@
	454	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_la-encpicbuf.Plo@am__quote@
	455	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_la-sop.Plo@am__quote@
	456
	457	.cc.o:
	458	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	459	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	460	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	461	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	462	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
	463
	464	.cc.obj:
	465	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
	466	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	467	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	468	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	469	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
	470
	471	.cc.lo:
	472	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	473	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Plo
	474	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
	475	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	476	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
	477
	478	libde265_encoder_la-analyze.lo: analyze.cc
	479	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_la-analyze.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_la-analyze.Tpo -c -o libde265_encoder_la-analyze.lo `test -f 'analyze.cc' \|\| echo '$(srcdir)/'`analyze.cc
	480	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_la-analyze.Tpo $(DEPDIR)/libde265_encoder_la-analyze.Plo
	481	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='analyze.cc' object='libde265_encoder_la-analyze.lo' libtool=yes @AMDEPBACKSLASH@
	482	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	483	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_la-analyze.lo `test -f 'analyze.cc' \|\| echo '$(srcdir)/'`analyze.cc
	484
	485	libde265_encoder_la-encode.lo: encode.cc
	486	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_la-encode.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_la-encode.Tpo -c -o libde265_encoder_la-encode.lo `test -f 'encode.cc' \|\| echo '$(srcdir)/'`encode.cc
	487	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_la-encode.Tpo $(DEPDIR)/libde265_encoder_la-encode.Plo
	488	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='encode.cc' object='libde265_encoder_la-encode.lo' libtool=yes @AMDEPBACKSLASH@
	489	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	490	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_la-encode.lo `test -f 'encode.cc' \|\| echo '$(srcdir)/'`encode.cc
	491
	492	libde265_encoder_la-encoder-params.lo: encoder-params.cc
	493	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_la-encoder-params.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_la-encoder-params.Tpo -c -o libde265_encoder_la-encoder-params.lo `test -f 'encoder-params.cc' \|\| echo '$(srcdir)/'`encoder-params.cc
	494	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_la-encoder-params.Tpo $(DEPDIR)/libde265_encoder_la-encoder-params.Plo
	495	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='encoder-params.cc' object='libde265_encoder_la-encoder-params.lo' libtool=yes @AMDEPBACKSLASH@
	496	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	497	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_la-encoder-params.lo `test -f 'encoder-params.cc' \|\| echo '$(srcdir)/'`encoder-params.cc
	498
	499	libde265_encoder_la-encoder-context.lo: encoder-context.cc
	500	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_la-encoder-context.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_la-encoder-context.Tpo -c -o libde265_encoder_la-encoder-context.lo `test -f 'encoder-context.cc' \|\| echo '$(srcdir)/'`encoder-context.cc
	501	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_la-encoder-context.Tpo $(DEPDIR)/libde265_encoder_la-encoder-context.Plo
	502	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='encoder-context.cc' object='libde265_encoder_la-encoder-context.lo' libtool=yes @AMDEPBACKSLASH@
	503	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	504	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_la-encoder-context.lo `test -f 'encoder-context.cc' \|\| echo '$(srcdir)/'`encoder-context.cc
	505
	506	libde265_encoder_la-encpicbuf.lo: encpicbuf.cc
	507	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_la-encpicbuf.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_la-encpicbuf.Tpo -c -o libde265_encoder_la-encpicbuf.lo `test -f 'encpicbuf.cc' \|\| echo '$(srcdir)/'`encpicbuf.cc
	508	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_la-encpicbuf.Tpo $(DEPDIR)/libde265_encoder_la-encpicbuf.Plo
	509	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='encpicbuf.cc' object='libde265_encoder_la-encpicbuf.lo' libtool=yes @AMDEPBACKSLASH@
	510	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	511	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_la-encpicbuf.lo `test -f 'encpicbuf.cc' \|\| echo '$(srcdir)/'`encpicbuf.cc
	512
	513	libde265_encoder_la-sop.lo: sop.cc
	514	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_la-sop.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_la-sop.Tpo -c -o libde265_encoder_la-sop.lo `test -f 'sop.cc' \|\| echo '$(srcdir)/'`sop.cc
	515	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_la-sop.Tpo $(DEPDIR)/libde265_encoder_la-sop.Plo
	516	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='sop.cc' object='libde265_encoder_la-sop.lo' libtool=yes @AMDEPBACKSLASH@
	517	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	518	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_la-sop.lo `test -f 'sop.cc' \|\| echo '$(srcdir)/'`sop.cc
	519
	520	mostlyclean-libtool:
	521	-rm -f *.lo
	522
	523	clean-libtool:
	524	-rm -rf .libs _libs
	525
	526	# This directory's subdirectories are mostly independent; you can cd
	527	# into them and run 'make' without going through this Makefile.
	528	# To change the values of 'make' variables: instead of editing Makefiles,
	529	# (1) if the variable is set in 'config.status', edit 'config.status'
	530	# (which will cause the Makefiles to be regenerated when you run 'make');
	531	# (2) otherwise, pass the desired values on the 'make' command line.
	532	$(am__recursive_targets):
	533	@fail=; \
	534	if $(am__make_keepgoing); then \
	535	failcom='fail=yes'; \
	536	else \
	537	failcom='exit 1'; \
	538	fi; \
	539	dot_seen=no; \
	540	target=`echo $@ \| sed s/-recursive//`; \
	541	case "$@" in \
	542	distclean-* \| maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
	543	*) list='$(SUBDIRS)' ;; \
	544	esac; \
	545	for subdir in $$list; do \
	546	echo "Making $$target in $$subdir"; \
	547	if test "$$subdir" = "."; then \
	548	dot_seen=yes; \
	549	local_target="$$target-am"; \
	550	else \
	551	local_target="$$target"; \
	552	fi; \
	553	($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
	554	\|\| eval $$failcom; \
	555	done; \
	556	if test "$$dot_seen" = "no"; then \
	557	$(MAKE) $(AM_MAKEFLAGS) "$$target-am" \|\| exit 1; \
	558	fi; test -z "$$fail"
	559
	560	ID: $(am__tagged_files)
	561	$(am__define_uniq_tagged_files); mkid -fID $$unique
	562	tags: tags-recursive
	563	TAGS: tags
	564
	565	tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	566	set x; \
	567	here=`pwd`; \
	568	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
	569	include_option=--etags-include; \
	570	empty_fix=.; \
	571	else \
	572	include_option=--include; \
	573	empty_fix=; \
	574	fi; \
	575	list='$(SUBDIRS)'; for subdir in $$list; do \
	576	if test "$$subdir" = .; then :; else \
	577	test ! -f $$subdir/TAGS \|\| \
	578	set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
	579	fi; \
	580	done; \
	581	$(am__define_uniq_tagged_files); \
	582	shift; \
	583	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
	584	test -n "$$unique" \|\| unique=$$empty_fix; \
	585	if test $$# -gt 0; then \
	586	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	587	"$$@" $$unique; \
	588	else \
	589	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	590	$$unique; \
	591	fi; \
	592	fi
	593	ctags: ctags-recursive
	594
	595	CTAGS: ctags
	596	ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	597	$(am__define_uniq_tagged_files); \
	598	test -z "$(CTAGS_ARGS)$$unique" \
	599	\|\| $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
	600	$$unique
	601
	602	GTAGS:
	603	here=`$(am__cd) $(top_builddir) && pwd` \
	604	&& $(am__cd) $(top_srcdir) \
	605	&& gtags -i $(GTAGS_ARGS) "$$here"
	606	cscopelist: cscopelist-recursive
	607
	608	cscopelist-am: $(am__tagged_files)
	609	list='$(am__tagged_files)'; \
	610	case "$(srcdir)" in \
	611	[\\/]* \| ?:[\\/]*) sdir="$(srcdir)" ;; \
	612	*) sdir=$(subdir)/$(srcdir) ;; \
	613	esac; \
	614	for i in $$list; do \
	615	if test -f "$$i"; then \
	616	echo "$(subdir)/$$i"; \
	617	else \
	618	echo "$$sdir/$$i"; \
	619	fi; \
	620	done >> $(top_builddir)/cscope.files
	621
	622	distclean-tags:
	623	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
	624
	625	distdir: $(DISTFILES)
	626	@srcdirstrip=`echo "$(srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	627	topsrcdirstrip=`echo "$(top_srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	628	list='$(DISTFILES)'; \
	629	dist_files=`for file in $$list; do echo $$file; done \| \
	630	sed -e "s\|^$$srcdirstrip/\|\|;t" \
	631	-e "s\|^$$topsrcdirstrip/\|$(top_builddir)/\|;t"`; \
	632	case $$dist_files in \
	633	/) $(MKDIR_P) `echo "$$dist_files" \| \
	634	sed '/\//!d;s\|^\|$(distdir)/\|;s,/[^/]*$$,,' \| \
	635	sort -u` ;; \
	636	esac; \
	637	for file in $$dist_files; do \
	638	if test -f $$file \|\| test -d $$file; then d=.; else d=$(srcdir); fi; \
	639	if test -d $$d/$$file; then \
	640	dir=`echo "/$$file" \| sed -e 's,/[^/]*$$,,'`; \
	641	if test -d "$(distdir)/$$file"; then \
	642	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	643	fi; \
	644	if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
	645	cp -fpR $(srcdir)/$$file "$(distdir)$$dir" \|\| exit 1; \
	646	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	647	fi; \
	648	cp -fpR $$d/$$file "$(distdir)$$dir" \|\| exit 1; \
	649	else \
	650	test -f "$(distdir)/$$file" \
	651	\|\| cp -p $$d/$$file "$(distdir)/$$file" \
	652	\|\| exit 1; \
	653	fi; \
	654	done
	655	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
	656	if test "$$subdir" = .; then :; else \
	657	$(am__make_dryrun) \
	658	\|\| test -d "$(distdir)/$$subdir" \
	659	\|\| $(MKDIR_P) "$(distdir)/$$subdir" \
	660	\|\| exit 1; \
	661	dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
	662	$(am__relativize); \
	663	new_distdir=$$reldir; \
	664	dir1=$$subdir; dir2="$(top_distdir)"; \
	665	$(am__relativize); \
	666	new_top_distdir=$$reldir; \
	667	echo " (cd $$subdir && $(MAKE) $(AM_MAKEFLAGS) top_distdir="$$new_top_distdir" distdir="$$new_distdir" \\"; \
	668	echo " am__remove_distdir=: am__skip_length_check=: am__skip_mode_fix=: distdir)"; \
	669	($(am__cd) $$subdir && \
	670	$(MAKE) $(AM_MAKEFLAGS) \
	671	top_distdir="$$new_top_distdir" \
	672	distdir="$$new_distdir" \
	673	am__remove_distdir=: \
	674	am__skip_length_check=: \
	675	am__skip_mode_fix=: \
	676	distdir) \
	677	\|\| exit 1; \
	678	fi; \
	679	done
	680	check-am: all-am
	681	check: check-recursive
	682	all-am: Makefile $(LTLIBRARIES)
	683	installdirs: installdirs-recursive
	684	installdirs-am:
	685	install: install-recursive
	686	install-exec: install-exec-recursive
	687	install-data: install-data-recursive
	688	uninstall: uninstall-recursive
	689
	690	install-am: all-am
	691	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
	692
	693	installcheck: installcheck-recursive
	694	install-strip:
	695	if test -z '$(STRIP)'; then \
	696	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	697	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	698	install; \
	699	else \
	700	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	701	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	702	"INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
	703	fi
	704	mostlyclean-generic:
	705
	706	clean-generic:
	707
	708	distclean-generic:
	709	-test -z "$(CONFIG_CLEAN_FILES)" \|\| rm -f $(CONFIG_CLEAN_FILES)
	710	-test . = "$(srcdir)" \|\| test -z "$(CONFIG_CLEAN_VPATH_FILES)" \|\| rm -f $(CONFIG_CLEAN_VPATH_FILES)
	711
	712	maintainer-clean-generic:
	713	@echo "This command is intended for maintainers to use"
	714	@echo "it deletes files that may require special tools to rebuild."
	715	clean: clean-recursive
	716
	717	clean-am: clean-generic clean-libtool clean-noinstLTLIBRARIES \
	718	mostlyclean-am
	719
	720	distclean: distclean-recursive
	721	-rm -rf ./$(DEPDIR)
	722	-rm -f Makefile
	723	distclean-am: clean-am distclean-compile distclean-generic \
	724	distclean-tags
	725
	726	dvi: dvi-recursive
	727
	728	dvi-am:
	729
	730	html: html-recursive
	731
	732	html-am:
	733
	734	info: info-recursive
	735
	736	info-am:
	737
	738	install-data-am:
	739
	740	install-dvi: install-dvi-recursive
	741
	742	install-dvi-am:
	743
	744	install-exec-am:
	745
	746	install-html: install-html-recursive
	747
	748	install-html-am:
	749
	750	install-info: install-info-recursive
	751
	752	install-info-am:
	753
	754	install-man:
	755
	756	install-pdf: install-pdf-recursive
	757
	758	install-pdf-am:
	759
	760	install-ps: install-ps-recursive
	761
	762	install-ps-am:
	763
	764	installcheck-am:
	765
	766	maintainer-clean: maintainer-clean-recursive
	767	-rm -rf ./$(DEPDIR)
	768	-rm -f Makefile
	769	maintainer-clean-am: distclean-am maintainer-clean-generic
	770
	771	mostlyclean: mostlyclean-recursive
	772
	773	mostlyclean-am: mostlyclean-compile mostlyclean-generic \
	774	mostlyclean-libtool
	775
	776	pdf: pdf-recursive
	777
	778	pdf-am:
	779
	780	ps: ps-recursive
	781
	782	ps-am:
	783
	784	uninstall-am:
	785
	786	.MAKE: $(am__recursive_targets) install-am install-strip
	787
	788	.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am check \
	789	check-am clean clean-generic clean-libtool \
	790	clean-noinstLTLIBRARIES cscopelist-am ctags ctags-am distclean \
	791	distclean-compile distclean-generic distclean-libtool \
	792	distclean-tags distdir dvi dvi-am html html-am info info-am \
	793	install install-am install-data install-data-am install-dvi \
	794	install-dvi-am install-exec install-exec-am install-html \
	795	install-html-am install-info install-info-am install-man \
	796	install-pdf install-pdf-am install-ps install-ps-am \
	797	install-strip installcheck installcheck-am installdirs \
	798	installdirs-am maintainer-clean maintainer-clean-generic \
	799	mostlyclean mostlyclean-compile mostlyclean-generic \
	800	mostlyclean-libtool pdf pdf-am ps ps-am tags tags-am uninstall \
	801	uninstall-am
	802
	803
	804	# Tell versions [3.59,3.63) of GNU make to not export all variables.
	805	# Otherwise a system limit (for SysV at least) may be exceeded.
	806	.NOEXPORT:

+20

-0

libde265/encoder/algo/CMakeLists.txt less more

	0	set (algo_sources
	1	algo.h algo.cc
	2	coding-options.h coding-options.cc
	3	ctb-qscale.h ctb-qscale.cc
	4	cb-split.h cb-split.cc
	5	cb-intrapartmode.h cb-intrapartmode.cc
	6	cb-interpartmode.h cb-interpartmode.cc
	7	cb-skip.h cb-skip.cc
	8	cb-intra-inter.h cb-intra-inter.cc
	9	cb-mergeindex.h cb-mergeindex.cc
	10	tb-split.h tb-split.cc
	11	tb-intrapredmode.h tb-intrapredmode.cc
	12	pb-mv.h pb-mv.cc
	13	)
	14
	15	add_library(algo STATIC ${algo_sources})
	16
	17	if(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
	18	SET_TARGET_PROPERTIES(algo PROPERTIES COMPILE_FLAGS "-fPIC")
	19	endif(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")

+19

-0

libde265/encoder/algo/Makefile.am less more

	0	noinst_LTLIBRARIES = libde265_encoder_algo.la
	1
	2	libde265_encoder_algo_la_CXXFLAGS = -I../..
	3	libde265_encoder_algo_la_SOURCES = \
	4	algo.h algo.cc \
	5	coding-options.h coding-options.cc \
	6	ctb-qscale.h ctb-qscale.cc \
	7	cb-split.h cb-split.cc \
	8	cb-intrapartmode.h cb-intrapartmode.cc \
	9	cb-interpartmode.h cb-interpartmode.cc \
	10	cb-skip.h cb-skip.cc \
	11	cb-intra-inter.h cb-intra-inter.cc \
	12	cb-mergeindex.h cb-mergeindex.cc \
	13	tb-split.h tb-split.cc \
	14	tb-intrapredmode.h tb-intrapredmode.cc \
	15	pb-mv.h pb-mv.cc
	16
	17	EXTRA_DIST = \
	18	CMakeLists.txt

+752

-0

libde265/encoder/algo/Makefile.in less more

	0	# Makefile.in generated by automake 1.14.1 from Makefile.am.
	1	# @configure_input@
	2
	3	# Copyright (C) 1994-2013 Free Software Foundation, Inc.
	4
	5	# This Makefile.in is free software; the Free Software Foundation
	6	# gives unlimited permission to copy and/or distribute it,
	7	# with or without modifications, as long as this notice is preserved.
	8
	9	# This program is distributed in the hope that it will be useful,
	10	# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
	11	# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
	12	# PARTICULAR PURPOSE.
	13
	14	@SET_MAKE@
	15
	16	VPATH = @srcdir@
	17	am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
	18	am__make_running_with_option = \
	19	case $${target_option-} in \
	20	?) ;; \
	21	*) echo "am__make_running_with_option: internal error: invalid" \
	22	"target option '$${target_option-}' specified" >&2; \
	23	exit 1;; \
	24	esac; \
	25	has_opt=no; \
	26	sane_makeflags=$$MAKEFLAGS; \
	27	if $(am__is_gnu_make); then \
	28	sane_makeflags=$$MFLAGS; \
	29	else \
	30	case $$MAKEFLAGS in \
	31	\\[\ \ ]) \
	32	bs=\\; \
	33	sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
	34	\| sed "s/$$bs$$bs[$$bs $$bs ]*//g"`;; \
	35	esac; \
	36	fi; \
	37	skip_next=no; \
	38	strip_trailopt () \
	39	{ \
	40	flg=`printf '%s\n' "$$flg" \| sed "s/$$1.*$$//"`; \
	41	}; \
	42	for flg in $$sane_makeflags; do \
	43	test $$skip_next = yes && { skip_next=no; continue; }; \
	44	case $$flg in \
	45	=\|--*) continue;; \
	46	-*I) strip_trailopt 'I'; skip_next=yes;; \
	47	-I?) strip_trailopt 'I';; \
	48	-*O) strip_trailopt 'O'; skip_next=yes;; \
	49	-O?) strip_trailopt 'O';; \
	50	-*l) strip_trailopt 'l'; skip_next=yes;; \
	51	-l?) strip_trailopt 'l';; \
	52	-[dEDm]) skip_next=yes;; \
	53	-[JT]) skip_next=yes;; \
	54	esac; \
	55	case $$flg in \
	56	$$target_option) has_opt=yes; break;; \
	57	esac; \
	58	done; \
	59	test $$has_opt = yes
	60	am__make_dryrun = (target_option=n; $(am__make_running_with_option))
	61	am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
	62	pkgdatadir = $(datadir)/@PACKAGE@
	63	pkgincludedir = $(includedir)/@PACKAGE@
	64	pkglibdir = $(libdir)/@PACKAGE@
	65	pkglibexecdir = $(libexecdir)/@PACKAGE@
	66	am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
	67	install_sh_DATA = $(install_sh) -c -m 644
	68	install_sh_PROGRAM = $(install_sh) -c
	69	install_sh_SCRIPT = $(install_sh) -c
	70	INSTALL_HEADER = $(INSTALL_DATA)
	71	transform = $(program_transform_name)
	72	NORMAL_INSTALL = :
	73	PRE_INSTALL = :
	74	POST_INSTALL = :
	75	NORMAL_UNINSTALL = :
	76	PRE_UNINSTALL = :
	77	POST_UNINSTALL = :
	78	build_triplet = @build@
	79	host_triplet = @host@
	80	target_triplet = @target@
	81	subdir = libde265/encoder/algo
	82	DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/Makefile.am \
	83	$(top_srcdir)/depcomp
	84	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
	85	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	86	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
	87	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
	88	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
	89	$(top_srcdir)/m4/lt~obsolete.m4 \
	90	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
	91	$(top_srcdir)/configure.ac
	92	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
	93	$(ACLOCAL_M4)
	94	mkinstalldirs = $(install_sh) -d
	95	CONFIG_HEADER = $(top_builddir)/config.h
	96	CONFIG_CLEAN_FILES =
	97	CONFIG_CLEAN_VPATH_FILES =
	98	LTLIBRARIES = $(noinst_LTLIBRARIES)
	99	libde265_encoder_algo_la_LIBADD =
	100	am_libde265_encoder_algo_la_OBJECTS = \
	101	libde265_encoder_algo_la-algo.lo \
	102	libde265_encoder_algo_la-coding-options.lo \
	103	libde265_encoder_algo_la-ctb-qscale.lo \
	104	libde265_encoder_algo_la-cb-split.lo \
	105	libde265_encoder_algo_la-cb-intrapartmode.lo \
	106	libde265_encoder_algo_la-cb-interpartmode.lo \
	107	libde265_encoder_algo_la-cb-skip.lo \
	108	libde265_encoder_algo_la-cb-intra-inter.lo \
	109	libde265_encoder_algo_la-cb-mergeindex.lo \
	110	libde265_encoder_algo_la-tb-split.lo \
	111	libde265_encoder_algo_la-tb-intrapredmode.lo \
	112	libde265_encoder_algo_la-pb-mv.lo
	113	libde265_encoder_algo_la_OBJECTS = \
	114	$(am_libde265_encoder_algo_la_OBJECTS)
	115	AM_V_lt = $(am__v_lt_@AM_V@)
	116	am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
	117	am__v_lt_0 = --silent
	118	am__v_lt_1 =
	119	libde265_encoder_algo_la_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
	120	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
	121	$(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) \
	122	$(LDFLAGS) -o $@
	123	AM_V_P = $(am__v_P_@AM_V@)
	124	am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
	125	am__v_P_0 = false
	126	am__v_P_1 = :
	127	AM_V_GEN = $(am__v_GEN_@AM_V@)
	128	am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
	129	am__v_GEN_0 = @echo " GEN " $@;
	130	am__v_GEN_1 =
	131	AM_V_at = $(am__v_at_@AM_V@)
	132	am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
	133	am__v_at_0 = @
	134	am__v_at_1 =
	135	DEFAULT_INCLUDES = -I.@am__isrc@ -I$(top_builddir)
	136	depcomp = $(SHELL) $(top_srcdir)/depcomp
	137	am__depfiles_maybe = depfiles
	138	am__mv = mv -f
	139	CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
	140	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
	141	LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	142	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
	143	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	144	$(AM_CXXFLAGS) $(CXXFLAGS)
	145	AM_V_CXX = $(am__v_CXX_@AM_V@)
	146	am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
	147	am__v_CXX_0 = @echo " CXX " $@;
	148	am__v_CXX_1 =
	149	CXXLD = $(CXX)
	150	CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	151	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
	152	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
	153	AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
	154	am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
	155	am__v_CXXLD_0 = @echo " CXXLD " $@;
	156	am__v_CXXLD_1 =
	157	COMPILE = $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) \
	158	$(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS)
	159	LTCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CC $(AM_LIBTOOLFLAGS) \
	160	$(LIBTOOLFLAGS) --mode=compile $(CC) $(DEFS) \
	161	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	162	$(AM_CFLAGS) $(CFLAGS)
	163	AM_V_CC = $(am__v_CC_@AM_V@)
	164	am__v_CC_ = $(am__v_CC_@AM_DEFAULT_V@)
	165	am__v_CC_0 = @echo " CC " $@;
	166	am__v_CC_1 =
	167	CCLD = $(CC)
	168	LINK = $(LIBTOOL) $(AM_V_lt) --tag=CC $(AM_LIBTOOLFLAGS) \
	169	$(LIBTOOLFLAGS) --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
	170	$(AM_LDFLAGS) $(LDFLAGS) -o $@
	171	AM_V_CCLD = $(am__v_CCLD_@AM_V@)
	172	am__v_CCLD_ = $(am__v_CCLD_@AM_DEFAULT_V@)
	173	am__v_CCLD_0 = @echo " CCLD " $@;
	174	am__v_CCLD_1 =
	175	SOURCES = $(libde265_encoder_algo_la_SOURCES)
	176	DIST_SOURCES = $(libde265_encoder_algo_la_SOURCES)
	177	am__can_run_installinfo = \
	178	case $$AM_UPDATE_INFO_DIR in \
	179	n\|no\|NO) false;; \
	180	*) (install-info --version) >/dev/null 2>&1;; \
	181	esac
	182	am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
	183	# Read a list of newline-separated strings from the standard input,
	184	# and print each of them once, without duplicates. Input order is
	185	# not preserved.
	186	am__uniquify_input = $(AWK) '\
	187	BEGIN { nonempty = 0; } \
	188	{ items[$$0] = 1; nonempty = 1; } \
	189	END { if (nonempty) { for (i in items) print i; }; } \
	190	'
	191	# Make sure the list of sources is unique. This is necessary because,
	192	# e.g., the same source file might be shared among _SOURCES variables
	193	# for different programs/libraries.
	194	am__define_uniq_tagged_files = \
	195	list='$(am__tagged_files)'; \
	196	unique=`for i in $$list; do \
	197	if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
	198	done \| $(am__uniquify_input)`
	199	ETAGS = etags
	200	CTAGS = ctags
	201	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
	202	ACLOCAL = @ACLOCAL@
	203	ALLOCA = @ALLOCA@
	204	AMTAR = @AMTAR@
	205	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
	206	AR = @AR@
	207	AUTOCONF = @AUTOCONF@
	208	AUTOHEADER = @AUTOHEADER@
	209	AUTOMAKE = @AUTOMAKE@
	210	AWK = @AWK@
	211	CC = @CC@
	212	CCAS = @CCAS@
	213	CCASDEPMODE = @CCASDEPMODE@
	214	CCASFLAGS = @CCASFLAGS@
	215	CCDEPMODE = @CCDEPMODE@
	216	CFLAGS = @CFLAGS@
	217	CPP = @CPP@
	218	CPPFLAGS = @CPPFLAGS@
	219	CXX = @CXX@
	220	CXXCPP = @CXXCPP@
	221	CXXDEPMODE = @CXXDEPMODE@
	222	CXXFLAGS = @CXXFLAGS@
	223	CYGPATH_W = @CYGPATH_W@
	224	DEFS = @DEFS@
	225	DEPDIR = @DEPDIR@
	226	DLLTOOL = @DLLTOOL@
	227	DSYMUTIL = @DSYMUTIL@
	228	DUMPBIN = @DUMPBIN@
	229	ECHO_C = @ECHO_C@
	230	ECHO_N = @ECHO_N@
	231	ECHO_T = @ECHO_T@
	232	EGREP = @EGREP@
	233	EXEEXT = @EXEEXT@
	234	FGREP = @FGREP@
	235	GREP = @GREP@
	236	HAVE_CXX11 = @HAVE_CXX11@
	237	INSTALL = @INSTALL@
	238	INSTALL_DATA = @INSTALL_DATA@
	239	INSTALL_PROGRAM = @INSTALL_PROGRAM@
	240	INSTALL_SCRIPT = @INSTALL_SCRIPT@
	241	INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
	242	LD = @LD@
	243	LDFLAGS = @LDFLAGS@
	244	LIBDE265_AGE = @LIBDE265_AGE@
	245	LIBDE265_CURRENT = @LIBDE265_CURRENT@
	246	LIBDE265_REVISION = @LIBDE265_REVISION@
	247	LIBOBJS = @LIBOBJS@
	248	LIBS = @LIBS@
	249	LIBTOOL = @LIBTOOL@
	250	LIPO = @LIPO@
	251	LN_S = @LN_S@
	252	LTLIBOBJS = @LTLIBOBJS@
	253	MAKEINFO = @MAKEINFO@
	254	MANIFEST_TOOL = @MANIFEST_TOOL@
	255	MKDIR_P = @MKDIR_P@
	256	NM = @NM@
	257	NMEDIT = @NMEDIT@
	258	NUMERIC_VERSION = @NUMERIC_VERSION@
	259	OBJDUMP = @OBJDUMP@
	260	OBJEXT = @OBJEXT@
	261	OTOOL = @OTOOL@
	262	OTOOL64 = @OTOOL64@
	263	PACKAGE = @PACKAGE@
	264	PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
	265	PACKAGE_NAME = @PACKAGE_NAME@
	266	PACKAGE_STRING = @PACKAGE_STRING@
	267	PACKAGE_TARNAME = @PACKAGE_TARNAME@
	268	PACKAGE_URL = @PACKAGE_URL@
	269	PACKAGE_VERSION = @PACKAGE_VERSION@
	270	PATH_SEPARATOR = @PATH_SEPARATOR@
	271	PKG_CONFIG = @PKG_CONFIG@
	272	PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
	273	PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
	274	QTCHOOSER = @QTCHOOSER@
	275	QTMOC = @QTMOC@
	276	QT_CFLAGS = @QT_CFLAGS@
	277	QT_LIBS = @QT_LIBS@
	278	RANLIB = @RANLIB@
	279	SDL_CFLAGS = @SDL_CFLAGS@
	280	SDL_LIBS = @SDL_LIBS@
	281	SED = @SED@
	282	SET_MAKE = @SET_MAKE@
	283	SHELL = @SHELL@
	284	STRIP = @STRIP@
	285	SWSCALE_CFLAGS = @SWSCALE_CFLAGS@
	286	SWSCALE_LIBS = @SWSCALE_LIBS@
	287	VERSION = @VERSION@
	288	VIDEOGFX_CFLAGS = @VIDEOGFX_CFLAGS@
	289	VIDEOGFX_LIBS = @VIDEOGFX_LIBS@
	290	abs_builddir = @abs_builddir@
	291	abs_srcdir = @abs_srcdir@
	292	abs_top_builddir = @abs_top_builddir@
	293	abs_top_srcdir = @abs_top_srcdir@
	294	ac_ct_AR = @ac_ct_AR@
	295	ac_ct_CC = @ac_ct_CC@
	296	ac_ct_CXX = @ac_ct_CXX@
	297	ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
	298	am__include = @am__include@
	299	am__leading_dot = @am__leading_dot@
	300	am__quote = @am__quote@
	301	am__tar = @am__tar@
	302	am__untar = @am__untar@
	303	bindir = @bindir@
	304	build = @build@
	305	build_alias = @build_alias@
	306	build_cpu = @build_cpu@
	307	build_os = @build_os@
	308	build_vendor = @build_vendor@
	309	builddir = @builddir@
	310	datadir = @datadir@
	311	datarootdir = @datarootdir@
	312	docdir = @docdir@
	313	dvidir = @dvidir@
	314	exec_prefix = @exec_prefix@
	315	host = @host@
	316	host_alias = @host_alias@
	317	host_cpu = @host_cpu@
	318	host_os = @host_os@
	319	host_vendor = @host_vendor@
	320	htmldir = @htmldir@
	321	includedir = @includedir@
	322	infodir = @infodir@
	323	install_sh = @install_sh@
	324	libdir = @libdir@
	325	libexecdir = @libexecdir@
	326	localedir = @localedir@
	327	localstatedir = @localstatedir@
	328	mandir = @mandir@
	329	mkdir_p = @mkdir_p@
	330	oldincludedir = @oldincludedir@
	331	pdfdir = @pdfdir@
	332	prefix = @prefix@
	333	program_transform_name = @program_transform_name@
	334	psdir = @psdir@
	335	sbindir = @sbindir@
	336	sharedstatedir = @sharedstatedir@
	337	srcdir = @srcdir@
	338	sysconfdir = @sysconfdir@
	339	target = @target@
	340	target_alias = @target_alias@
	341	target_cpu = @target_cpu@
	342	target_os = @target_os@
	343	target_vendor = @target_vendor@
	344	top_build_prefix = @top_build_prefix@
	345	top_builddir = @top_builddir@
	346	top_srcdir = @top_srcdir@
	347	noinst_LTLIBRARIES = libde265_encoder_algo.la
	348	libde265_encoder_algo_la_CXXFLAGS = -I../..
	349	libde265_encoder_algo_la_SOURCES = \
	350	algo.h algo.cc \
	351	coding-options.h coding-options.cc \
	352	ctb-qscale.h ctb-qscale.cc \
	353	cb-split.h cb-split.cc \
	354	cb-intrapartmode.h cb-intrapartmode.cc \
	355	cb-interpartmode.h cb-interpartmode.cc \
	356	cb-skip.h cb-skip.cc \
	357	cb-intra-inter.h cb-intra-inter.cc \
	358	cb-mergeindex.h cb-mergeindex.cc \
	359	tb-split.h tb-split.cc \
	360	tb-intrapredmode.h tb-intrapredmode.cc \
	361	pb-mv.h pb-mv.cc
	362
	363	EXTRA_DIST = \
	364	CMakeLists.txt
	365
	366	all: all-am
	367
	368	.SUFFIXES:
	369	.SUFFIXES: .cc .lo .o .obj
	370	$(srcdir)/Makefile.in: $(srcdir)/Makefile.am $(am__configure_deps)
	371	@for dep in $?; do \
	372	case '$(am__configure_deps)' in \
	373	$$dep) \
	374	( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
	375	&& { if test -f $@; then exit 0; else break; fi; }; \
	376	exit 1;; \
	377	esac; \
	378	done; \
	379	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu libde265/encoder/algo/Makefile'; \
	380	$(am__cd) $(top_srcdir) && \
	381	$(AUTOMAKE) --gnu libde265/encoder/algo/Makefile
	382	.PRECIOUS: Makefile
	383	Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
	384	@case '$?' in \
	385	config.status) \
	386	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
	387	*) \
	388	echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
	389	cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
	390	esac;
	391
	392	$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
	393	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	394
	395	$(top_srcdir)/configure: $(am__configure_deps)
	396	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	397	$(ACLOCAL_M4): $(am__aclocal_m4_deps)
	398	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	399	$(am__aclocal_m4_deps):
	400
	401	clean-noinstLTLIBRARIES:
	402	-test -z "$(noinst_LTLIBRARIES)" \|\| rm -f $(noinst_LTLIBRARIES)
	403	@list='$(noinst_LTLIBRARIES)'; \
	404	locs=`for p in $$list; do echo $$p; done \| \
	405	sed 's\|^[^/]$$\|.\|; s\|/[^/]$$\|\|; s\|$$\|/so_locations\|' \| \
	406	sort -u`; \
	407	test -z "$$locs" \|\| { \
	408	echo rm -f $${locs}; \
	409	rm -f $${locs}; \
	410	}
	411
	412	libde265_encoder_algo.la: $(libde265_encoder_algo_la_OBJECTS) $(libde265_encoder_algo_la_DEPENDENCIES) $(EXTRA_libde265_encoder_algo_la_DEPENDENCIES)
	413	$(AM_V_CXXLD)$(libde265_encoder_algo_la_LINK) $(libde265_encoder_algo_la_OBJECTS) $(libde265_encoder_algo_la_LIBADD) $(LIBS)
	414
	415	mostlyclean-compile:
	416	-rm -f *.$(OBJEXT)
	417
	418	distclean-compile:
	419	-rm -f *.tab.c
	420
	421	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-algo.Plo@am__quote@
	422	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-cb-interpartmode.Plo@am__quote@
	423	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-cb-intra-inter.Plo@am__quote@
	424	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-cb-intrapartmode.Plo@am__quote@
	425	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-cb-mergeindex.Plo@am__quote@
	426	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-cb-skip.Plo@am__quote@
	427	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-cb-split.Plo@am__quote@
	428	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-coding-options.Plo@am__quote@
	429	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-ctb-qscale.Plo@am__quote@
	430	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-pb-mv.Plo@am__quote@
	431	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-tb-intrapredmode.Plo@am__quote@
	432	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libde265_encoder_algo_la-tb-split.Plo@am__quote@
	433
	434	.cc.o:
	435	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	436	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	437	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	438	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	439	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
	440
	441	.cc.obj:
	442	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
	443	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	444	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	445	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	446	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
	447
	448	.cc.lo:
	449	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	450	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Plo
	451	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
	452	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	453	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
	454
	455	libde265_encoder_algo_la-algo.lo: algo.cc
	456	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-algo.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-algo.Tpo -c -o libde265_encoder_algo_la-algo.lo `test -f 'algo.cc' \|\| echo '$(srcdir)/'`algo.cc
	457	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-algo.Tpo $(DEPDIR)/libde265_encoder_algo_la-algo.Plo
	458	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='algo.cc' object='libde265_encoder_algo_la-algo.lo' libtool=yes @AMDEPBACKSLASH@
	459	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	460	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-algo.lo `test -f 'algo.cc' \|\| echo '$(srcdir)/'`algo.cc
	461
	462	libde265_encoder_algo_la-coding-options.lo: coding-options.cc
	463	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-coding-options.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-coding-options.Tpo -c -o libde265_encoder_algo_la-coding-options.lo `test -f 'coding-options.cc' \|\| echo '$(srcdir)/'`coding-options.cc
	464	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-coding-options.Tpo $(DEPDIR)/libde265_encoder_algo_la-coding-options.Plo
	465	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='coding-options.cc' object='libde265_encoder_algo_la-coding-options.lo' libtool=yes @AMDEPBACKSLASH@
	466	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	467	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-coding-options.lo `test -f 'coding-options.cc' \|\| echo '$(srcdir)/'`coding-options.cc
	468
	469	libde265_encoder_algo_la-ctb-qscale.lo: ctb-qscale.cc
	470	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-ctb-qscale.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-ctb-qscale.Tpo -c -o libde265_encoder_algo_la-ctb-qscale.lo `test -f 'ctb-qscale.cc' \|\| echo '$(srcdir)/'`ctb-qscale.cc
	471	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-ctb-qscale.Tpo $(DEPDIR)/libde265_encoder_algo_la-ctb-qscale.Plo
	472	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='ctb-qscale.cc' object='libde265_encoder_algo_la-ctb-qscale.lo' libtool=yes @AMDEPBACKSLASH@
	473	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	474	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-ctb-qscale.lo `test -f 'ctb-qscale.cc' \|\| echo '$(srcdir)/'`ctb-qscale.cc
	475
	476	libde265_encoder_algo_la-cb-split.lo: cb-split.cc
	477	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-cb-split.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-cb-split.Tpo -c -o libde265_encoder_algo_la-cb-split.lo `test -f 'cb-split.cc' \|\| echo '$(srcdir)/'`cb-split.cc
	478	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-cb-split.Tpo $(DEPDIR)/libde265_encoder_algo_la-cb-split.Plo
	479	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='cb-split.cc' object='libde265_encoder_algo_la-cb-split.lo' libtool=yes @AMDEPBACKSLASH@
	480	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	481	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-cb-split.lo `test -f 'cb-split.cc' \|\| echo '$(srcdir)/'`cb-split.cc
	482
	483	libde265_encoder_algo_la-cb-intrapartmode.lo: cb-intrapartmode.cc
	484	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-cb-intrapartmode.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-cb-intrapartmode.Tpo -c -o libde265_encoder_algo_la-cb-intrapartmode.lo `test -f 'cb-intrapartmode.cc' \|\| echo '$(srcdir)/'`cb-intrapartmode.cc
	485	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-cb-intrapartmode.Tpo $(DEPDIR)/libde265_encoder_algo_la-cb-intrapartmode.Plo
	486	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='cb-intrapartmode.cc' object='libde265_encoder_algo_la-cb-intrapartmode.lo' libtool=yes @AMDEPBACKSLASH@
	487	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	488	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-cb-intrapartmode.lo `test -f 'cb-intrapartmode.cc' \|\| echo '$(srcdir)/'`cb-intrapartmode.cc
	489
	490	libde265_encoder_algo_la-cb-interpartmode.lo: cb-interpartmode.cc
	491	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-cb-interpartmode.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-cb-interpartmode.Tpo -c -o libde265_encoder_algo_la-cb-interpartmode.lo `test -f 'cb-interpartmode.cc' \|\| echo '$(srcdir)/'`cb-interpartmode.cc
	492	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-cb-interpartmode.Tpo $(DEPDIR)/libde265_encoder_algo_la-cb-interpartmode.Plo
	493	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='cb-interpartmode.cc' object='libde265_encoder_algo_la-cb-interpartmode.lo' libtool=yes @AMDEPBACKSLASH@
	494	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	495	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-cb-interpartmode.lo `test -f 'cb-interpartmode.cc' \|\| echo '$(srcdir)/'`cb-interpartmode.cc
	496
	497	libde265_encoder_algo_la-cb-skip.lo: cb-skip.cc
	498	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-cb-skip.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-cb-skip.Tpo -c -o libde265_encoder_algo_la-cb-skip.lo `test -f 'cb-skip.cc' \|\| echo '$(srcdir)/'`cb-skip.cc
	499	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-cb-skip.Tpo $(DEPDIR)/libde265_encoder_algo_la-cb-skip.Plo
	500	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='cb-skip.cc' object='libde265_encoder_algo_la-cb-skip.lo' libtool=yes @AMDEPBACKSLASH@
	501	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	502	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-cb-skip.lo `test -f 'cb-skip.cc' \|\| echo '$(srcdir)/'`cb-skip.cc
	503
	504	libde265_encoder_algo_la-cb-intra-inter.lo: cb-intra-inter.cc
	505	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-cb-intra-inter.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-cb-intra-inter.Tpo -c -o libde265_encoder_algo_la-cb-intra-inter.lo `test -f 'cb-intra-inter.cc' \|\| echo '$(srcdir)/'`cb-intra-inter.cc
	506	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-cb-intra-inter.Tpo $(DEPDIR)/libde265_encoder_algo_la-cb-intra-inter.Plo
	507	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='cb-intra-inter.cc' object='libde265_encoder_algo_la-cb-intra-inter.lo' libtool=yes @AMDEPBACKSLASH@
	508	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	509	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-cb-intra-inter.lo `test -f 'cb-intra-inter.cc' \|\| echo '$(srcdir)/'`cb-intra-inter.cc
	510
	511	libde265_encoder_algo_la-cb-mergeindex.lo: cb-mergeindex.cc
	512	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-cb-mergeindex.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-cb-mergeindex.Tpo -c -o libde265_encoder_algo_la-cb-mergeindex.lo `test -f 'cb-mergeindex.cc' \|\| echo '$(srcdir)/'`cb-mergeindex.cc
	513	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-cb-mergeindex.Tpo $(DEPDIR)/libde265_encoder_algo_la-cb-mergeindex.Plo
	514	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='cb-mergeindex.cc' object='libde265_encoder_algo_la-cb-mergeindex.lo' libtool=yes @AMDEPBACKSLASH@
	515	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	516	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-cb-mergeindex.lo `test -f 'cb-mergeindex.cc' \|\| echo '$(srcdir)/'`cb-mergeindex.cc
	517
	518	libde265_encoder_algo_la-tb-split.lo: tb-split.cc
	519	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-tb-split.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-tb-split.Tpo -c -o libde265_encoder_algo_la-tb-split.lo `test -f 'tb-split.cc' \|\| echo '$(srcdir)/'`tb-split.cc
	520	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-tb-split.Tpo $(DEPDIR)/libde265_encoder_algo_la-tb-split.Plo
	521	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='tb-split.cc' object='libde265_encoder_algo_la-tb-split.lo' libtool=yes @AMDEPBACKSLASH@
	522	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	523	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-tb-split.lo `test -f 'tb-split.cc' \|\| echo '$(srcdir)/'`tb-split.cc
	524
	525	libde265_encoder_algo_la-tb-intrapredmode.lo: tb-intrapredmode.cc
	526	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-tb-intrapredmode.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-tb-intrapredmode.Tpo -c -o libde265_encoder_algo_la-tb-intrapredmode.lo `test -f 'tb-intrapredmode.cc' \|\| echo '$(srcdir)/'`tb-intrapredmode.cc
	527	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-tb-intrapredmode.Tpo $(DEPDIR)/libde265_encoder_algo_la-tb-intrapredmode.Plo
	528	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='tb-intrapredmode.cc' object='libde265_encoder_algo_la-tb-intrapredmode.lo' libtool=yes @AMDEPBACKSLASH@
	529	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	530	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-tb-intrapredmode.lo `test -f 'tb-intrapredmode.cc' \|\| echo '$(srcdir)/'`tb-intrapredmode.cc
	531
	532	libde265_encoder_algo_la-pb-mv.lo: pb-mv.cc
	533	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -MT libde265_encoder_algo_la-pb-mv.lo -MD -MP -MF $(DEPDIR)/libde265_encoder_algo_la-pb-mv.Tpo -c -o libde265_encoder_algo_la-pb-mv.lo `test -f 'pb-mv.cc' \|\| echo '$(srcdir)/'`pb-mv.cc
	534	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/libde265_encoder_algo_la-pb-mv.Tpo $(DEPDIR)/libde265_encoder_algo_la-pb-mv.Plo
	535	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='pb-mv.cc' object='libde265_encoder_algo_la-pb-mv.lo' libtool=yes @AMDEPBACKSLASH@
	536	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	537	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libde265_encoder_algo_la_CXXFLAGS) $(CXXFLAGS) -c -o libde265_encoder_algo_la-pb-mv.lo `test -f 'pb-mv.cc' \|\| echo '$(srcdir)/'`pb-mv.cc
	538
	539	mostlyclean-libtool:
	540	-rm -f *.lo
	541
	542	clean-libtool:
	543	-rm -rf .libs _libs
	544
	545	ID: $(am__tagged_files)
	546	$(am__define_uniq_tagged_files); mkid -fID $$unique
	547	tags: tags-am
	548	TAGS: tags
	549
	550	tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	551	set x; \
	552	here=`pwd`; \
	553	$(am__define_uniq_tagged_files); \
	554	shift; \
	555	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
	556	test -n "$$unique" \|\| unique=$$empty_fix; \
	557	if test $$# -gt 0; then \
	558	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	559	"$$@" $$unique; \
	560	else \
	561	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	562	$$unique; \
	563	fi; \
	564	fi
	565	ctags: ctags-am
	566
	567	CTAGS: ctags
	568	ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	569	$(am__define_uniq_tagged_files); \
	570	test -z "$(CTAGS_ARGS)$$unique" \
	571	\|\| $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
	572	$$unique
	573
	574	GTAGS:
	575	here=`$(am__cd) $(top_builddir) && pwd` \
	576	&& $(am__cd) $(top_srcdir) \
	577	&& gtags -i $(GTAGS_ARGS) "$$here"
	578	cscopelist: cscopelist-am
	579
	580	cscopelist-am: $(am__tagged_files)
	581	list='$(am__tagged_files)'; \
	582	case "$(srcdir)" in \
	583	[\\/]* \| ?:[\\/]*) sdir="$(srcdir)" ;; \
	584	*) sdir=$(subdir)/$(srcdir) ;; \
	585	esac; \
	586	for i in $$list; do \
	587	if test -f "$$i"; then \
	588	echo "$(subdir)/$$i"; \
	589	else \
	590	echo "$$sdir/$$i"; \
	591	fi; \
	592	done >> $(top_builddir)/cscope.files
	593
	594	distclean-tags:
	595	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
	596
	597	distdir: $(DISTFILES)
	598	@srcdirstrip=`echo "$(srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	599	topsrcdirstrip=`echo "$(top_srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	600	list='$(DISTFILES)'; \
	601	dist_files=`for file in $$list; do echo $$file; done \| \
	602	sed -e "s\|^$$srcdirstrip/\|\|;t" \
	603	-e "s\|^$$topsrcdirstrip/\|$(top_builddir)/\|;t"`; \
	604	case $$dist_files in \
	605	/) $(MKDIR_P) `echo "$$dist_files" \| \
	606	sed '/\//!d;s\|^\|$(distdir)/\|;s,/[^/]*$$,,' \| \
	607	sort -u` ;; \
	608	esac; \
	609	for file in $$dist_files; do \
	610	if test -f $$file \|\| test -d $$file; then d=.; else d=$(srcdir); fi; \
	611	if test -d $$d/$$file; then \
	612	dir=`echo "/$$file" \| sed -e 's,/[^/]*$$,,'`; \
	613	if test -d "$(distdir)/$$file"; then \
	614	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	615	fi; \
	616	if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
	617	cp -fpR $(srcdir)/$$file "$(distdir)$$dir" \|\| exit 1; \
	618	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	619	fi; \
	620	cp -fpR $$d/$$file "$(distdir)$$dir" \|\| exit 1; \
	621	else \
	622	test -f "$(distdir)/$$file" \
	623	\|\| cp -p $$d/$$file "$(distdir)/$$file" \
	624	\|\| exit 1; \
	625	fi; \
	626	done
	627	check-am: all-am
	628	check: check-am
	629	all-am: Makefile $(LTLIBRARIES)
	630	installdirs:
	631	install: install-am
	632	install-exec: install-exec-am
	633	install-data: install-data-am
	634	uninstall: uninstall-am
	635
	636	install-am: all-am
	637	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
	638
	639	installcheck: installcheck-am
	640	install-strip:
	641	if test -z '$(STRIP)'; then \
	642	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	643	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	644	install; \
	645	else \
	646	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	647	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	648	"INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
	649	fi
	650	mostlyclean-generic:
	651
	652	clean-generic:
	653
	654	distclean-generic:
	655	-test -z "$(CONFIG_CLEAN_FILES)" \|\| rm -f $(CONFIG_CLEAN_FILES)
	656	-test . = "$(srcdir)" \|\| test -z "$(CONFIG_CLEAN_VPATH_FILES)" \|\| rm -f $(CONFIG_CLEAN_VPATH_FILES)
	657
	658	maintainer-clean-generic:
	659	@echo "This command is intended for maintainers to use"
	660	@echo "it deletes files that may require special tools to rebuild."
	661	clean: clean-am
	662
	663	clean-am: clean-generic clean-libtool clean-noinstLTLIBRARIES \
	664	mostlyclean-am
	665
	666	distclean: distclean-am
	667	-rm -rf ./$(DEPDIR)
	668	-rm -f Makefile
	669	distclean-am: clean-am distclean-compile distclean-generic \
	670	distclean-tags
	671
	672	dvi: dvi-am
	673
	674	dvi-am:
	675
	676	html: html-am
	677
	678	html-am:
	679
	680	info: info-am
	681
	682	info-am:
	683
	684	install-data-am:
	685
	686	install-dvi: install-dvi-am
	687
	688	install-dvi-am:
	689
	690	install-exec-am:
	691
	692	install-html: install-html-am
	693
	694	install-html-am:
	695
	696	install-info: install-info-am
	697
	698	install-info-am:
	699
	700	install-man:
	701
	702	install-pdf: install-pdf-am
	703
	704	install-pdf-am:
	705
	706	install-ps: install-ps-am
	707
	708	install-ps-am:
	709
	710	installcheck-am:
	711
	712	maintainer-clean: maintainer-clean-am
	713	-rm -rf ./$(DEPDIR)
	714	-rm -f Makefile
	715	maintainer-clean-am: distclean-am maintainer-clean-generic
	716
	717	mostlyclean: mostlyclean-am
	718
	719	mostlyclean-am: mostlyclean-compile mostlyclean-generic \
	720	mostlyclean-libtool
	721
	722	pdf: pdf-am
	723
	724	pdf-am:
	725
	726	ps: ps-am
	727
	728	ps-am:
	729
	730	uninstall-am:
	731
	732	.MAKE: install-am install-strip
	733
	734	.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean clean-generic \
	735	clean-libtool clean-noinstLTLIBRARIES cscopelist-am ctags \
	736	ctags-am distclean distclean-compile distclean-generic \
	737	distclean-libtool distclean-tags distdir dvi dvi-am html \
	738	html-am info info-am install install-am install-data \
	739	install-data-am install-dvi install-dvi-am install-exec \
	740	install-exec-am install-html install-html-am install-info \
	741	install-info-am install-man install-pdf install-pdf-am \
	742	install-ps install-ps-am install-strip installcheck \
	743	installcheck-am installdirs maintainer-clean \
	744	maintainer-clean-generic mostlyclean mostlyclean-compile \
	745	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
	746	tags tags-am uninstall uninstall-am
	747
	748
	749	# Tell versions [3.59,3.63) of GNU make to not export all variables.
	750	# Otherwise a system limit (for SysV at least) may be exceeded.
	751	.NOEXPORT:

+24

-0

libde265/encoder/algo/algo.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "libde265/encoder/algo/algo.h"
	23	#include "libde265/encoder/encoder-context.h"

+74

-0

libde265/encoder/algo/algo.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef ALGO_H
	23	#define ALGO_H
	24
	25	#include "libde265/encoder/encode.h"
	26
	27
	28	/* When entering the next recursion level, it is assumed that
	29	a valid CB structure is passed down. Ownership is transferred to
	30	the new algorithm. That algorithm passes back a (possibly different)
	31	CB structure that the first algorithm should use. The original CB
	32	structure might have been deleted in the called algorithm.
	33
	34	The context_model_table passed down is at the current state.
	35	When the algorithm returns, the state should represent the state
	36	after running this algorithm.
	37
	38	When returning from the algorithm, it is also assumed that the
	39	ectx->img content (reconstruction and metadata) represent the
	40	current state. When the algorithm tries several variants, it
	41	has to restore the state to the selected variant.
	42	*/
	43
	44	class Algo_CB
	45	{
	46	public:
	47	virtual ~Algo_CB() { }
	48
	49	/* The context_model_table that is provided can be modified and
	50	even released in the function. On exit, it should be filled with
	51	a (optionally new) context_model_table that represents the state
	52	after encoding the syntax element. However, to speed up computation,
	53	it is also allowed to not modify the context_model_table at all.
	54	*/
	55	virtual enc_cb* analyze(encoder_context*,
	56	context_model_table&,
	57	enc_cb* cb) = 0;
	58	};
	59
	60
	61	class Algo_PB
	62	{
	63	public:
	64	virtual ~Algo_PB() { }
	65
	66	virtual enc_cb* analyze(encoder_context*,
	67	context_model_table&,
	68	enc_cb* cb,
	69	int PBidx, int x,int y,int w,int h) = 0;
	70	};
	71
	72
	73	#endif

+113

-0

libde265/encoder/algo/cb-interpartmode.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/cb-interpartmode.h"
	24	#include "libde265/encoder/algo/coding-options.h"
	25	#include "libde265/encoder/encoder-context.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29
	30
	31
	32	enc_cb* Algo_CB_InterPartMode::codeAllPBs(encoder_context* ectx,
	33	context_model_table& ctxModel,
	34	enc_cb* cb)
	35	{
	36	int x = cb->x;
	37	int y = cb->y;
	38	int log2Size = cb->log2Size;
	39	int w = 1<<log2Size;
	40	int s; // splitSize;
	41
	42	int nPB;
	43	switch (cb->PartMode) {
	44	case PART_2Nx2N:
	45	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x,y,1<<log2Size,1<<log2Size);
	46	break;
	47
	48	case PART_NxN:
	49	s = 1<<(log2Size-1);
	50	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x ,y ,s,s);
	51	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x+s,y ,s,s);
	52	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 2, x ,y+s,s,s);
	53	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 3, x+s,y+s,s,s);
	54	break;
	55
	56	case PART_2NxN:
	57	s = 1<<(log2Size-1);
	58	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x,y ,w,s);
	59	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x,y+s,w,s);
	60	break;
	61
	62	case PART_Nx2N:
	63	s = 1<<(log2Size-1);
	64	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x ,y,s,w);
	65	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x+s,y,s,w);
	66	break;
	67
	68	case PART_2NxnU:
	69	s = 1<<(log2Size-2);
	70	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x,y ,w,s);
	71	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x,y+s,w,w-s);
	72	break;
	73
	74	case PART_2NxnD:
	75	s = 1<<(log2Size-2);
	76	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x,y ,w,w-s);
	77	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x,y+w-s,w,s);
	78	break;
	79
	80	case PART_nLx2N:
	81	s = 1<<(log2Size-2);
	82	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x ,y,s ,w);
	83	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x+s,y,w-s,w);
	84	break;
	85
	86	case PART_nRx2N:
	87	s = 1<<(log2Size-2);
	88	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 0, x ,y,w-s,w);
	89	cb = mChildAlgo->analyze(ectx, ctxModel, cb, 1, x+w-s,y,s ,w);
	90	break;
	91	}
	92
	93	return cb;
	94	}
	95
	96
	97	enc_cb* Algo_CB_InterPartMode_Fixed::analyze(encoder_context* ectx,
	98	context_model_table& ctxModel,
	99	enc_cb* cb)
	100	{
	101	const int x = cb->x;
	102	const int y = cb->y;
	103
	104	enum PartMode partMode = mParams.partMode();
	105
	106	cb->PartMode = partMode;
	107	ectx->img->set_PartMode(x,y, partMode);
	108
	109	cb = codeAllPBs(ectx,ctxModel,cb);
	110
	111	return cb;
	112	}

+105

-0

libde265/encoder/algo/cb-interpartmode.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CB_INTERPARTMODE_H
	23	#define CB_INTERPARTMODE_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38	#include "libde265/encoder/algo/tb-intrapredmode.h"
	39	#include "libde265/encoder/algo/tb-split.h"
	40	#include "libde265/encoder/algo/cb-intrapartmode.h"
	41
	42
	43	// ========== CB Intra/Inter decision ==========
	44
	45	class Algo_CB_InterPartMode : public Algo_CB
	46	{
	47	public:
	48	virtual ~Algo_CB_InterPartMode() { }
	49
	50	void setChildAlgo(Algo_PB* algo) { mChildAlgo = algo; }
	51
	52	protected:
	53	Algo_PB* mChildAlgo;
	54
	55	enc_cb* codeAllPBs(encoder_context*,
	56	context_model_table&,
	57	enc_cb* cb);
	58	};
	59
	60
	61
	62
	63	class option_InterPartMode : public choice_option<enum PartMode> // choice_option
	64	{
	65	public:
	66	option_InterPartMode() {
	67	add_choice("2Nx2N", PART_2Nx2N, true);
	68	add_choice("NxN", PART_NxN);
	69	add_choice("Nx2N", PART_Nx2N);
	70	add_choice("2NxN", PART_2NxN);
	71	add_choice("2NxnU", PART_2NxnU);
	72	add_choice("2NxnD", PART_2NxnD);
	73	add_choice("nLx2N", PART_nLx2N);
	74	add_choice("nRx2N", PART_nRx2N);
	75	}
	76	};
	77
	78	class Algo_CB_InterPartMode_Fixed : public Algo_CB_InterPartMode
	79	{
	80	public:
	81	struct params
	82	{
	83	params() {
	84	partMode.set_ID("CB-InterPartMode-Fixed-partMode");
	85	}
	86
	87	option_InterPartMode partMode;
	88	};
	89
	90	void registerParams(config_parameters& config) {
	91	config.add_option(&mParams.partMode);
	92	}
	93
	94	void setParams(const params& p) { mParams=p; }
	95
	96	virtual enc_cb* analyze(encoder_context*,
	97	context_model_table&,
	98	enc_cb* cb);
	99
	100	private:
	101	params mParams;
	102	};
	103
	104	#endif

+128

-0

libde265/encoder/algo/cb-intra-inter.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/cb-intra-inter.h"
	24	#include "libde265/encoder/algo/coding-options.h"
	25	#include "libde265/encoder/encoder-context.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29
	30
	31
	32	enc_cb* Algo_CB_IntraInter_BruteForce::analyze(encoder_context* ectx,
	33	context_model_table& ctxModel,
	34	enc_cb* cb)
	35	{
	36	assert(cb->pcm_flag==0);
	37
	38	bool try_intra = true;
	39	bool try_inter = (ectx->shdr->slice_type != SLICE_TYPE_I);
	40
	41	bool debug_halt = try_inter;
	42	//try_inter = false;
	43	//try_intra = !try_inter; // TODO HACK: no intra in inter frames
	44
	45	if (ectx->imgdata->frame_number > 0) {
	46	//printf("%d\n",ectx->imgdata->frame_number);
	47	}
	48
	49	// 0: intra
	50	// 1: inter
	51
	52	CodingOptions options(ectx,cb,ctxModel);
	53
	54	CodingOption option_intra = options.new_option(try_intra);
	55	CodingOption option_inter = options.new_option(try_inter);
	56
	57	options.start();
	58
	59	enc_cb* cb_inter = NULL;
	60	enc_cb* cb_intra = NULL;
	61
	62	const int log2CbSize = cb->log2Size;
	63	const int x = cb->x;
	64	const int y = cb->y;
	65
	66
	67	// try encoding with inter
	68
	69	if (option_inter) {
	70	option_inter.begin();
	71	cb_inter = option_inter.get_cb();
	72
	73	cb_inter->PredMode = MODE_INTER;
	74	ectx->img->set_pred_mode(x,y, log2CbSize, MODE_INTER);
	75
	76	enc_cb* cb_result;
	77	cb_result=mInterAlgo->analyze(ectx, option_inter.get_context(), cb_inter);
	78
	79	if (cb_result->PredMode != MODE_SKIP) {
	80	CABAC_encoder_estim* cabac = option_inter.get_cabac();
	81	cabac->reset();
	82
	83	cabac->write_CABAC_bit(CONTEXT_MODEL_PRED_MODE_FLAG, 0); // 0 - inter
	84	float rate_pred_mode_flag = cabac->getRDBits();
	85	//printf("inter bits: %f\n", rate_pred_mode_flag);
	86
	87	cb_result->rate += rate_pred_mode_flag;
	88	}
	89
	90	option_inter.set_cb(cb_result);
	91
	92	option_inter.end();
	93	}
	94
	95
	96	// try intra
	97
	98	if (option_intra) {
	99	option_intra.begin();
	100	cb_intra = option_intra.get_cb();
	101
	102	cb_intra->PredMode = MODE_INTRA;
	103	ectx->img->set_pred_mode(x,y, log2CbSize, MODE_INTRA);
	104
	105	enc_cb* cb_result;
	106	cb_result=mIntraAlgo->analyze(ectx, option_intra.get_context(), cb_intra);
	107
	108	if (ectx->shdr->slice_type != SLICE_TYPE_I) {
	109	CABAC_encoder_estim* cabac = option_intra.get_cabac();
	110	cabac->reset();
	111
	112	cabac->write_CABAC_bit(CONTEXT_MODEL_PRED_MODE_FLAG, 1); // 1 - intra
	113	float rate_pred_mode_flag = cabac->getRDBits();
	114	//printf("intra bits: %f\n", rate_pred_mode_flag);
	115
	116	cb_result->rate += rate_pred_mode_flag;
	117	}
	118
	119	option_intra.set_cb(cb_result);
	120
	121	option_intra.end();
	122	}
	123
	124
	125	options.compute_rdo_costs();
	126	return options.return_best_rdo();
	127	}

+67

-0

libde265/encoder/algo/cb-intra-inter.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CB_INTRA_INTER_H
	23	#define CB_INTRA_INTER_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38	#include "libde265/encoder/algo/tb-intrapredmode.h"
	39	#include "libde265/encoder/algo/tb-split.h"
	40	#include "libde265/encoder/algo/cb-intrapartmode.h"
	41
	42
	43	// ========== CB Intra/Inter decision ==========
	44
	45	class Algo_CB_IntraInter : public Algo_CB
	46	{
	47	public:
	48	virtual ~Algo_CB_IntraInter() { }
	49
	50	void setIntraChildAlgo(Algo_CB* algo) { mIntraAlgo = algo; }
	51	void setInterChildAlgo(Algo_CB* algo) { mInterAlgo = algo; }
	52
	53	protected:
	54	Algo_CB* mIntraAlgo;
	55	Algo_CB* mInterAlgo;
	56	};
	57
	58	class Algo_CB_IntraInter_BruteForce : public Algo_CB_IntraInter
	59	{
	60	public:
	61	virtual enc_cb* analyze(encoder_context*,
	62	context_model_table&,
	63	enc_cb* cb);
	64	};
	65
	66	#endif

+180

-0

libde265/encoder/algo/cb-intrapartmode.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/cb-intrapartmode.h"
	24	#include "libde265/encoder/algo/coding-options.h"
	25	#include "libde265/encoder/encoder-context.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29	#include <iostream>
	30
	31
	32	#define ENCODER_DEVELOPMENT 1
	33
	34
	35
	36	enc_cb* Algo_CB_IntraPartMode_BruteForce::analyze(encoder_context* ectx,
	37	context_model_table& ctxModel,
	38	enc_cb* cb_in)
	39	{
	40	const int log2CbSize = cb_in->log2Size;
	41	const int x = cb_in->x;
	42	const int y = cb_in->y;
	43
	44	const bool can_use_NxN = ((log2CbSize == ectx->sps.Log2MinCbSizeY) &&
	45	(log2CbSize > ectx->sps.Log2MinTrafoSize));
	46
	47
	48	// test all modes
	49
	50	assert(cb_in->pcm_flag==0);
	51
	52
	53	// 0: 2Nx2N (always checked)
	54	// 1: NxN (only checked at MinCbSize)
	55
	56	CodingOptions options(ectx,cb_in,ctxModel);
	57	CodingOption option[2];
	58	option[0] = options.new_option(true);
	59	option[1] = options.new_option(can_use_NxN);
	60
	61	options.start();
	62
	63	for (int p=0;p<2;p++)
	64	if (option[p]) {
	65	option[p].begin();
	66
	67	enc_cb* cb = option[p].get_cb();
	68
	69	// --- set intra prediction mode ---
	70
	71	cb->PartMode = (p==0 ? PART_2Nx2N : PART_NxN);
	72
	73	ectx->img->set_pred_mode(x,y, log2CbSize, cb->PredMode); // TODO: probably unnecessary
	74	ectx->img->set_PartMode (x,y, cb->PartMode);
	75
	76	// encode transform tree
	77
	78	int IntraSplitFlag= (cb->PredMode == MODE_INTRA && cb->PartMode == PART_NxN);
	79	int MaxTrafoDepth = ectx->sps.max_transform_hierarchy_depth_intra + IntraSplitFlag;
	80
	81	cb->transform_tree = mTBIntraPredModeAlgo->analyze(ectx, option[p].get_context(),
	82	ectx->imgdata->input, NULL, cb,
	83	x,y, x,y, log2CbSize,
	84	0,
	85	0, MaxTrafoDepth, IntraSplitFlag);
	86
	87	cb->distortion = cb->transform_tree->distortion;
	88	cb->rate = cb->transform_tree->rate;
	89
	90
	91	// rate for cu syntax
	92
	93	logtrace(LogSymbols,"$1 part_mode=%d\n",cb->PartMode);
	94	if (log2CbSize == ectx->sps.Log2MinCbSizeY) {
	95	int bin = (cb->PartMode==PART_2Nx2N);
	96	option[p].get_cabac()->reset();
	97	option[p].get_cabac()->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+0, bin);
	98	cb->rate += option[p].get_cabac()->getRDBits();
	99	}
	100
	101	option[p].end();
	102	}
	103
	104	options.compute_rdo_costs();
	105	return options.return_best_rdo();
	106	}
	107
	108
	109	enc_cb* Algo_CB_IntraPartMode_Fixed::analyze(encoder_context* ectx,
	110	context_model_table& ctxModel,
	111	enc_cb* cb)
	112	{
	113	std::cout << "CB-IntraPartMode in size=" << (1<<cb->log2Size)
	114	<< " hash=" << ctxModel.debug_dump() << "\n";
	115
	116	enum PartMode PartMode = mParams.partMode();
	117
	118
	119	const int log2CbSize = cb->log2Size;
	120	const int x = cb->x;
	121	const int y = cb->y;
	122
	123
	124	// NxN can only be applied at minimum CB size.
	125	// If we are not at the minimum size, we have to use 2Nx2N.
	126
	127	if (PartMode==PART_NxN && log2CbSize != ectx->sps.Log2MinCbSizeY) {
	128	PartMode = PART_2Nx2N;
	129	}
	130
	131
	132	// --- set intra prediction mode ---
	133
	134	cb->PartMode = PartMode;
	135	ectx->img->set_PartMode(x,y, PartMode);
	136
	137
	138	// encode transform tree
	139
	140	int IntraSplitFlag= (cb->PredMode == MODE_INTRA && cb->PartMode == PART_NxN);
	141	int MaxTrafoDepth = ectx->sps.max_transform_hierarchy_depth_intra + IntraSplitFlag;
	142
	143	cb->transform_tree = mTBIntraPredModeAlgo->analyze(ectx, ctxModel,
	144	ectx->imgdata->input, NULL, cb,
	145	cb->x,cb->y, cb->x,cb->y, log2CbSize,
	146	0,
	147	0, MaxTrafoDepth, IntraSplitFlag);
	148
	149
	150	// rate and distortion for this CB
	151
	152	cb->distortion = cb->transform_tree->distortion;
	153	cb->rate = cb->transform_tree->rate;
	154
	155	std::cout << "SUM TB-tree hinter PartMode " << cb->rate << "\n";
	156
	157
	158	// rate for cu syntax
	159
	160	CABAC_encoder_estim estim;
	161	estim.set_context_models(&ctxModel);
	162
	163	//encode_coding_unit(ectx,&estim,cb,x,y,log2CbSize, false);
	164
	165	//encode_part_mode(ectx,&estim, MODE_INTRA, PartMode, 0);
	166
	167	logtrace(LogSymbols,"$1 part_mode=%d\n",PartMode);
	168	if (log2CbSize == ectx->sps.Log2MinCbSizeY) {
	169	int bin = (PartMode==PART_2Nx2N);
	170	estim.write_CABAC_bit(CONTEXT_MODEL_PART_MODE+0, bin);
	171	}
	172
	173	cb->rate += estim.getRDBits();
	174
	175	std::cout << "CB-IntraPartMode out size=" << (1<<cb->log2Size)
	176	<< " hash=" << ctxModel.debug_dump() << "\n";
	177
	178	return cb;
	179	}

+143

-0

libde265/encoder/algo/cb-intrapartmode.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CB_INTRAPARTMODE_H
	23	#define CB_INTRAPARTMODE_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38	#include "libde265/encoder/algo/tb-intrapredmode.h"
	39	#include "libde265/encoder/algo/tb-split.h"
	40
	41
	42	/* Encoder search tree, bottom up:
	43
	44	- Algo_TB_Split - whether TB is split or not
	45
	46	- Algo_TB_IntraPredMode - choose the intra prediction mode (or NOP, if at the wrong tree level)
	47
	48	- Algo_CB_IntraPartMode - choose between NxN and 2Nx2N intra parts
	49
	50	- Algo_CB_Split - whether CB is split or not
	51
	52	- Algo_CTB_QScale - select QScale on CTB granularity
	53	*/
	54
	55
	56	// ========== CB intra NxN vs. 2Nx2N decision ==========
	57
	58	enum ALGO_CB_IntraPartMode {
	59	ALGO_CB_IntraPartMode_BruteForce,
	60	ALGO_CB_IntraPartMode_Fixed
	61	};
	62
	63	class option_ALGO_CB_IntraPartMode : public choice_option<enum ALGO_CB_IntraPartMode>
	64	{
	65	public:
	66	option_ALGO_CB_IntraPartMode() {
	67	add_choice("fixed", ALGO_CB_IntraPartMode_Fixed);
	68	add_choice("brute-force",ALGO_CB_IntraPartMode_BruteForce, true);
	69	}
	70	};
	71
	72
	73	class Algo_CB_IntraPartMode : public Algo_CB
	74	{
	75	public:
	76	Algo_CB_IntraPartMode() : mTBIntraPredModeAlgo(NULL) { }
	77	virtual ~Algo_CB_IntraPartMode() { }
	78
	79	virtual enc_cb* analyze(encoder_context*,
	80	context_model_table&,
	81	enc_cb* cb) = 0;
	82
	83	void setChildAlgo(Algo_TB_IntraPredMode* algo) { mTBIntraPredModeAlgo = algo; }
	84
	85	protected:
	86	Algo_TB_IntraPredMode* mTBIntraPredModeAlgo;
	87	};
	88
	89	/* Try both NxN, 2Nx2N and choose better one.
	90	*/
	91	class Algo_CB_IntraPartMode_BruteForce : public Algo_CB_IntraPartMode
	92	{
	93	public:
	94	virtual enc_cb* analyze(encoder_context*,
	95	context_model_table&,
	96	enc_cb* cb);
	97	};
	98
	99
	100	class option_PartMode : public choice_option<enum PartMode> // choice_option
	101	{
	102	public:
	103	option_PartMode() {
	104	add_choice("NxN", PART_NxN);
	105	add_choice("2Nx2N", PART_2Nx2N, true);
	106	}
	107	};
	108
	109
	110	/* Always use choose selected part mode.
	111	If NxN is chosen but cannot be applied (CB tree not at maximum depth), 2Nx2N is used instead.
	112	*/
	113	class Algo_CB_IntraPartMode_Fixed : public Algo_CB_IntraPartMode
	114	{
	115	public:
	116	Algo_CB_IntraPartMode_Fixed() { }
	117
	118	struct params
	119	{
	120	params() {
	121	partMode.set_ID("CB-IntraPartMode-Fixed-partMode");
	122	}
	123
	124	option_PartMode partMode;
	125	};
	126
	127	void registerParams(config_parameters& config) {
	128	config.add_option(&mParams.partMode);
	129	}
	130
	131	void setParams(const params& p) { mParams=p; }
	132
	133	virtual enc_cb* analyze(encoder_context* ectx,
	134	context_model_table& ctxModel,
	135	enc_cb* cb);
	136
	137	private:
	138	params mParams;
	139	};
	140
	141
	142	#endif

+138

-0

libde265/encoder/algo/cb-mergeindex.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/cb-mergeindex.h"
	24	#include "libde265/encoder/encoder-context.h"
	25	#include <assert.h>
	26	#include <limits>
	27	#include <math.h>
	28
	29
	30
	31
	32	enc_cb* Algo_CB_MergeIndex_Fixed::analyze(encoder_context* ectx,
	33	context_model_table& ctxModel,
	34	enc_cb* cb)
	35	{
	36	assert(cb->split_cu_flag==false);
	37	assert(cb->PredMode==MODE_SKIP); // TODO: \|\| (cb->PredMode==MODE_INTER && cb->inter.skip_flag));
	38
	39
	40	MotionVectorSpec mergeCandList[5];
	41
	42	int partIdx = 0;
	43
	44	int cbSize = 1 << cb->log2Size;
	45
	46	get_merge_candidate_list(ectx, ectx->shdr, ectx->img,
	47	cb->x, cb->y, // xC/yC
	48	cb->x, cb->y, // xP/yP
	49	cbSize, // nCS
	50	cbSize,cbSize, // nPbW/nPbH
	51	partIdx, // partIdx
	52	mergeCandList);
	53
	54	motion_spec& spec = cb->inter.pb[partIdx].spec;
	55
	56	spec.merge_flag = 1; // we use merge mode
	57	spec.merge_idx = 0;
	58
	59
	60	// build prediction
	61
	62	// previous frame (TODO)
	63	const de265_image* refimg = ectx->get_image(ectx->imgdata->frame_number -1);
	64
	65	//printf("prev frame: %p %d\n",refimg,ectx->imgdata->frame_number);
	66
	67	/*
	68	printf("#l0: %d\n",ectx->imgdata->shdr.num_ref_idx_l0_active);
	69	printf("#l1: %d\n",ectx->imgdata->shdr.num_ref_idx_l1_active);
	70
	71	for (int i=0;i<ectx->imgdata->shdr.num_ref_idx_l0_active;i++)
	72	printf("RefPixList[0][%d] = %d\n", i, ectx->imgdata->shdr.RefPicList[0][i]);
	73	*/
	74
	75	// TODO: fake motion data
	76
	77	const MotionVectorSpec& vec = mergeCandList[spec.merge_idx];
	78	cb->inter.pb[partIdx].motion = vec;
	79
	80	ectx->img->set_mv_info(cb->x,cb->y, 1<<cb->log2Size,1<<cb->log2Size, vec);
	81
	82	generate_inter_prediction_samples(ectx, ectx->shdr, ectx->prediction,
	83	cb->x,cb->y, // int xC,int yC,
	84	0,0, // int xB,int yB,
	85	1<<cb->log2Size, // int nCS,
	86	1<<cb->log2Size,
	87	1<<cb->log2Size, // int nPbW,int nPbH,
	88	&vec);
	89
	90	generate_inter_prediction_samples(ectx, ectx->shdr, ectx->img,
	91	cb->x,cb->y, // int xC,int yC,
	92	0,0, // int xB,int yB,
	93	1<<cb->log2Size, // int nCS,
	94	1<<cb->log2Size,
	95	1<<cb->log2Size, // int nPbW,int nPbH,
	96	&vec);
	97
	98	// estimate rate for sending merge index
	99
	100	//CABAC_encoder_estim cabac;
	101	//cabac.write_bits();
	102
	103	int IntraSplitFlag = 0;
	104	int MaxTrafoDepth = ectx->sps.max_transform_hierarchy_depth_inter;
	105
	106	if (mCodeResidual) {
	107	assert(false);
	108	cb->transform_tree = mTBSplit->analyze(ectx,ctxModel, ectx->imgdata->input, NULL, cb,
	109	cb->x,cb->y,cb->x,cb->y, cb->log2Size,0,
	110	0, MaxTrafoDepth, IntraSplitFlag);
	111
	112	cb->inter.rqt_root_cbf = ! cb->transform_tree->isZeroBlock();
	113
	114	cb->distortion = cb->transform_tree->distortion;
	115	cb->rate = cb->transform_tree->rate;
	116	}
	117	else {
	118	const de265_image* input = ectx->imgdata->input;
	119	de265_image* img = ectx->prediction;
	120	int x0 = cb->x;
	121	int y0 = cb->y;
	122	int tbSize = 1<<cb->log2Size;
	123
	124	CABAC_encoder_estim cabac;
	125	cabac.set_context_models(&ctxModel);
	126	encode_merge_idx(ectx, &cabac, spec.merge_idx);
	127
	128	cb->distortion = compute_distortion_ssd(input, img, x0,y0, cb->log2Size, 0);
	129	cb->rate = cabac.getRDBits();
	130
	131	cb->inter.rqt_root_cbf = 0;
	132	}
	133
	134	//printf("%d;%d rqt_root_cbf=%d\n",cb->x,cb->y,cb->inter.rqt_root_cbf);
	135
	136	return cb;
	137	}

+69

-0

libde265/encoder/algo/cb-mergeindex.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CB_MERGEINDEX_H
	23	#define CB_MERGEINDEX_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38	#include "libde265/encoder/algo/tb-split.h"
	39
	40
	41	// ========== CB Skip/Inter decision ==========
	42
	43	class Algo_CB_MergeIndex : public Algo_CB
	44	{
	45	public:
	46	Algo_CB_MergeIndex() : mCodeResidual(false) { }
	47	virtual ~Algo_CB_MergeIndex() { }
	48
	49	void set_code_residual(bool flag=true) { mCodeResidual=flag; }
	50
	51	void setChildAlgo(Algo_TB_Split* algo) { mTBSplit = algo; }
	52	// TODO void setInterChildAlgo(Algo_CB_IntraPartMode* algo) { mInterPartModeAlgo = algo; }
	53
	54	protected:
	55	Algo_TB_Split* mTBSplit;
	56
	57	bool mCodeResidual;
	58	};
	59
	60	class Algo_CB_MergeIndex_Fixed : public Algo_CB_MergeIndex
	61	{
	62	public:
	63	virtual enc_cb* analyze(encoder_context*,
	64	context_model_table&,
	65	enc_cb* cb);
	66	};
	67
	68	#endif

+106

-0

libde265/encoder/algo/cb-skip.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/cb-skip.h"
	24	#include "libde265/encoder/algo/coding-options.h"
	25	#include "libde265/encoder/encoder-context.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29
	30
	31
	32
	33	enc_cb* Algo_CB_Skip_BruteForce::analyze(encoder_context* ectx,
	34	context_model_table& ctxModel,
	35	enc_cb* cb)
	36	{
	37	bool try_skip = (ectx->shdr->slice_type != SLICE_TYPE_I);
	38	bool try_nonskip = true;
	39
	40	CodingOptions options(ectx,cb,ctxModel);
	41	CodingOption option_skip = options.new_option(try_skip);
	42	CodingOption option_nonskip = options.new_option(try_nonskip);
	43	options.start();
	44
	45	for (int i=0;i<CONTEXT_MODEL_TABLE_LENGTH;i++) {
	46	//printf("%i: %d/%d\n",i, ctxModel[i].state, ctxModel[i].MPSbit);
	47	}
	48
	49	if (option_skip) {
	50	CodingOption& opt = option_skip;
	51	opt.begin();
	52
	53	enc_cb* cb = opt.get_cb();
	54
	55	// calc rate for skip flag (=true)
	56
	57	CABAC_encoder_estim* cabac = opt.get_cabac();
	58	encode_cu_skip_flag(ectx, cabac, cb, true);
	59	float rate_pred_mode = cabac->getRDBits();
	60	cabac->reset();
	61
	62	// set skip flag
	63
	64	cb->PredMode = MODE_SKIP;
	65	ectx->img->set_pred_mode(cb->x,cb->y, cb->log2Size, cb->PredMode);
	66
	67	// encode CB
	68
	69	cb = mSkipAlgo->analyze(ectx, opt.get_context(), cb);
	70
	71	// add rate for PredMode
	72
	73	cb->rate += rate_pred_mode;
	74	opt.set_cb(cb);
	75	opt.end();
	76	}
	77
	78	if (option_nonskip) {
	79	CodingOption& opt = option_nonskip;
	80	enc_cb* cb = opt.get_cb();
	81
	82	opt.begin();
	83
	84	// calc rate for skip flag (=true)
	85
	86	float rate_pred_mode = 0;
	87
	88	if (try_skip) {
	89	CABAC_encoder_estim* cabac = opt.get_cabac();
	90	encode_cu_skip_flag(ectx, cabac, cb, false);
	91	rate_pred_mode = cabac->getRDBits();
	92	cabac->reset();
	93	}
	94
	95	cb = mNonSkipAlgo->analyze(ectx, opt.get_context(), cb);
	96	// add rate for PredMode
	97
	98	cb->rate += rate_pred_mode;
	99	opt.set_cb(cb);
	100	opt.end();
	101	}
	102
	103	options.compute_rdo_costs();
	104	return options.return_best_rdo();
	105	}

+69

-0

libde265/encoder/algo/cb-skip.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CB_SKIP_H
	23	#define CB_SKIP_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38	#include "libde265/encoder/algo/cb-mergeindex.h"
	39
	40
	41	// ========== CB Skip/Inter decision ==========
	42
	43	class Algo_CB_Skip : public Algo_CB
	44	{
	45	public:
	46	virtual ~Algo_CB_Skip() { }
	47
	48	void setSkipAlgo(Algo_CB_MergeIndex* algo) {
	49	mSkipAlgo = algo;
	50	mSkipAlgo->set_code_residual(false);
	51	}
	52
	53	void setNonSkipAlgo(Algo_CB* algo) { mNonSkipAlgo = algo; }
	54
	55	protected:
	56	Algo_CB_MergeIndex* mSkipAlgo;
	57	Algo_CB* mNonSkipAlgo;
	58	};
	59
	60	class Algo_CB_Skip_BruteForce : public Algo_CB_Skip
	61	{
	62	public:
	63	virtual enc_cb* analyze(encoder_context*,
	64	context_model_table&,
	65	enc_cb* cb);
	66	};
	67
	68	#endif

+155

-0

libde265/encoder/algo/cb-split.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/cb-split.h"
	24	#include "libde265/encoder/algo/coding-options.h"
	25	#include "libde265/encoder/encoder-context.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29	#include <iostream>
	30
	31
	32	// Utility function to encode all four children in a splitted CB.
	33	// Children are coded with the specified algo_cb_split.
	34	enc_cb* Algo_CB_Split::encode_cb_split(encoder_context* ectx,
	35	context_model_table& ctxModel,
	36	enc_cb* cb)
	37	{
	38	int w = ectx->imgdata->input->get_width();
	39	int h = ectx->imgdata->input->get_height();
	40
	41
	42	cb->split_cu_flag = true;
	43
	44
	45	// encode all 4 children and sum their distortions and rates
	46
	47	for (int i=0;i<4;i++) {
	48	int child_x = cb->x + ((i&1) << (cb->log2Size-1));
	49	int child_y = cb->y + ((i>>1) << (cb->log2Size-1));
	50
	51	if (child_x>=w \|\| child_y>=h) {
	52	cb->children[i] = NULL;
	53	}
	54	else {
	55	enc_cb* childCB = new enc_cb;
	56	childCB->log2Size = cb->log2Size-1;
	57	childCB->ctDepth = cb->ctDepth+1;
	58
	59	childCB->x = child_x;
	60	childCB->y = child_y;
	61
	62	cb->children[i] = analyze(ectx, ctxModel, childCB);
	63
	64	cb->distortion += cb->children[i]->distortion;
	65	cb->rate += cb->children[i]->rate;
	66	}
	67	}
	68
	69	return cb;
	70	}
	71
	72
	73
	74
	75	enc_cb* Algo_CB_Split_BruteForce::analyze(encoder_context* ectx,
	76	context_model_table& ctxModel,
	77	enc_cb* cb_input)
	78	{
	79	assert(cb_input->pcm_flag==0);
	80
	81	// --- prepare coding options ---
	82
	83	const SplitType split_type = get_split_type(&ectx->sps,
	84	cb_input->x, cb_input->y,
	85	cb_input->log2Size);
	86
	87
	88	bool can_split_CB = (split_type != ForcedNonSplit);
	89	bool can_nosplit_CB = (split_type != ForcedSplit);
	90
	91	//if (can_split_CB) { can_nosplit_CB=false; } // TODO TMP
	92	//if (can_nosplit_CB) { can_split_CB=false; } // TODO TMP
	93
	94	CodingOptions options(ectx, cb_input, ctxModel);
	95
	96	CodingOption option_split = options.new_option(can_split_CB);
	97	CodingOption option_no_split = options.new_option(can_nosplit_CB);
	98
	99	options.start();
	100
	101	// --- encode without splitting ---
	102
	103	if (option_no_split) {
	104	CodingOption& opt = option_no_split; // abbrev.
	105
	106	opt.begin();
	107
	108	enc_cb* cb = opt.get_cb();
	109
	110	// set CB size in image data-structure
	111	ectx->img->set_ctDepth(cb->x,cb->y,cb->log2Size, cb->ctDepth);
	112	ectx->img->set_log2CbSize(cb->x,cb->y,cb->log2Size, true);
	113
	114	/* We set QP here, because this is required at in non-split CBs only.
	115	*/
	116	cb->qp = ectx->active_qp;
	117
	118	// analyze subtree
	119	assert(mChildAlgo);
	120	cb = mChildAlgo->analyze(ectx, opt.get_context(), cb);
	121
	122	// add rate for split flag
	123	if (split_type == OptionalSplit) {
	124	encode_split_cu_flag(ectx,opt.get_cabac(), cb->x,cb->y, cb->ctDepth, 0);
	125	cb->rate += opt.get_cabac_rate();
	126	}
	127
	128	opt.set_cb(cb);
	129	opt.end();
	130	}
	131
	132	// --- encode with splitting ---
	133
	134	if (option_split) {
	135	option_split.begin();
	136
	137	enc_cb* cb = option_split.get_cb();
	138	cb = encode_cb_split(ectx, option_split.get_context(), cb);
	139
	140	// add rate for split flag
	141	if (split_type == OptionalSplit) {
	142	encode_split_cu_flag(ectx,option_split.get_cabac(), cb->x,cb->y, cb->ctDepth, 1);
	143	cb->rate += option_split.get_cabac_rate();
	144	}
	145
	146	option_split.set_cb(cb);
	147	option_split.end();
	148	}
	149
	150	options.compute_rdo_costs();
	151	enc_cb* bestCB = options.return_best_rdo();
	152
	153	return bestCB;
	154	}

+85

-0

libde265/encoder/algo/cb-split.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CB_SPLIT_H
	23	#define CB_SPLIT_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38	#include "libde265/encoder/algo/tb-intrapredmode.h"
	39	#include "libde265/encoder/algo/tb-split.h"
	40
	41
	42	/* Encoder search tree, bottom up:
	43
	44	- Algo_TB_Split - whether TB is split or not
	45
	46	- Algo_TB_IntraPredMode - choose the intra prediction mode (or NOP, if at the wrong tree level)
	47
	48	- Algo_CB_IntraPartMode - choose between NxN and 2Nx2N intra parts
	49
	50	- Algo_CB_Split - whether CB is split or not
	51
	52	- Algo_CTB_QScale - select QScale on CTB granularity
	53	*/
	54
	55
	56	// ========== CB split decision ==========
	57
	58	class Algo_CB_Split : public Algo_CB
	59	{
	60	public:
	61	virtual ~Algo_CB_Split() { }
	62
	63	// TODO: probably, this will later be a intra/inter decision which again
	64	// has two child algorithms, depending on the coding mode.
	65	void setChildAlgo(Algo_CB* algo) { mChildAlgo = algo; }
	66
	67	protected:
	68	Algo_CB* mChildAlgo;
	69
	70	enc_cb* encode_cb_split(encoder_context* ectx,
	71	context_model_table& ctxModel,
	72	enc_cb* cb);
	73	};
	74
	75
	76	class Algo_CB_Split_BruteForce : public Algo_CB_Split
	77	{
	78	public:
	79	virtual enc_cb* analyze(encoder_context*,
	80	context_model_table&,
	81	enc_cb* cb);
	82	};
	83
	84	#endif

+170

-0

libde265/encoder/algo/coding-options.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "libde265/encoder/algo/coding-options.h"
	23	#include "libde265/encoder/encoder-context.h"
	24
	25
	26	CodingOptions::CodingOptions(encoder_context* ectx, enc_cb* cb, context_model_table& tab)
	27	{
	28	mCBInput = cb;
	29	mContextModelInput = &tab;
	30
	31	mCurrentlyReconstructedOption=-1;
	32	mBestRDO=-1;
	33
	34	mECtx = ectx;
	35	}
	36
	37	CodingOptions::~CodingOptions()
	38	{
	39	}
	40
	41	CodingOption CodingOptions::new_option(bool active)
	42	{
	43	if (!active) {
	44	return CodingOption();
	45	}
	46
	47
	48	CodingOptionData opt;
	49
	50	bool firstOption = mOptions.empty();
	51	if (firstOption) {
	52	opt.cb = mCBInput;
	53	}
	54	else {
	55	opt.cb = new enc_cb(*mCBInput);
	56	}
	57
	58	opt.context = *mContextModelInput;
	59
	60	CodingOption option(this, mOptions.size());
	61
	62	mOptions.push_back( std::move(opt) );
	63
	64	return option;
	65	}
	66
	67
	68	void CodingOptions::start(enum RateEstimationMethod rateMethod)
	69	{
	70	/* We don't need the input context model anymore.
	71	Releasing it now may save a copy during a later decouple().
	72	*/
	73	mContextModelInput->release();
	74
	75	bool adaptiveContext;
	76	switch (rateMethod) {
	77	case Rate_Default:
	78	adaptiveContext = mECtx->use_adaptive_context;
	79	break;
	80	case Rate_FixedContext:
	81	adaptiveContext = false;
	82	break;
	83	case Rate_AdaptiveContext:
	84	adaptiveContext = true;
	85	break;
	86	}
	87
	88	if (adaptiveContext) {
	89	/* If we modify the context models in this algorithm,
	90	we need separate models for each option.
	91	*/
	92	for (auto& option : mOptions) {
	93	option.context.decouple();
	94	}
	95
	96	cabac = &cabac_adaptive;
	97	}
	98	else {
	99	cabac = &cabac_constant;
	100	}
	101	}
	102
	103
	104	void CodingOption::begin()
	105	{
	106	mParent->cabac->reset();
	107	mParent->cabac->set_context_models( &get_context() );
	108
	109	if (mParent->mCurrentlyReconstructedOption >= 0) {
	110	mParent->mOptions[mParent->mCurrentlyReconstructedOption].cb->save(mParent->mECtx->img);
	111	}
	112
	113	mParent->mCurrentlyReconstructedOption = mOptionIdx;
	114	}
	115
	116	void CodingOption::end()
	117	{
	118	assert(mParent->mCurrentlyReconstructedOption == mOptionIdx);
	119	}
	120
	121
	122	void CodingOptions::compute_rdo_costs()
	123	{
	124	for (int i=0;i<mOptions.size();i++) {
	125	mOptions[i].rdoCost = mOptions[i].cb->distortion + mECtx->lambda * mOptions[i].cb->rate;
	126	}
	127	}
	128
	129
	130	enc_cb* CodingOptions::return_best_rdo()
	131	{
	132	assert(mOptions.size()>0);
	133
	134
	135	float bestRDOCost = 0;
	136	bool first=true;
	137	int bestRDO=-1;
	138
	139	for (int i=0;i<mOptions.size();i++) {
	140	float cost = mOptions[i].rdoCost;
	141	if (first \|\| cost < bestRDOCost) {
	142	bestRDOCost = cost;
	143	first = false;
	144	bestRDO = i;
	145	}
	146	}
	147
	148
	149	assert(bestRDO>=0);
	150
	151	if (bestRDO != mCurrentlyReconstructedOption) {
	152	mOptions[bestRDO].cb->restore(mECtx->img);
	153	}
	154
	155	*mContextModelInput = mOptions[bestRDO].context;
	156
	157
	158	// delete all CBs except the best one
	159
	160	for (int i=0;i<mOptions.size();i++) {
	161	if (i != bestRDO)
	162	{
	163	delete mOptions[i].cb;
	164	mOptions[i].cb = NULL;
	165	}
	166	}
	167
	168	return mOptions[bestRDO].cb;
	169	}

+134

-0

libde265/encoder/algo/coding-options.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CODING_OPTIONS_H
	23	#define CODING_OPTIONS_H
	24
	25	#include "libde265/encoder/encode.h"
	26
	27
	28	class CodingOption;
	29
	30
	31	class CodingOptions
	32	{
	33	public:
	34	CodingOptions(encoder_context, enc_cb, context_model_table& tab);
	35	~CodingOptions();
	36
	37	// --- init --- call before object use
	38
	39	CodingOption new_option(bool active=true);
	40
	41	enum RateEstimationMethod
	42	{
	43	Rate_Default,
	44	Rate_AdaptiveContext,
	45	Rate_FixedContext
	46	};
	47
	48	void start(enum RateEstimationMethod = Rate_Default);
	49
	50
	51	// --- processing ---
	52
	53	// compute RDO cost (D + lambda*R) for all options
	54	void compute_rdo_costs();
	55
	56
	57	// --- end processing --- do not call any function after this one
	58
	59	/* Return the CB with the lowest RDO cost. All other CBs are destroyed.
	60	If the current reconstruction and metadata are not from the returned CB,
	61	the data from the returned CB is reconstructed.
	62	*/
	63	enc_cb* return_best_rdo();
	64
	65	private:
	66	struct CodingOptionData
	67	{
	68	enc_cb* cb;
	69	context_model_table context;
	70	bool mOptionActive;
	71	float rdoCost;
	72	};
	73
	74
	75	encoder_context* mECtx;
	76
	77	enc_cb* mCBInput;
	78	context_model_table* mContextModelInput;
	79
	80	int mCurrentlyReconstructedOption;
	81	int mBestRDO;
	82
	83	std::vector<CodingOptionData> mOptions;
	84
	85	CABAC_encoder_estim cabac_adaptive;
	86	CABAC_encoder_estim_constant cabac_constant;
	87	CABAC_encoder_estim* cabac;
	88
	89	friend class CodingOption;
	90	};
	91
	92
	93	class CodingOption
	94	{
	95	public:
	96	CodingOption() {
	97	mParent = nullptr;
	98	mOptionIdx = 0;
	99	}
	100
	101	enc_cb* get_cb() { return mParent->mOptions[mOptionIdx].cb; }
	102	void set_cb(enc_cb* cb) { mParent->mOptions[mOptionIdx].cb = cb; }
	103
	104	context_model_table& get_context() { return mParent->mOptions[mOptionIdx].context; }
	105
	106	operator bool() const { return mParent; }
	107
	108	/* When modifying the reconstruction image or metadata, you have to
	109	encapsulate the modification between these two functions to ensure
	110	that the correct reconstruction will be active after return_best_rdo().
	111	*/
	112	void begin();
	113	void end();
	114
	115	// Manually set RDO costs instead of computing them with compute_rdo_costs.
	116	// Only required when using custom costs.
	117	void set_rdo_cost(float rdo) { mParent->mOptions[mOptionIdx].rdoCost=rdo; }
	118
	119	CABAC_encoder_estim* get_cabac() { return mParent->cabac; }
	120	float get_cabac_rate() const { return mParent->cabac->getRDBits(); }
	121
	122	private:
	123	CodingOption(class CodingOptions* parent, int idx)
	124	: mParent(parent), mOptionIdx(idx) { }
	125
	126	class CodingOptions* mParent;
	127	int mOptionIdx;
	128
	129	friend class CodingOptions;
	130	};
	131
	132
	133	#endif

+53

-0

libde265/encoder/algo/ctb-qscale.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/ctb-qscale.h"
	24	#include "libde265/encoder/encoder-context.h"
	25	#include <assert.h>
	26	#include <limits>
	27	#include <math.h>
	28
	29
	30	#define ENCODER_DEVELOPMENT 1
	31
	32
	33	enc_cb* Algo_CTB_QScale_Constant::analyze(encoder_context* ectx,
	34	context_model_table& ctxModel,
	35	int ctb_x,int ctb_y)
	36	{
	37	enc_cb* cb = new enc_cb();
	38
	39	cb->log2Size = ectx->sps.Log2CtbSizeY;
	40	cb->ctDepth = 0;
	41	cb->x = ctb_x;
	42	cb->y = ctb_y;
	43
	44	ectx->img->set_QPY(ctb_x,ctb_y, cb->log2Size, ectx->active_qp);
	45
	46	// write currently unused coding options to image
	47	ectx->img->set_cu_transquant_bypass(ctb_x,ctb_y,cb->log2Size, cb->cu_transquant_bypass_flag);
	48	ectx->img->set_pcm_flag(ctb_x,ctb_y,cb->log2Size, cb->pcm_flag);
	49
	50	assert(mChildAlgo);
	51	return mChildAlgo->analyze(ectx,ctxModel,cb);
	52	}

+107

-0

libde265/encoder/algo/ctb-qscale.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef CTB_QSCALE_H
	23	#define CTB_QSCALE_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/cb-split.h"
	38
	39
	40	/* Encoder search tree, bottom up:
	41
	42	- Algo_TB_Split - whether TB is split or not
	43
	44	- Algo_TB_IntraPredMode - choose the intra prediction mode (or NOP, if at the wrong tree level)
	45
	46	- Algo_CB_IntraPartMode - choose between NxN and 2Nx2N intra parts
	47
	48	- Algo_CB_Split - whether CB is split or not
	49
	50	- Algo_CTB_QScale - select QScale on CTB granularity
	51	*/
	52
	53
	54	// ========== choose a qscale at CTB level ==========
	55
	56	class Algo_CTB_QScale
	57	{
	58	public:
	59	Algo_CTB_QScale() : mChildAlgo(NULL) { }
	60	virtual ~Algo_CTB_QScale() { }
	61
	62	virtual enc_cb* analyze(encoder_context*,
	63	context_model_table&,
	64	int ctb_x,int ctb_y) = 0;
	65
	66	void setChildAlgo(Algo_CB_Split* algo) { mChildAlgo = algo; }
	67
	68	protected:
	69	Algo_CB_Split* mChildAlgo;
	70	};
	71
	72
	73
	74	class Algo_CTB_QScale_Constant : public Algo_CTB_QScale
	75	{
	76	public:
	77	struct params
	78	{
	79	params() {
	80	mQP.set_range(1,51);
	81	mQP.set_default(27);
	82	mQP.set_ID("CTB-QScale-Constant");
	83	mQP.set_cmd_line_options("qp",'q');
	84	}
	85
	86	option_int mQP;
	87	};
	88
	89	void setParams(const params& p) { mParams=p; }
	90
	91	void registerParams(config_parameters& config) {
	92	config.add_option(&mParams.mQP);
	93	}
	94
	95	virtual enc_cb* analyze(encoder_context*,
	96	context_model_table&,
	97	int ctb_x,int ctb_y);
	98
	99	int getQP() const { return mParams.mQP; }
	100
	101	private:
	102	params mParams;
	103	};
	104
	105
	106	#endif

+306

-0

libde265/encoder/algo/pb-mv.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/algo/pb-mv.h"
	24	#include "libde265/encoder/algo/coding-options.h"
	25	#include "libde265/encoder/encoder-context.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29
	30
	31
	32	enc_cb* Algo_PB_MV_Test::analyze(encoder_context* ectx,
	33	context_model_table& ctxModel,
	34	enc_cb* cb,
	35	int PBidx, int x,int y,int w,int h)
	36	{
	37	enum MVTestMode testMode = mParams.testMode();
	38
	39
	40	MotionVector mvp[2];
	41
	42	fill_luma_motion_vector_predictors(ectx, ectx->shdr, ectx->img,
	43	cb->x,cb->y,1<<cb->log2Size, x,y,w,h,
	44	0, // l
	45	0, 0, // int refIdx, int partIdx,
	46	mvp);
	47
	48	//printf("%d/%d: [%d;%d] [%d;%d]\n",cb->x,cb->y, mvp[0].x,mvp[0].y, mvp[1].x,mvp[1].y);
	49
	50
	51	motion_spec& spec = cb->inter.pb[PBidx].spec;
	52	MotionVectorSpec& vec = cb->inter.pb[PBidx].motion;
	53
	54	spec.merge_flag = 0;
	55	spec.merge_idx = 0;
	56
	57	spec.inter_pred_idc = PRED_L0;
	58	spec.refIdx[0] = vec.refIdx[0] = 0;
	59	spec.mvp_l0_flag = 0;
	60
	61	int value = mParams.range();
	62
	63	switch (testMode) {
	64	case MVTestMode_Zero:
	65	spec.mvd[0][0]=0;
	66	spec.mvd[0][1]=0;
	67	break;
	68
	69	case MVTestMode_Random:
	70	spec.mvd[0][0] = (rand() % (2*value+1)) - value;
	71	spec.mvd[0][1] = (rand() % (2*value+1)) - value;
	72	break;
	73
	74	case MVTestMode_Horizontal:
	75	spec.mvd[0][0]=value;
	76	spec.mvd[0][1]=0;
	77	break;
	78
	79	case MVTestMode_Vertical:
	80	spec.mvd[0][0]=0;
	81	spec.mvd[0][1]=value;
	82	break;
	83	}
	84
	85	spec.mvd[0][0] -= mvp[0].x;
	86	spec.mvd[0][1] -= mvp[0].y;
	87
	88	vec.mv[0].x = mvp[0].x + spec.mvd[0][0];
	89	vec.mv[0].y = mvp[0].y + spec.mvd[0][1];
	90	vec.predFlag[0] = 1;
	91	vec.predFlag[1] = 0;
	92
	93	ectx->img->set_mv_info(x,y,w,h, vec);
	94
	95	generate_inter_prediction_samples(ectx, ectx->shdr, ectx->prediction,
	96	cb->x,cb->y, // int xC,int yC,
	97	0,0, // int xB,int yB,
	98	1<<cb->log2Size, // int nCS,
	99	1<<cb->log2Size,
	100	1<<cb->log2Size, // int nPbW,int nPbH,
	101	&vec);
	102
	103	// TODO estimate rate for sending MV
	104
	105	int IntraSplitFlag = 0;
	106	int MaxTrafoDepth = ectx->sps.max_transform_hierarchy_depth_inter;
	107
	108	mCodeResidual=true;
	109	if (mCodeResidual) {
	110	assert(mTBSplitAlgo);
	111	cb->transform_tree = mTBSplitAlgo->analyze(ectx,ctxModel, ectx->imgdata->input, NULL, cb,
	112	cb->x,cb->y,cb->x,cb->y, cb->log2Size,0,
	113	0, MaxTrafoDepth, IntraSplitFlag);
	114
	115	cb->inter.rqt_root_cbf = ! cb->transform_tree->isZeroBlock();
	116
	117	cb->distortion = cb->transform_tree->distortion;
	118	cb->rate = cb->transform_tree->rate;
	119	}
	120	else {
	121	const de265_image* input = ectx->imgdata->input;
	122	de265_image* img = ectx->prediction;
	123	int x0 = cb->x;
	124	int y0 = cb->y;
	125	int tbSize = 1<<cb->log2Size;
	126
	127	cb->distortion = compute_distortion_ssd(input, img, x0,y0, cb->log2Size, 0);
	128	cb->rate = 5; // fake (MV)
	129
	130	cb->inter.rqt_root_cbf = 0;
	131	}
	132
	133	return cb;
	134	}
	135
	136
	137
	138
	139	int sad(const uint8_t* p1,int stride1,
	140	const uint8_t* p2,int stride2,
	141	int w,int h)
	142	{
	143	int cost=0;
	144
	145	for (int y=0;y<h;y++) {
	146	for (int x=0;x<w;x++) {
	147	cost += abs_value(p1 - p2);
	148	p1++;
	149	p2++;
	150	}
	151
	152	p1 += stride1-w;
	153	p2 += stride2-w;
	154	}
	155
	156	return cost;
	157	}
	158
	159
	160	enc_cb* Algo_PB_MV_Search::analyze(encoder_context* ectx,
	161	context_model_table& ctxModel,
	162	enc_cb* cb,
	163	int PBidx, int x,int y,int pbW,int pbH)
	164	{
	165	enum MVSearchAlgo searchAlgo = mParams.mvSearchAlgo();
	166
	167
	168	MotionVector mvp[2];
	169
	170	fill_luma_motion_vector_predictors(ectx, ectx->shdr, ectx->img,
	171	cb->x,cb->y,1<<cb->log2Size, x,y,pbW,pbH,
	172	0, // l
	173	0, 0, // int refIdx, int partIdx,
	174	mvp);
	175
	176	motion_spec& spec = cb->inter.pb[PBidx].spec;
	177	MotionVectorSpec& vec = cb->inter.pb[PBidx].motion;
	178
	179	spec.merge_flag = 0;
	180	spec.merge_idx = 0;
	181
	182	spec.inter_pred_idc = PRED_L0;
	183	spec.refIdx[0] = vec.refIdx[0] = 0;
	184	spec.mvp_l0_flag = 0;
	185
	186	int hrange = mParams.hrange();
	187	int vrange = mParams.vrange();
	188
	189	// previous frame (TODO)
	190	const de265_image* refimg = ectx->get_image(ectx->imgdata->frame_number -1);
	191	const de265_image* inputimg = ectx->imgdata->input;
	192
	193	int w = refimg->get_width();
	194	int h = refimg->get_height();
	195
	196	int mincost = 0x7fffffff;
	197
	198	double lambda = 10.0;
	199
	200	double bits_h = new double[2hrange+1];
	201	double bits_v = new double[2vrange+1];
	202
	203	for (int i=-hrange;i<=hrange;i++) {
	204	int diff = (i - mvp[0].x);
	205	int b;
	206
	207	if (diff==0) { b=0; }
	208	else if (diff==1 \|\| diff==-1) { b=2; }
	209	else { b=abs_value(b+2); }
	210
	211	bits_h[i+hrange]=b;
	212	}
	213
	214	for (int i=-vrange;i<=vrange;i++) {
	215	int diff = (i - mvp[0].y);
	216	int b;
	217
	218	if (diff==0) { b=0; }
	219	else if (diff==1 \|\| diff==-1) { b=2; }
	220	else { b=abs_value(b+2); }
	221
	222	bits_v[i+vrange]=b;
	223	}
	224
	225	for (int my = y-vrange; my<=y+vrange; my++)
	226	for (int mx = x-hrange; mx<=x+hrange; mx++)
	227	{
	228	if (mx<0 \|\| mx+pbW>w \|\| my<0 \|\| my+pbH>h) continue;
	229
	230	int cost = sad(refimg->get_image_plane_at_pos(0,mx,my),
	231	refimg->get_image_stride(0),
	232	inputimg->get_image_plane_at_pos(0,x,y),
	233	inputimg->get_image_stride(0),
	234	pbW,pbH);
	235
	236	int bits = bits_h[mx-x+hrange] + bits_v[my-y+vrange];
	237
	238	cost += lambda * bits;
	239
	240	//printf("%d %d : %d\n",mx,my,cost);
	241
	242	if (cost<mincost) {
	243	mincost=cost;
	244
	245	spec.mvd[0][0]=(mx-x)<<2;
	246	spec.mvd[0][1]=(my-y)<<2;
	247	}
	248	}
	249
	250	spec.mvd[0][0] -= mvp[0].x;
	251	spec.mvd[0][1] -= mvp[0].y;
	252
	253	vec.mv[0].x = mvp[0].x + spec.mvd[0][0];
	254	vec.mv[0].y = mvp[0].y + spec.mvd[0][1];
	255	vec.predFlag[0] = 1;
	256	vec.predFlag[1] = 0;
	257
	258	ectx->img->set_mv_info(x,y,pbW,pbH, vec);
	259
	260	generate_inter_prediction_samples(ectx, ectx->shdr, ectx->prediction,
	261	cb->x,cb->y, // int xC,int yC,
	262	0,0, // int xB,int yB,
	263	1<<cb->log2Size, // int nCS,
	264	1<<cb->log2Size,
	265	1<<cb->log2Size, // int nPbW,int nPbH,
	266	&vec);
	267
	268	// --- create residual ---
	269
	270
	271
	272	// TODO estimate rate for sending MV
	273
	274	int IntraSplitFlag = 0;
	275	int MaxTrafoDepth = ectx->sps.max_transform_hierarchy_depth_inter;
	276
	277	mCodeResidual=true;
	278	if (mCodeResidual) {
	279	cb->transform_tree = mTBSplitAlgo->analyze(ectx,ctxModel, ectx->imgdata->input, NULL, cb,
	280	cb->x,cb->y,cb->x,cb->y, cb->log2Size,0,
	281	0, MaxTrafoDepth, IntraSplitFlag);
	282
	283	cb->inter.rqt_root_cbf = ! cb->transform_tree->isZeroBlock();
	284
	285	cb->distortion = cb->transform_tree->distortion;
	286	cb->rate = cb->transform_tree->rate;
	287	}
	288	else {
	289	const de265_image* input = ectx->imgdata->input;
	290	de265_image* img = ectx->img;
	291	int x0 = cb->x;
	292	int y0 = cb->y;
	293	int tbSize = 1<<cb->log2Size;
	294
	295	cb->distortion = compute_distortion_ssd(input, img, x0,y0, cb->log2Size, 0);
	296	cb->rate = 5; // fake (MV)
	297
	298	cb->inter.rqt_root_cbf = 0;
	299	}
	300
	301	delete[] bits_h;
	302	delete[] bits_v;
	303
	304	return cb;
	305	}

+178

-0

libde265/encoder/algo/pb-mv.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef PB_MV_H
	23	#define PB_MV_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/algo.h"
	38
	39
	40	// ========== CB Intra/Inter decision ==========
	41
	42	class Algo_TB_Split;
	43
	44
	45	class Algo_PB_MV : public Algo_PB
	46	{
	47	public:
	48	Algo_PB_MV() : mTBSplitAlgo(NULL) { }
	49	virtual ~Algo_PB_MV() { }
	50
	51	void setChildAlgo(Algo_TB_Split* algo) { mTBSplitAlgo = algo; }
	52
	53	protected:
	54	Algo_TB_Split* mTBSplitAlgo;
	55	};
	56
	57
	58
	59
	60	enum MVTestMode
	61	{
	62	MVTestMode_Zero,
	63	MVTestMode_Random,
	64	MVTestMode_Horizontal,
	65	MVTestMode_Vertical
	66	};
	67
	68	class option_MVTestMode : public choice_option<enum MVTestMode>
	69	{
	70	public:
	71	option_MVTestMode() {
	72	add_choice("zero", MVTestMode_Zero);
	73	add_choice("random", MVTestMode_Random);
	74	add_choice("horiz", MVTestMode_Horizontal, true);
	75	add_choice("verti", MVTestMode_Vertical);
	76	}
	77	};
	78
	79
	80	class Algo_PB_MV_Test : public Algo_PB_MV
	81	{
	82	public:
	83	Algo_PB_MV_Test() : mCodeResidual(false) { }
	84
	85	struct params
	86	{
	87	params() {
	88	testMode.set_ID("PB-MV-TestMode");
	89	range.set_ID ("PB-MV-Range");
	90	range.set_default(4);
	91	}
	92
	93	option_MVTestMode testMode;
	94	option_int range;
	95	};
	96
	97	void registerParams(config_parameters& config) {
	98	config.add_option(&mParams.testMode);
	99	config.add_option(&mParams.range);
	100	}
	101
	102	void setParams(const params& p) { mParams=p; }
	103
	104	virtual enc_cb* analyze(encoder_context*,
	105	context_model_table&,
	106	enc_cb* cb,
	107	int PBidx, int x,int y,int w,int h);
	108
	109	private:
	110	params mParams;
	111
	112	bool mCodeResidual;
	113	};
	114
	115
	116
	117
	118	enum MVSearchAlgo
	119	{
	120	MVSearchAlgo_Zero,
	121	MVSearchAlgo_Full,
	122	MVSearchAlgo_Diamond,
	123	MVSearchAlgo_PMVFast
	124	};
	125
	126	class option_MVSearchAlgo : public choice_option<enum MVSearchAlgo>
	127	{
	128	public:
	129	option_MVSearchAlgo() {
	130	add_choice("zero", MVSearchAlgo_Zero);
	131	add_choice("full", MVSearchAlgo_Full, true);
	132	add_choice("diamond",MVSearchAlgo_Diamond);
	133	add_choice("pmvfast",MVSearchAlgo_PMVFast);
	134	}
	135	};
	136
	137
	138	class Algo_PB_MV_Search : public Algo_PB_MV
	139	{
	140	public:
	141	Algo_PB_MV_Search() : mCodeResidual(false) { }
	142
	143	struct params
	144	{
	145	params() {
	146	mvSearchAlgo.set_ID("PB-MV-Search-Algo");
	147	hrange.set_ID ("PB-MV-Search-HRange");
	148	vrange.set_ID ("PB-MV-Search-VRange");
	149	hrange.set_default(8);
	150	vrange.set_default(8);
	151	}
	152
	153	option_MVSearchAlgo mvSearchAlgo;
	154	option_int hrange;
	155	option_int vrange;
	156	};
	157
	158	void registerParams(config_parameters& config) {
	159	config.add_option(&mParams.mvSearchAlgo);
	160	config.add_option(&mParams.hrange);
	161	config.add_option(&mParams.vrange);
	162	}
	163
	164	void setParams(const params& p) { mParams=p; }
	165
	166	virtual enc_cb* analyze(encoder_context*,
	167	context_model_table&,
	168	enc_cb* cb,
	169	int PBidx, int x,int y,int w,int h);
	170
	171	private:
	172	params mParams;
	173
	174	bool mCodeResidual;
	175	};
	176
	177	#endif

+524

-0

libde265/encoder/algo/tb-intrapredmode.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/analyze.h"
	24	#include "libde265/encoder/encoder-context.h"
	25	#include "libde265/encoder/algo/tb-split.h"
	26	#include <assert.h>
	27	#include <limits>
	28	#include <math.h>
	29	#include <algorithm>
	30	#include <iostream>
	31
	32
	33	enum IntraPredMode find_best_intra_mode(de265_image& img,int x0,int y0, int log2BlkSize, int cIdx,
	34	const uint8_t* ref, int stride)
	35	{
	36	//return INTRA_ANGULAR_20;
	37
	38	enum IntraPredMode best_mode;
	39	int min_sad=-1;
	40
	41	int candidates[3];
	42
	43	const seq_parameter_set* sps = &img.sps;
	44
	45
	46	fillIntraPredModeCandidates(candidates, x0,y0,
	47	sps->getPUIndexRS(x0,y0),
	48	x0>0, y0>0, &img);
	49
	50	// --- test candidates first ---
	51
	52	for (int idx=0;idx<3;idx++) {
	53	enum IntraPredMode mode = (enum IntraPredMode)candidates[idx];
	54	decode_intra_prediction(&img, x0,y0, (enum IntraPredMode)mode, 1<<log2BlkSize, cIdx);
	55
	56	uint32_t distortion = SSD(ref,stride,
	57	img.get_image_plane_at_pos(cIdx, x0,y0), img.get_image_stride(cIdx),
	58	1<<log2BlkSize, 1<<log2BlkSize);
	59
	60	int sad=distortion;
	61
	62	sad *= 0.5;
	63	//sad *= 0.9;
	64
	65	if (mode==0 \|\| sad<min_sad) {
	66	min_sad = sad;
	67	best_mode = (enum IntraPredMode)mode;
	68	}
	69	}
	70
	71
	72	// --- test all modes ---
	73
	74	for (int idx=0;idx<35;idx++) {
	75	enum IntraPredMode mode = (enum IntraPredMode)idx; //candidates[idx];
	76	decode_intra_prediction(&img, x0,y0, (enum IntraPredMode)mode, 1<<log2BlkSize, cIdx);
	77
	78
	79	uint32_t distortion = SSD(ref,stride,
	80	img.get_image_plane_at_pos(cIdx, x0,y0), img.get_image_stride(cIdx),
	81	1<<log2BlkSize, 1<<log2BlkSize);
	82
	83	int sad=distortion;
	84
	85	if (min_sad<0 \|\| sad<min_sad) {
	86	min_sad = sad;
	87	best_mode = (enum IntraPredMode)mode;
	88	}
	89	}
	90
	91	return best_mode;
	92	}
	93
	94
	95
	96
	97	float estim_TB_bitrate(const encoder_context* ectx,
	98	const de265_image* input,
	99	int x0,int y0, int log2BlkSize,
	100	enum TBBitrateEstimMethod method)
	101	{
	102	int blkSize = 1<<log2BlkSize;
	103
	104	float distortion;
	105
	106	switch (method)
	107	{
	108	case TBBitrateEstim_SSD:
	109	return SSD(input->get_image_plane_at_pos(0, x0,y0),
	110	input->get_image_stride(0),
	111	ectx->img->get_image_plane_at_pos(0, x0,y0),
	112	ectx->img->get_image_stride(0),
	113	1<<log2BlkSize, 1<<log2BlkSize);
	114	break;
	115
	116	case TBBitrateEstim_SAD:
	117	return SAD(input->get_image_plane_at_pos(0, x0,y0),
	118	input->get_image_stride(0),
	119	ectx->img->get_image_plane_at_pos(0, x0,y0),
	120	ectx->img->get_image_stride(0),
	121	1<<log2BlkSize, 1<<log2BlkSize);
	122	break;
	123
	124	case TBBitrateEstim_SATD_DCT:
	125	case TBBitrateEstim_SATD_Hadamard:
	126	{
	127	int16_t coeffs[32*32];
	128	int16_t diff[32*32];
	129
	130	diff_blk(diff,blkSize,
	131	input->get_image_plane_at_pos(0, x0,y0), input->get_image_stride(0),
	132	ectx->img->get_image_plane_at_pos(0, x0,y0), ectx->img->get_image_stride(0),
	133	blkSize);
	134
	135	if (method == TBBitrateEstim_SATD_Hadamard) {
	136	ectx->acceleration.hadamard_transform_8[log2BlkSize-2](coeffs, diff, &diff[blkSize] - &diff[0]);
	137	}
	138	else {
	139	ectx->acceleration.fwd_transform_8[log2BlkSize-2](coeffs, diff, &diff[blkSize] - &diff[0]);
	140	}
	141
	142	float distortion=0;
	143	for (int i=0;i<blkSize*blkSize;i++) {
	144	distortion += abs_value((int)coeffs[i]);
	145	}
	146
	147	return distortion;
	148	}
	149	break;
	150
	151	/*
	152	case TBBitrateEstim_AccurateBits:
	153	assert(false);
	154	return 0;
	155	*/
	156	}
	157
	158	assert(false);
	159	return 0;
	160	}
	161
	162
	163
	164	enc_tb*
	165	Algo_TB_IntraPredMode_BruteForce::analyze(encoder_context* ectx,
	166	context_model_table& ctxModel,
	167	const de265_image* input,
	168	const enc_tb* parent,
	169	enc_cb* cb,
	170	int x0,int y0, int xBase,int yBase,
	171	int log2TbSize, int blkIdx,
	172	int TrafoDepth, int MaxTrafoDepth,
	173	int IntraSplitFlag)
	174	{
	175	//printf("encode_transform_tree_may_split %d %d (%d %d) size %d\n",x0,y0,xBase,yBase,1<<log2TbSize);
	176
	177	/*
	178	enum IntraPredMode pre_intraMode = find_best_intra_mode(ectx->img,x0,y0, log2TbSize, 0,
	179	input->get_image_plane_at_pos(0,x0,y0),
	180	input->get_image_stride(0));
	181	*/
	182
	183	bool selectIntraPredMode = false;
	184	selectIntraPredMode \|= (cb->PredMode==MODE_INTRA && cb->PartMode==PART_2Nx2N && TrafoDepth==0);
	185	selectIntraPredMode \|= (cb->PredMode==MODE_INTRA && cb->PartMode==PART_NxN && TrafoDepth==1);
	186
	187	if (selectIntraPredMode) {
	188	enc_tb* tb[35];
	189
	190	float minCost = std::numeric_limits<float>::max();
	191	int minCostIdx=0;
	192	float minCandCost;
	193
	194	const de265_image* img = ectx->img;
	195	const seq_parameter_set* sps = &img->sps;
	196	int candidates[3];
	197	fillIntraPredModeCandidates(candidates, x0,y0,
	198	sps->getPUIndexRS(x0,y0),
	199	x0>0, y0>0, img);
	200
	201
	202	for (int i = 0; i<35; i++) {
	203	if (!mPredMode_enabled[i]) {
	204	tb[i]=NULL;
	205	continue;
	206	}
	207
	208
	209	context_model_table ctxIntra = ctxModel.copy();
	210	//copy_context_model_table(ctxIntra, ctxModel);
	211
	212	enum IntraPredMode intraMode = (IntraPredMode)i;
	213
	214	cb->intra.pred_mode[blkIdx] = intraMode;
	215	if (blkIdx==0) { cb->intra.chroma_mode = intraMode; }
	216
	217	ectx->img->set_IntraPredMode(x0,y0,log2TbSize, intraMode);
	218
	219	tb[intraMode] = mTBSplitAlgo->analyze(ectx,ctxIntra,input,parent,
	220	cb, x0,y0, xBase,yBase, log2TbSize, blkIdx,
	221	TrafoDepth, MaxTrafoDepth, IntraSplitFlag);
	222
	223
	224	float sad;
	225	if ((1<<log2TbSize)==8) {
	226	decode_intra_prediction(ectx->img, x0,y0, intraMode, 1<<log2TbSize, 0);
	227	sad = estim_TB_bitrate(ectx,input, x0,y0, log2TbSize, TBBitrateEstim_SAD);
	228	}
	229
	230
	231	float rate = tb[intraMode]->rate;
	232	int enc_bin;
	233
	234	if (log2TbSize==3) {
	235	// printf("RATE2 %d %f %f\n",log2TbSize,tb[intraMode]->rate,sad);
	236	}
	237
	238	/**/ if (candidates[0]==intraMode) { rate += 1; enc_bin=1; }
	239	else if (candidates[1]==intraMode) { rate += 2; enc_bin=1; }
	240	else if (candidates[2]==intraMode) { rate += 2; enc_bin=1; }
	241	else { rate += 5; enc_bin=0; }
	242
	243	CABAC_encoder_estim estim;
	244	estim.set_context_models(&ctxIntra);
	245	rate += estim.RDBits_for_CABAC_bin(CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG, enc_bin);
	246
	247	float cost = tb[intraMode]->distortion + ectx->lambda * rate;
	248	if (cost<minCost) {
	249	minCost=cost;
	250	minCostIdx=intraMode;
	251	//minCandCost=c;
	252	}
	253	}
	254
	255
	256	enum IntraPredMode intraMode = (IntraPredMode)minCostIdx;
	257
	258	cb->intra.pred_mode[blkIdx] = intraMode;
	259	if (blkIdx==0) { cb->intra.chroma_mode = intraMode; } //INTRA_CHROMA_LIKE_LUMA;
	260	ectx->img->set_IntraPredMode(x0,y0,log2TbSize, intraMode);
	261
	262	tb[minCostIdx]->reconstruct(ectx, ectx->img,
	263	cb, blkIdx);
	264
	265
	266	//printf("INTRA %d %d %d\n",pre_intraMode,intraMode,minCandCost);
	267
	268	for (int i = 0; i<35; i++) {
	269	if (i != minCostIdx) {
	270	delete tb[i];
	271	}
	272	}
	273
	274	return tb[minCostIdx];
	275	}
	276	else {
	277	return mTBSplitAlgo->analyze(ectx, ctxModel, input, parent, cb,
	278	x0,y0,xBase,yBase, log2TbSize,
	279	blkIdx, TrafoDepth, MaxTrafoDepth,
	280	IntraSplitFlag);
	281	}
	282
	283	assert(false);
	284	return nullptr;
	285	}
	286
	287
	288
	289	enc_tb*
	290	Algo_TB_IntraPredMode_MinResidual::analyze(encoder_context* ectx,
	291	context_model_table& ctxModel,
	292	const de265_image* input,
	293	const enc_tb* parent,
	294	enc_cb* cb,
	295	int x0,int y0, int xBase,int yBase,
	296	int log2TbSize, int blkIdx,
	297	int TrafoDepth, int MaxTrafoDepth,
	298	int IntraSplitFlag)
	299	{
	300
	301	bool selectIntraPredMode = false;
	302	selectIntraPredMode \|= (cb->PredMode==MODE_INTRA && cb->PartMode==PART_2Nx2N && TrafoDepth==0);
	303	selectIntraPredMode \|= (cb->PredMode==MODE_INTRA && cb->PartMode==PART_NxN && TrafoDepth==1);
	304
	305	if (selectIntraPredMode) {
	306
	307	enum IntraPredMode intraMode;
	308	float minDistortion;
	309
	310	for (int idx=0;idx<35;idx++) {
	311	enum IntraPredMode mode = (enum IntraPredMode)idx;
	312	decode_intra_prediction(ectx->img, x0,y0, (enum IntraPredMode)mode, 1<<log2TbSize, 0);
	313
	314	float distortion;
	315	distortion = estim_TB_bitrate(ectx, input, x0,y0, log2TbSize,
	316	mParams.bitrateEstimMethod());
	317
	318	if (idx==0 \|\| distortion<minDistortion) {
	319	minDistortion = distortion;
	320	intraMode = mode;
	321	}
	322	}
	323
	324
	325	cb->intra.pred_mode[blkIdx] = intraMode;
	326	if (blkIdx==0) { cb->intra.chroma_mode = intraMode; }
	327
	328	ectx->img->set_IntraPredMode(x0,y0,log2TbSize, intraMode);
	329
	330	/*
	331	decode_intra_prediction(ectx->img, x0,y0, intraMode, 1<< log2TbSize, 0);
	332	decode_intra_prediction(ectx->img, x0>>1,y0>>1, intraMode, 1<<(log2TbSize-1), 1);
	333	decode_intra_prediction(ectx->img, x0>>1,y0>>1, intraMode, 1<<(log2TbSize-1), 2);
	334	*/
	335
	336	// Note: cannot prepare intra prediction pixels here, because this has to
	337	// be done at the lowest TB split level.
	338
	339	enc_tb* tb = mTBSplitAlgo->analyze(ectx,ctxModel,input,parent,
	340	cb, x0,y0, xBase,yBase, log2TbSize, blkIdx,
	341	TrafoDepth, MaxTrafoDepth, IntraSplitFlag);
	342
	343	debug_show_image(ectx->img, 0);
	344
	345	return tb;
	346	}
	347	else {
	348	return mTBSplitAlgo->analyze(ectx, ctxModel, input, parent, cb,
	349	x0,y0,xBase,yBase, log2TbSize,
	350	blkIdx, TrafoDepth, MaxTrafoDepth,
	351	IntraSplitFlag);
	352	}
	353
	354	assert(false);
	355	return nullptr;
	356	}
	357
	358	static bool sortDistortions(std::pair<enum IntraPredMode,float> i,
	359	std::pair<enum IntraPredMode,float> j)
	360	{
	361	return i.second < j.second;
	362	}
	363
	364	enc_tb*
	365	Algo_TB_IntraPredMode_FastBrute::analyze(encoder_context* ectx,
	366	context_model_table& ctxModel,
	367	const de265_image* input,
	368	const enc_tb* parent,
	369	enc_cb* cb,
	370	int x0,int y0, int xBase,int yBase,
	371	int log2TbSize, int blkIdx,
	372	int TrafoDepth, int MaxTrafoDepth,
	373	int IntraSplitFlag)
	374	{
	375	//printf("encode_transform_tree_may_split %d %d (%d %d) size %d\n",x0,y0,xBase,yBase,1<<log2TbSize);
	376
	377	/*
	378	enum IntraPredMode pre_intraMode = find_best_intra_mode(ectx->img,x0,y0, log2TbSize, 0,
	379	input->get_image_plane_at_pos(0,x0,y0),
	380	input->get_image_stride(0));
	381	*/
	382
	383	bool selectIntraPredMode = false;
	384	selectIntraPredMode \|= (cb->PredMode==MODE_INTRA && cb->PartMode==PART_2Nx2N && TrafoDepth==0);
	385	selectIntraPredMode \|= (cb->PredMode==MODE_INTRA && cb->PartMode==PART_NxN && TrafoDepth==1);
	386
	387	if (selectIntraPredMode) {
	388	float minCost = std::numeric_limits<float>::max();
	389	int minCostIdx=0;
	390	float minCandCost;
	391
	392	const de265_image* img = ectx->img;
	393	const seq_parameter_set* sps = &img->sps;
	394	int candidates[3];
	395	fillIntraPredModeCandidates(candidates, x0,y0,
	396	sps->getPUIndexRS(x0,y0),
	397	x0>0, y0>0, img);
	398
	399
	400
	401	std::vector< std::pair<enum IntraPredMode,float> > distortions;
	402
	403	for (int idx=0;idx<35;idx++)
	404	if (idx!=candidates[0] && idx!=candidates[1] && idx!=candidates[2] && mPredMode_enabled[idx])
	405	{
	406	enum IntraPredMode mode = (enum IntraPredMode)idx;
	407	decode_intra_prediction(ectx->img, x0,y0, (enum IntraPredMode)mode, 1<<log2TbSize, 0);
	408
	409	float distortion;
	410	distortion = estim_TB_bitrate(ectx, input, x0,y0, log2TbSize,
	411	mParams.bitrateEstimMethod());
	412
	413	distortions.push_back( std::make_pair((enum IntraPredMode)idx, distortion) );
	414	}
	415
	416	std::sort( distortions.begin(), distortions.end(), sortDistortions );
	417
	418
	419	for (int i=0;i<distortions.size();i++)
	420	{
	421	//printf("%d -> %f\n",i,distortions[i].second);
	422	}
	423
	424	int keepNBest=std::min((int)mParams.keepNBest, (int)distortions.size());
	425	distortions.resize(keepNBest);
	426	distortions.push_back(std::make_pair((enum IntraPredMode)candidates[0],0));
	427	distortions.push_back(std::make_pair((enum IntraPredMode)candidates[1],0));
	428	distortions.push_back(std::make_pair((enum IntraPredMode)candidates[2],0));
	429
	430
	431	enc_tb* tb[35];
	432	context_model_table contexts[35];
	433
	434	for (int i=0;i<35;i++) tb[i]=NULL;
	435
	436	for (int i=0;i<distortions.size();i++) {
	437
	438	//copy_context_model_table(ctxIntra, ctxModel);
	439
	440	enum IntraPredMode intraMode = (IntraPredMode)distortions[i].first;
	441
	442	if (!mPredMode_enabled[intraMode]) { continue; }
	443
	444	cb->intra.pred_mode[blkIdx] = intraMode;
	445	if (blkIdx==0) { cb->intra.chroma_mode = intraMode; }
	446
	447	ectx->img->set_IntraPredMode(x0,y0,log2TbSize, intraMode);
	448
	449	contexts[intraMode] = ctxModel.copy();
	450	tb[intraMode] = mTBSplitAlgo->analyze(ectx,contexts[intraMode],input,parent,
	451	cb, x0,y0, xBase,yBase, log2TbSize, blkIdx,
	452	TrafoDepth, MaxTrafoDepth, IntraSplitFlag);
	453
	454	float rate = tb[intraMode]->rate_withoutCbfChroma;
	455	int enc_bin;
	456
	457	/**/ if (candidates[0]==intraMode) { rate += 1; enc_bin=1; }
	458	else if (candidates[1]==intraMode) { rate += 2; enc_bin=1; }
	459	else if (candidates[2]==intraMode) { rate += 2; enc_bin=1; }
	460	else { rate += 5; enc_bin=0; }
	461
	462	CABAC_encoder_estim estim;
	463	estim.set_context_models(&contexts[intraMode]);
	464	//rate += estim.RDBits_for_CABAC_bin(CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG, enc_bin);
	465	logtrace(LogSymbols,"$1 prev_intra_luma_pred_flag=%d\n",enc_bin);
	466	estim.write_CABAC_bit(CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG, enc_bin);
	467
	468	// TODO: currently we make the chroma-pred-mode decision for each part even
	469	// in NxN part mode. Since we always set this to the same value, it does not
	470	// matter. However, we should only add the rate for it once (for blkIdx=0).
	471
	472	if (blkIdx==0) {
	473	logtrace(LogSymbols,"$1 intra_chroma_pred_mode=%d\n",0);
	474	estim.write_CABAC_bit(CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE,0);
	475	}
	476	rate += estim.getRDBits();
	477
	478	float cbfRate = tb[intraMode]->rate - tb[intraMode]->rate_withoutCbfChroma;
	479	tb[intraMode]->rate_withoutCbfChroma = rate;
	480	tb[intraMode]->rate = tb[intraMode]->rate_withoutCbfChroma + cbfRate;
	481
	482	//printf("QQQ %f %f\n", b, estim.getRDBits());
	483
	484	float cost = tb[intraMode]->distortion + ectx->lambda * rate;
	485
	486	//printf("idx:%d mode:%d cost:%f\n",i,intraMode,cost);
	487
	488	if (cost<minCost) {
	489	minCost=cost;
	490	minCostIdx=intraMode;
	491	//minCandCost=c;
	492	}
	493	}
	494
	495
	496	enum IntraPredMode intraMode = (IntraPredMode)minCostIdx;
	497
	498	cb->intra.pred_mode[blkIdx] = intraMode;
	499	if (blkIdx==0) { cb->intra.chroma_mode = intraMode; } //INTRA_CHROMA_LIKE_LUMA;
	500	ectx->img->set_IntraPredMode(x0,y0,log2TbSize, intraMode);
	501
	502	tb[minCostIdx]->reconstruct(ectx, ectx->img, cb, blkIdx);
	503	ctxModel = contexts[minCostIdx];
	504
	505	for (int i = 0; i<35; i++) {
	506	if (i != minCostIdx) {
	507	delete tb[i];
	508	}
	509	}
	510
	511	return tb[minCostIdx];
	512	}
	513	else {
	514	return mTBSplitAlgo->analyze(ectx, ctxModel, input, parent, cb,
	515	x0,y0,xBase,yBase, log2TbSize,
	516	blkIdx, TrafoDepth, MaxTrafoDepth,
	517	IntraSplitFlag);
	518
	519	}
	520
	521	assert(false);
	522	return nullptr;
	523	}

+281

-0

libde265/encoder/algo/tb-intrapredmode.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef TB_INTRAPREDMODE_H
	23	#define TB_INTRAPREDMODE_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37
	38	/* Encoder search tree, bottom up:
	39
	40	- Algo_TB_Split - whether TB is split or not
	41
	42	- Algo_TB_IntraPredMode - choose the intra prediction mode (or NOP, if at the wrong tree level)
	43
	44	- Algo_CB_IntraPartMode - choose between NxN and 2Nx2N intra parts
	45
	46	- Algo_CB_Split - whether CB is split or not
	47
	48	- Algo_CTB_QScale - select QScale on CTB granularity
	49	*/
	50
	51
	52	// ========== TB intra prediction mode ==========
	53
	54	enum ALGO_TB_IntraPredMode {
	55	ALGO_TB_IntraPredMode_BruteForce,
	56	ALGO_TB_IntraPredMode_FastBrute,
	57	ALGO_TB_IntraPredMode_MinResidual
	58	};
	59
	60	class option_ALGO_TB_IntraPredMode : public choice_option<enum ALGO_TB_IntraPredMode>
	61	{
	62	public:
	63	option_ALGO_TB_IntraPredMode() {
	64	add_choice("min-residual",ALGO_TB_IntraPredMode_MinResidual);
	65	add_choice("brute-force" ,ALGO_TB_IntraPredMode_BruteForce);
	66	add_choice("fast-brute" ,ALGO_TB_IntraPredMode_FastBrute, true);
	67	}
	68	};
	69
	70
	71	enum TBBitrateEstimMethod {
	72	//TBBitrateEstim_AccurateBits,
	73	TBBitrateEstim_SSD,
	74	TBBitrateEstim_SAD,
	75	TBBitrateEstim_SATD_DCT,
	76	TBBitrateEstim_SATD_Hadamard
	77	};
	78
	79	class option_TBBitrateEstimMethod : public choice_option<enum TBBitrateEstimMethod>
	80	{
	81	public:
	82	option_TBBitrateEstimMethod() {
	83	add_choice("ssd",TBBitrateEstim_SSD);
	84	add_choice("sad",TBBitrateEstim_SAD);
	85	add_choice("satd-dct",TBBitrateEstim_SATD_DCT);
	86	add_choice("satd",TBBitrateEstim_SATD_Hadamard, true);
	87	}
	88	};
	89
	90	class Algo_TB_Split;
	91
	92
	93	/** Base class for intra prediction-mode algorithms.
	94	Selects one of the 35 prediction modes.
	95	*/
	96	class Algo_TB_IntraPredMode
	97	{
	98	public:
	99	Algo_TB_IntraPredMode() : mTBSplitAlgo(NULL) { }
	100	virtual ~Algo_TB_IntraPredMode() { }
	101
	102	virtual enc_tb* analyze(encoder_context*,
	103	context_model_table&,
	104	const de265_image* input,
	105	const enc_tb* parent,
	106	enc_cb* cb,
	107	int x0,int y0, int xBase,int yBase, int log2TbSize,
	108	int blkIdx,
	109	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag) = 0;
	110
	111	void setChildAlgo(Algo_TB_Split* algo) { mTBSplitAlgo = algo; }
	112
	113	protected:
	114	Algo_TB_Split* mTBSplitAlgo;
	115	};
	116
	117
	118	enum ALGO_TB_IntraPredMode_Subset {
	119	ALGO_TB_IntraPredMode_Subset_All,
	120	ALGO_TB_IntraPredMode_Subset_HVPlus,
	121	ALGO_TB_IntraPredMode_Subset_DC,
	122	ALGO_TB_IntraPredMode_Subset_Planar
	123	};
	124
	125	class option_ALGO_TB_IntraPredMode_Subset : public choice_option<enum ALGO_TB_IntraPredMode_Subset>
	126	{
	127	public:
	128	option_ALGO_TB_IntraPredMode_Subset() {
	129	add_choice("all" ,ALGO_TB_IntraPredMode_Subset_All, true);
	130	add_choice("HV+" ,ALGO_TB_IntraPredMode_Subset_HVPlus);
	131	add_choice("DC" ,ALGO_TB_IntraPredMode_Subset_DC);
	132	add_choice("planar",ALGO_TB_IntraPredMode_Subset_Planar);
	133	}
	134	};
	135
	136
	137	/** Utility class for intra prediction-mode algorithm that uses a subset of modes.
	138	*/
	139	class Algo_TB_IntraPredMode_ModeSubset : public Algo_TB_IntraPredMode
	140	{
	141	public:
	142	Algo_TB_IntraPredMode_ModeSubset() {
	143	for (int i=0;i<35;i++) {
	144	mPredMode_enabled[i] = true;
	145	}
	146	}
	147
	148	void disableAllIntraPredModes() {
	149	for (int i=0;i<35;i++) {
	150	mPredMode_enabled[i] = false;
	151	}
	152	}
	153
	154	void enableIntraPredMode(int mode, bool flag=true) {
	155	mPredMode_enabled[mode] = flag;
	156	}
	157
	158	void enableIntraPredModeSubset(enum ALGO_TB_IntraPredMode_Subset subset) {
	159	switch (subset)
	160	{
	161	case ALGO_TB_IntraPredMode_Subset_All: // activate all is the default
	162	for (int i=0;i<35;i++) { enableIntraPredMode(i); }
	163	break;
	164	case ALGO_TB_IntraPredMode_Subset_DC:
	165	disableAllIntraPredModes();
	166	enableIntraPredMode(INTRA_DC);
	167	break;
	168	case ALGO_TB_IntraPredMode_Subset_Planar:
	169	disableAllIntraPredModes();
	170	enableIntraPredMode(INTRA_PLANAR);
	171	break;
	172	case ALGO_TB_IntraPredMode_Subset_HVPlus:
	173	disableAllIntraPredModes();
	174	enableIntraPredMode(INTRA_DC);
	175	enableIntraPredMode(INTRA_PLANAR);
	176	enableIntraPredMode(INTRA_ANGULAR_10);
	177	enableIntraPredMode(INTRA_ANGULAR_26);
	178	break;
	179	}
	180	}
	181
	182	protected:
	183	bool mPredMode_enabled[35];
	184	};
	185
	186
	187	/** Algorithm that brute-forces through all intra prediction mode.
	188	*/
	189	class Algo_TB_IntraPredMode_BruteForce : public Algo_TB_IntraPredMode_ModeSubset
	190	{
	191	public:
	192
	193	virtual enc_tb* analyze(encoder_context*,
	194	context_model_table&,
	195	const de265_image* input,
	196	const enc_tb* parent,
	197	enc_cb* cb,
	198	int x0,int y0, int xBase,int yBase, int log2TbSize,
	199	int blkIdx,
	200	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag);
	201	};
	202
	203
	204	/** Algorithm that makes a quick pre-selection of modes and then brute-forces through them.
	205	*/
	206	class Algo_TB_IntraPredMode_FastBrute : public Algo_TB_IntraPredMode_ModeSubset
	207	{
	208	public:
	209
	210	struct params
	211	{
	212	params() {
	213	keepNBest.set_ID("IntraPredMode-FastBrute-keepNBest");
	214	keepNBest.set_range(0,32);
	215	keepNBest.set_default(5);
	216
	217	bitrateEstimMethod.set_ID("IntraPredMode-FastBrute-estimator");
	218	}
	219
	220	option_TBBitrateEstimMethod bitrateEstimMethod;
	221	option_int keepNBest;
	222	};
	223
	224	void registerParams(config_parameters& config) {
	225	config.add_option(&mParams.keepNBest);
	226	config.add_option(&mParams.bitrateEstimMethod);
	227	}
	228
	229	void setParams(const params& p) { mParams=p; }
	230
	231
	232	virtual enc_tb* analyze(encoder_context*,
	233	context_model_table&,
	234	const de265_image* input,
	235	const enc_tb* parent,
	236	enc_cb* cb,
	237	int x0,int y0, int xBase,int yBase, int log2TbSize,
	238	int blkIdx,
	239	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag);
	240
	241	private:
	242	params mParams;
	243	};
	244
	245
	246	/** Algorithm that selects the intra prediction mode on minimum residual only.
	247	*/
	248	class Algo_TB_IntraPredMode_MinResidual : public Algo_TB_IntraPredMode_ModeSubset
	249	{
	250	public:
	251
	252	struct params
	253	{
	254	params() {
	255	bitrateEstimMethod.set_ID("IntraPredMode-MinResidual-estimator");
	256	}
	257
	258	option_TBBitrateEstimMethod bitrateEstimMethod;
	259	};
	260
	261	void setParams(const params& p) { mParams=p; }
	262
	263	void registerParams(config_parameters& config) {
	264	config.add_option(&mParams.bitrateEstimMethod);
	265	}
	266
	267	virtual enc_tb* analyze(encoder_context*,
	268	context_model_table&,
	269	const de265_image* input,
	270	const enc_tb* parent,
	271	enc_cb* cb,
	272	int x0,int y0, int xBase,int yBase, int log2TbSize,
	273	int blkIdx,
	274	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag);
	275
	276	private:
	277	params mParams;
	278	};
	279
	280	#endif

+525

-0

libde265/encoder/algo/tb-split.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/analyze.h"
	24	#include "libde265/encoder/encoder-context.h"
	25	#include <assert.h>
	26	#include <limits>
	27	#include <math.h>
	28	#include <iostream>
	29
	30
	31	#define ENCODER_DEVELOPMENT 1
	32
	33
	34	void diff_blk(int16_t* out,int out_stride,
	35	const uint8_t* a_ptr, int a_stride,
	36	const uint8_t* b_ptr, int b_stride,
	37	int blkSize)
	38	{
	39	for (int by=0;by<blkSize;by++)
	40	for (int bx=0;bx<blkSize;bx++)
	41	{
	42	out[byout_stride+bx] = a_ptr[bya_stride+bx] - b_ptr[by*b_stride+bx];
	43	}
	44	}
	45
	46
	47	static bool has_nonzero_value(const int16_t* data, int n)
	48	{
	49	for (int i=0;i<n;i++)
	50	if (data[i]) return true;
	51
	52	return false;
	53	}
	54
	55
	56	void show_debug_image(const de265_image* input, int slot);
	57
	58	/*
	59	void encode_transform_unit(encoder_context* ectx,
	60	enc_tb* tb,
	61	const de265_image* input,
	62	//int16_t* residual, int stride,
	63	int x0,int y0, // luma position
	64	int log2TbSize, // chroma adapted
	65	const enc_cb* cb,
	66	int cIdx)
	67	{
	68	}
	69	*/
	70
	71	void analyze_transform_unit(encoder_context* ectx,
	72	enc_tb* tb,
	73	const de265_image* input, // TODO: probably pass pixels/stride directly
	74	//int16_t* residual, int stride,
	75	int x0,int y0, // luma position
	76	int log2TbSize, // chroma adapted
	77	const enc_cb* cb,
	78	int cIdx)
	79	{
	80	int xC = x0;
	81	int yC = y0;
	82	int tbSize = 1<<log2TbSize;
	83	if (cIdx>0) { xC>>=1; yC>>=1; }
	84
	85	enum PredMode predMode = cb->PredMode;
	86
	87	int16_t blk[32*32]; // residual
	88
	89	// --- do intra prediction ---
	90
	91	if (predMode==MODE_INTRA) {
	92	enum IntraPredMode intraPredMode;
	93
	94	if (cIdx==0) {
	95	intraPredMode = ectx->img->get_IntraPredMode(x0,y0);
	96	}
	97	else {
	98	intraPredMode = cb->intra.chroma_mode;
	99	}
	100
	101	decode_intra_prediction(ectx->img, xC, yC, intraPredMode, tbSize , cIdx);
	102
	103
	104	// --- subtract prediction from prediction ---
	105
	106	uint8_t* pred = ectx->img->get_image_plane(cIdx);
	107	int stride = ectx->img->get_image_stride(cIdx);
	108
	109	diff_blk(blk,tbSize,
	110	input->get_image_plane_at_pos(cIdx,xC,yC), input->get_image_stride(cIdx),
	111	&pred[yC*stride+xC],stride, tbSize);
	112	}
	113	else {
	114	// --- subtract prediction from prediction ---
	115
	116	uint8_t* pred = ectx->prediction->get_image_plane(cIdx);
	117	int stride = ectx->prediction->get_image_stride(cIdx);
	118
	119	//printBlk("input",input->get_image_plane_at_pos(cIdx,xC,yC), tbSize, input->get_image_stride(cIdx));
	120	//printBlk("prediction", pred,tbSize, stride);
	121
	122	diff_blk(blk,tbSize,
	123	input->get_image_plane_at_pos(cIdx,xC,yC), input->get_image_stride(cIdx),
	124	&pred[yC*stride+xC],stride, tbSize);
	125
	126	//printBlk("residual", blk,tbSize,tbSize);
	127	}
	128
	129
	130	//show_debug_image(ectx->img, 0);
	131
	132
	133
	134
	135	// --- forward transform ---
	136
	137	tb->alloc_coeff_memory(cIdx, tbSize);
	138
	139	int trType;
	140	if (cIdx==0 && log2TbSize==2 && predMode==MODE_INTRA) trType=1; // TODO: inter mode
	141	else trType=0;
	142
	143	fwd_transform(&ectx->acceleration, tb->coeff[cIdx], tbSize, log2TbSize, trType, blk, tbSize);
	144
	145
	146	// --- quantization ---
	147
	148	quant_coefficients(tb->coeff[cIdx], tb->coeff[cIdx], log2TbSize, cb->qp, true);
	149
	150	tb->cbf[cIdx] = has_nonzero_value(tb->coeff[cIdx], 1<<(log2TbSize<<1));
	151	}
	152
	153
	154	static void recursive_cbfChroma(CABAC_encoder_estim* cabac,
	155	enc_tb* tb, int log2TrafoSize, int trafoDepth)
	156	{
	157	float bits_pre = cabac->getRDBits();
	158
	159	// --- CBF CB/CR ---
	160
	161	// For 4x4 luma, there is no signaling of chroma CBF, because only the
	162	// chroma CBF for 8x8 is relevant.
	163	if (log2TrafoSize>2) {
	164	if (trafoDepth==0 \|\| tb->parent->cbf[1]) {
	165	encode_cbf_chroma(cabac, trafoDepth, tb->cbf[1]);
	166	}
	167	if (trafoDepth==0 \|\| tb->parent->cbf[2]) {
	168	encode_cbf_chroma(cabac, trafoDepth, tb->cbf[2]);
	169	}
	170	}
	171
	172	if (tb->split_transform_flag) {
	173	for (int i=0;i<4;i++) {
	174	recursive_cbfChroma(cabac, tb->children[i], log2TrafoSize-1, trafoDepth+1);
	175	}
	176	}
	177
	178	float bits_post = cabac->getRDBits();
	179
	180	tb->rate = tb->rate_withoutCbfChroma + (bits_post - bits_pre);
	181	}
	182
	183
	184	enc_tb* encode_transform_tree_no_split(encoder_context* ectx,
	185	context_model_table& ctxModel,
	186	const de265_image* input,
	187	const enc_tb* parent,
	188	enc_cb* cb,
	189	int x0,int y0, int xBase,int yBase, int log2TbSize,
	190	int blkIdx,
	191	int trafoDepth, int MaxTrafoDepth, int IntraSplitFlag)
	192	{
	193	de265_image* img = ectx->img;
	194
	195	int stride = ectx->img->get_image_stride(0);
	196
	197	uint8_t* luma_plane = ectx->img->get_image_plane(0);
	198	uint8_t* cb_plane = ectx->img->get_image_plane(1);
	199	uint8_t* cr_plane = ectx->img->get_image_plane(2);
	200
	201	// --- compute transform coefficients ---
	202
	203	enc_tb* tb = new enc_tb();
	204
	205	tb->parent = parent;
	206	tb->split_transform_flag = false;
	207	tb->log2Size = log2TbSize;
	208	tb->x = x0;
	209	tb->y = y0;
	210	tb->cbf[0] = tb->cbf[1] = tb->cbf[2] = 0;
	211
	212
	213	// luma block
	214
	215	analyze_transform_unit(ectx, tb, input, x0,y0, log2TbSize, cb, 0 /* Y */);
	216
	217
	218	// chroma blocks
	219
	220	if (log2TbSize > 2) {
	221	// if TB is > 4x4, do chroma transform of half size
	222	analyze_transform_unit(ectx, tb, input, x0,y0, log2TbSize-1, cb, 1 /* Cb */);
	223	analyze_transform_unit(ectx, tb, input, x0,y0, log2TbSize-1, cb, 2 /* Cr */);
	224	}
	225	else if (blkIdx==3) {
	226	// if TB size is 4x4, do chroma transform for last sub-block
	227	analyze_transform_unit(ectx, tb, input, xBase,yBase, log2TbSize, cb, 1 /* Cb */);
	228	analyze_transform_unit(ectx, tb, input, xBase,yBase, log2TbSize, cb, 2 /* Cr */);
	229	}
	230
	231
	232	// reconstruction
	233
	234	tb->reconstruct(ectx, ectx->img, cb, blkIdx);
	235
	236
	237
	238	// measure rate
	239
	240	CABAC_encoder_estim estim;
	241	estim.set_context_models(&ctxModel);
	242
	243
	244	tb->rate_withoutCbfChroma = 0;
	245
	246	const seq_parameter_set* sps = &ectx->img->sps;
	247
	248
	249	if (log2TbSize <= sps->Log2MaxTrafoSize &&
	250	log2TbSize > sps->Log2MinTrafoSize &&
	251	trafoDepth < MaxTrafoDepth &&
	252	!(IntraSplitFlag && trafoDepth==0))
	253	{
	254	encode_split_transform_flag(ectx, &estim, log2TbSize, 0);
	255	tb->rate_withoutCbfChroma += estim.getRDBits();
	256	estim.reset();
	257	}
	258
	259	// --- CBF CB/CR ---
	260
	261	if (cb->PredMode == MODE_INTRA \|\| trafoDepth != 0 \|\|
	262	tb->cbf[1] \|\| tb->cbf[2]) {
	263	encode_cbf_luma(&estim, trafoDepth==0, tb->cbf[0]);
	264	}
	265
	266	encode_transform_unit(ectx,&estim, tb,cb, x0,y0, xBase,yBase, log2TbSize, trafoDepth, blkIdx);
	267
	268	tb->rate_withoutCbfChroma += estim.getRDBits();
	269
	270	estim.reset(); // TODO: not needed ?
	271
	272	recursive_cbfChroma(&estim,tb,log2TbSize,trafoDepth);
	273
	274	//float rate_cbfChroma = estim.getRDBits();
	275	//tb->rate = tb->rate_withoutCbfChroma + rate_cbfChroma;
	276
	277
	278	// measure distortion
	279
	280	int tbSize = 1<<log2TbSize;
	281	tb->distortion = SSD(input->get_image_plane_at_pos(0, x0,y0), input->get_image_stride(0),
	282	img ->get_image_plane_at_pos(0, x0,y0), img ->get_image_stride(0),
	283	tbSize, tbSize);
	284
	285	return tb;
	286	}
	287
	288
	289	enc_tb* Algo_TB_Split::encode_transform_tree_split(encoder_context* ectx,
	290	context_model_table& ctxModel,
	291	const de265_image* input,
	292	const enc_tb* parent,
	293	enc_cb* cb,
	294	int x0,int y0, int log2TbSize,
	295	int TrafoDepth, int MaxTrafoDepth,
	296	int IntraSplitFlag)
	297	{
	298	const de265_image* img = ectx->img;
	299
	300	enc_tb* tb = new enc_tb();
	301
	302	tb->parent = parent;
	303	tb->split_transform_flag = true;
	304	tb->log2Size = log2TbSize;
	305	tb->x = x0;
	306	tb->y = y0;
	307
	308	tb->distortion = 0;
	309	tb->rate = 0;
	310	tb->rate_withoutCbfChroma = 0;
	311
	312
	313	// Since we try to code all sub-blocks, we enable all CBF flags.
	314	// Should we see later that the child TBs are zero, we clear those flags later.
	315
	316	tb->cbf[0]=1;
	317	tb->cbf[1]=1;
	318	tb->cbf[2]=1;
	319
	320
	321	context_model ctxModelCbfChroma[4];
	322	for (int i=0;i<4;i++) {
	323	ctxModelCbfChroma[i] = ctxModel[CONTEXT_MODEL_CBF_CHROMA+i];
	324	}
	325
	326
	327	// --- encode all child nodes ---
	328
	329	for (int i=0;i<4;i++) {
	330	int dx = (i&1) << (log2TbSize-1);
	331	int dy = (i>>1) << (log2TbSize-1);
	332
	333	if (cb->PredMode == MODE_INTRA) {
	334	tb->children[i] = mAlgo_TB_IntraPredMode->analyze(ectx, ctxModel, input, tb, cb,
	335	x0+dx, y0+dy, x0,y0,
	336	log2TbSize-1, i,
	337	TrafoDepth+1, MaxTrafoDepth, IntraSplitFlag);
	338	}
	339	else {
	340	tb->children[i] = this->analyze(ectx, ctxModel, input, tb, cb,
	341	x0+dx, y0+dy, x0,y0,
	342	log2TbSize-1, i,
	343	TrafoDepth+1, MaxTrafoDepth, IntraSplitFlag);
	344	}
	345
	346	tb->distortion += tb->children[i]->distortion;
	347	tb->rate_withoutCbfChroma += tb->children[i]->rate_withoutCbfChroma;
	348	}
	349
	350	tb->set_cbf_flags_from_children();
	351
	352
	353	// --- add rate for this TB level ---
	354
	355	CABAC_encoder_estim estim;
	356	estim.set_context_models(&ctxModel);
	357
	358
	359
	360
	361	const seq_parameter_set* sps = &ectx->img->sps;
	362
	363	if (log2TbSize <= sps->Log2MaxTrafoSize &&
	364	log2TbSize > sps->Log2MinTrafoSize &&
	365	TrafoDepth < MaxTrafoDepth &&
	366	!(IntraSplitFlag && TrafoDepth==0))
	367	{
	368	encode_split_transform_flag(ectx, &estim, log2TbSize, 1);
	369	tb->rate_withoutCbfChroma += estim.getRDBits();
	370	estim.reset();
	371	}
	372
	373	// restore chroma CBF context models
	374
	375	for (int i=0;i<4;i++) {
	376	ctxModel[CONTEXT_MODEL_CBF_CHROMA+i] = ctxModelCbfChroma[i];
	377	}
	378
	379	recursive_cbfChroma(&estim,tb, log2TbSize, TrafoDepth);
	380	//tb->rate = tb->rate_withoutCbfChroma + estim.getRDBits();
	381
	382	return tb;
	383	}
	384
	385
	386
	387	struct Logging_TB_Split : public Logging
	388	{
	389	int skipTBSplit, noskipTBSplit;
	390	int zeroBlockCorrelation[6][2][5];
	391
	392	const char* name() const { return "tb-split"; }
	393
	394	void print(const encoder_context* ectx, const char* filename)
	395	{
	396	printf("%d %d\n\n",skipTBSplit, noskipTBSplit);
	397
	398	for (int tb=3;tb<=5;tb++) {
	399	for (int z=0;z<=1;z++) {
	400	float total = 0;
	401
	402	for (int c=0;c<5;c++)
	403	total += zeroBlockCorrelation[tb][z][c];
	404
	405	for (int c=0;c<5;c++) {
	406	printf("%d %d %d : %d %5.2f\n", tb,z,c,
	407	zeroBlockCorrelation[tb][z][c],
	408	total==0 ? 0 : zeroBlockCorrelation[tb][z][c]/total*100);
	409	}
	410	}
	411	}
	412
	413
	414	for (int z=0;z<2;z++) {
	415	printf("\n");
	416	for (int tb=3;tb<=5;tb++) {
	417	float total = 0;
	418
	419	for (int c=0;c<5;c++)
	420	total += zeroBlockCorrelation[tb][z][c];
	421
	422	printf("%dx%d ",1<<tb,1<<tb);
	423
	424	for (int c=0;c<5;c++) {
	425	printf("%5.2f ", total==0 ? 0 : zeroBlockCorrelation[tb][z][c]/total*100);
	426	}
	427	printf("\n");
	428	}
	429	}
	430	}
	431	} logging_tb_split;
	432
	433
	434
	435	enc_tb*
	436	Algo_TB_Split_BruteForce::analyze(encoder_context* ectx,
	437	context_model_table& ctxModel,
	438	const de265_image* input,
	439	const enc_tb* parent,
	440	enc_cb* cb,
	441	int x0,int y0, int xBase,int yBase, int log2TbSize,
	442	int blkIdx,
	443	int TrafoDepth, int MaxTrafoDepth,
	444	int IntraSplitFlag)
	445	{
	446	bool test_split = (log2TbSize > 2 &&
	447	TrafoDepth < MaxTrafoDepth &&
	448	log2TbSize > ectx->sps.Log2MinTrafoSize);
	449
	450	bool test_no_split = true;
	451	if (IntraSplitFlag && TrafoDepth==0) test_no_split=false; // we have to split
	452	if (log2TbSize > ectx->sps.Log2MaxTrafoSize) test_no_split=false;
	453
	454	context_model_table ctxSplit;
	455	if (test_split) {
	456	ctxSplit = ctxModel.copy();
	457	}
	458
	459
	460	enc_tb* tb_no_split = NULL;
	461	enc_tb* tb_split = NULL;
	462	float rd_cost_no_split = std::numeric_limits<float>::max();
	463	float rd_cost_split = std::numeric_limits<float>::max();
	464
	465	if (test_no_split) {
	466	tb_no_split = encode_transform_tree_no_split(ectx, ctxModel, input, parent,
	467	cb, x0,y0, xBase,yBase, log2TbSize,
	468	blkIdx,
	469	TrafoDepth,MaxTrafoDepth,IntraSplitFlag);
	470
	471	rd_cost_no_split = tb_no_split->distortion + ectx->lambda * tb_no_split->rate;
	472
	473	if (log2TbSize <= mParams.zeroBlockPrune()) {
	474	bool zeroBlock = tb_no_split->isZeroBlock();
	475
	476	if (zeroBlock) {
	477	test_split = false;
	478	logging_tb_split.skipTBSplit++;
	479	}
	480	else
	481	logging_tb_split.noskipTBSplit++;
	482	}
	483	}
	484
	485
	486	if (test_split) {
	487	tb_split = encode_transform_tree_split(ectx, ctxSplit, input, parent, cb,
	488	x0,y0, log2TbSize,
	489	TrafoDepth, MaxTrafoDepth, IntraSplitFlag);
	490
	491	rd_cost_split = tb_split->distortion + ectx->lambda * tb_split->rate;
	492	}
	493
	494
	495	if (test_split && test_no_split) {
	496	bool zero_block = tb_no_split->isZeroBlock();
	497
	498	int nChildZero = 0;
	499	for (int i=0;i<4;i++) {
	500	if (tb_split->children[i]->isZeroBlock()) nChildZero++;
	501	}
	502
	503	logging_tb_split.zeroBlockCorrelation[log2TbSize][zero_block ? 0 : 1][nChildZero]++;
	504	}
	505
	506
	507	bool split = (rd_cost_split < rd_cost_no_split);
	508
	509	if (split) {
	510	ctxModel = ctxSplit;
	511
	512	delete tb_no_split;
	513	assert(tb_split);
	514	return tb_split;
	515	}
	516	else {
	517	delete tb_split;
	518	assert(tb_no_split);
	519	tb_no_split->reconstruct(ectx, ectx->img,
	520	cb, blkIdx);
	521
	522	return tb_no_split;
	523	}
	524	}

+152

-0

libde265/encoder/algo/tb-split.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef TB_SPLIT_H
	23	#define TB_SPLIT_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/tb-intrapredmode.h"
	38
	39
	40	/* Encoder search tree, bottom up:
	41
	42	- Algo_TB_Split - whether TB is split or not
	43
	44	- Algo_TB_IntraPredMode - choose the intra prediction mode (or NOP, if at the wrong tree level)
	45
	46	- Algo_CB_IntraPartMode - choose between NxN and 2Nx2N intra parts
	47
	48	- Algo_CB_Split - whether CB is split or not
	49
	50	- Algo_CTB_QScale - select QScale on CTB granularity
	51	*/
	52
	53	/*
	54	struct ResidualBlock
	55	{
	56	const int16_t* data[4];
	57	int stride[4];
	58	};
	59	*/
	60
	61	void diff_blk(int16_t* out,int out_stride,
	62	const uint8_t* a_ptr, int a_stride,
	63	const uint8_t* b_ptr, int b_stride,
	64	int blkSize);
	65
	66
	67	// ========== TB split decision ==========
	68
	69	class Algo_TB_Split
	70	{
	71	public:
	72	Algo_TB_Split() : mAlgo_TB_IntraPredMode(NULL) { }
	73	virtual ~Algo_TB_Split() { }
	74
	75	virtual enc_tb* analyze(encoder_context*,
	76	context_model_table&,
	77	const de265_image* input,
	78	const enc_tb* parent,
	79	enc_cb* cb,
	80	int x0,int y0, int xBase,int yBase, int log2TbSize,
	81	int blkIdx,
	82	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag) = 0;
	83
	84	void setAlgo_TB_IntraPredMode(Algo_TB_IntraPredMode* algo) { mAlgo_TB_IntraPredMode=algo; }
	85
	86	protected:
	87	enc_tb* encode_transform_tree_split(encoder_context* ectx,
	88	context_model_table& ctxModel,
	89	const de265_image* input,
	90	const enc_tb* parent,
	91	enc_cb* cb,
	92	int x0,int y0, int log2TbSize,
	93	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag);
	94
	95	Algo_TB_IntraPredMode* mAlgo_TB_IntraPredMode;
	96	};
	97
	98
	99
	100	enum ALGO_TB_Split_BruteForce_ZeroBlockPrune {
	101	// numeric value specifies the maximum size for log2Tb for which the pruning is applied
	102	ALGO_TB_BruteForce_ZeroBlockPrune_off = 0,
	103	ALGO_TB_BruteForce_ZeroBlockPrune_8x8 = 3,
	104	ALGO_TB_BruteForce_ZeroBlockPrune_8x8_16x16 = 4,
	105	ALGO_TB_BruteForce_ZeroBlockPrune_all = 5
	106	};
	107
	108	class option_ALGO_TB_Split_BruteForce_ZeroBlockPrune
	109	: public choice_option<enum ALGO_TB_Split_BruteForce_ZeroBlockPrune>
	110	{
	111	public:
	112	option_ALGO_TB_Split_BruteForce_ZeroBlockPrune() {
	113	add_choice("off" ,ALGO_TB_BruteForce_ZeroBlockPrune_off);
	114	add_choice("8x8" ,ALGO_TB_BruteForce_ZeroBlockPrune_8x8);
	115	add_choice("8-16" ,ALGO_TB_BruteForce_ZeroBlockPrune_8x8_16x16);
	116	add_choice("all" ,ALGO_TB_BruteForce_ZeroBlockPrune_all, true);
	117	}
	118	};
	119
	120	class Algo_TB_Split_BruteForce : public Algo_TB_Split
	121	{
	122	public:
	123	struct params
	124	{
	125	params() {
	126	zeroBlockPrune.set_ID("TB-Split-BruteForce-ZeroBlockPrune");
	127	}
	128
	129	option_ALGO_TB_Split_BruteForce_ZeroBlockPrune zeroBlockPrune;
	130	};
	131
	132	void setParams(const params& p) { mParams=p; }
	133
	134	void registerParams(config_parameters& config) {
	135	config.add_option(&mParams.zeroBlockPrune);
	136	}
	137
	138	virtual enc_tb* analyze(encoder_context*,
	139	context_model_table&,
	140	const de265_image* input,
	141	const enc_tb* parent,
	142	enc_cb* cb,
	143	int x0,int y0, int xBase,int yBase, int log2TbSize,
	144	int blkIdx,
	145	int TrafoDepth, int MaxTrafoDepth, int IntraSplitFlag);
	146
	147	private:
	148	params mParams;
	149	};
	150
	151	#endif

+371

-0

libde265/encoder/analyze.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22
	23	#include "libde265/encoder/analyze.h"
	24	#include "libde265/encoder/encoder-context.h"
	25	#include <assert.h>
	26	#include <limits>
	27	#include <math.h>
	28	#include <iostream>
	29
	30
	31	#define ENCODER_DEVELOPMENT 0
	32	#define COMPARE_ESTIMATED_RATE_TO_REAL_RATE 0
	33
	34
	35	static int IntraPredModeCnt[7][35];
	36	static int MPM_used[7][35];
	37
	38	static int IntraPredModeCnt_total[35];
	39	static int MPM_used_total[35];
	40
	41	void statistics_IntraPredMode(const encoder_context* ectx, int x,int y, const enc_cb* cb)
	42	{
	43	if (cb->split_cu_flag) {
	44	for (int i=0;i<4;i++)
	45	if (cb->children[i]) {
	46	statistics_IntraPredMode(ectx, childX(x,i,cb->log2Size), childY(y,i,cb->log2Size), cb->children[i]);
	47	}
	48	}
	49	else {
	50	int cnt;
	51	int size = cb->log2Size;
	52
	53	if (cb->PartMode == PART_NxN) { cnt=4; size--; } else cnt=1;
	54
	55	for (int i=0;i<cnt;i++) {
	56	IntraPredModeCnt[size][ cb->intra.pred_mode[i] ]++;
	57	IntraPredModeCnt_total[ cb->intra.pred_mode[i] ]++;
	58
	59	int xi = childX(x,i,cb->log2Size);
	60	int yi = childY(y,i,cb->log2Size);
	61
	62	int candModeList[3];
	63	fillIntraPredModeCandidates(candModeList,xi,yi, xi>0, yi>0, ectx->img);
	64
	65	int predmode = cb->intra.pred_mode[i];
	66	if (candModeList[0]==predmode \|\|
	67	candModeList[1]==predmode \|\|
	68	candModeList[2]==predmode) {
	69	MPM_used[size][predmode]++;
	70	MPM_used_total[predmode]++;
	71	}
	72	}
	73	}
	74	}
	75
	76	void statistics_print()
	77	{
	78	for (int i=0;i<35;i++) {
	79	printf("%d",i);
	80	printf(" %d %d",IntraPredModeCnt_total[i], MPM_used_total[i]);
	81
	82	for (int k=2;k<=6;k++) {
	83	printf(" %d %d",IntraPredModeCnt[k][i], MPM_used[k][i]);
	84	}
	85
	86	printf("\n");
	87	}
	88	}
	89
	90
	91	void print_tb_tree_rates(const enc_tb* tb, int level)
	92	{
	93	for (int i=0;i<level;i++)
	94	std::cout << " ";
	95
	96	std::cout << "TB rate=" << tb->rate << " (" << tb->rate_withoutCbfChroma << ")\n";
	97	if (tb->split_transform_flag) {
	98	for (int i=0;i<4;i++)
	99	print_tb_tree_rates(tb->children[i], level+1);
	100	}
	101	}
	102
	103
	104	void print_cb_tree_rates(const enc_cb* cb, int level)
	105	{
	106	for (int i=0;i<level;i++)
	107	std::cout << " ";
	108
	109	std::cout << "CB rate=" << cb->rate << "\n";
	110	if (cb->split_cu_flag) {
	111	for (int i=0;i<4;i++)
	112	print_cb_tree_rates(cb->children[i], level+1);
	113	}
	114	else {
	115	print_tb_tree_rates(cb->transform_tree, level+1);
	116	}
	117	}
	118
	119
	120	double encode_image(encoder_context* ectx,
	121	const de265_image* input,
	122	EncodingAlgorithm& algo)
	123	{
	124	int stride=input->get_image_stride(0);
	125
	126	int w = ectx->sps.pic_width_in_luma_samples;
	127	int h = ectx->sps.pic_height_in_luma_samples;
	128
	129	// --- create reconstruction image ---
	130	ectx->img = new de265_image;
	131	ectx->img->vps = ectx->vps;
	132	ectx->img->sps = ectx->sps;
	133	ectx->img->pps = ectx->pps;
	134	ectx->img->PicOrderCntVal = input->PicOrderCntVal;
	135
	136	ectx->img->alloc_image(w,h, de265_chroma_420, &ectx->sps, true,
	137	NULL /* no decctx */, ectx, 0,NULL,false);
	138	//ectx->img->alloc_encoder_data(&ectx->sps);
	139	ectx->img->clear_metadata();
	140
	141	#if 1
	142	if (1) {
	143	ectx->prediction = new de265_image;
	144	ectx->prediction->alloc_image(w,h, de265_chroma_420, &ectx->sps, false /* no metadata */,
	145	NULL /* no decctx /, NULL / no encctx */, 0,NULL,false);
	146	ectx->prediction->vps = ectx->vps;
	147	ectx->prediction->sps = ectx->sps;
	148	ectx->prediction->pps = ectx->pps;
	149	}
	150	#endif
	151
	152	ectx->active_qp = ectx->pps.pic_init_qp; // TODO take current qp from slice
	153
	154
	155	ectx->cabac_ctx_models.init(ectx->shdr->initType, ectx->shdr->SliceQPY);
	156	ectx->cabac_encoder.set_context_models(&ectx->cabac_ctx_models);
	157
	158
	159	context_model_table modelEstim;
	160	CABAC_encoder_estim cabacEstim;
	161
	162	modelEstim.init(ectx->shdr->initType, ectx->shdr->SliceQPY);
	163	cabacEstim.set_context_models(&modelEstim);
	164
	165
	166	int Log2CtbSize = ectx->sps.Log2CtbSizeY;
	167
	168	uint8_t* luma_plane = ectx->img->get_image_plane(0);
	169	uint8_t* cb_plane = ectx->img->get_image_plane(1);
	170	uint8_t* cr_plane = ectx->img->get_image_plane(2);
	171
	172
	173	// encode CTB by CTB
	174
	175	for (int y=0;y<ectx->sps.PicHeightInCtbsY;y++)
	176	for (int x=0;x<ectx->sps.PicWidthInCtbsY;x++)
	177	{
	178	ectx->img->set_SliceAddrRS(x, y, ectx->shdr->SliceAddrRS);
	179
	180	int x0 = x<<Log2CtbSize;
	181	int y0 = y<<Log2CtbSize;
	182
	183	logtrace(LogSlice,"encode CTB at %d %d\n",x0,y0);
	184
	185	// make a copy of the context model that we can modify for testing alternatives
	186
	187	context_model_table ctxModel;
	188	//copy_context_model_table(ctxModel, ectx->ctx_model_bitstream);
	189	ctxModel = ectx->cabac_ctx_models.copy();
	190	ctxModel = modelEstim.copy(); // TODO TMP
	191
	192	disable_logging(LogSymbols);
	193	enable_logging(LogSymbols); // TODO TMP
	194
	195	//printf("================================================== ANALYZE\n");
	196
	197	#if 1
	198	/*
	199	enc_cb* cb = encode_cb_may_split(ectx, ctxModel,
	200	input, x0,y0, Log2CtbSize, 0, qp);
	201	*/
	202
	203	enc_cb* cb = algo.getAlgoCTBQScale()->analyze(ectx,ctxModel, x0,y0);
	204	#else
	205	float minCost = std::numeric_limits<float>::max();
	206	int bestQ = 0;
	207	int qp = ectx->params.constant_QP;
	208
	209	enc_cb* cb;
	210	for (int q=1;q<51;q++) {
	211	copy_context_model_table(ctxModel, ectx->ctx_model_bitstream);
	212
	213	enc_cb* cbq = encode_cb_may_split(ectx, ctxModel,
	214	input, x0,y0, Log2CtbSize, 0, q);
	215
	216	float cost = cbq->distortion + ectx->lambda * cbq->rate;
	217	if (cost<minCost) { minCost=cost; bestQ=q; }
	218
	219	if (q==qp) { cb=cbq; }
	220	}
	221
	222	printf("Q %d\n",bestQ);
	223	fflush(stdout);
	224	#endif
	225
	226	//print_cb_tree_rates(cb,0);
	227
	228	//statistics_IntraPredMode(ectx, x0,y0, cb);
	229
	230
	231	// --- write bitstream ---
	232
	233	//ectx->switch_CABAC_to_bitstream();
	234
	235	enable_logging(LogSymbols);
	236
	237	encode_ctb(ectx, &ectx->cabac_encoder, cb, x,y);
	238
	239	//printf("================================================== WRITE\n");
	240
	241
	242	if (COMPARE_ESTIMATED_RATE_TO_REAL_RATE) {
	243	float realPre = cabacEstim.getRDBits();
	244	encode_ctb(ectx, &cabacEstim, cb, x,y);
	245	float realPost = cabacEstim.getRDBits();
	246
	247	printf("estim: %f real: %f diff: %f\n",
	248	cb->rate,
	249	realPost-realPre,
	250	cb->rate - (realPost-realPre));
	251	}
	252
	253
	254	int last = (y==ectx->sps.PicHeightInCtbsY-1 &&
	255	x==ectx->sps.PicWidthInCtbsY-1);
	256	ectx->cabac_encoder.write_CABAC_term_bit(last);
	257
	258
	259	delete cb;
	260
	261	//ectx->free_all_pools();
	262	}
	263
	264
	265	//statistics_print();
	266
	267
	268	delete ectx->prediction;
	269
	270
	271	// frame PSNR
	272
	273	double psnr = PSNR(MSE(input->get_image_plane(0), input->get_image_stride(0),
	274	luma_plane, ectx->img->get_image_stride(0),
	275	input->get_width(), input->get_height()));
	276	return psnr;
	277	}
	278
	279
	280
	281	void EncodingAlgorithm_Custom::setParams(encoder_params& params)
	282	{
	283	// build algorithm tree
	284
	285	mAlgo_CTB_QScale_Constant.setChildAlgo(&mAlgo_CB_Split_BruteForce);
	286	mAlgo_CB_Split_BruteForce.setChildAlgo(&mAlgo_CB_Skip_BruteForce);
	287
	288	mAlgo_CB_Skip_BruteForce.setSkipAlgo(&mAlgo_CB_MergeIndex_Fixed);
	289	mAlgo_CB_Skip_BruteForce.setNonSkipAlgo(&mAlgo_CB_IntraInter_BruteForce);
	290	//&mAlgo_CB_InterPartMode_Fixed);
	291
	292	Algo_CB_IntraPartMode* algo_CB_IntraPartMode = NULL;
	293	switch (params.mAlgo_CB_IntraPartMode()) {
	294	case ALGO_CB_IntraPartMode_BruteForce:
	295	algo_CB_IntraPartMode = &mAlgo_CB_IntraPartMode_BruteForce;
	296	break;
	297	case ALGO_CB_IntraPartMode_Fixed:
	298	algo_CB_IntraPartMode = &mAlgo_CB_IntraPartMode_Fixed;
	299	break;
	300	}
	301
	302	mAlgo_CB_IntraInter_BruteForce.setIntraChildAlgo(algo_CB_IntraPartMode);
	303	mAlgo_CB_IntraInter_BruteForce.setInterChildAlgo(&mAlgo_CB_InterPartMode_Fixed);
	304
	305	mAlgo_CB_MergeIndex_Fixed.setChildAlgo(&mAlgo_TB_Split_BruteForce);
	306
	307	Algo_PB_MV* pbAlgo = NULL;
	308	switch (params.mAlgo_MEMode()) {
	309	case MEMode_Test:
	310	pbAlgo = &mAlgo_PB_MV_Test;
	311	break;
	312	case MEMode_Search:
	313	pbAlgo = &mAlgo_PB_MV_Search;
	314	break;
	315	}
	316
	317	mAlgo_CB_InterPartMode_Fixed.setChildAlgo(pbAlgo);
	318	pbAlgo->setChildAlgo(&mAlgo_TB_Split_BruteForce);
	319
	320
	321	Algo_TB_IntraPredMode_ModeSubset* algo_TB_IntraPredMode = NULL;
	322	switch (params.mAlgo_TB_IntraPredMode()) {
	323	case ALGO_TB_IntraPredMode_BruteForce:
	324	algo_TB_IntraPredMode = &mAlgo_TB_IntraPredMode_BruteForce;
	325	break;
	326	case ALGO_TB_IntraPredMode_FastBrute:
	327	algo_TB_IntraPredMode = &mAlgo_TB_IntraPredMode_FastBrute;
	328	break;
	329	case ALGO_TB_IntraPredMode_MinResidual:
	330	algo_TB_IntraPredMode = &mAlgo_TB_IntraPredMode_MinResidual;
	331	break;
	332	}
	333
	334	algo_CB_IntraPartMode->setChildAlgo(algo_TB_IntraPredMode);
	335
	336	mAlgo_TB_Split_BruteForce.setAlgo_TB_IntraPredMode(algo_TB_IntraPredMode);
	337	//mAlgo_TB_Split_BruteForce.setParams(params.TB_Split_BruteForce);
	338
	339	algo_TB_IntraPredMode->setChildAlgo(&mAlgo_TB_Split_BruteForce);
	340
	341
	342	// ===== set algorithm parameters ======
	343
	344	//mAlgo_CB_IntraPartMode_Fixed.setParams(params.CB_IntraPartMode_Fixed);
	345
	346	//mAlgo_TB_IntraPredMode_FastBrute.setParams(params.TB_IntraPredMode_FastBrute);
	347	//mAlgo_TB_IntraPredMode_MinResidual.setParams(params.TB_IntraPredMode_MinResidual);
	348
	349
	350	//mAlgo_CTB_QScale_Constant.setParams(params.CTB_QScale_Constant);
	351
	352
	353	algo_TB_IntraPredMode->enableIntraPredModeSubset( params.mAlgo_TB_IntraPredMode_Subset() );
	354	}
	355
	356
	357	void Logging::print_logging(const encoder_context* ectx, const char* id, const char* filename)
	358	{
	359	#if 000
	360	if (strcmp(id,logging_tb_split.name())==0) {
	361	logging_tb_split.print(ectx,filename);
	362	}
	363	#endif
	364	}
	365
	366
	367	void en265_print_logging(const encoder_context* ectx, const char* id, const char* filename)
	368	{
	369	Logging::print_logging(ectx,id,filename);
	370	}

+146

-0

libde265/encoder/analyze.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef ANALYZE_H
	23	#define ANALYZE_H
	24
	25	#include "libde265/nal-parser.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/encoder/encode.h"
	28	#include "libde265/slice.h"
	29	#include "libde265/scan.h"
	30	#include "libde265/intrapred.h"
	31	#include "libde265/transform.h"
	32	#include "libde265/fallback-dct.h"
	33	#include "libde265/quality.h"
	34	#include "libde265/fallback.h"
	35	#include "libde265/configparam.h"
	36
	37	#include "libde265/encoder/algo/tb-intrapredmode.h"
	38	#include "libde265/encoder/algo/tb-split.h"
	39	#include "libde265/encoder/algo/cb-intrapartmode.h"
	40	#include "libde265/encoder/algo/cb-interpartmode.h"
	41	#include "libde265/encoder/algo/cb-split.h"
	42	#include "libde265/encoder/algo/ctb-qscale.h"
	43	#include "libde265/encoder/algo/cb-mergeindex.h"
	44	//#include "libde265/encoder/algo/cb-skip-or-inter.h"
	45	#include "libde265/encoder/algo/pb-mv.h"
	46	#include "libde265/encoder/algo/cb-skip.h"
	47	#include "libde265/encoder/algo/cb-intra-inter.h"
	48
	49
	50	/* Encoder search tree, bottom up:
	51
	52	- Algo_TB_Split - whether TB is split or not
	53
	54	- Algo_TB_IntraPredMode - choose the intra prediction mode (or NOP, if at the wrong tree level)
	55
	56	- Algo_CB_IntraPartMode - choose between NxN and 2Nx2N intra parts
	57
	58	- Algo_CB_PredMode - intra / inter
	59
	60	- Algo_CB_Split - whether CB is split or not
	61
	62	- Algo_CTB_QScale - select QScale on CTB granularity
	63	*/
	64
	65
	66	// ========== an encoding algorithm combines a set of algorithm modules ==========
	67
	68	class EncodingAlgorithm
	69	{
	70	public:
	71	virtual ~EncodingAlgorithm() { }
	72
	73	virtual Algo_CTB_QScale* getAlgoCTBQScale() = 0;
	74
	75	virtual int getPPS_QP() const = 0;
	76	virtual int getSlice_QPDelta() const { return 0; }
	77	};
	78
	79
	80	class EncodingAlgorithm_Custom : public EncodingAlgorithm
	81	{
	82	public:
	83
	84	void setParams(struct encoder_params& params);
	85
	86	void registerParams(config_parameters& config) {
	87	mAlgo_CTB_QScale_Constant.registerParams(config);
	88	mAlgo_CB_IntraPartMode_Fixed.registerParams(config);
	89	mAlgo_CB_InterPartMode_Fixed.registerParams(config);
	90	mAlgo_PB_MV_Test.registerParams(config);
	91	mAlgo_PB_MV_Search.registerParams(config);
	92	mAlgo_TB_IntraPredMode_FastBrute.registerParams(config);
	93	mAlgo_TB_IntraPredMode_MinResidual.registerParams(config);
	94	mAlgo_TB_Split_BruteForce.registerParams(config);
	95	}
	96
	97	virtual Algo_CTB_QScale* getAlgoCTBQScale() { return &mAlgo_CTB_QScale_Constant; }
	98
	99	virtual int getPPS_QP() const { return mAlgo_CTB_QScale_Constant.getQP(); }
	100
	101	private:
	102	Algo_CTB_QScale_Constant mAlgo_CTB_QScale_Constant;
	103
	104	Algo_CB_Split_BruteForce mAlgo_CB_Split_BruteForce;
	105	Algo_CB_Skip_BruteForce mAlgo_CB_Skip_BruteForce;
	106	Algo_CB_IntraInter_BruteForce mAlgo_CB_IntraInter_BruteForce;
	107
	108	Algo_CB_IntraPartMode_BruteForce mAlgo_CB_IntraPartMode_BruteForce;
	109	Algo_CB_IntraPartMode_Fixed mAlgo_CB_IntraPartMode_Fixed;
	110
	111	Algo_CB_InterPartMode_Fixed mAlgo_CB_InterPartMode_Fixed;
	112	Algo_CB_MergeIndex_Fixed mAlgo_CB_MergeIndex_Fixed;
	113
	114	Algo_PB_MV_Test mAlgo_PB_MV_Test;
	115	Algo_PB_MV_Search mAlgo_PB_MV_Search;
	116
	117	Algo_TB_Split_BruteForce mAlgo_TB_Split_BruteForce;
	118
	119	Algo_TB_IntraPredMode_BruteForce mAlgo_TB_IntraPredMode_BruteForce;
	120	Algo_TB_IntraPredMode_FastBrute mAlgo_TB_IntraPredMode_FastBrute;
	121	Algo_TB_IntraPredMode_MinResidual mAlgo_TB_IntraPredMode_MinResidual;
	122	};
	123
	124
	125
	126	double encode_image(encoder_context, const de265_image input, EncodingAlgorithm&);
	127
	128	void encode_sequence(encoder_context*);
	129
	130
	131	class Logging
	132	{
	133	public:
	134	virtual ~Logging() { }
	135
	136	static void print_logging(const encoder_context* ectx, const char* id, const char* filename);
	137
	138	virtual const char* name() const = 0;
	139	virtual void print(const encoder_context* ectx, const char* filename) = 0;
	140	};
	141
	142
	143	LIBDE265_API void en265_print_logging(const encoder_context* ectx, const char* id, const char* filename);
	144
	145	#endif

+1981

-0

libde265/encoder/encode.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "encode.h"
	23	#include "encoder-context.h"
	24	#include "slice.h"
	25	#include "scan.h"
	26	#include "intrapred.h"
	27	#include "libde265/transform.h"
	28	#include "libde265/fallback-dct.h"
	29	#include <iostream>
	30
	31
	32	int allocTB = 0;
	33	int allocCB = 0;
	34
	35	#define DEBUG_ALLOCS 0
	36
	37
	38
	39	void enc_node::save(const de265_image* img)
	40	{
	41	delete[] mReconstruction;
	42
	43	int blkSize = Log2SizeToArea(log2Size);
	44	mReconstruction = new uint8_t[blkSize * 3/2];
	45
	46	int w = 1<<log2Size;
	47
	48	copy_subimage(mReconstruction, w,
	49	img->get_image_plane_at_pos(0, x,y),
	50	img->get_image_stride(0),
	51	w,w);
	52
	53	copy_subimage(mReconstruction + blkSize, w>>1,
	54	img->get_image_plane_at_pos(1, x>>1,y>>1),
	55	img->get_image_stride(1),
	56	w>>1,w>>1);
	57
	58	copy_subimage(mReconstruction + blkSize*5/4, w>>1,
	59	img->get_image_plane_at_pos(2, x>>1,y>>1),
	60	img->get_image_stride(2),
	61	w>>1,w>>1);
	62	}
	63
	64
	65	void enc_node::restore(de265_image* img)
	66	{
	67	assert(mReconstruction);
	68
	69	int blkSize = Log2SizeToArea(log2Size);
	70	int w = 1<<log2Size;
	71
	72	copy_subimage(img->get_image_plane_at_pos(0, x,y),
	73	img->get_image_stride(0),
	74	mReconstruction, w,
	75	w,w);
	76
	77	copy_subimage(img->get_image_plane_at_pos(1, x>>1,y>>1),
	78	img->get_image_stride(1),
	79	mReconstruction + blkSize, w>>1,
	80	w>>1,w>>1);
	81
	82	copy_subimage(img->get_image_plane_at_pos(2, x>>1,y>>1),
	83	img->get_image_stride(2),
	84	mReconstruction + blkSize*5/4, w>>1,
	85	w>>1,w>>1);
	86	}
	87
	88
	89	void enc_cb::save(const de265_image* img)
	90	{
	91	enc_node::save(img);
	92
	93	// TODO: save metadata in node buffer memory
	94	}
	95
	96
	97	void enc_cb::restore(de265_image* img)
	98	{
	99	enc_node::restore(img);
	100
	101	// write back all the metadata
	102
	103	write_to_image(img);
	104	}
	105
	106
	107	void enc_cb::set_rqt_root_bf_from_children_cbf()
	108	{
	109	assert(transform_tree);
	110	inter.rqt_root_cbf = (transform_tree->cbf[0] \|
	111	transform_tree->cbf[1] \|
	112	transform_tree->cbf[2]);
	113	}
	114
	115
	116
	117
	118	alloc_pool enc_tb::mMemPool(sizeof(enc_tb));
	119
	120	enc_tb::enc_tb()
	121	: split_transform_flag(false)
	122	{
	123	coeff[0]=coeff[1]=coeff[2]=NULL;
	124
	125	if (DEBUG_ALLOCS) { allocTB++; printf("TB : %d\n",allocTB); }
	126	}
	127
	128
	129	enc_tb::~enc_tb()
	130	{
	131	if (split_transform_flag) {
	132	for (int i=0;i<4;i++) {
	133	delete children[i];
	134	}
	135	}
	136	else {
	137	for (int i=0;i<3;i++) {
	138	delete[] coeff[i];
	139	}
	140	}
	141
	142	if (DEBUG_ALLOCS) { allocTB--; printf("TB ~: %d\n",allocTB); }
	143	}
	144
	145
	146	void enc_tb::alloc_coeff_memory(int cIdx, int tbSize)
	147	{
	148	assert(coeff[cIdx]==NULL);
	149	coeff[cIdx] = new int16_t[tbSize*tbSize];
	150	}
	151
	152
	153	void enc_tb::reconstruct_tb(encoder_context* ectx,
	154	de265_image* img,
	155	int x0,int y0, // luma
	156	int log2TbSize, // chroma adapted
	157	const enc_cb* cb, int cIdx) const
	158	{
	159	// chroma adapted position
	160	int xC=x0;
	161	int yC=y0;
	162
	163	if (cIdx>0) {
	164	xC>>=1;
	165	yC>>=1;
	166	}
	167
	168	if (cb->PredMode == MODE_INTRA) {
	169
	170	enum IntraPredMode intraPredMode = img->get_IntraPredMode(x0,y0);
	171
	172	if (cIdx>0) {
	173	intraPredMode = cb->intra.chroma_mode;
	174	//intraPredMode = lumaPredMode_to_chromaPredMode(intraPredMode, cb->intra.chroma_mode);
	175	}
	176
	177	decode_intra_prediction(img, xC,yC, intraPredMode, 1<< log2TbSize , cIdx);
	178	}
	179	else {
	180	int size = 1<<log2TbSize;
	181
	182	uint8_t* dst_ptr = img->get_image_plane_at_pos(cIdx, xC, yC );
	183	int dst_stride = img->get_image_stride(cIdx);
	184
	185	uint8_t* src_ptr = ectx->prediction->get_image_plane_at_pos(cIdx, xC, yC );
	186	int src_stride = ectx->prediction->get_image_stride(cIdx);
	187
	188	for (int y=0;y<size;y++) {
	189	for (int x=0;x<size;x++) {
	190	dst_ptr[ydst_stride+x] = src_ptr[ysrc_stride+x];
	191	}
	192	}
	193	}
	194
	195	ALIGNED_16(int16_t) dequant_coeff[32*32];
	196
	197	if (cbf[cIdx]) dequant_coefficients(dequant_coeff, coeff[cIdx], log2TbSize, cb->qp);
	198
	199	//printf("--- quantized coeffs ---\n");
	200	//printBlk(coeff[0],1<<log2BlkSize,1<<log2BlkSize);
	201
	202	//printf("--- dequantized coeffs ---\n");
	203	//printBlk(dequant_coeff[0],1<<log2BlkSize,1<<log2BlkSize);
	204
	205	//printf("--- plane at %d %d / %d ---\n",x0,y0,cIdx);
	206
	207	uint8_t* ptr = img->get_image_plane_at_pos(cIdx, xC, yC );
	208	int stride = img->get_image_stride(cIdx);
	209
	210	int trType = (cIdx==0 && log2TbSize==2); // TODO: inter
	211
	212	//printf("--- prediction %d %d / %d ---\n",x0,y0,cIdx);
	213	//printBlk("prediction",ptr,1<<log2TbSize,stride);
	214
	215	if (cbf[cIdx]) inv_transform(&ectx->acceleration,
	216	ptr,stride, dequant_coeff, log2TbSize, trType);
	217
	218
	219	//printf("--- RECO intra prediction %d %d ---\n",x0,y0);
	220	//printBlk("prediction",ptr,1<<log2TbSize,stride);
	221
	222	//dequant_and_add_transform(accel, img, x0,y0, qp);
	223
	224	//printf("--- RECO add residual %d %d ---\n",x0,y0);
	225	//img->printBlk(x0,y0,0,log2CbSize);
	226	}
	227
	228
	229	void enc_tb::reconstruct(encoder_context* ectx,
	230	de265_image* img,
	231	const enc_cb* cb,
	232	int blkIdx) const
	233	{
	234	if (split_transform_flag) {
	235	for (int i=0;i<4;i++) {
	236	children[i]->reconstruct(ectx,img,
	237	cb, i);
	238	}
	239	}
	240	else {
	241	reconstruct_tb(ectx, img, x,y, log2Size, cb, 0);
	242
	243	if (log2Size>2) {
	244	reconstruct_tb(ectx, img, x,y, log2Size-1, cb, 1);
	245	reconstruct_tb(ectx, img, x,y, log2Size-1, cb, 2);
	246	}
	247	else if (blkIdx==3) {
	248	int xBase = x - (1<<log2Size);
	249	int yBase = y - (1<<log2Size);
	250
	251	reconstruct_tb(ectx, img, xBase,yBase, log2Size, cb, 1);
	252	reconstruct_tb(ectx, img, xBase,yBase, log2Size, cb, 2);
	253	}
	254	}
	255	}
	256
	257
	258	void enc_tb::set_cbf_flags_from_children()
	259	{
	260	assert(split_transform_flag);
	261
	262	cbf[0] = 0;
	263	cbf[1] = 0;
	264	cbf[2] = 0;
	265
	266	for (int i=0;i<4;i++) {
	267	cbf[0] \|= children[i]->cbf[0];
	268	cbf[1] \|= children[i]->cbf[1];
	269	cbf[2] \|= children[i]->cbf[2];
	270	}
	271	}
	272
	273
	274
	275
	276	alloc_pool enc_cb::mMemPool(sizeof(enc_cb), 200);
	277
	278
	279	enc_cb::enc_cb()
	280	: split_cu_flag(false),
	281	cu_transquant_bypass_flag(false),
	282	pcm_flag(false),
	283	transform_tree(NULL),
	284	distortion(0),
	285	rate(0)
	286	{
	287	if (DEBUG_ALLOCS) { allocCB++; printf("CB : %d\n",allocCB); }
	288	}
	289
	290	enc_cb::~enc_cb()
	291	{
	292	if (split_cu_flag) {
	293	for (int i=0;i<4;i++) {
	294	delete children[i];
	295	}
	296	}
	297	else {
	298	delete transform_tree;
	299	}
	300
	301	if (DEBUG_ALLOCS) { allocCB--; printf("CB ~: %d\n",allocCB); }
	302	}
	303
	304
	305	void enc_cb::write_to_image(de265_image* img) const
	306	{
	307	//printf("write_to_image %d %d size:%d\n",x,y,1<<log2Size);
	308
	309
	310	if (!split_cu_flag) {
	311	img->set_log2CbSize(x,y,log2Size, true);
	312	img->set_ctDepth(x,y,log2Size, ctDepth);
	313	assert(pcm_flag==0);
	314	img->set_pcm_flag(x,y,log2Size, pcm_flag);
	315	img->set_cu_transquant_bypass(x,y,log2Size, cu_transquant_bypass_flag);
	316	img->set_QPY(x,y,log2Size, qp);
	317	img->set_pred_mode(x,y, log2Size, PredMode);
	318	img->set_PartMode(x,y, PartMode);
	319
	320	if (PredMode == MODE_INTRA) {
	321	//img->set_ChromaIntraPredMode(x,y,log2Size, intra.chroma_mode);
	322
	323	if (PartMode == PART_NxN) {
	324	int h = 1<<(log2Size-1);
	325	img->set_IntraPredMode(x ,y ,log2Size-1, intra.pred_mode[0]);
	326	img->set_IntraPredMode(x+h,y ,log2Size-1, intra.pred_mode[1]);
	327	img->set_IntraPredMode(x ,y+h,log2Size-1, intra.pred_mode[2]);
	328	img->set_IntraPredMode(x+h,y+h,log2Size-1, intra.pred_mode[3]);
	329	}
	330	else {
	331	img->set_IntraPredMode(x,y,log2Size, intra.pred_mode[0]);
	332	}
	333	}
	334	else {
	335	int nC = 1<<log2Size;
	336	int nC2 = nC>>1;
	337	int nC4 = nC>>2;
	338	int nC3 = nC-nC4;
	339	switch (PartMode) {
	340	case PART_2Nx2N:
	341	img->set_mv_info(x,y,nC,nC, inter.pb[0].motion);
	342	break;
	343	case PART_NxN:
	344	img->set_mv_info(x ,y ,nC2,nC2, inter.pb[0].motion);
	345	img->set_mv_info(x+nC2,y ,nC2,nC2, inter.pb[1].motion);
	346	img->set_mv_info(x ,y+nC2,nC2,nC2, inter.pb[2].motion);
	347	img->set_mv_info(x+nC2,y+nC2,nC2,nC2, inter.pb[3].motion);
	348	break;
	349	case PART_2NxN:
	350	img->set_mv_info(x,y ,nC,nC2, inter.pb[0].motion);
	351	img->set_mv_info(x,y+nC2,nC,nC2, inter.pb[1].motion);
	352	break;
	353	case PART_Nx2N:
	354	img->set_mv_info(x ,y,nC2,nC, inter.pb[0].motion);
	355	img->set_mv_info(x+nC2,y,nC2,nC, inter.pb[1].motion);
	356	break;
	357	case PART_2NxnU:
	358	img->set_mv_info(x,y ,nC,nC4, inter.pb[0].motion);
	359	img->set_mv_info(x,y+nC4,nC,nC3, inter.pb[1].motion);
	360	break;
	361	case PART_2NxnD:
	362	img->set_mv_info(x,y ,nC,nC3, inter.pb[0].motion);
	363	img->set_mv_info(x,y+nC3,nC,nC4, inter.pb[1].motion);
	364	break;
	365	case PART_nLx2N:
	366	img->set_mv_info(x ,y,nC4,nC, inter.pb[0].motion);
	367	img->set_mv_info(x+nC4,y,nC3,nC, inter.pb[1].motion);
	368	break;
	369	case PART_nRx2N:
	370	img->set_mv_info(x ,y,nC3,nC, inter.pb[0].motion);
	371	img->set_mv_info(x+nC3,y,nC4,nC, inter.pb[1].motion);
	372	break;
	373	}
	374	}
	375	}
	376	else {
	377	for (int i=0;i<4;i++) {
	378	if (children[i]) {
	379	children[i]->write_to_image(img);
	380	}
	381	}
	382	}
	383	}
	384
	385
	386	void enc_cb::reconstruct(encoder_context* ectx, de265_image* img) const
	387	{
	388	if (split_cu_flag) {
	389	for (int i=0;i<4;i++) {
	390	children[i]->reconstruct(ectx, img);
	391	}
	392	}
	393	else {
	394	write_to_image(img);
	395	transform_tree->reconstruct(ectx,img,this,0);
	396	}
	397	}
	398
	399
	400
	401	void encode_split_cu_flag(encoder_context* ectx,
	402	CABAC_encoder* cabac,
	403	int x0, int y0, int ctDepth, int split_flag)
	404	{
	405	logtrace(LogSymbols,"$1 split_cu_flag=%d\n",split_flag);
	406
	407	// check if neighbors are available
	408
	409	int availableL = check_CTB_available(ectx->img, x0,y0, x0-1,y0);
	410	int availableA = check_CTB_available(ectx->img, x0,y0, x0,y0-1);
	411
	412	int condL = 0;
	413	int condA = 0;
	414
	415	if (availableL && ectx->img->get_ctDepth(x0-1,y0) > ctDepth) condL=1;
	416	if (availableA && ectx->img->get_ctDepth(x0,y0-1) > ctDepth) condA=1;
	417
	418	int contextOffset = condL + condA;
	419	int context = contextOffset;
	420
	421	// decode bit
	422
	423	logtrace(LogSlice,"> split_cu_flag = %d (context=%d)\n",split_flag,context);
	424
	425	cabac->write_CABAC_bit(CONTEXT_MODEL_SPLIT_CU_FLAG + context, split_flag);
	426	}
	427
	428
	429	void encode_part_mode(encoder_context* ectx,
	430	CABAC_encoder* cabac,
	431	enum PredMode PredMode, enum PartMode PartMode, int cLog2CbSize)
	432	{
	433	logtrace(LogSymbols,"$1 part_mode=%d\n",PartMode);
	434	logtrace(LogSlice,"> part_mode = %d\n",PartMode);
	435
	436	if (PredMode == MODE_INTRA) {
	437	int bin = (PartMode==PART_2Nx2N);
	438	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+0, bin);
	439	}
	440	else {
	441	if (PartMode==PART_2Nx2N) {
	442	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+0, 1);
	443	return;
	444	}
	445	else {
	446	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+0, 0);
	447	}
	448
	449	if (cLog2CbSize > ectx->sps.Log2MinCbSizeY) {
	450	if (ectx->sps.amp_enabled_flag) {
	451	switch (PartMode) {
	452	case PART_2NxN:
	453	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 1);
	454	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 1);
	455	break;
	456	case PART_Nx2N:
	457	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 0);
	458	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 1);
	459	break;
	460	case PART_2NxnU:
	461	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 1);
	462	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 0);
	463	cabac->write_CABAC_bypass(0);
	464	break;
	465	case PART_2NxnD:
	466	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 1);
	467	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 0);
	468	cabac->write_CABAC_bypass(1);
	469	break;
	470	case PART_nLx2N:
	471	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 0);
	472	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 0);
	473	cabac->write_CABAC_bypass(0);
	474	break;
	475	case PART_nRx2N:
	476	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 0);
	477	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 0);
	478	cabac->write_CABAC_bypass(1);
	479	break;
	480	case PART_NxN:
	481	case PART_2Nx2N:
	482	assert(false);
	483	break;
	484	}
	485	}
	486	else {
	487	if (PartMode==PART_2NxN) {
	488	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 1);
	489	}
	490	else {
	491	assert(PartMode==PART_Nx2N);
	492	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 0);
	493	}
	494	}
	495	}
	496	else {
	497	if (PartMode==PART_2NxN) {
	498	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 1);
	499	}
	500	else {
	501	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+1, 0);
	502
	503	if (cLog2CbSize==3) {
	504	assert(PartMode==PART_Nx2N);
	505	}
	506	else {
	507	if (PartMode==PART_Nx2N) {
	508	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 1);
	509	}
	510	else {
	511	assert(PartMode==PART_NxN);
	512	cabac->write_CABAC_bit(CONTEXT_MODEL_PART_MODE+3, 0);
	513	}
	514	}
	515	}
	516	}
	517	}
	518	}
	519
	520
	521	static void encode_pred_mode_flag(encoder_context* ectx,
	522	CABAC_encoder* cabac,
	523	enum PredMode PredMode)
	524	{
	525	logtrace(LogSlice,"> pred_mode = %d\n",PredMode);
	526
	527	int flag = (PredMode == MODE_INTRA) ? 1 : 0;
	528
	529	logtrace(LogSymbols,"$1 pred_mode=%d\n",flag);
	530
	531	cabac->write_CABAC_bit(CONTEXT_MODEL_PRED_MODE_FLAG, flag);
	532	}
	533
	534
	535	static void encode_prev_intra_luma_pred_flag(encoder_context* ectx,
	536	CABAC_encoder* cabac,
	537	int intraPred)
	538	{
	539	logtrace(LogSymbols,"$1 prev_intra_luma_pred_flag=%d\n",intraPred>=0);
	540	int bin = (intraPred>=0);
	541
	542	logtrace(LogSlice,"> prev_intra_luma_pred_flag = %d\n",bin);
	543
	544	cabac->write_CABAC_bit(CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG, bin);
	545	}
	546
	547	static void encode_intra_mpm_or_rem(encoder_context* ectx,
	548	CABAC_encoder* cabac,
	549	int intraPred)
	550	{
	551	if (intraPred>=0) {
	552	logtrace(LogSymbols,"$1 mpm_idx=%d\n",intraPred);
	553	logtrace(LogSlice,"> mpm_idx = %d\n",intraPred);
	554	assert(intraPred<=2);
	555	cabac->write_CABAC_TU_bypass(intraPred, 2);
	556	}
	557	else {
	558	logtrace(LogSymbols,"$1 rem_intra_luma_pred_mode=%d\n",-intraPred-1);
	559	logtrace(LogSlice,"> rem_intra_luma_pred_mode = %d\n",-intraPred-1);
	560	cabac->write_CABAC_FL_bypass(-intraPred-1, 5);
	561	}
	562	}
	563
	564
	565	static void encode_intra_chroma_pred_mode(encoder_context* ectx,
	566	CABAC_encoder* cabac,
	567	int mode)
	568	{
	569	logtrace(LogSymbols,"$1 intra_chroma_pred_mode=%d\n",mode);
	570	logtrace(LogSlice,"> intra_chroma_pred_mode = %d\n",mode);
	571
	572	if (mode==4) {
	573	cabac->write_CABAC_bit(CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE,0);
	574	}
	575	else {
	576	assert(mode<4);
	577
	578	cabac->write_CABAC_bit(CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE,1);
	579	cabac->write_CABAC_FL_bypass(mode, 2);
	580	}
	581	}
	582
	583
	584	/* Optimized variant that tests most likely branch first.
	585	*/
	586	enum IntraChromaPredMode find_chroma_pred_mode(enum IntraPredMode chroma_mode,
	587	enum IntraPredMode luma_mode)
	588	{
	589	// most likely mode: chroma mode = luma mode
	590
	591	if (luma_mode==chroma_mode) {
	592	return INTRA_CHROMA_LIKE_LUMA;
	593	}
	594
	595
	596	// check remaining candidates
	597
	598	IntraPredMode mode = chroma_mode;
	599
	600	// angular-34 is coded by setting the coded mode equal to the luma_mode
	601	if (chroma_mode == INTRA_ANGULAR_34) {
	602	mode = luma_mode;
	603	}
	604
	605	switch (mode) {
	606	case INTRA_PLANAR: return INTRA_CHROMA_PLANAR_OR_34;
	607	case INTRA_ANGULAR_26: return INTRA_CHROMA_ANGULAR_26_OR_34;
	608	case INTRA_ANGULAR_10: return INTRA_CHROMA_ANGULAR_10_OR_34;
	609	case INTRA_DC: return INTRA_CHROMA_DC_OR_34;
	610	default:
	611	assert(false);
	612	return INTRA_CHROMA_DC_OR_34;
	613	}
	614	}
	615
	616
	617
	618	void encode_split_transform_flag(encoder_context* ectx,
	619	CABAC_encoder* cabac,
	620	int log2TrafoSize, int split_flag)
	621	{
	622	logtrace(LogSymbols,"$1 split_transform_flag=%d\n",split_flag);
	623	logtrace(LogSlice,"> split_transform_flag = %d\n",split_flag);
	624
	625	int context = 5-log2TrafoSize;
	626	assert(context >= 0 && context <= 2);
	627
	628	cabac->write_CABAC_bit(CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG + context, split_flag);
	629	}
	630
	631
	632	void encode_cbf_luma(CABAC_encoder* cabac,
	633	bool zeroTrafoDepth, int cbf_luma)
	634	{
	635	logtrace(LogSymbols,"$1 cbf_luma=%d\n",cbf_luma);
	636	logtrace(LogSlice,"> cbf_luma = %d\n",cbf_luma);
	637
	638	int context = (zeroTrafoDepth ? 1 : 0);
	639
	640	cabac->write_CABAC_bit(CONTEXT_MODEL_CBF_LUMA + context, cbf_luma);
	641	}
	642
	643
	644	void encode_cbf_chroma(CABAC_encoder* cabac,
	645	int trafoDepth, int cbf_chroma)
	646	{
	647	logtrace(LogSymbols,"$1 cbf_chroma=%d\n",cbf_chroma);
	648	logtrace(LogSlice,"> cbf_chroma = %d\n",cbf_chroma);
	649
	650	int context = trafoDepth;
	651	assert(context >= 0 && context <= 3);
	652
	653	cabac->write_CABAC_bit(CONTEXT_MODEL_CBF_CHROMA + context, cbf_chroma);
	654	}
	655
	656	static inline void encode_coded_sub_block_flag(encoder_context* ectx,
	657	CABAC_encoder* cabac,
	658	int cIdx,
	659	uint8_t coded_sub_block_neighbors,
	660	int flag)
	661	{
	662	logtrace(LogSymbols,"$1 coded_sub_block_flag=%d\n",flag);
	663	logtrace(LogSlice,"# coded_sub_block_flag = %d\n",flag);
	664
	665	// tricky computation of csbfCtx
	666	int csbfCtx = ((coded_sub_block_neighbors & 1) \| // right neighbor set or
	667	(coded_sub_block_neighbors >> 1)); // bottom neighbor set -> csbfCtx=1
	668
	669	int ctxIdxInc = csbfCtx;
	670	if (cIdx!=0) {
	671	ctxIdxInc += 2;
	672	}
	673
	674	cabac->write_CABAC_bit(CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG + ctxIdxInc, flag);
	675	}
	676
	677	static inline void encode_significant_coeff_flag_lookup(encoder_context* ectx,
	678	CABAC_encoder* cabac,
	679	uint8_t ctxIdxInc,
	680	int significantFlag)
	681	{
	682	logtrace(LogSymbols,"$1 significant_coeff_flag=%d\n",significantFlag);
	683	logtrace(LogSlice,"# significant_coeff_flag = significantFlag\n");
	684	logtrace(LogSlice,"context: %d\n",ctxIdxInc);
	685
	686	cabac->write_CABAC_bit(CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG + ctxIdxInc, significantFlag);
	687	}
	688
	689	static inline void encode_coeff_abs_level_greater1(encoder_context* ectx,
	690	CABAC_encoder* cabac,
	691	int cIdx, int i,
	692	bool firstCoeffInSubblock,
	693	bool firstSubblock,
	694	int lastSubblock_greater1Ctx,
	695	int* lastInvocation_greater1Ctx,
	696	int* lastInvocation_coeff_abs_level_greater1_flag,
	697	int* lastInvocation_ctxSet, int c1,
	698	int value)
	699	{
	700	logtrace(LogSymbols,"$1 coeff_abs_level_greater1=%d\n",value);
	701	logtrace(LogSlice,"# coeff_abs_level_greater1 = %d\n",value);
	702
	703	logtrace(LogSlice," cIdx:%d i:%d firstCoeffInSB:%d firstSB:%d lastSB>1:%d last>1Ctx:%d lastLev>1:%d lastCtxSet:%d\n", cIdx,i,firstCoeffInSubblock,firstSubblock,lastSubblock_greater1Ctx,
	704	*lastInvocation_greater1Ctx,
	705	*lastInvocation_coeff_abs_level_greater1_flag,
	706	*lastInvocation_ctxSet);
	707
	708	int lastGreater1Ctx;
	709	int greater1Ctx;
	710	int ctxSet;
	711
	712	logtrace(LogSlice,"c1: %d\n",c1);
	713
	714	if (firstCoeffInSubblock) {
	715	// block with real DC -> ctx 0
	716	if (i==0 \|\| cIdx>0) { ctxSet=0; }
	717	else { ctxSet=2; }
	718
	719	if (firstSubblock) { lastGreater1Ctx=1; }
	720	else { lastGreater1Ctx = lastSubblock_greater1Ctx; }
	721
	722	if (lastGreater1Ctx==0) { ctxSet++; }
	723
	724	logtrace(LogSlice,"ctxSet: %d\n",ctxSet);
	725
	726	greater1Ctx=1;
	727	}
	728	else { // !firstCoeffInSubblock
	729	ctxSet = *lastInvocation_ctxSet;
	730	logtrace(LogSlice,"ctxSet (old): %d\n",ctxSet);
	731
	732	greater1Ctx = *lastInvocation_greater1Ctx;
	733	if (greater1Ctx>0) {
	734	int lastGreater1Flag=*lastInvocation_coeff_abs_level_greater1_flag;
	735	if (lastGreater1Flag==1) greater1Ctx=0;
	736	else { /if (greater1Ctx>0)/ greater1Ctx++; }
	737	}
	738	}
	739
	740	ctxSet = c1; // use HM algo
	741
	742	int ctxIdxInc = (ctxSet*4) + (greater1Ctx>=3 ? 3 : greater1Ctx);
	743
	744	if (cIdx>0) { ctxIdxInc+=16; }
	745
	746	cabac->write_CABAC_bit(CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG + ctxIdxInc, value);
	747
	748	*lastInvocation_greater1Ctx = greater1Ctx;
	749	*lastInvocation_coeff_abs_level_greater1_flag = value;
	750	*lastInvocation_ctxSet = ctxSet;
	751	}
	752
	753	static void encode_coeff_abs_level_greater2(encoder_context* ectx,
	754	CABAC_encoder* cabac,
	755	int cIdx, // int i,int n,
	756	int ctxSet,
	757	int value)
	758	{
	759	logtrace(LogSymbols,"$1 coeff_abs_level_greater2=%d\n",value);
	760	logtrace(LogSlice,"# coeff_abs_level_greater2 = %d\n",value);
	761
	762	int ctxIdxInc = ctxSet;
	763
	764	if (cIdx>0) ctxIdxInc+=4;
	765
	766	cabac->write_CABAC_bit(CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG + ctxIdxInc, value);
	767	}
	768
	769
	770	bool TU(int val, int maxi)
	771	{
	772	for (int i=0;i<val;i++) {
	773	printf("1");
	774	}
	775	if (val<maxi) { printf("0"); return false; }
	776	else return true;
	777	}
	778
	779	void bin(int val, int bits)
	780	{
	781	for (int i=0;i<bits;i++) {
	782	int bit = (1<<(bits-1-i));
	783	if (val&bit) printf("1"); else printf("0");
	784	}
	785	}
	786
	787	void ExpG(int level, int riceParam)
	788	{
	789	int prefix = level >> riceParam;
	790	int suffix = level - (prefix<<riceParam);
	791
	792	//printf("%d %d ",prefix,suffix);
	793
	794	int base=0;
	795	int range=1;
	796	int nBits=0;
	797	while (prefix >= base+range) {
	798	printf("1");
	799	base+=range;
	800	range*=2;
	801	nBits++;
	802	}
	803
	804	printf("0.");
	805	bin(prefix-base, nBits);
	806	printf(":");
	807	bin(suffix,riceParam);
	808	}
	809
	810	int blamain()
	811	{
	812	int riceParam=2;
	813	int TRMax = 4<<riceParam;
	814
	815	for (int level=0;level<128;level++)
	816	{
	817	printf("%d: ",level);
	818
	819	int prefixPart = std::min(TRMax, level);
	820
	821	// code TR prefix
	822
	823	bool isMaxi = TU(prefixPart>>riceParam, TRMax>>riceParam);
	824	printf(":");
	825	if (TRMax>prefixPart) {
	826	int remain = prefixPart & ((1<<riceParam)-1);
	827	bin(remain, riceParam);
	828	}
	829	printf("\|");
	830
	831	if (isMaxi) {
	832	ExpG(level-TRMax, riceParam+1);
	833	}
	834
	835	printf("\n");
	836	}
	837
	838	return 0;
	839	}
	840
	841
	842	static void encode_coeff_abs_level_remaining(encoder_context* ectx,
	843	CABAC_encoder* cabac,
	844	int cRiceParam,
	845	int level)
	846	{
	847	logtrace(LogSymbols,"$1 coeff_abs_level_remaining=%d\n",level);
	848	logtrace(LogSlice,"# encode_coeff_abs_level_remaining = %d\n",level);
	849
	850	int cTRMax = 4<<cRiceParam;
	851	int prefixPart = std::min(level, cTRMax);
	852
	853	// --- code prefix with TR ---
	854
	855	// TU part, length 4 (cTRMax>>riceParam)
	856
	857	int nOnes = (prefixPart>>cRiceParam);
	858	cabac->write_CABAC_TU_bypass(nOnes, 4);
	859
	860	// TR suffix
	861
	862	if (cTRMax > prefixPart) {
	863	int remain = prefixPart & ((1<<cRiceParam)-1);
	864	cabac->write_CABAC_FL_bypass(remain, cRiceParam);
	865	}
	866
	867
	868	// --- remainder suffix ---
	869
	870	if (nOnes==4) {
	871	int remain = level-cTRMax;
	872	int ExpGRiceParam = cRiceParam+1;
	873
	874	int prefix = remain >> ExpGRiceParam;
	875	int suffix = remain - (prefix<<ExpGRiceParam);
	876
	877	int base=0;
	878	int range=1;
	879	int nBits=0;
	880	while (prefix >= base+range) {
	881	cabac->write_CABAC_bypass(1);
	882	base+=range;
	883	range*=2;
	884	nBits++;
	885	}
	886
	887	cabac->write_CABAC_bypass(0);
	888	cabac->write_CABAC_FL_bypass(prefix-base, nBits);
	889	cabac->write_CABAC_FL_bypass(suffix, ExpGRiceParam);
	890	}
	891	}
	892
	893	// ---------------------------------------------------------------------------
	894
	895	void findLastSignificantCoeff(const position* sbScan, const position* cScan,
	896	const int16_t* coeff, int log2TrafoSize,
	897	int* lastSignificantX, int* lastSignificantY,
	898	int* lastSb, int* lastPos)
	899	{
	900	int nSb = 1<<((log2TrafoSize-2)<<1); // number of sub-blocks
	901
	902	// find last significant coefficient
	903
	904	for (int i=nSb ; i-->0 ;) {
	905	int x0 = sbScan[i].x << 2;
	906	int y0 = sbScan[i].y << 2;
	907	for (int c=16 ; c-->0 ;) {
	908	int x = x0 + cScan[c].x;
	909	int y = y0 + cScan[c].y;
	910
	911	if (coeff[x+(y<<log2TrafoSize)]) {
	912	*lastSignificantX = x;
	913	*lastSignificantY = y;
	914	*lastSb = i;
	915	*lastPos= c;
	916
	917	logtrace(LogSlice,"last significant coeff at: %d;%d, Sb:%d Pos:%d\n", x,y,i,c);
	918
	919	return;
	920	}
	921	}
	922	}
	923
	924	// all coefficients == 0 ? cannot be since cbf should be false in this case
	925	assert(false);
	926	}
	927
	928
	929	bool subblock_has_nonzero_coefficient(const int16_t* coeff, int coeffStride,
	930	const position& sbPos)
	931	{
	932	int x0 = sbPos.x << 2;
	933	int y0 = sbPos.y << 2;
	934
	935	coeff += x0 + y0*coeffStride;
	936
	937	for (int y=0;y<4;y++) {
	938	if (coeff[0] \|\| coeff[1] \|\| coeff[2] \|\| coeff[3]) { return true; }
	939	coeff += coeffStride;
	940	}
	941
	942	return false;
	943	}
	944
	945	/*
	946	Example 16x16: prefix in [0;7]
	947
	948	prefix \| last pos
	949	=============\|=============
	950	0 \| 0
	951	1 \| 1
	952	2 \| 2
	953	3 \| 3
	954	-------------+-------------
	955	lsb nBits \|
	956	4 0 1 \| 4, 5
	957	5 1 1 \| 6, 7
	958	6 0 2 \| 8, 9,10,11
	959	7 1 2 \| 12,13,14,15
	960	*/
	961	void encode_last_signficiant_coeff_prefix(encoder_context* ectx,
	962	CABAC_encoder* cabac,
	963	int log2TrafoSize,
	964	int cIdx, int lastSignificant,
	965	int context_model_index)
	966	{
	967	logtrace(LogSlice,"> last_significant_coeff_prefix=%d log2TrafoSize:%d cIdx:%d\n",
	968	lastSignificant,log2TrafoSize,cIdx);
	969
	970	int cMax = (log2TrafoSize<<1)-1;
	971
	972	int ctxOffset, ctxShift;
	973	if (cIdx==0) {
	974	ctxOffset = 3*(log2TrafoSize-2) + ((log2TrafoSize-1)>>2);
	975	ctxShift = (log2TrafoSize+1)>>2;
	976	}
	977	else {
	978	ctxOffset = 15;
	979	ctxShift = log2TrafoSize-2;
	980	}
	981
	982	for (int binIdx=0;binIdx<lastSignificant;binIdx++)
	983	{
	984	int ctxIdxInc = (binIdx >> ctxShift);
	985	cabac->write_CABAC_bit(context_model_index + ctxOffset + ctxIdxInc, 1);
	986	}
	987
	988	if (lastSignificant != cMax) {
	989	int binIdx = lastSignificant;
	990	int ctxIdxInc = (binIdx >> ctxShift);
	991	cabac->write_CABAC_bit(context_model_index + ctxOffset + ctxIdxInc, 0);
	992	}
	993	}
	994
	995
	996	void split_last_significant_position(int pos, int* prefix, int* suffix, int* nSuffixBits)
	997	{
	998	logtrace(LogSlice,"split position %d : ",pos);
	999
	1000	// most frequent case
	1001
	1002	if (pos<=3) {
	1003	*prefix=pos;
	1004	*suffix=-1; // just to have some defined value
	1005	*nSuffixBits=0;
	1006	logtrace(LogSlice,"prefix=%d suffix=%d (%d bits)\n",prefix,suffix,*nSuffixBits);
	1007	return;
	1008	}
	1009
	1010	pos -= 4;
	1011	int nBits=1;
	1012	int range=4;
	1013	while (pos>=range) {
	1014	nBits++;
	1015	pos-=range;
	1016	range<<=1;
	1017	}
	1018
	1019	*prefix = (1+nBits)<<1;
	1020	if (pos >= (range>>1)) {
	1021	*prefix \|= 1;
	1022	pos -= (range>>1);
	1023	}
	1024	*suffix = pos;
	1025	*nSuffixBits = nBits;
	1026
	1027	logtrace(LogSlice,"prefix=%d suffix=%d (%d bits)\n",prefix,suffix,*nSuffixBits);
	1028	}
	1029
	1030
	1031	extern uint8_t* ctxIdxLookup[4 /* 4-log2-32 /][2 / !!cIdx /][2 / !!scanIdx /][4 / prevCsbf */];
	1032
	1033	/* These values are read from the image metadata:
	1034	- intra prediction mode (x0;y0)
	1035	*/
	1036	void encode_residual(encoder_context* ectx,
	1037	CABAC_encoder* cabac,
	1038	const enc_tb* tb, const enc_cb* cb,
	1039	int x0,int y0,int log2TrafoSize,int cIdx)
	1040	{
	1041	const de265_image* img = ectx->img;
	1042	const seq_parameter_set& sps = img->sps;
	1043	const pic_parameter_set& pps = img->pps;
	1044
	1045	int16_t* coeff = tb->coeff[cIdx];
	1046
	1047	if (pps.transform_skip_enabled_flag && true /* TODO */) {
	1048	}
	1049
	1050
	1051	#if 0
	1052	printf("write coefficients\n");
	1053	for (int y=0;y<(1<<log2TrafoSize);y++)
	1054	{
	1055	for (int x=0;x<(1<<log2TrafoSize);x++)
	1056	{
	1057	printf("%4d ",coeff[x+y*(1<<log2TrafoSize)]);
	1058	}
	1059	printf("\n");
	1060	}
	1061	#endif
	1062
	1063
	1064	// --- get scan orders ---
	1065
	1066	enum PredMode PredMode = cb->PredMode;
	1067	int scanIdx;
	1068
	1069	if (PredMode == MODE_INTRA) {
	1070	if (cIdx==0) {
	1071	scanIdx = get_intra_scan_idx_luma(log2TrafoSize, img->get_IntraPredMode(x0,y0));
	1072	}
	1073	else {
	1074	enum IntraPredMode chromaMode = cb->intra.chroma_mode;
	1075	/*
	1076	enum IntraPredMode chromaMode = lumaPredMode_to_chromaPredMode(img->get_IntraPredMode(x0,y0),
	1077	cb->intra.chroma_mode);
	1078	*/
	1079	scanIdx = get_intra_scan_idx_chroma(log2TrafoSize, chromaMode);
	1080	}
	1081	}
	1082	else {
	1083	scanIdx=0;
	1084	}
	1085
	1086
	1087	const position* ScanOrderSub = get_scan_order(log2TrafoSize-2, scanIdx);
	1088	const position* ScanOrderPos = get_scan_order(2, scanIdx);
	1089
	1090	int lastSignificantX, lastSignificantY;
	1091	int lastScanPos;
	1092	int lastSubBlock;
	1093	findLastSignificantCoeff(ScanOrderSub, ScanOrderPos,
	1094	coeff, log2TrafoSize,
	1095	&lastSignificantX, &lastSignificantY,
	1096	&lastSubBlock, &lastScanPos);
	1097
	1098	int codedSignificantX = lastSignificantX;
	1099	int codedSignificantY = lastSignificantY;
	1100
	1101	if (scanIdx==2) {
	1102	std::swap(codedSignificantX, codedSignificantY);
	1103	}
	1104
	1105
	1106
	1107	int prefixX, suffixX, suffixBitsX;
	1108	int prefixY, suffixY, suffixBitsY;
	1109
	1110	split_last_significant_position(codedSignificantX, &prefixX,&suffixX,&suffixBitsX);
	1111	split_last_significant_position(codedSignificantY, &prefixY,&suffixY,&suffixBitsY);
	1112
	1113	encode_last_signficiant_coeff_prefix(ectx, cabac, log2TrafoSize, cIdx, prefixX,
	1114	CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX);
	1115
	1116	encode_last_signficiant_coeff_prefix(ectx, cabac, log2TrafoSize, cIdx, prefixY,
	1117	CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX);
	1118
	1119
	1120	if (codedSignificantX > 3) {
	1121	cabac->write_CABAC_FL_bypass(suffixX, suffixBitsX);
	1122	}
	1123	if (codedSignificantY > 3) {
	1124	cabac->write_CABAC_FL_bypass(suffixY, suffixBitsY);
	1125	}
	1126
	1127
	1128
	1129	int sbWidth = 1<<(log2TrafoSize-2);
	1130	int CoeffStride = 1<<log2TrafoSize;
	1131
	1132	uint8_t coded_sub_block_neighbors[32/432/4]; // 642 flags
	1133	memset(coded_sub_block_neighbors,0,sbWidth*sbWidth);
	1134
	1135	int c1 = 1;
	1136	bool firstSubblock = true; // for coeff_abs_level_greater1_flag context model
	1137	int lastSubblock_greater1Ctx=false; /* for coeff_abs_level_greater1_flag context model
	1138	(initialization not strictly needed)
	1139	*/
	1140
	1141	int lastInvocation_greater1Ctx=0;
	1142	int lastInvocation_coeff_abs_level_greater1_flag=0;
	1143	int lastInvocation_ctxSet=0;
	1144
	1145
	1146
	1147	// ----- encode coefficients -----
	1148
	1149	//tctx->nCoeff[cIdx] = 0;
	1150
	1151
	1152	// i - subblock index
	1153	// n - coefficient index in subblock
	1154
	1155	for (int i=lastSubBlock;i>=0;i--) {
	1156	position S = ScanOrderSub[i];
	1157	int inferSbDcSigCoeffFlag=0;
	1158
	1159	logtrace(LogSlice,"sub block scan idx: %d\n",i);
	1160
	1161
	1162	// --- check whether this sub-block has to be coded ---
	1163
	1164	int sub_block_is_coded = 0;
	1165
	1166	if ((i<lastSubBlock) && (i>0)) {
	1167	sub_block_is_coded = subblock_has_nonzero_coefficient(coeff, CoeffStride, S);
	1168	encode_coded_sub_block_flag(ectx, cabac, cIdx,
	1169	coded_sub_block_neighbors[S.x+S.y*sbWidth],
	1170	sub_block_is_coded);
	1171	inferSbDcSigCoeffFlag=1;
	1172	}
	1173	else if (i==0 \|\| i==lastSubBlock) {
	1174	// first (DC) and last sub-block are always coded
	1175	// - the first will most probably contain coefficients
	1176	// - the last obviously contains the last coded coefficient
	1177
	1178	sub_block_is_coded = 1;
	1179	}
	1180
	1181	if (sub_block_is_coded) {
	1182	if (S.x > 0) coded_sub_block_neighbors[S.x-1 + S.y *sbWidth] \|= 1;
	1183	if (S.y > 0) coded_sub_block_neighbors[S.x + (S.y-1)*sbWidth] \|= 2;
	1184	}
	1185
	1186	logtrace(LogSlice,"subblock is coded: %s\n", sub_block_is_coded ? "yes":"no");
	1187
	1188
	1189	// --- write significant coefficient flags ---
	1190
	1191	int16_t coeff_value[16];
	1192	int16_t coeff_baseLevel[16];
	1193	int8_t coeff_scan_pos[16];
	1194	int8_t coeff_sign[16];
	1195	int8_t coeff_has_max_base_level[16];
	1196	int nCoefficients=0;
	1197
	1198
	1199	if (sub_block_is_coded) {
	1200	int x0 = S.x<<2;
	1201	int y0 = S.y<<2;
	1202
	1203	int log2w = log2TrafoSize-2;
	1204	int prevCsbf = coded_sub_block_neighbors[S.x+S.y*sbWidth];
	1205	uint8_t* ctxIdxMap = ctxIdxLookup[log2w][!!cIdx][!!scanIdx][prevCsbf];
	1206
	1207
	1208
	1209
	1210	// set the last coded coefficient in the last subblock
	1211
	1212	if (i==lastSubBlock) {
	1213	coeff_value[nCoefficients] = coeff[lastSignificantX+(lastSignificantY<<log2TrafoSize)];
	1214	coeff_has_max_base_level[nCoefficients] = 1; // TODO
	1215	coeff_scan_pos[nCoefficients] = lastScanPos;
	1216	nCoefficients++;
	1217	}
	1218
	1219
	1220	// --- encode all coefficients' significant_coeff flags except for the DC coefficient ---
	1221
	1222	int last_coeff = (i==lastSubBlock) ? lastScanPos-1 : 15;
	1223
	1224	for (int n= last_coeff ; n>0 ; n--) {
	1225	int subX = ScanOrderPos[n].x;
	1226	int subY = ScanOrderPos[n].y;
	1227	int xC = x0 + subX;
	1228	int yC = y0 + subY;
	1229
	1230
	1231	// for all AC coefficients in sub-block, a significant_coeff flag is coded
	1232
	1233	int isSignificant = !!tb->coeff[cIdx][xC + (yC<<log2TrafoSize)];
	1234
	1235	logtrace(LogSlice,"coeff %d is significant: %d\n", n, isSignificant);
	1236
	1237	logtrace(LogSlice,"context idx: %d;%d\n",xC,yC);
	1238
	1239	encode_significant_coeff_flag_lookup(ectx, cabac,
	1240	ctxIdxMap[xC+(yC<<log2TrafoSize)],
	1241	isSignificant);
	1242	//ctxIdxMap[(i<<4)+n]);
	1243
	1244	if (isSignificant) {
	1245	coeff_value[nCoefficients] = coeff[xC+(yC<<log2TrafoSize)];
	1246	coeff_has_max_base_level[nCoefficients] = 1;
	1247	coeff_scan_pos[nCoefficients] = n;
	1248	nCoefficients++;
	1249
	1250	// since we have a coefficient in the sub-block,
	1251	// we cannot infer the DC coefficient anymore
	1252	inferSbDcSigCoeffFlag = 0;
	1253	}
	1254	}
	1255
	1256
	1257	// --- decode DC coefficient significance ---
	1258
	1259	if (last_coeff>=0) // last coded coefficient (always set to 1) is not the DC coefficient
	1260	{
	1261	if (inferSbDcSigCoeffFlag==0) {
	1262	// if we cannot infert the DC coefficient, it is coded
	1263	int isSignificant = !!tb->coeff[cIdx][x0 + (y0<<log2TrafoSize)];
	1264
	1265	logtrace(LogSlice,"DC coeff is significant: %d\n", isSignificant);
	1266
	1267	encode_significant_coeff_flag_lookup(ectx, cabac,
	1268	ctxIdxMap[x0+(y0<<log2TrafoSize)],
	1269	isSignificant);
	1270
	1271	if (isSignificant) {
	1272	coeff_value[nCoefficients] = coeff[x0+(y0<<log2TrafoSize)];
	1273	coeff_has_max_base_level[nCoefficients] = 1;
	1274	coeff_scan_pos[nCoefficients] = 0;
	1275	nCoefficients++;
	1276	}
	1277	}
	1278	else {
	1279	// we can infer that the DC coefficient must be present
	1280	coeff_value[nCoefficients] = coeff[x0+(y0<<log2TrafoSize)];
	1281	coeff_has_max_base_level[nCoefficients] = 1;
	1282	coeff_scan_pos[nCoefficients] = 0;
	1283	nCoefficients++;
	1284	}
	1285	}
	1286	}
	1287
	1288
	1289
	1290	// --- encode coefficient values ---
	1291
	1292	if (nCoefficients) {
	1293
	1294	// separate absolute coefficient value and sign
	1295
	1296	logtrace(LogSlice,"coefficients to code: ");
	1297
	1298	for (int l=0;l<nCoefficients;l++) {
	1299	logtrace(LogSlice,"%d ",coeff_value[l]);
	1300
	1301	if (coeff_value[l]<0) {
	1302	coeff_value[l] = -coeff_value[l];
	1303	coeff_sign[l] = 1;
	1304	}
	1305	else {
	1306	coeff_sign[l] = 0;
	1307	}
	1308
	1309	coeff_baseLevel[l] = 1;
	1310
	1311	logtrace(LogSlice,"(%d) ",coeff_scan_pos[l]);
	1312	}
	1313
	1314	logtrace(LogSlice,"\n");
	1315
	1316
	1317	int ctxSet;
	1318	if (i==0 \|\| cIdx>0) { ctxSet=0; }
	1319	else { ctxSet=2; }
	1320
	1321	if (c1==0) { ctxSet++; }
	1322	c1=1;
	1323
	1324
	1325	// --- encode greater-1 flags ---
	1326
	1327	int newLastGreater1ScanPos=-1;
	1328
	1329	int lastGreater1Coefficient = libde265_min(8,nCoefficients);
	1330	for (int c=0;c<lastGreater1Coefficient;c++) {
	1331	int greater1_flag = (coeff_value[c]>1);
	1332
	1333	encode_coeff_abs_level_greater1(ectx, cabac, cIdx,i,
	1334	c==0,
	1335	firstSubblock,
	1336	lastSubblock_greater1Ctx,
	1337	&lastInvocation_greater1Ctx,
	1338	&lastInvocation_coeff_abs_level_greater1_flag,
	1339	&lastInvocation_ctxSet, ctxSet,
	1340	greater1_flag);
	1341
	1342	if (greater1_flag) {
	1343	coeff_baseLevel[c]++;
	1344
	1345	c1=0;
	1346
	1347	if (newLastGreater1ScanPos == -1) {
	1348	newLastGreater1ScanPos=c;
	1349	}
	1350	}
	1351	else {
	1352	coeff_has_max_base_level[c] = 0;
	1353
	1354	if (c1<3 && c1>0) {
	1355	c1++;
	1356	}
	1357	}
	1358	}
	1359
	1360	firstSubblock = false;
	1361	lastSubblock_greater1Ctx = lastInvocation_greater1Ctx;
	1362
	1363
	1364	// --- decode greater-2 flag ---
	1365
	1366	if (newLastGreater1ScanPos != -1) {
	1367	int greater2_flag = (coeff_value[newLastGreater1ScanPos]>2);
	1368	encode_coeff_abs_level_greater2(ectx,cabac, cIdx, lastInvocation_ctxSet, greater2_flag);
	1369	coeff_baseLevel[newLastGreater1ScanPos] += greater2_flag;
	1370	coeff_has_max_base_level[newLastGreater1ScanPos] = greater2_flag;
	1371	}
	1372
	1373
	1374	// --- encode coefficient signs ---
	1375
	1376	int signHidden = (coeff_scan_pos[0]-coeff_scan_pos[nCoefficients-1] > 3 &&
	1377	!cb->cu_transquant_bypass_flag);
	1378
	1379	for (int n=0;n<nCoefficients-1;n++) {
	1380	cabac->write_CABAC_bypass(coeff_sign[n]);
	1381	//logtrace(LogSlice,"a) sign[%d] = %d\n", n, coeff_sign[n]);
	1382	}
	1383
	1384	// n==nCoefficients-1
	1385	if (!pps.sign_data_hiding_flag \|\| !signHidden) {
	1386	cabac->write_CABAC_bypass(coeff_sign[nCoefficients-1]);
	1387	//logtrace(LogSlice,"b) sign[%d] = %d\n", nCoefficients-1, coeff_sign[nCoefficients-1]);
	1388	}
	1389	else {
	1390	assert(coeff_sign[nCoefficients-1] == 0);
	1391	}
	1392
	1393	// --- decode coefficient value ---
	1394
	1395	int sumAbsLevel=0;
	1396	int uiGoRiceParam=0;
	1397
	1398	for (int n=0;n<nCoefficients;n++) {
	1399	int baseLevel = coeff_baseLevel[n];
	1400
	1401	int coeff_abs_level_remaining;
	1402
	1403	if (coeff_has_max_base_level[n]) {
	1404	logtrace(LogSlice,"value[%d]=%d, base level: %d\n",n,coeff_value[n],coeff_baseLevel[n]);
	1405
	1406	coeff_abs_level_remaining = coeff_value[n] - coeff_baseLevel[n];
	1407
	1408	encode_coeff_abs_level_remaining(ectx, cabac, uiGoRiceParam,
	1409	coeff_abs_level_remaining);
	1410
	1411	// (9-462)
	1412	if (baseLevel + coeff_abs_level_remaining > 3*(1<<uiGoRiceParam)) {
	1413	uiGoRiceParam++;
	1414	if (uiGoRiceParam>4) uiGoRiceParam=4;
	1415	}
	1416	}
	1417	else {
	1418	coeff_abs_level_remaining = 0;
	1419	}
	1420
	1421
	1422	// --- DEBUG: check coefficient ---
	1423
	1424	#if 0
	1425	int16_t currCoeff = baseLevel + coeff_abs_level_remaining;
	1426	if (coeff_sign[n]) {
	1427	currCoeff = -currCoeff;
	1428	}
	1429
	1430	if (pps.sign_data_hiding_flag && signHidden) {
	1431	sumAbsLevel += baseLevel + coeff_abs_level_remaining;
	1432
	1433	if (n==nCoefficients-1 && (sumAbsLevel & 1)) {
	1434	currCoeff = -currCoeff;
	1435	}
	1436	}
	1437
	1438	assert(currCoeff == coeff_value[n]);
	1439	#endif
	1440	} // iterate through coefficients in sub-block
	1441	} // if nonZero
	1442
	1443	}
	1444	}
	1445
	1446
	1447	void encode_transform_unit(encoder_context* ectx,
	1448	CABAC_encoder* cabac,
	1449	const enc_tb* tb, const enc_cb* cb,
	1450	int x0,int y0, int xBase,int yBase,
	1451	int log2TrafoSize, int trafoDepth, int blkIdx)
	1452	{
	1453	if (tb->cbf[0] \|\| tb->cbf[1] \|\| tb->cbf[2]) {
	1454	if (ectx->img->pps.cu_qp_delta_enabled_flag &&
	1455	true /!ectx->IsCuQpDeltaCoded/) {
	1456	assert(0);
	1457	}
	1458
	1459	if (tb->cbf[0]) {
	1460	encode_residual(ectx,cabac, tb,cb,x0,y0,log2TrafoSize,0);
	1461	}
	1462
	1463	// larger than 4x4
	1464	if (log2TrafoSize>2) {
	1465	if (tb->cbf[1]) {
	1466	encode_residual(ectx,cabac,tb,cb,x0,y0,log2TrafoSize-1,1);
	1467	}
	1468	if (tb->cbf[2]) {
	1469	encode_residual(ectx,cabac,tb,cb,x0,y0,log2TrafoSize-1,2);
	1470	}
	1471	}
	1472	else if (blkIdx==3) {
	1473	// cannot check for tb->parent->cbf[], because this may not yet be set
	1474	if (tb->cbf[1]) {
	1475	encode_residual(ectx,cabac,tb,cb,xBase,yBase,log2TrafoSize,1);
	1476	}
	1477	if (tb->cbf[2]) {
	1478	encode_residual(ectx,cabac,tb,cb,xBase,yBase,log2TrafoSize,2);
	1479	}
	1480	}
	1481	}
	1482	}
	1483
	1484
	1485	void encode_transform_tree(encoder_context* ectx,
	1486	CABAC_encoder* cabac,
	1487	const enc_tb* tb, const enc_cb* cb,
	1488	int x0,int y0, int xBase,int yBase,
	1489	int log2TrafoSize, int trafoDepth, int blkIdx,
	1490	int MaxTrafoDepth, int IntraSplitFlag, bool recurse)
	1491	{
	1492	//de265_image* img = ectx->img;
	1493	const seq_parameter_set* sps = &ectx->img->sps;
	1494
	1495	if (log2TrafoSize <= sps->Log2MaxTrafoSize &&
	1496	log2TrafoSize > sps->Log2MinTrafoSize &&
	1497	trafoDepth < MaxTrafoDepth &&
	1498	!(IntraSplitFlag && trafoDepth==0))
	1499	{
	1500	int split_transform_flag = tb->split_transform_flag;
	1501	encode_split_transform_flag(ectx, cabac, log2TrafoSize, split_transform_flag);
	1502	}
	1503	else
	1504	{
	1505	int interSplitFlag=0; // TODO
	1506
	1507	bool split_transform_flag = (log2TrafoSize > sps->Log2MaxTrafoSize \|\|
	1508	(IntraSplitFlag==1 && trafoDepth==0) \|\|
	1509	interSplitFlag==1) ? 1:0;
	1510
	1511	/*
	1512	printf("split_transform_flag log2TrafoSize:%d Log2MaxTrafoSize:%d "
	1513	"IntraSplitFlag:%d trafoDepth:%d -> %d\n",
	1514	log2TrafoSize,sps->Log2MaxTrafoSize,
	1515	IntraSplitFlag, trafoDepth,
	1516	split_transform_flag);
	1517	*/
	1518
	1519	assert(tb->split_transform_flag == split_transform_flag);
	1520	}
	1521
	1522	// --- CBF CB/CR ---
	1523
	1524	// For 4x4 luma, there is no signaling of chroma CBF, because only the
	1525	// chroma CBF for 8x8 is relevant.
	1526	if (log2TrafoSize>2) {
	1527	if (trafoDepth==0 \|\| tb->parent->cbf[1]) {
	1528	encode_cbf_chroma(cabac, trafoDepth, tb->cbf[1]);
	1529	}
	1530	if (trafoDepth==0 \|\| tb->parent->cbf[2]) {
	1531	encode_cbf_chroma(cabac, trafoDepth, tb->cbf[2]);
	1532	}
	1533	}
	1534
	1535	if (tb->split_transform_flag) {
	1536	if (recurse) {
	1537	int x1 = x0 + (1<<(log2TrafoSize-1));
	1538	int y1 = y0 + (1<<(log2TrafoSize-1));
	1539
	1540	encode_transform_tree(ectx, cabac, tb->children[0], cb, x0,y0,x0,y0,log2TrafoSize-1,
	1541	trafoDepth+1, 0, MaxTrafoDepth, IntraSplitFlag, true);
	1542	encode_transform_tree(ectx, cabac, tb->children[1], cb, x1,y0,x0,y0,log2TrafoSize-1,
	1543	trafoDepth+1, 1, MaxTrafoDepth, IntraSplitFlag, true);
	1544	encode_transform_tree(ectx, cabac, tb->children[2], cb, x0,y1,x0,y0,log2TrafoSize-1,
	1545	trafoDepth+1, 2, MaxTrafoDepth, IntraSplitFlag, true);
	1546	encode_transform_tree(ectx, cabac, tb->children[3], cb, x1,y1,x0,y0,log2TrafoSize-1,
	1547	trafoDepth+1, 3, MaxTrafoDepth, IntraSplitFlag, true);
	1548	}
	1549	}
	1550	else {
	1551	if (cb->PredMode == MODE_INTRA \|\| trafoDepth != 0 \|\|
	1552	tb->cbf[1] \|\| tb->cbf[2]) {
	1553	encode_cbf_luma(cabac, trafoDepth==0, tb->cbf[0]);
	1554	}
	1555	else {
	1556	/* Note: usually, cbf[0] should be TRUE, but while estimating the bitrate, this
	1557	function can also be called with all CBFs FALSE. Usually, this is handled by
	1558	the rqt_root_cbf flag, but during analysis, this is set after the bitrate is estimated.
	1559	*/
	1560	// assert(tb->cbf[0]==true);
	1561	}
	1562
	1563	encode_transform_unit(ectx,cabac, tb,cb, x0,y0, xBase,yBase, log2TrafoSize, trafoDepth, blkIdx);
	1564	}
	1565	}
	1566
	1567
	1568	void encode_cu_skip_flag(encoder_context* ectx,
	1569	CABAC_encoder* cabac,
	1570	const enc_cb* cb,
	1571	bool skip)
	1572	{
	1573	logtrace(LogSymbols,"$1 cu_skip_flag=%d\n",skip);
	1574
	1575	const de265_image* img = ectx->img;
	1576
	1577	int x0 = cb->x;
	1578	int y0 = cb->y;
	1579
	1580	// check if neighbors are available
	1581
	1582	int availableL = check_CTB_available(img, x0,y0, x0-1,y0);
	1583	int availableA = check_CTB_available(img, x0,y0, x0,y0-1);
	1584
	1585	int condL = 0;
	1586	int condA = 0;
	1587
	1588	if (availableL && img->get_cu_skip_flag(x0-1,y0)) condL=1;
	1589	if (availableA && img->get_cu_skip_flag(x0,y0-1)) condA=1;
	1590
	1591	int contextOffset = condL + condA;
	1592	int context = contextOffset;
	1593
	1594	// decode bit
	1595
	1596	int bit = skip;
	1597
	1598	logtrace(LogSlice,"> cu_skip_flag ctx=%d, bit=%d\n", context,bit);
	1599
	1600	cabac->write_CABAC_bit(CONTEXT_MODEL_CU_SKIP_FLAG + context, bit);
	1601	}
	1602
	1603
	1604	void encode_merge_idx(encoder_context* ectx,
	1605	CABAC_encoder* cabac,
	1606	int mergeIdx)
	1607	{
	1608	logtrace(LogSymbols,"$1 merge_idx=%d\n",mergeIdx);
	1609	logtrace(LogSlice,"# merge_idx %d\n", mergeIdx);
	1610
	1611	if (ectx->shdr->MaxNumMergeCand <= 1) {
	1612	return; // code nothing, we use only a single merge candidate
	1613	}
	1614
	1615	// TU coding, first bin is CABAC, remaining are bypass.
	1616	// cMax = MaxNumMergeCand-1
	1617
	1618	cabac->write_CABAC_bit(CONTEXT_MODEL_MERGE_IDX, mergeIdx ? 1 : 0);
	1619
	1620	if (mergeIdx>0) {
	1621	int idx=1;
	1622
	1623	while (idx<ectx->shdr->MaxNumMergeCand-1) {
	1624	int increase = (idx < mergeIdx);
	1625
	1626	cabac->write_CABAC_bypass(increase);
	1627	if (increase) {
	1628	idx++;
	1629	}
	1630	else {
	1631	break;
	1632	}
	1633	}
	1634	}
	1635	}
	1636
	1637
	1638	static inline void encode_rqt_root_cbf(encoder_context* ectx,
	1639	CABAC_encoder* cabac,
	1640	int rqt_root_cbf)
	1641	{
	1642	logtrace(LogSymbols,"$1 rqt_root_cbf=%d\n",rqt_root_cbf);
	1643	cabac->write_CABAC_bit(CONTEXT_MODEL_RQT_ROOT_CBF, rqt_root_cbf);
	1644	}
	1645
	1646
	1647	void encode_mvd(encoder_context* ectx,
	1648	CABAC_encoder* cabac,
	1649	const int16_t mvd[2])
	1650	{
	1651	int mvd0abs = abs_value(mvd[0]);
	1652	int mvd1abs = abs_value(mvd[1]);
	1653
	1654	int mvd0_greater_0 = !!(mvd0abs);
	1655	int mvd1_greater_0 = !!(mvd1abs);
	1656
	1657	cabac->write_CABAC_bit(CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG+0, mvd0_greater_0);
	1658	cabac->write_CABAC_bit(CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG+0, mvd1_greater_0);
	1659
	1660	if (mvd0_greater_0) {
	1661	cabac->write_CABAC_bit(CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG+1, mvd0abs>1);
	1662	}
	1663	if (mvd1_greater_0) {
	1664	cabac->write_CABAC_bit(CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG+1, mvd1abs>1);
	1665	}
	1666
	1667	if (mvd0abs) {
	1668	if (mvd0abs>1) {
	1669	cabac->write_CABAC_EGk(mvd0abs-2,1);
	1670	}
	1671	cabac->write_CABAC_bypass(mvd[0]<0);
	1672	}
	1673
	1674	if (mvd1abs) {
	1675	if (mvd1abs>1) {
	1676	cabac->write_CABAC_EGk(mvd1abs-2,1);
	1677	}
	1678	cabac->write_CABAC_bypass(mvd[1]<0);
	1679	}
	1680	}
	1681
	1682
	1683	void encode_prediction_unit(encoder_context* ectx,
	1684	CABAC_encoder* cabac,
	1685	const enc_cb* cb, int pbIdx,
	1686	int x0,int y0, int w, int h)
	1687	{
	1688	const enc_pb_inter& pb = cb->inter.pb[pbIdx];
	1689
	1690	logtrace(LogSymbols,"$1 merge_flag=%d\n",pb.spec.merge_flag);
	1691	cabac->write_CABAC_bit(CONTEXT_MODEL_MERGE_FLAG, pb.spec.merge_flag);
	1692
	1693	if (pb.spec.merge_flag) {
	1694	assert(false); // TODO
	1695	}
	1696	else {
	1697	if (ectx->shdr->slice_type == SLICE_TYPE_B) {
	1698	assert(false); // TODO
	1699	}
	1700
	1701	if (pb.spec.inter_pred_idc != PRED_L1) {
	1702	if (ectx->shdr->num_ref_idx_l0_active > 1) {
	1703	assert(false); // TODO
	1704	//cabac->write_CABAC_bit(CONTEXT_MODEL_REF_IDX_LX, pb.spec.mvp_l0_flag);
	1705	}
	1706
	1707	encode_mvd(ectx,cabac, pb.spec.mvd[0]);
	1708
	1709	logtrace(LogSymbols,"$1 mvp_lx_flag=%d\n",pb.spec.mvp_l0_flag);
	1710	cabac->write_CABAC_bit(CONTEXT_MODEL_MVP_LX_FLAG, pb.spec.mvp_l0_flag);
	1711	}
	1712
	1713	if (pb.spec.inter_pred_idc != PRED_L0) {
	1714	assert(false); // TODO
	1715	}
	1716
	1717	/*
	1718	enum InterPredIdc
	1719	PRED_L0=0,
	1720	PRED_L1=1,
	1721	PRED_BI=2
	1722	*/
	1723	}
	1724	}
	1725
	1726
	1727	void encode_coding_unit(encoder_context* ectx,
	1728	CABAC_encoder* cabac,
	1729	const enc_cb* cb, int x0,int y0, int log2CbSize, bool recurse)
	1730	{
	1731	logtrace(LogSlice,"--- encode CU (%d;%d) ---\n",x0,y0);
	1732
	1733	de265_image* img = ectx->img;
	1734	const slice_segment_header* shdr = &ectx->imgdata->shdr;
	1735	const seq_parameter_set* sps = &ectx->img->sps;
	1736
	1737
	1738	int nCbS = 1<<log2CbSize;
	1739
	1740
	1741	// write skip_flag
	1742
	1743	if (shdr->slice_type != SLICE_TYPE_I) {
	1744	encode_cu_skip_flag(ectx,cabac, cb, cb->PredMode==MODE_SKIP);
	1745	}
	1746
	1747	if (cb->PredMode==MODE_SKIP) {
	1748	assert(cb->inter.pb[0].spec.merge_flag);
	1749	encode_merge_idx(ectx,cabac, cb->inter.pb[0].spec.merge_idx);
	1750	}
	1751	else {
	1752
	1753	enum PredMode PredMode = cb->PredMode;
	1754	enum PartMode PartMode = PART_2Nx2N;
	1755	int IntraSplitFlag=0;
	1756
	1757	if (shdr->slice_type != SLICE_TYPE_I) {
	1758	encode_pred_mode_flag(ectx,cabac, PredMode);
	1759	}
	1760
	1761	if (PredMode != MODE_INTRA \|\|
	1762	log2CbSize == sps->Log2MinCbSizeY) {
	1763	PartMode = cb->PartMode;
	1764	encode_part_mode(ectx,cabac, PredMode, PartMode, log2CbSize);
	1765	}
	1766
	1767	if (PredMode == MODE_INTRA) {
	1768
	1769	int availableA0 = check_CTB_available(img, x0,y0, x0-1,y0);
	1770	int availableB0 = check_CTB_available(img, x0,y0, x0,y0-1);
	1771
	1772	if (PartMode==PART_2Nx2N) {
	1773	logtrace(LogSlice,"x0,y0: %d,%d\n",x0,y0);
	1774	int PUidx = (x0>>sps->Log2MinPUSize) + (y0>>sps->Log2MinPUSize)*sps->PicWidthInMinPUs;
	1775
	1776	int candModeList[3];
	1777	fillIntraPredModeCandidates(candModeList,x0,y0,PUidx,
	1778	availableA0,availableB0, img);
	1779
	1780	for (int i=0;i<3;i++)
	1781	logtrace(LogSlice,"candModeList[%d] = %d\n", i, candModeList[i]);
	1782
	1783	enum IntraPredMode mode = cb->intra.pred_mode[0];
	1784	int intraPred = find_intra_pred_mode(mode, candModeList);
	1785	encode_prev_intra_luma_pred_flag(ectx,cabac, intraPred);
	1786	encode_intra_mpm_or_rem(ectx,cabac, intraPred);
	1787
	1788	logtrace(LogSlice,"IntraPredMode: %d (candidates: %d %d %d)\n", mode,
	1789	candModeList[0], candModeList[1], candModeList[2]);
	1790	logtrace(LogSlice," MPM/REM = %d\n",intraPred);
	1791	}
	1792	else {
	1793	IntraSplitFlag=1;
	1794
	1795	int pbOffset = nCbS/2;
	1796	int PUidx;
	1797
	1798	int intraPred[4];
	1799	int childIdx=0;
	1800
	1801	for (int j=0;j<nCbS;j+=pbOffset)
	1802	for (int i=0;i<nCbS;i+=pbOffset, childIdx++)
	1803	{
	1804	int x=x0+i, y=y0+j;
	1805
	1806	int availableA = availableA0 \|\| (i>0); // left candidate always available for right blk
	1807	int availableB = availableB0 \|\| (j>0); // top candidate always available for bottom blk
	1808
	1809	PUidx = (x>>sps->Log2MinPUSize) + (y>>sps->Log2MinPUSize)*sps->PicWidthInMinPUs;
	1810
	1811	int candModeList[3];
	1812	fillIntraPredModeCandidates(candModeList,x,y,PUidx,
	1813	availableA,availableB, img);
	1814
	1815	enum IntraPredMode mode = cb->intra.pred_mode[childIdx];
	1816
	1817	assert(ectx->img->get_IntraPredMode(x,y) == mode);
	1818
	1819	intraPred[childIdx] = find_intra_pred_mode(mode, candModeList);
	1820	}
	1821
	1822	for (int i=0;i<4;i++)
	1823	encode_prev_intra_luma_pred_flag(ectx,cabac, intraPred[i]);
	1824
	1825	for (int i=0;i<4;i++)
	1826	encode_intra_mpm_or_rem(ectx,cabac, intraPred[i]);
	1827	}
	1828
	1829	IntraChromaPredMode chromaPredMode = find_chroma_pred_mode(cb->intra.chroma_mode,
	1830	cb->intra.pred_mode[0]);
	1831	encode_intra_chroma_pred_mode(ectx,cabac, chromaPredMode);
	1832	}
	1833	else {
	1834	switch (cb->PartMode) {
	1835	case PART_2Nx2N:
	1836	encode_prediction_unit(ectx,cabac,cb, 0, cb->x,cb->y,1<<cb->log2Size,1<<cb->log2Size);
	1837	break;
	1838	case PART_2NxN:
	1839	case PART_Nx2N:
	1840	case PART_NxN:
	1841	case PART_2NxnU:
	1842	case PART_2NxnD:
	1843	case PART_nLx2N:
	1844	case PART_nRx2N:
	1845	assert(false); // TODO
	1846	}
	1847	}
	1848
	1849
	1850	if (true) { // !pcm
	1851
	1852	if (cb->PredMode != MODE_INTRA &&
	1853	!(cb->PartMode == PART_2Nx2N && cb->inter.pb[0].spec.merge_flag)) {
	1854
	1855	//printf("%d %d %d\n",cb->PredMode,cb->PartMode,cb->inter.pb[0].merge_flag);
	1856
	1857	encode_rqt_root_cbf(ectx,cabac, cb->inter.rqt_root_cbf);
	1858	}
	1859
	1860	//printf("%d;%d encode rqt_root_cbf=%d\n",x0,y0,cb->inter.rqt_root_cbf);
	1861
	1862	if (cb->PredMode == MODE_INTRA \|\| cb->inter.rqt_root_cbf) {
	1863	int MaxTrafoDepth;
	1864	if (PredMode == MODE_INTRA)
	1865	{ MaxTrafoDepth = sps->max_transform_hierarchy_depth_intra + IntraSplitFlag; }
	1866	else
	1867	{ MaxTrafoDepth = sps->max_transform_hierarchy_depth_inter; }
	1868
	1869
	1870	if (recurse) {
	1871	//printf("%d;%d store transform tree\n",x0,y0);
	1872
	1873	encode_transform_tree(ectx,cabac, cb->transform_tree, cb,
	1874	x0,y0, x0,y0, log2CbSize, 0, 0, MaxTrafoDepth, IntraSplitFlag, true);
	1875	}
	1876	}
	1877	}
	1878	}
	1879	}
	1880
	1881
	1882	SplitType get_split_type(const seq_parameter_set* sps,
	1883	int x0,int y0, int log2CbSize)
	1884	{
	1885	/*
	1886	CU split flag:
	1887
	1888	\| overlaps \| minimum \|\|
	1889	case \| border \| size \|\| split
	1890	-----+----------+---------++----------
	1891	A \| 0 \| 0 \|\| optional
	1892	B \| 0 \| 1 \|\| 0
	1893	C \| 1 \| 0 \|\| 1
	1894	D \| 1 \| 1 \|\| 0
	1895	*/
	1896	if (x0+(1<<log2CbSize) <= sps->pic_width_in_luma_samples &&
	1897	y0+(1<<log2CbSize) <= sps->pic_height_in_luma_samples &&
	1898	log2CbSize > sps->Log2MinCbSizeY) {
	1899
	1900	// case A
	1901
	1902	return OptionalSplit;
	1903	} else {
	1904	// case B/C/D
	1905
	1906	if (log2CbSize > sps->Log2MinCbSizeY) { return ForcedSplit; }
	1907	else { return ForcedNonSplit; }
	1908	}
	1909	}
	1910
	1911
	1912	void encode_quadtree(encoder_context* ectx,
	1913	CABAC_encoder* cabac,
	1914	const enc_cb* cb, int x0,int y0, int log2CbSize, int ctDepth,
	1915	bool recurse)
	1916	{
	1917	//de265_image* img = ectx->img;
	1918	const seq_parameter_set* sps = &ectx->img->sps;
	1919
	1920	int split_flag = get_split_type(sps,x0,y0,log2CbSize);
	1921
	1922	// if it is an optional split, take the decision from the CU flag
	1923	if (split_flag == OptionalSplit) {
	1924	split_flag = cb->split_cu_flag;
	1925
	1926	encode_split_cu_flag(ectx,cabac, x0,y0, ctDepth, split_flag);
	1927	}
	1928
	1929
	1930	if (split_flag) {
	1931	if (recurse) {
	1932	int x1 = x0 + (1<<(log2CbSize-1));
	1933	int y1 = y0 + (1<<(log2CbSize-1));
	1934
	1935	encode_quadtree(ectx,cabac, cb->children[0], x0,y0, log2CbSize-1, ctDepth+1, true);
	1936
	1937	if (x1<sps->pic_width_in_luma_samples)
	1938	encode_quadtree(ectx,cabac, cb->children[1], x1,y0, log2CbSize-1, ctDepth+1, true);
	1939
	1940	if (y1<sps->pic_height_in_luma_samples)
	1941	encode_quadtree(ectx,cabac, cb->children[2], x0,y1, log2CbSize-1, ctDepth+1, true);
	1942
	1943	if (x1<sps->pic_width_in_luma_samples &&
	1944	y1<sps->pic_height_in_luma_samples)
	1945	encode_quadtree(ectx,cabac, cb->children[3], x1,y1, log2CbSize-1, ctDepth+1, true);
	1946	}
	1947	}
	1948	else {
	1949	encode_coding_unit(ectx,cabac, cb,x0,y0, log2CbSize, true);
	1950	}
	1951	}
	1952
	1953
	1954	void encode_ctb(encoder_context* ectx,
	1955	CABAC_encoder* cabac,
	1956	enc_cb* cb, int ctbX,int ctbY)
	1957	{
	1958	logtrace(LogSlice,"----- encode CTB (%d;%d) -----\n",ctbX,ctbY);
	1959
	1960	#if 0
	1961	printf("MODEL:\n");
	1962	for (int i=0;i<CONTEXT_MODEL_TABLE_LENGTH;i++)
	1963	{
	1964	printf("%d;%d ",
	1965	ectx->ctx_model[i].state,
	1966	ectx->ctx_model[i].MPSbit);
	1967
	1968	if ((i%16)==15) printf("\n");
	1969	}
	1970	printf("\n");
	1971	#endif
	1972
	1973	de265_image* img = ectx->img;
	1974	int log2ctbSize = img->sps.Log2CtbSizeY;
	1975
	1976	encode_quadtree(ectx,cabac, cb, ctbX<<log2ctbSize, ctbY<<log2ctbSize, log2ctbSize, 0, true);
	1977	}
	1978
	1979
	1980	// ---------------------------------------------------------------------------

+289

-0

libde265/encoder/encode.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef ENCODE_H
	23	#define ENCODE_H
	24
	25	#include "libde265/image.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/image-io.h"
	28	#include "libde265/alloc_pool.h"
	29
	30	class encoder_context;
	31	class enc_cb;
	32
	33
	34	class enc_node
	35	{
	36	public:
	37	enc_node() { mReconstruction=NULL; }
	38	virtual ~enc_node() { delete[] mReconstruction; }
	39
	40	uint16_t x,y;
	41	uint8_t log2Size : 3;
	42
	43	virtual void save(const de265_image*);
	44	virtual void restore(de265_image*);
	45
	46	private:
	47	uint8_t* mReconstruction;
	48	};
	49
	50
	51	class enc_tb : public enc_node
	52	{
	53	public:
	54	enc_tb();
	55	~enc_tb();
	56
	57	const enc_tb* parent;
	58
	59	uint8_t split_transform_flag : 1;
	60	uint8_t TrafoDepth : 2; // 2 bits enough ? (TODO)
	61
	62	uint8_t cbf[3];
	63
	64	union {
	65	// split
	66	struct {
	67	enc_tb* children[4];
	68	};
	69
	70	// non-split
	71	struct {
	72	int16_t* coeff[3];
	73	};
	74	};
	75
	76	float distortion; // total distortion for this level of the TB tree (including all children)
	77	float rate; // total rate for coding this TB level and all children
	78	float rate_withoutCbfChroma;
	79
	80	void set_cbf_flags_from_children();
	81
	82	void reconstruct(encoder_context* ectx,
	83	de265_image* img,
	84	const enc_cb* cb, int blkIdx=0) const;
	85
	86	bool isZeroBlock() const { return cbf[0]==false && cbf[1]==false && cbf[2]==false; }
	87
	88	void alloc_coeff_memory(int cIdx, int tbSize);
	89
	90	static void* operator new(const size_t size) { return mMemPool.new_obj(size); }
	91	static void operator delete(void* obj) { mMemPool.delete_obj(obj); }
	92
	93	private:
	94	static alloc_pool mMemPool;
	95
	96	void reconstruct_tb(encoder_context* ectx,
	97	de265_image* img, int x0,int y0, int log2TbSize,
	98	const enc_cb* cb, int cIdx) const;
	99	};
	100
	101
	102	struct enc_pb_inter
	103	{
	104	/* absolute motion information (for MV-prediction candidates)
	105	*/
	106	MotionVectorSpec motion;
	107
	108	/* specification how to code the motion vector in the bitstream
	109	*/
	110	motion_spec spec;
	111
	112
	113	// NOT TRUE: refIdx in 'spec' is not used. It is taken from 'motion'
	114	// Currently, information is duplicated. Same as with inter_pred_idc/predFlag[].
	115
	116	/* SPEC:
	117	int8_t refIdx[2]; // not used
	118	int16_t mvd[2][2];
	119
	120	uint8_t inter_pred_idc : 2; // enum InterPredIdc
	121	uint8_t mvp_l0_flag : 1;
	122	uint8_t mvp_l1_flag : 1;
	123	uint8_t merge_flag : 1;
	124	uint8_t merge_idx : 3;
	125	*/
	126	};
	127
	128
	129	class enc_cb : public enc_node
	130	{
	131	public:
	132	enc_cb();
	133	~enc_cb();
	134
	135	const enc_cb* parent;
	136
	137	uint8_t split_cu_flag : 1;
	138	uint8_t ctDepth : 2;
	139
	140	union {
	141	// split
	142	struct {
	143	enc_cb* children[4]; // undefined when split_cu_flag==false
	144	};
	145
	146	// non-split
	147	struct {
	148	uint8_t qp : 6;
	149	uint8_t cu_transquant_bypass_flag : 1; // currently unused
	150	uint8_t pcm_flag : 1;
	151
	152	enum PredMode PredMode; // : 6;
	153	enum PartMode PartMode; // : 3;
	154
	155	union {
	156	struct {
	157	enum IntraPredMode pred_mode[4];
	158	enum IntraPredMode chroma_mode;
	159	} intra;
	160
	161	struct {
	162	enc_pb_inter pb[4];
	163
	164	uint8_t rqt_root_cbf : 1;
	165	} inter;
	166	};
	167
	168	const enc_tb* transform_tree;
	169	};
	170	};
	171
	172
	173	float distortion;
	174	float rate;
	175
	176
	177	void set_rqt_root_bf_from_children_cbf();
	178
	179	/* Save CB reconstruction in the node and restore it again to the image.
	180	Pixel data and metadata.
	181	*/
	182	virtual void save(const de265_image*);
	183	virtual void restore(de265_image*);
	184
	185
	186	/* Decode this CB: pixel data and write metadata to image.
	187	*/
	188	void reconstruct(encoder_context* ectx,de265_image* img) const;
	189
	190
	191	// memory management
	192
	193	static void* operator new(const size_t size) { return mMemPool.new_obj(size); }
	194	static void operator delete(void* obj) { mMemPool.delete_obj(obj); }
	195
	196	private:
	197	void write_to_image(de265_image*) const;
	198
	199	static alloc_pool mMemPool;
	200	};
	201
	202
	203
	204	inline int childX(int x0, int idx, int log2CbSize)
	205	{
	206	return x0 + ((idx&1) << (log2CbSize-1));
	207	}
	208
	209	inline int childY(int y0, int idx, int log2CbSize)
	210	{
	211	return y0 + ((idx>>1) << (log2CbSize-1));
	212	}
	213
	214
	215
	216	void encode_split_cu_flag(encoder_context* ectx,
	217	CABAC_encoder* cabac,
	218	int x0, int y0, int ctDepth, int split_flag);
	219
	220	void encode_transform_tree(encoder_context* ectx,
	221	CABAC_encoder* cabac,
	222	const enc_tb* tb, const enc_cb* cb,
	223	int x0,int y0, int xBase,int yBase,
	224	int log2TrafoSize, int trafoDepth, int blkIdx,
	225	int MaxTrafoDepth, int IntraSplitFlag, bool recurse);
	226
	227	void encode_coding_unit(encoder_context* ectx,
	228	CABAC_encoder* cabac,
	229	const enc_cb* cb, int x0,int y0, int log2CbSize, bool recurse);
	230
	231	/* returns
	232	1 - forced split
	233	0 - forced non-split
	234	-1 - optional split
	235	*/
	236	enum SplitType {
	237	ForcedNonSplit = 0,
	238	ForcedSplit = 1,
	239	OptionalSplit = 2
	240	};
	241
	242	SplitType get_split_type(const seq_parameter_set* sps,
	243	int x0,int y0, int log2CbSize);
	244
	245
	246	void encode_split_transform_flag(encoder_context* ectx,
	247	CABAC_encoder* cabac,
	248	int log2TrafoSize, int split_flag);
	249
	250	void encode_merge_idx(encoder_context* ectx,
	251	CABAC_encoder* cabac,
	252	int mergeIdx);
	253
	254	void encode_cu_skip_flag(encoder_context* ectx,
	255	CABAC_encoder* cabac,
	256	const enc_cb* cb,
	257	bool skip);
	258
	259	void encode_cbf_luma(CABAC_encoder* cabac,
	260	bool zeroTrafoDepth, int cbf_luma);
	261
	262	void encode_cbf_chroma(CABAC_encoder* cabac,
	263	int trafoDepth, int cbf_chroma);
	264
	265	void encode_transform_unit(encoder_context* ectx,
	266	CABAC_encoder* cabac,
	267	const enc_tb* tb, const enc_cb* cb,
	268	int x0,int y0, int xBase,int yBase,
	269	int log2TrafoSize, int trafoDepth, int blkIdx);
	270
	271
	272	void encode_quadtree(encoder_context* ectx,
	273	CABAC_encoder* cabac,
	274	const enc_cb* cb, int x0,int y0, int log2CbSize, int ctDepth,
	275	bool recurse);
	276
	277	void encode_ctb(encoder_context* ectx,
	278	CABAC_encoder* cabac,
	279	enc_cb* cb, int ctbX,int ctbY);
	280
	281
	282	class de265_encoder
	283	{
	284	public:
	285	virtual ~de265_encoder() { }
	286	};
	287
	288	#endif

+294

-0

libde265/encoder/encoder-context.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "encoder/encoder-context.h"
	23	#include "encoder/analyze.h"
	24	#include "libde265/util.h"
	25
	26	#include <math.h>
	27
	28
	29	encoder_context::encoder_context()
	30	{
	31	encoder_started=false;
	32
	33	//img_source = NULL;
	34	//reconstruction_sink = NULL;
	35	//packet_sink = NULL;
	36
	37	image_spec_is_defined = false;
	38	parameters_have_been_set = false;
	39	headers_have_been_sent = false;
	40
	41	param_image_allocation_userdata = NULL;
	42	release_func = NULL;
	43
	44	use_adaptive_context = true; //false;
	45
	46	//enc_coeff_pool.set_blk_size(646420); // TODO: this a guess
	47
	48	//switch_CABAC_to_bitstream();
	49
	50
	51	params.registerParams(params_config);
	52	algo.registerParams(params_config);
	53	}
	54
	55
	56	encoder_context::~encoder_context()
	57	{
	58	while (!output_packets.empty()) {
	59	en265_free_packet(this, output_packets.front());
	60	output_packets.pop_front();
	61	}
	62	}
	63
	64
	65	void encoder_context::start_encoder()
	66	{
	67	if (encoder_started) {
	68	return;
	69	}
	70
	71
	72	if (params.sop_structure() == SOP_Intra) {
	73	sop = std::shared_ptr<sop_creator_intra_only>(new sop_creator_intra_only());
	74	}
	75	else {
	76	auto s = std::shared_ptr<sop_creator_trivial_low_delay>(new sop_creator_trivial_low_delay());
	77	s->setParams(params.mSOP_LowDelay);
	78	sop = s;
	79	}
	80
	81	sop->set_encoder_context(this);
	82	sop->set_encoder_picture_buffer(&picbuf);
	83
	84
	85	encoder_started=true;
	86	}
	87
	88
	89	en265_packet* encoder_context::create_packet(en265_packet_content_type t)
	90	{
	91	en265_packet* pck = new en265_packet;
	92
	93	uint8_t* data = new uint8_t[cabac_encoder.size()];
	94	memcpy(data, cabac_encoder.data(), cabac_encoder.size());
	95
	96	pck->version = 1;
	97
	98	pck->data = data;
	99	pck->length = cabac_encoder.size();
	100
	101	pck->frame_number = -1;
	102	pck->content_type = t;
	103	pck->complete_picture = 0;
	104	pck->final_slice = 0;
	105	pck->dependent_slice = 0;
	106	//pck->pts = 0;
	107	//pck->user_data = NULL;
	108	pck->nuh_layer_id = 0;
	109	pck->nuh_temporal_id = 0;
	110
	111	pck->encoder_context = this;
	112
	113	pck->input_image = NULL;
	114	pck->reconstruction = NULL;
	115
	116	cabac_encoder.reset();
	117
	118	return pck;
	119	}
	120
	121
	122	de265_error encoder_context::encode_headers()
	123	{
	124	nal_header nal;
	125
	126	// VPS
	127
	128	vps.set_defaults(Profile_Main, 6,2);
	129
	130
	131	// SPS
	132
	133	sps.set_defaults();
	134	sps.set_CB_log2size_range( Log2(params.min_cb_size), Log2(params.max_cb_size));
	135	sps.set_TB_log2size_range( Log2(params.min_tb_size), Log2(params.max_tb_size));
	136	sps.max_transform_hierarchy_depth_intra = params.max_transform_hierarchy_depth_intra;
	137
	138	sps.set_resolution(image_width, image_height);
	139	sop->set_SPS_header_values();
	140	sps.compute_derived_values();
	141
	142
	143	// PPS
	144
	145	pps.set_defaults();
	146	pps.pic_init_qp = algo.getPPS_QP();
	147
	148	// turn off deblocking filter
	149	pps.deblocking_filter_control_present_flag = true;
	150	pps.deblocking_filter_override_enabled_flag = false;
	151	pps.pic_disable_deblocking_filter_flag = true;
	152	pps.pps_loop_filter_across_slices_enabled_flag = false;
	153
	154	pps.set_derived_values(&sps);
	155
	156
	157
	158	// write headers
	159
	160	en265_packet* pck;
	161
	162	nal.set(NAL_UNIT_VPS_NUT);
	163	nal.write(cabac_encoder);
	164	vps.write(this, cabac_encoder);
	165	cabac_encoder.add_trailing_bits();
	166	cabac_encoder.flush_VLC();
	167	pck = create_packet(EN265_PACKET_VPS);
	168	pck->nal_unit_type = EN265_NUT_VPS;
	169	output_packets.push_back(pck);
	170
	171	nal.set(NAL_UNIT_SPS_NUT);
	172	nal.write(cabac_encoder);
	173	sps.write(this, cabac_encoder);
	174	cabac_encoder.add_trailing_bits();
	175	cabac_encoder.flush_VLC();
	176	pck = create_packet(EN265_PACKET_SPS);
	177	pck->nal_unit_type = EN265_NUT_SPS;
	178	output_packets.push_back(pck);
	179
	180	nal.set(NAL_UNIT_PPS_NUT);
	181	nal.write(cabac_encoder);
	182	pps.write(this, cabac_encoder, &sps);
	183	cabac_encoder.add_trailing_bits();
	184	cabac_encoder.flush_VLC();
	185	pck = create_packet(EN265_PACKET_PPS);
	186	pck->nal_unit_type = EN265_NUT_PPS;
	187	output_packets.push_back(pck);
	188
	189
	190
	191	headers_have_been_sent = true;
	192
	193	return DE265_OK;
	194	}
	195
	196
	197	de265_error encoder_context::encode_picture_from_input_buffer()
	198	{
	199	if (!picbuf.have_more_frames_to_encode()) {
	200	return DE265_OK;
	201	}
	202
	203
	204	if (!image_spec_is_defined) {
	205	const image_data* id = picbuf.peek_next_picture_to_encode();
	206	image_width = id->input->get_width();
	207	image_height = id->input->get_height();
	208	image_spec_is_defined = true;
	209	}
	210
	211
	212	if (!parameters_have_been_set) {
	213	algo.setParams(params);
	214
	215
	216	// TODO: must be <30, because Y->C mapping (tab8_22) is not implemented yet
	217	int qp = algo.getPPS_QP();
	218
	219	//lambda = ectx->params.lambda;
	220	lambda = 0.0242 * pow(1.27245, qp);
	221
	222	parameters_have_been_set = true;
	223	}
	224
	225
	226	if (!headers_have_been_sent) {
	227	encode_headers();
	228	}
	229
	230
	231
	232
	233
	234	image_data* imgdata;
	235	imgdata = picbuf.get_next_picture_to_encode();
	236	assert(imgdata);
	237	picbuf.mark_encoding_started(imgdata->frame_number);
	238
	239	this->imgdata = imgdata;
	240	this->shdr = &imgdata->shdr;
	241	loginfo(LogEncoder,"encoding frame %d\n",imgdata->frame_number);
	242
	243
	244	// write slice header
	245
	246	// slice
	247
	248	imgdata->shdr.slice_deblocking_filter_disabled_flag = true;
	249	imgdata->shdr.slice_loop_filter_across_slices_enabled_flag = false;
	250	imgdata->shdr.compute_derived_values(&pps);
	251
	252	//shdr.slice_pic_order_cnt_lsb = poc & 0xFF;
	253
	254	imgdata->nal.write(cabac_encoder);
	255	imgdata->shdr.write(this, cabac_encoder, &sps, &pps, imgdata->nal.nal_unit_type);
	256	cabac_encoder.add_trailing_bits();
	257	cabac_encoder.flush_VLC();
	258
	259
	260	// encode image
	261
	262	cabac_encoder.init_CABAC();
	263	double psnr = encode_image(this,imgdata->input, algo);
	264	loginfo(LogEncoder," PSNR-Y: %f\n", psnr);
	265	cabac_encoder.flush_CABAC();
	266	cabac_encoder.add_trailing_bits();
	267	cabac_encoder.flush_VLC();
	268
	269
	270	// set reconstruction image
	271
	272	picbuf.set_reconstruction_image(imgdata->frame_number, img);
	273	//picbuf.set_prediction_image(imgdata->frame_number, prediction);
	274	img=NULL;
	275	this->imgdata = NULL;
	276	this->shdr = NULL;
	277
	278	// build output packet
	279
	280	en265_packet* pck = create_packet(EN265_PACKET_SLICE);
	281	pck->input_image = imgdata->input;
	282	pck->reconstruction = imgdata->reconstruction;
	283	pck->frame_number = imgdata->frame_number;
	284	pck->nal_unit_type = (enum en265_nal_unit_type)imgdata->nal.nal_unit_type;
	285	pck->nuh_layer_id = imgdata->nal.nuh_layer_id;
	286	pck->nuh_temporal_id= imgdata->nal.nuh_temporal_id;
	287	output_packets.push_back(pck);
	288
	289
	290	picbuf.mark_encoding_finished(imgdata->frame_number);
	291
	292	return DE265_OK;
	293	}

+157

-0

libde265/encoder/encoder-context.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef ENCODER_CONTEXT_H
	23	#define ENCODER_CONTEXT_H
	24
	25	#include "libde265/image.h"
	26	#include "libde265/decctx.h"
	27	#include "libde265/image-io.h"
	28	#include "libde265/encoder/encoder-params.h"
	29	#include "libde265/encoder/encpicbuf.h"
	30	#include "libde265/encoder/sop.h"
	31	#include "libde265/en265.h"
	32	#include "libde265/util.h"
	33
	34	#include <memory>
	35
	36
	37	class encoder_context : public base_context
	38	{
	39	public:
	40	encoder_context();
	41	~encoder_context();
	42
	43	virtual const de265_image* get_image(int frame_id) const {
	44	return picbuf.get_picture(frame_id)->reconstruction;
	45	}
	46
	47	virtual bool has_image(int frame_id) const {
	48	return picbuf.has_picture(frame_id);
	49	}
	50
	51	bool encoder_started;
	52
	53	encoder_params params;
	54	config_parameters params_config;
	55
	56	EncodingAlgorithm_Custom algo;
	57
	58	int image_width, image_height;
	59	bool image_spec_is_defined; // whether we know the input image size
	60
	61	void* param_image_allocation_userdata;
	62	void (release_func)(en265_encoder_context,
	63	de265_image*,
	64	void* userdata);
	65
	66	//error_queue errqueue;
	67	//acceleration_functions accel;
	68
	69	// quick links
	70	de265_image* img; // reconstruction
	71	de265_image* prediction;
	72	image_data* imgdata; // input image
	73	slice_segment_header* shdr;
	74
	75	// temporary memory for motion compensated pixels (when CB-algo passes this down to TB-algo)
	76	//uint8_t prediction[3][64*64]; // stride: 1<<(cb->log2Size)
	77	//int prediction_x0,prediction_y0;
	78
	79
	80	int active_qp; // currently active QP
	81	/int target_qp;/ /* QP we want to code at.
	82	(Not actually the real QP. Check image.get_QPY() for that.) */
	83
	84	video_parameter_set vps;
	85	seq_parameter_set sps;
	86	pic_parameter_set pps;
	87	//slice_segment_header shdr;
	88
	89	bool parameters_have_been_set;
	90	bool headers_have_been_sent;
	91
	92	encoder_picture_buffer picbuf;
	93	std::shared_ptr<sop_creator> sop;
	94
	95	std::deque<en265_packet*> output_packets;
	96
	97
	98	// --- rate-control ---
	99
	100	float lambda;
	101
	102
	103	// --- CABAC output and rate estimation ---
	104
	105	//CABAC_encoder* cabac; // currently active CABAC output (estim or bitstream)
	106	//context_model_table2* ctx_model; // currently active ctx models (estim or bitstream)
	107
	108	// CABAC bitstream writer
	109	CABAC_encoder_bitstream cabac_encoder;
	110	context_model_table cabac_ctx_models;
	111
	112	//std::shared_ptr<CABAC_encoder> cabac_estim;
	113
	114	bool use_adaptive_context;
	115
	116
	117	/*** TODO: CABAC_encoder direkt an encode-Funktion übergeben, anstatt hier
	118	aussenrum zwischenzuspeichern (mit undefinierter Lifetime).
	119	Das Context-Model kann dann gleich mit in den Encoder rein cabac_encoder(ctxtable).
	120	write_bits() wird dann mit dem context-index aufgerufen, nicht mit dem model direkt.
	121	***/
	122
	123
	124	/*
	125	void switch_CABAC(context_model_table2* model) {
	126	cabac = cabac_estim.get();
	127	ctx_model = model;
	128	}
	129
	130	void switch_CABAC_to_bitstream() {
	131	cabac = &cabac_bitstream;
	132	ctx_model = &ctx_model_bitstream;
	133	}
	134	*/
	135
	136	en265_packet* create_packet(en265_packet_content_type t);
	137
	138
	139	// --- encoding control ---
	140
	141	void start_encoder();
	142	de265_error encode_headers();
	143	de265_error encode_picture_from_input_buffer();
	144
	145
	146	// Input images can be released after encoding and when the output packet is released.
	147	// This is important to do as soon as possible, as the image might actually wrap
	148	// scarce resources like camera picture buffers.
	149	// This function does release (only) the raw input data.
	150	void release_input_image(int frame_number) { picbuf.release_input_image(frame_number); }
	151
	152	void mark_image_is_outputted(int frame_number) { picbuf.mark_image_is_outputted(frame_number); }
	153	};
	154
	155
	156	#endif

+75

-0

libde265/encoder/encoder-params.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "encoder-params.h"
	23
	24
	25
	26	static std::vector<int> power2range(int low,int high)
	27	{
	28	std::vector<int> vals;
	29	for (int i=low; i<=high; i*=2)
	30	vals.push_back(i);
	31	return vals;
	32	}
	33
	34	encoder_params::encoder_params()
	35	{
	36	//rateControlMethod = RateControlMethod_ConstantQP;
	37
	38	min_cb_size.set_ID("min-cb-size"); min_cb_size.set_valid_values(power2range(8,64)); min_cb_size.set_default(8);
	39	max_cb_size.set_ID("max-cb-size"); max_cb_size.set_valid_values(power2range(8,64)); max_cb_size.set_default(32);
	40	min_tb_size.set_ID("min-tb-size"); min_tb_size.set_valid_values(power2range(4,32)); min_tb_size.set_default(4);
	41	max_tb_size.set_ID("max-tb-size"); max_tb_size.set_valid_values(power2range(8,32)); max_tb_size.set_default(32);
	42
	43	max_transform_hierarchy_depth_intra.set_ID("max-transform-hierarchy-depth-intra");
	44	max_transform_hierarchy_depth_intra.set_range(0,4);
	45	max_transform_hierarchy_depth_intra.set_default(3);
	46
	47	sop_structure.set_ID("sop-structure");
	48
	49	mAlgo_TB_IntraPredMode.set_ID("TB-IntraPredMode");
	50	mAlgo_TB_IntraPredMode_Subset.set_ID("TB-IntraPredMode-subset");
	51	mAlgo_CB_IntraPartMode.set_ID("CB-IntraPartMode");
	52
	53	mAlgo_MEMode.set_ID("MEMode");
	54	}
	55
	56
	57	void encoder_params::registerParams(config_parameters& config)
	58	{
	59	config.add_option(&min_cb_size);
	60	config.add_option(&max_cb_size);
	61	config.add_option(&min_tb_size);
	62	config.add_option(&max_tb_size);
	63	config.add_option(&max_transform_hierarchy_depth_intra);
	64
	65	config.add_option(&sop_structure);
	66
	67	config.add_option(&mAlgo_TB_IntraPredMode);
	68	config.add_option(&mAlgo_TB_IntraPredMode_Subset);
	69	config.add_option(&mAlgo_CB_IntraPartMode);
	70
	71	config.add_option(&mAlgo_MEMode);
	72
	73	mSOP_LowDelay.registerParams(config);
	74	}

+141

-0

libde265/encoder/encoder-params.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef ENCODER_PARAMS_H
	23	#define ENCODER_PARAMS_H
	24
	25	#include "libde265/encoder/encode.h"
	26	#include "libde265/encoder/analyze.h"
	27	#include "libde265/encoder/sop.h"
	28
	29
	30	enum RateControlMethod
	31	{
	32	RateControlMethod_ConstantQP,
	33	RateControlMethod_ConstantLambda
	34	};
	35
	36	enum IntraPredSearch
	37	{
	38	IntraPredSearch_Complete
	39	};
	40
	41
	42	enum SOP_Structure
	43	{
	44	SOP_Intra,
	45	SOP_LowDelay
	46	};
	47
	48	class option_SOP_Structure : public choice_option<enum SOP_Structure>
	49	{
	50	public:
	51	option_SOP_Structure() {
	52	add_choice("intra", SOP_Intra);
	53	add_choice("low-delay", SOP_LowDelay, true);
	54	}
	55	};
	56
	57
	58	enum MEMode
	59	{
	60	MEMode_Test,
	61	MEMode_Search
	62	};
	63
	64	class option_MEMode : public choice_option<enum MEMode>
	65	{
	66	public:
	67	option_MEMode() {
	68	add_choice("test", MEMode_Test, true);
	69	add_choice("search", MEMode_Search);
	70	}
	71	};
	72
	73
	74	struct encoder_params
	75	{
	76	encoder_params();
	77
	78	void registerParams(config_parameters& config);
	79
	80
	81	// CB quad-tree
	82
	83	option_int min_cb_size;
	84	option_int max_cb_size;
	85
	86	option_int min_tb_size;
	87	option_int max_tb_size;
	88
	89	option_int max_transform_hierarchy_depth_intra;
	90
	91
	92	option_SOP_Structure sop_structure;
	93
	94	sop_creator_trivial_low_delay::params mSOP_LowDelay;
	95
	96
	97	// --- Algo_TB_IntraPredMode
	98
	99	option_ALGO_TB_IntraPredMode mAlgo_TB_IntraPredMode;
	100	option_ALGO_TB_IntraPredMode_Subset mAlgo_TB_IntraPredMode_Subset;
	101
	102	//Algo_TB_IntraPredMode_FastBrute::params TB_IntraPredMode_FastBrute;
	103	//Algo_TB_IntraPredMode_MinResidual::params TB_IntraPredMode_MinResidual;
	104
	105
	106	// --- Algo_TB_Split_BruteForce
	107
	108	//Algo_TB_Split_BruteForce::params TB_Split_BruteForce;
	109
	110
	111	// --- Algo_CB_IntraPartMode
	112
	113	option_ALGO_CB_IntraPartMode mAlgo_CB_IntraPartMode;
	114
	115	//Algo_CB_IntraPartMode_Fixed::params CB_IntraPartMode_Fixed;
	116
	117	// --- Algo_CB_Split
	118
	119	// --- Algo_CTB_QScale
	120
	121	//Algo_CTB_QScale_Constant::params CTB_QScale_Constant;
	122
	123	option_MEMode mAlgo_MEMode;
	124
	125
	126	// intra-prediction
	127
	128	enum IntraPredSearch intraPredSearch;
	129
	130
	131	// rate-control
	132
	133	enum RateControlMethod rateControlMethod;
	134
	135	//int constant_QP;
	136	//int lambda;
	137	};
	138
	139
	140	#endif

+321

-0

libde265/encoder/encpicbuf.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "libde265/encoder/encpicbuf.h"
	21	#include "libde265/util.h"
	22
	23
	24	encoder_picture_buffer::encoder_picture_buffer()
	25	{
	26	}
	27
	28	encoder_picture_buffer::~encoder_picture_buffer()
	29	{
	30	flush_images();
	31	}
	32
	33
	34	image_data::image_data()
	35	{
	36	//printf("new %p\n",this);
	37
	38	frame_number = 0;
	39
	40	input = NULL;
	41	prediction = NULL;
	42	reconstruction = NULL;
	43
	44	// SOP metadata
	45
	46	sps_index = -1;
	47	skip_priority = 0;
	48	is_intra = true;
	49
	50	state = state_unprocessed;
	51
	52	is_in_output_queue = true;
	53	}
	54
	55	image_data::~image_data()
	56	{
	57	//printf("delete %p\n",this);
	58
	59	delete input;
	60	// TODO: this could still be referenced in the packet output queue, so the
	61	// images should really be refcounted. release for now to prevent leaks
	62	delete reconstruction;
	63	delete prediction;
	64	}
	65
	66
	67	// --- input pushed by the input process ---
	68
	69	void encoder_picture_buffer::reset()
	70	{
	71	flush_images();
	72
	73	mEndOfStream = false;
	74	}
	75
	76
	77	void encoder_picture_buffer::flush_images()
	78	{
	79	while (!mImages.empty()) {
	80	delete mImages.front();
	81	mImages.pop_front();
	82	}
	83	}
	84
	85
	86	image_data* encoder_picture_buffer::insert_next_image_in_encoding_order(const de265_image* img,
	87	int frame_number)
	88	{
	89	image_data* data = new image_data();
	90	data->frame_number = frame_number;
	91	data->input = img;
	92	data->shdr.set_defaults();
	93
	94	mImages.push_back(data);
	95
	96	return data;
	97	}
	98
	99	void encoder_picture_buffer::insert_end_of_stream()
	100	{
	101	mEndOfStream = true;
	102	}
	103
	104
	105	// --- SOP structure ---
	106
	107	void image_data::set_intra()
	108	{
	109	is_intra = true;
	110	}
	111
	112	void image_data::set_NAL_type(uint8_t nalType)
	113	{
	114	nal.nal_unit_type = nalType;
	115	}
	116
	117	void image_data::set_references(int sps_index, // -1 -> custom
	118	const std::vector<int>& l0,
	119	const std::vector<int>& l1,
	120	const std::vector<int>& lt,
	121	const std::vector<int>& keepMoreReferences)
	122	{
	123	this->sps_index = sps_index;
	124	ref0 = l0;
	125	ref1 = l1;
	126	longterm = lt;
	127	keep = keepMoreReferences;
	128
	129
	130	// TODO: pps.num_ref_idx_l0_default_active
	131
	132	shdr.num_ref_idx_l0_active = l0.size();
	133	//shdr.num_ref_idx_l1_active = l1.size();
	134
	135	assert(l0.size() < MAX_NUM_REF_PICS);
	136	for (int i=0;i<l0.size();i++) {
	137	shdr.RefPicList[0][i] = l0[i];
	138	}
	139
	140	/*
	141	assert(l1.size() < MAX_NUM_REF_PICS);
	142	for (int i=0;i<l1.size();i++) {
	143	shdr.RefPicList[1][i] = l1[i];
	144	}
	145	*/
	146	}
	147
	148	void image_data::set_NAL_temporal_id(int temporal_id)
	149	{
	150	this->nal.nuh_temporal_id = temporal_id;
	151	}
	152
	153	void image_data::set_skip_priority(int skip_priority)
	154	{
	155	this->skip_priority = skip_priority;
	156	}
	157
	158	void encoder_picture_buffer::sop_metadata_commit(int frame_number)
	159	{
	160	image_data* data = mImages.back();
	161	assert(data->frame_number == frame_number);
	162
	163	data->state = image_data::state_sop_metadata_available;
	164	}
	165
	166
	167
	168	// --- infos pushed by encoder ---
	169
	170	void encoder_picture_buffer::mark_encoding_started(int frame_number)
	171	{
	172	image_data* data = get_picture(frame_number);
	173
	174	data->state = image_data::state_encoding;
	175	}
	176
	177	void encoder_picture_buffer::set_prediction_image(int frame_number, de265_image* pred)
	178	{
	179	image_data* data = get_picture(frame_number);
	180
	181	data->prediction = pred;
	182	}
	183
	184	void encoder_picture_buffer::set_reconstruction_image(int frame_number, de265_image* reco)
	185	{
	186	image_data* data = get_picture(frame_number);
	187
	188	data->reconstruction = reco;
	189	}
	190
	191	void encoder_picture_buffer::mark_encoding_finished(int frame_number)
	192	{
	193	image_data* data = get_picture(frame_number);
	194
	195	data->state = image_data::state_keep_for_reference;
	196
	197
	198	// --- delete images that are not required anymore ---
	199
	200	// first, mark all images unused
	201
	202	#ifdef FOR_LOOP_AUTO_SUPPORT
	203	FOR_LOOP(auto, imgdata, mImages) {
	204	#else
	205	FOR_LOOP(image_data *, imgdata, mImages) {
	206	#endif
	207	imgdata->mark_used = false;
	208	}
	209
	210	// mark all images that will be used later
	211
	212	FOR_LOOP(int, f, data->ref0) { get_picture(f)->mark_used=true; }
	213	FOR_LOOP(int, f, data->ref1) { get_picture(f)->mark_used=true; }
	214	FOR_LOOP(int, f, data->longterm) { get_picture(f)->mark_used=true; }
	215	FOR_LOOP(int, f, data->keep) { get_picture(f)->mark_used=true; }
	216	data->mark_used=true;
	217
	218	// copy over all images that we still keep
	219
	220	std::deque<image_data*> newImageSet;
	221	#ifdef FOR_LOOP_AUTO_SUPPORT
	222	FOR_LOOP(auto, imgdata, mImages) {
	223	#else
	224	FOR_LOOP(image_data *, imgdata, mImages) {
	225	#endif
	226	if (imgdata->mark_used \|\| imgdata->is_in_output_queue) {
	227	imgdata->reconstruction->PicState = UsedForShortTermReference; // TODO: this is only a hack
	228
	229	newImageSet.push_back(imgdata);
	230	}
	231	else {
	232	// image is not needed anymore for reference, remove it from EncPicBuf
	233
	234	delete imgdata;
	235	}
	236	}
	237
	238	mImages = newImageSet;
	239	}
	240
	241
	242
	243	// --- data access ---
	244
	245	bool encoder_picture_buffer::have_more_frames_to_encode() const
	246	{
	247	for (int i=0;i<mImages.size();i++) {
	248	if (mImages[i]->state < image_data::state_encoding) {
	249	return true;
	250	}
	251	}
	252
	253	return false;
	254	}
	255
	256
	257	image_data* encoder_picture_buffer::get_next_picture_to_encode()
	258	{
	259	for (int i=0;i<mImages.size();i++) {
	260	if (mImages[i]->state < image_data::state_encoding) {
	261	return mImages[i];
	262	}
	263	}
	264
	265	return NULL;
	266	}
	267
	268
	269	const image_data* encoder_picture_buffer::get_picture(int frame_number) const
	270	{
	271	for (int i=0;i<mImages.size();i++) {
	272	if (mImages[i]->frame_number == frame_number)
	273	return mImages[i];
	274	}
	275
	276	assert(false);
	277	return NULL;
	278	}
	279
	280
	281	image_data* encoder_picture_buffer::get_picture(int frame_number)
	282	{
	283	for (int i=0;i<mImages.size();i++) {
	284	if (mImages[i]->frame_number == frame_number)
	285	return mImages[i];
	286	}
	287
	288	assert(false);
	289	return NULL;
	290	}
	291
	292
	293	bool encoder_picture_buffer::has_picture(int frame_number) const
	294	{
	295	for (int i=0;i<mImages.size();i++) {
	296	if (mImages[i]->frame_number == frame_number)
	297	return true;
	298	}
	299
	300	return false;
	301	}
	302
	303
	304	void encoder_picture_buffer::mark_image_is_outputted(int frame_number)
	305	{
	306	image_data* idata = get_picture(frame_number);
	307	assert(idata);
	308
	309	idata->is_in_output_queue = false;
	310	}
	311
	312
	313	void encoder_picture_buffer::release_input_image(int frame_number)
	314	{
	315	image_data* idata = get_picture(frame_number);
	316	assert(idata);
	317
	318	delete idata->input;
	319	idata->input = NULL;
	320	}

+144

-0

libde265/encoder/encpicbuf.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifndef DE265_ENCPICBUF_H
	21	#define DE265_ENCPICBUF_H
	22
	23	#include "libde265/image.h"
	24	#include "libde265/sps.h"
	25
	26	#include <deque>
	27	#include <vector>
	28
	29
	30	/* TODO: we need a way to quickly access pictures with a stable ID, like in the DPB.
	31	*/
	32
	33	struct image_data
	34	{
	35	image_data();
	36	~image_data();
	37
	38	int frame_number;
	39
	40	const de265_image* input; // owner
	41	de265_image* prediction; // owner
	42	de265_image* reconstruction; // owner
	43
	44	// SOP metadata
	45
	46	nal_header nal; // TODO: image split into several NALs (always same NAL header?)
	47
	48	slice_segment_header shdr; // TODO: multi-slice pictures
	49
	50	std::vector<int> ref0;
	51	std::vector<int> ref1;
	52	std::vector<int> longterm;
	53	std::vector<int> keep;
	54	int sps_index;
	55	int skip_priority;
	56	bool is_intra; // TODO: remove, use shdr.slice_type instead
	57
	58	/* unprocessed only input image has been inserted, no metadata
	59	sop_metadata_available sop-creator has filled in references and skipping metadata
	60	a) encoding encoding started for this frame, reconstruction image was created
	61	. keep_for_reference encoding finished, picture is kept in the buffer for reference
	62	b) skipped image was skipped, no encoding was done, no reconstruction image
	63	*/
	64	enum state {
	65	state_unprocessed,
	66	state_sop_metadata_available,
	67	state_encoding,
	68	state_keep_for_reference,
	69	state_skipped
	70	} state;
	71
	72	bool is_in_output_queue;
	73
	74	bool mark_used;
	75
	76
	77	// --- SOP structure ---
	78
	79	void set_intra();
	80	void set_NAL_type(uint8_t nalType);
	81	void set_NAL_temporal_id(int temporal_id);
	82	void set_references(int sps_index, // -1 -> custom
	83	const std::vector<int>& l0, const std::vector<int>& l1,
	84	const std::vector<int>& lt,
	85	const std::vector<int>& keepMoreReferences);
	86	void set_skip_priority(int skip_priority);
	87	};
	88
	89
	90	class encoder_picture_buffer
	91	{
	92	public:
	93	encoder_picture_buffer();
	94	~encoder_picture_buffer();
	95
	96
	97	// --- input pushed by the input process ---
	98
	99	void reset();
	100
	101	image_data* insert_next_image_in_encoding_order(const de265_image*, int frame_number);
	102	void insert_end_of_stream();
	103
	104
	105	// --- SOP structure ---
	106
	107	void sop_metadata_commit(int frame_number); // note: frame_number is only for consistency checking
	108
	109
	110	// --- infos pushed by encoder ---
	111
	112	void mark_encoding_started(int frame_number);
	113	void set_prediction_image(int frame_number, de265_image*); // store it just for debugging fun
	114	void set_reconstruction_image(int frame_number, de265_image*);
	115	void mark_encoding_finished(int frame_number);
	116
	117
	118
	119	// --- data access ---
	120
	121	bool have_more_frames_to_encode() const;
	122	image_data* get_next_picture_to_encode(); // or return NULL if no picture is available
	123	const image_data* get_picture(int frame_number) const;
	124	bool has_picture(int frame_number) const;
	125
	126	const image_data* peek_next_picture_to_encode() const {
	127	assert(!mImages.empty());
	128	return mImages.front();
	129	}
	130
	131	void mark_image_is_outputted(int frame_number);
	132	void release_input_image(int frame_number);
	133
	134	private:
	135	bool mEndOfStream;
	136	std::deque<image_data*> mImages;
	137
	138	void flush_images();
	139	image_data* get_picture(int frame_number);
	140	};
	141
	142
	143	#endif

+106

-0

libde265/encoder/sop.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "libde265/encoder/sop.h"
	21	#include "libde265/encoder/encoder-context.h"
	22
	23
	24	sop_creator_intra_only::sop_creator_intra_only()
	25	{
	26	}
	27
	28
	29	void sop_creator_intra_only::set_SPS_header_values()
	30	{
	31	mEncCtx->sps.log2_max_pic_order_cnt_lsb = get_num_poc_lsb_bits();
	32	}
	33
	34
	35	void sop_creator_intra_only::insert_new_input_image(de265_image* img)
	36	{
	37	img->PicOrderCntVal = get_pic_order_count();
	38
	39	reset_poc();
	40	int poc = get_pic_order_count();
	41
	42	assert(mEncPicBuf);
	43	image_data* imgdata = mEncPicBuf->insert_next_image_in_encoding_order(img, get_frame_number());
	44
	45	imgdata->set_intra();
	46	imgdata->set_NAL_type(NAL_UNIT_IDR_N_LP);
	47	imgdata->shdr.slice_type = SLICE_TYPE_I;
	48	imgdata->shdr.slice_pic_order_cnt_lsb = get_pic_order_count_lsb();
	49
	50	mEncPicBuf->sop_metadata_commit(get_frame_number());
	51
	52	advance_frame();
	53	}
	54
	55
	56	// ---------------------------------------------------------------------------
	57
	58
	59	sop_creator_trivial_low_delay::sop_creator_trivial_low_delay()
	60	{
	61	}
	62
	63
	64	void sop_creator_trivial_low_delay::set_SPS_header_values()
	65	{
	66	ref_pic_set rps;
	67	rps.DeltaPocS0[0] = -1;
	68	rps.UsedByCurrPicS0[0] = true;
	69	rps.NumNegativePics = 1;
	70	rps.NumPositivePics = 0;
	71	rps.compute_derived_values();
	72	mEncCtx->sps.ref_pic_sets.push_back(rps);
	73	mEncCtx->sps.log2_max_pic_order_cnt_lsb = get_num_poc_lsb_bits();
	74	}
	75
	76
	77	void sop_creator_trivial_low_delay::insert_new_input_image(de265_image* img)
	78	{
	79	img->PicOrderCntVal = get_pic_order_count();
	80
	81	int frame = get_frame_number();
	82
	83	std::vector<int> l0, l1, empty;
	84	if (!isIntra(frame)) {
	85	l0.push_back(frame-1);
	86	}
	87
	88	assert(mEncPicBuf);
	89	image_data* imgdata = mEncPicBuf->insert_next_image_in_encoding_order(img, get_frame_number());
	90
	91	if (isIntra(frame)) {
	92	reset_poc();
	93	imgdata->set_intra();
	94	imgdata->set_NAL_type(NAL_UNIT_IDR_N_LP);
	95	imgdata->shdr.slice_type = SLICE_TYPE_I;
	96	} else {
	97	imgdata->set_references(0, l0,l1, empty,empty);
	98	imgdata->set_NAL_type(NAL_UNIT_TRAIL_R);
	99	imgdata->shdr.slice_type = SLICE_TYPE_P;
	100	}
	101	imgdata->shdr.slice_pic_order_cnt_lsb = get_pic_order_count_lsb();
	102	mEncPicBuf->sop_metadata_commit(get_frame_number());
	103
	104	advance_frame();
	105	}

+143

-0

libde265/encoder/sop.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifndef DE265_SOP_H
	21	#define DE265_SOP_H
	22
	23	#include "libde265/image.h"
	24	#include "libde265/sps.h"
	25	#include "libde265/encoder/encpicbuf.h"
	26	#include "libde265/configparam.h"
	27
	28	#include <deque>
	29	#include <vector>
	30
	31	/*
	32	struct refpic_set
	33	{
	34	std::vector<int> l0;
	35	std::vector<int> l1;
	36	};
	37	*/
	38
	39	class pic_order_counter
	40	{
	41	public:
	42	pic_order_counter() { mFrameNumber=0; mPOC=0; mNumLsbBits=6; }
	43
	44	void reset_poc() { mPOC=0; }
	45
	46	int get_frame_number() const { return mFrameNumber; }
	47
	48	int get_pic_order_count() const { return mPOC; }
	49	int get_pic_order_count_lsb() const {
	50	return mPOC & ((1<<mNumLsbBits)-1);
	51	}
	52
	53	void advance_frame(int n=1) { mFrameNumber+=n; mPOC+=n; }
	54
	55	void set_num_poc_lsb_bits(int n) { mNumLsbBits=n; }
	56	int get_num_poc_lsb_bits() const { return mNumLsbBits; }
	57
	58	private:
	59	int mFrameNumber;
	60	int mPOC;
	61	int mNumLsbBits;
	62	};
	63
	64
	65	class sop_creator : public pic_order_counter
	66	{
	67	public:
	68	sop_creator() { mEncCtx=NULL; mEncPicBuf=NULL; }
	69	virtual ~sop_creator() { }
	70
	71	void set_encoder_context(encoder_context* encctx) { mEncCtx=encctx; }
	72	void set_encoder_picture_buffer(encoder_picture_buffer* encbuf) { mEncPicBuf=encbuf; }
	73
	74	/* Fills in the following fields:
	75	- SPS.ref_pic_sets
	76	- SPS.log2_max_pic_order_cnt_lsb
	77	*/
	78	virtual void set_SPS_header_values() = 0;
	79
	80	/* Fills in the following fields:
	81	- NAL.nal_type
	82	- SHDR.slice_type
	83	- SHDR.slice_pic_order_cnt_lsb
	84	- IMGDATA.references
	85	*/
	86	virtual void insert_new_input_image(de265_image*) = 0;
	87	virtual void insert_end_of_stream() { mEncPicBuf->insert_end_of_stream(); }
	88
	89	virtual int get_number_of_temporal_layers() const { return 1; }
	90
	91	//virtual std::vector<refpic_set> get_sps_refpic_sets() const = 0;
	92
	93	protected:
	94	encoder_context* mEncCtx;
	95	encoder_picture_buffer* mEncPicBuf;
	96	};
	97
	98
	99
	100	class sop_creator_intra_only : public sop_creator
	101	{
	102	public:
	103	sop_creator_intra_only();
	104
	105	virtual void set_SPS_header_values();
	106	virtual void insert_new_input_image(de265_image* img);
	107	};
	108
	109
	110
	111	class sop_creator_trivial_low_delay : public sop_creator
	112	{
	113	public:
	114	struct params {
	115	params() {
	116	intraPeriod.set_ID("sop-lowDelay-intraPeriod");
	117	intraPeriod.set_minimum(1);
	118	intraPeriod.set_default(250);
	119	}
	120
	121	void registerParams(config_parameters& config) {
	122	config.add_option(&intraPeriod);
	123	}
	124
	125	option_int intraPeriod;
	126	};
	127
	128	sop_creator_trivial_low_delay();
	129
	130	void setParams(const params& p) { mParams=p; }
	131
	132	virtual void set_SPS_header_values();
	133	virtual void insert_new_input_image(de265_image* img);
	134
	135	private:
	136	params mParams;
	137
	138	bool isIntra(int frame) const { return (frame % mParams.intraPeriod)==0; }
	139	};
	140
	141
	142	#endif

+966

-44

libde265/fallback-dct.cc less more

17	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
18	18	*/
19	19
20		#include "fallback-motion.h"
21		#include "util.h"
	20	#include "fallback-dct.h"
22	21
23	22	#if defined(_MSC_VER) \|\| defined(__MINGW32__)
24	23	# include <malloc.h>

27	26	#endif
28	27
29	28	#include <assert.h>
30
31
32		void transform_skip_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride)
	29	#include <algorithm>
	30
	31
	32	static void printMatrix(const char* name, const int16_t* v, int n)
	33	{
	34	printf("--- %s ---\n",name);
	35	for (int r=0;r<n;r++) {
	36	for (int c=0;c<n;c++) {
	37	printf("%4d ",v[c+r*n]);
	38	}
	39	printf("\n");
	40	}
	41	}
	42
	43
	44
	45	void transform_skip_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride)
33	46	{
34	47	int nT = 4;
35	48	int bdShift2 = 20-8;
	49
	50	assert(0); // DEPRECATED, should not be used anymore because of fixed 4x4 size
36	51
37	52	for (int y=0;y<nT;y++)
38	53	for (int x=0;x<nT;x++) {

44	59	}
45	60
46	61
47		void transform_bypass_8_fallback(uint8_t dst, int16_t coeffs, int nT, ptrdiff_t stride)
48		{
49		int bdShift2 = 20-8;
50
	62	void transform_skip_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth)
	63	{
	64	int nT = 4;
	65	int bdShift2 = 20-bit_depth;
	66
	67	assert(0); // DEPRECATED, should not be used anymore because of fixed 4x4 size
	68
	69	for (int y=0;y<nT;y++)
	70	for (int x=0;x<nT;x++) {
	71	int32_t c = coeffs[x+y*nT] << 7;
	72	c = (c+(1<<(bdShift2-1)))>>bdShift2;
	73
	74	dst[ystride+x] = Clip_BitDepth(dst[ystride+x] + c, bit_depth);
	75	}
	76	}
	77
	78
	79	void transform_skip_residual_fallback(int32_t residual, const int16_t coeffs, int nT,
	80	int tsShift,int bdShift)
	81	{
	82	const int rnd = 1<<(bdShift-1);
	83
	84	for (int y=0;y<nT;y++)
	85	for (int x=0;x<nT;x++) {
	86	int32_t c = coeffs[x+y*nT] << tsShift;
	87	residual[x+y*nT] = (c + rnd) >> bdShift;
	88	}
	89	}
	90
	91
	92	void transform_skip_rdpcm_v_8_fallback(uint8_t dst, const int16_t coeffs, int log2nT, ptrdiff_t stride)
	93	{
	94	int bitDepth = 8;
	95	int bdShift2 = 20-bitDepth;
	96	int offset = (1<<(bdShift2-1));
	97	int tsShift = 5 + log2nT; // TODO: extended_precision
	98	int nT = 1<<log2nT;
	99
	100	for (int x=0;x<nT;x++) {
	101	int32_t sum = 0;
	102
	103	for (int y=0;y<nT;y++) {
	104	int c = coeffs[x+y*nT] << tsShift;
	105	sum += (c+offset)>>bdShift2;
	106
	107	dst[ystride+x] = Clip1_8bit(dst[ystride+x] + sum);
	108	}
	109	}
	110	}
	111
	112	void transform_skip_rdpcm_h_8_fallback(uint8_t dst, const int16_t coeffs, int log2nT, ptrdiff_t stride)
	113	{
	114	int bitDepth = 8;
	115	int bdShift2 = 20-bitDepth;
	116	int offset = (1<<(bdShift2-1));
	117	int tsShift = 5 + log2nT; // TODO: extended_precision
	118	int nT = 1<<log2nT;
	119
	120	for (int y=0;y<nT;y++) {
	121	int32_t sum = 0;
	122
	123	for (int x=0;x<nT;x++) {
	124	int c = coeffs[x+y*nT] << tsShift;
	125	sum += (c+offset)>>bdShift2;
	126
	127	dst[ystride+x] = Clip1_8bit(dst[ystride+x] + sum);
	128	}
	129	}
	130	}
	131
	132
	133	void transform_bypass_rdpcm_v_8_fallback(uint8_t dst, const int16_t coeffs,int nT,ptrdiff_t stride)
	134	{
	135	for (int x=0;x<nT;x++) {
	136	int32_t sum=0;
	137	for (int y=0;y<nT;y++) {
	138	sum += coeffs[x+y*nT];
	139
	140	dst[ystride+x] = Clip1_8bit(dst[ystride+x] + sum);
	141	}
	142	}
	143	}
	144
	145
	146	void transform_bypass_rdpcm_h_8_fallback(uint8_t dst, const int16_t coeffs,int nT,ptrdiff_t stride)
	147	{
	148	for (int y=0;y<nT;y++) {
	149	int32_t sum=0;
	150	for (int x=0;x<nT;x++) {
	151	sum += coeffs[x+y*nT];
	152
	153	dst[ystride+x] = Clip1_8bit(dst[ystride+x] + sum);
	154	}
	155	}
	156	}
	157
	158
	159	void transform_bypass_rdpcm_v_fallback(int32_t dst, const int16_t coeffs,int nT)
	160	{
	161	for (int x=0;x<nT;x++) {
	162	int32_t sum=0;
	163	for (int y=0;y<nT;y++) {
	164	sum += coeffs[x+y*nT];
	165
	166	dst[y*nT+x] = sum;
	167	}
	168	}
	169	}
	170
	171
	172	void transform_bypass_rdpcm_h_fallback(int32_t dst, const int16_t coeffs,int nT)
	173	{
	174	for (int y=0;y<nT;y++) {
	175	int32_t sum=0;
	176	for (int x=0;x<nT;x++) {
	177	sum += coeffs[x+y*nT];
	178
	179	dst[y*nT+x] = sum;
	180	}
	181	}
	182	}
	183
	184
	185	void rdpcm_v_fallback(int32_t* residual, const int16_t* coeffs, int nT,int tsShift,int bdShift)
	186	{
	187	int rnd = (1<<(bdShift-1));
	188
	189	for (int x=0;x<nT;x++) {
	190	int sum=0;
	191	for (int y=0;y<nT;y++) {
	192	int c = coeffs[x+y*nT] << tsShift;
	193	sum += (c+rnd)>>bdShift;
	194	residual[y*nT+x] = sum;
	195	}
	196	}
	197	}
	198
	199
	200	void rdpcm_h_fallback(int32_t* residual, const int16_t* coeffs, int nT,int tsShift,int bdShift)
	201	{
	202	int rnd = (1<<(bdShift-1));
	203
	204	for (int y=0;y<nT;y++) {
	205	int sum=0;
	206	for (int x=0;x<nT;x++) {
	207	int c = coeffs[x+y*nT] << tsShift;
	208	sum += (c+rnd)>>bdShift;
	209	residual[y*nT+x] = sum;
	210	}
	211	}
	212	}
	213
	214
	215	void transform_bypass_fallback(int32_t dst, const int16_t coeffs, int nT)
	216	{
51	217	for (int y=0;y<nT;y++)
52	218	for (int x=0;x<nT;x++) {
53	219	int32_t c = coeffs[x+y*nT];
54	220
	221	dst[y*nT+x] = c;
	222	}
	223	}
	224
	225
	226	void transform_bypass_8_fallback(uint8_t dst, const int16_t coeffs, int nT, ptrdiff_t stride)
	227	{
	228	for (int y=0;y<nT;y++)
	229	for (int x=0;x<nT;x++) {
	230	int32_t c = coeffs[x+y*nT];
	231
55	232	dst[ystride+x] = Clip1_8bit(dst[ystride+x] + c);
56	233	}
57	234	}
58
	235
	236
	237	void transform_bypass_16_fallback(uint16_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth)
	238	{
	239	int bdShift2 = 20-bit_depth;
	240
	241	for (int y=0;y<nT;y++)
	242	for (int x=0;x<nT;x++) {
	243	int32_t c = coeffs[x+y*nT];
	244
	245	dst[ystride+x] = Clip_BitDepth(dst[ystride+x] + c, bit_depth);
	246	}
	247	}
	248
	249
	250	void rotate_coefficients_fallback(int16_t *coeff, int nT)
	251	{
	252	for (int y=0;y<nT/2;y++)
	253	for (int x=0;x<nT;x++) {
	254	std::swap(coeff[ynT+x], coeff[(nT-1-y)nT + nT-1-x]);
	255	}
	256	}
	257
	258
59	259
60	260	static int8_t mat_8_357[4][4] = {
61	261	{ 29, 55, 74, 84 },

66	266
67	267
68	268
69		void transform_4x4_luma_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride)
	269	void transform_4x4_luma_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride)
70	270	{
71	271	int16_t g[4][4];
72	272

78	278	// --- V ---
79	279
80	280	for (int c=0;c<4;c++) {
81
	281	/*
82	282	logtrace(LogTransform,"DST-V: ");
83	283	for (int r=0;r<4;r++) {
84	284	logtrace(LogTransform,"%d ",coeffs[c+r*4]);
85	285	}
86	286	logtrace(LogTransform,"* -> ");
87
	287	*/
88	288
89	289	for (int i=0;i<4;i++) {
90	290	int sum=0;

96	296	g[i][c] = Clip3(-32768,32767, (sum+rndV)>>7);
97	297	}
98	298
99
	299	/*
100	300	for (int y=0;y<4;y++) {
101	301	logtrace(LogTransform,"*%d ",g[y][c]);
102	302	}
103	303	logtrace(LogTransform,"*\n");
	304	*/
104	305	}
105	306
106	307

108	309
109	310	for (int y=0;y<4;y++) {
110	311
	312	/*
111	313	logtrace(LogTransform,"DST-H: ");
112	314	for (int c=0;c<4;c++) {
113	315	logtrace(LogTransform,"%d ",g[y][c]);
114	316	}
115	317	logtrace(LogTransform,"* -> ");
116
	318	*/
117	319
118	320	for (int i=0;i<4;i++) {
119	321	int sum=0;

130	332	}
131	333
132	334	logtrace(LogTransform,"*\n");
	335	}
	336	}
	337
	338
	339	void transform_4x4_luma_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride,
	340	int bit_depth)
	341	{
	342	int16_t g[4][4];
	343
	344	int postShift = 20-bit_depth;
	345	int rndV = 1<<(7-1);
	346	int rndH = 1<<(postShift-1);
	347
	348
	349	// --- V ---
	350
	351	for (int c=0;c<4;c++) {
	352	/*
	353	logtrace(LogTransform,"DST-V: ");
	354	for (int r=0;r<4;r++) {
	355	logtrace(LogTransform,"%d ",coeffs[c+r*4]);
	356	}
	357	logtrace(LogTransform,"* -> ");
	358	*/
	359
	360	for (int i=0;i<4;i++) {
	361	int sum=0;
	362
	363	for (int j=0;j<4;j++) {
	364	sum += mat_8_357[j][i] * coeffs[c+j*4];
	365	}
	366
	367	g[i][c] = Clip3(-32768,32767, (sum+rndV)>>7);
	368	}
	369
	370	/*
	371	for (int y=0;y<4;y++) {
	372	logtrace(LogTransform,"*%d ",g[y][c]);
	373	}
	374	logtrace(LogTransform,"*\n");
	375	*/
	376	}
	377
	378
	379	// --- H ---
	380
	381	for (int y=0;y<4;y++) {
	382
	383	/*
	384	logtrace(LogTransform,"DST-H: ");
	385	for (int c=0;c<4;c++) {
	386	logtrace(LogTransform,"%d ",g[y][c]);
	387	}
	388	logtrace(LogTransform,"* -> ");
	389	*/
	390
	391	for (int i=0;i<4;i++) {
	392	int sum=0;
	393
	394	for (int j=0;j<4;j++) {
	395	sum += mat_8_357[j][i] * g[y][j];
	396	}
	397
	398	int out = Clip3(-32768,32767, (sum+rndH)>>postShift);
	399
	400	dst[ystride+i] = Clip_BitDepth(dst[ystride+i] + out, bit_depth);
	401
	402	logtrace(LogTransform,"*%d ",out);
	403	}
	404
	405	logtrace(LogTransform,"*\n");
	406	}
	407	}
	408
	409
	410	void fdst_4x4_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	411	{
	412	int16_t g[4*4];
	413
	414	int BD = 8;
	415	int shift1 = Log2(4) + BD -9;
	416	int shift2 = Log2(4) + 6;
	417
	418	int rnd1 = 1<<(shift1-1);
	419	int rnd2 = 1<<(shift2-1);
	420
	421
	422	// --- V ---
	423
	424	for (int c=0;c<4;c++) {
	425
	426	/*
	427	logtrace(LogTransform,"DST-V: ");
	428	for (int r=0;r<4;r++) {
	429	logtrace(LogTransform,"%d ",coeffs[c+r*4]);
	430	}
	431	logtrace(LogTransform,"* -> ");
	432	*/
	433
	434	for (int i=0;i<4;i++) {
	435	int sum=0;
	436
	437	for (int j=0;j<4;j++) {
	438	sum += mat_8_357[i][j] * input[c+j*stride];
	439	}
	440
	441	g[c+4*i] = Clip3(-32768,32767, (sum+rnd1)>>shift1);
	442	}
	443	}
	444
	445
	446	// --- H ---
	447
	448	for (int y=0;y<4;y++) {
	449	for (int i=0;i<4;i++) {
	450	int sum=0;
	451
	452	for (int j=0;j<4;j++) {
	453	sum += mat_8_357[i][j] * g[y*4+j];
	454	}
	455
	456	// TODO: do we need clipping ?
	457	int out = (sum+rnd2)>>shift2; // Clip3(-32768,32767, (sum+rndH)>>postShift);
	458
	459	coeffs[y*4+i] = out;
	460
	461	logtrace(LogTransform,"*%d ",out);
	462	}
	463
	464	logtrace(LogTransform,"*\n");
	465	}
	466	}
	467
	468
	469	void transform_idst_4x4_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits)
	470	{
	471	int16_t g[4][4];
	472
	473	int rndV = 1<<(7-1);
	474	int rndH = 1<<(bdShift-1);
	475
	476	int CoeffMax = (1<<max_coeff_bits)-1;
	477	int CoeffMin = -(1<<max_coeff_bits);
	478
	479
	480	// --- V ---
	481
	482	for (int c=0;c<4;c++) {
	483	for (int i=0;i<4;i++) {
	484	int sum=0;
	485
	486	for (int j=0;j<4;j++) {
	487	sum += mat_8_357[j][i] * coeffs[c+j*4];
	488	}
	489
	490	g[i][c] = Clip3(CoeffMin,CoeffMax, (sum+rndV)>>7);
	491	}
	492	}
	493
	494
	495	// --- H ---
	496
	497	for (int y=0;y<4;y++) {
	498	for (int i=0;i<4;i++) {
	499	int sum=0;
	500
	501	for (int j=0;j<4;j++) {
	502	sum += mat_8_357[j][i] * g[y][j];
	503	}
	504
	505	dst[y*4+i] = (sum + rndH)>>bdShift;
	506	}
133	507	}
134	508	}
135	509

173	547
174	548
175	549
176		static void transform_dct_add_8(uint8_t *dst, ptrdiff_t stride,
177		int nT, int16_t *coeffs)
178		{
179		int postShift = 20-8; // 8 bit
	550	template <class pixel_t>
	551	void transform_idct_add(pixel_t *dst, ptrdiff_t stride,
	552	int nT, const int16_t *coeffs, int bit_depth)
	553	{
	554	/*
	555	The effective shift is
	556	7 bits right for bit-depth 8,
	557	6 bits right for bit-depth 9,
	558	5 bits right for bit-depth 10.
	559
	560	Computation is independent of the block size.
	561	Each multiplication with the table includes a left shift of 6 bits.
	562	Hence, we have 2* 6 bits = 12 bits left shift.
	563	V-pass has fixed 7 bit right shift.
	564	H-pass has 20-BitDepth bit right shift;
	565
	566	Effective shift 's' means: residual value 1 gives DC-coeff (1<<s).
	567	*/
	568
	569
	570	int postShift = 20-bit_depth;
180	571	int rnd1 = 1<<(7-1);
181	572	int rnd2 = 1<<(postShift-1);
182	573	int fact = (1<<(5-Log2(nT)));

199	590	printf("\n");
200	591	*/
201	592
	593	/*
	594	printf("--- input\n");
	595	for (int r=0;r<nT;r++, printf("\n"))
	596	for (int c=0;c<nT;c++) {
	597	printf("%3d ",coeffs[c+r*nT]);
	598	}
	599	*/
	600
202	601	for (int c=0;c<nT;c++) {
203	602
	603	/*
204	604	logtrace(LogTransform,"DCT-V: ");
205	605	for (int i=0;i<nT;i++) {
206	606	logtrace(LogTransform,"%d ",coeffs[c+inT]);
207	607	}
208	608	logtrace(LogTransform,"* -> ");
	609	*/
209	610
210	611
211	612	// find last non-zero coefficient to reduce computations carried out in DCT

217	618
218	619	for (int i=0;i<nT;i++) {
219	620	int sum=0;
220
	621
	622	/*
	623	printf("input: ");
	624	for (int j=0;j<nT;j++) {
	625	printf("%3d ",coeffs[c+j*nT]);
	626	}
	627	printf("\n");
	628
	629	printf("mat: ");
	630	for (int j=0;j<nT;j++) {
	631	printf("%3d ",mat_dct[fact*j][i]);
	632	}
	633	printf("\n");
	634	*/
	635
221	636	for (int j=0;j<=lastCol /nT/;j++) {
222	637	sum += mat_dct[factj][i] coeffs[c+j*nT];
223	638	}
224
	639
225	640	g[c+i*nT] = Clip3(-32768,32767, (sum+rnd1)>>7);
226	641
227	642	logtrace(LogTransform,"%d ",g[c+inT]);

229	644	logtrace(LogTransform,"*\n");
230	645	}
231	646
	647	/*
	648	printf("--- temp\n");
	649	for (int r=0;r<nT;r++, printf("\n"))
	650	for (int c=0;c<nT;c++) {
	651	printf("%3d ",g[c+r*nT]);
	652	}
	653	*/
232	654
233	655	for (int y=0;y<nT;y++) {
234
	656	/*
235	657	logtrace(LogTransform,"DCT-H: ");
236	658	for (int i=0;i<nT;i++) {
237	659	logtrace(LogTransform,"%d ",g[i+ynT]);
238	660	}
239	661	logtrace(LogTransform,"* -> ");
	662	*/
240	663
241	664
242	665	// find last non-zero coefficient to reduce computations carried out in DCT

249	672
250	673	for (int i=0;i<nT;i++) {
251	674	int sum=0;
252
	675
253	676	for (int j=0;j<=lastCol /nT/;j++) {
254	677	sum += mat_dct[factj][i] g[y*nT+j];
255	678	}
256
	679
257	680	//int out = Clip3(-32768,32767, (sum+rnd2)>>postShift);
258	681	int out = (sum+rnd2)>>postShift;
259	682
260	683	//fprintf(stderr,"%d%d+%d = %d\n",y,stride,i,ystride+i);
261	684	//fprintf(stderr,"[%p]=%d\n",&dst[ystride+i], Clip1_8bit(dst[ystride+i]));
262		dst[ystride+i] = Clip1_8bit(dst[ystride+i] + out);
	685	dst[ystride+i] = Clip_BitDepth(dst[ystride+i] + out, bit_depth);
263	686
264	687	logtrace(LogTransform,"*%d ",out);
265	688	}

268	691	}
269	692
270	693
271		void transform_4x4_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride)
272		{
273		transform_dct_add_8(dst,stride, 4, coeffs);
274		}
275
276		void transform_8x8_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride)
277		{
278		transform_dct_add_8(dst,stride, 8, coeffs);
279		}
280
281		void transform_16x16_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride)
282		{
283		transform_dct_add_8(dst,stride, 16, coeffs);
284		}
285
286		void transform_32x32_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride)
287		{
288		transform_dct_add_8(dst,stride, 32, coeffs);
289		}
	694
	695	void transform_idct_fallback(int32_t dst, int nT, const int16_t coeffs, int bdShift, int max_coeff_bits)
	696	{
	697	/*
	698	The effective shift is
	699	7 bits right for bit-depth 8,
	700	6 bits right for bit-depth 9,
	701	5 bits right for bit-depth 10.
	702
	703	One transformation with raw transform filter values increases range be 2048 (=32*64).
	704	This equals 11 bits.
	705
	706	Computation is independent of the block size.
	707	Each multiplication with the table includes a left shift of 6 bits.
	708	Hence, we have 2* 6 bits = 12 bits left shift.
	709	V-pass has fixed 7 bit right shift.
	710	H-pass has 20-BitDepth bit right shift;
	711
	712	Effective shift 's' means: residual value 1 gives DC-coeff (1<<s).
	713	*/
	714
	715
	716	int rnd1 = 1<<(7-1);
	717	int fact = (1<<(5-Log2(nT)));
	718
	719	//int bdShift = 20-bit_depth;
	720	int rnd2 = 1<<(bdShift-1);
	721
	722	int16_t g[3232]; // actually, only [nTnT] used
	723
	724	int CoeffMax = (1<<max_coeff_bits)-1;
	725	int CoeffMin = -(1<<max_coeff_bits);
	726
	727	// TODO: valgrind reports that dst[] contains uninitialized data.
	728	// Probably from intra-prediction.
	729
	730	/*
	731	for (int i=0;i<nT*nT;i++) {
	732	printf("%d\n",coeffs[i]);
	733	}
	734
	735	for (int y=0;y<nT;y++) {
	736	for (int i=0;i<nT;i++) {
	737	printf("%d ",dst[y*stride+i]);
	738	}
	739	}
	740	printf("\n");
	741	*/
	742
	743	/*
	744	printf("--- input\n");
	745	for (int r=0;r<nT;r++, printf("\n"))
	746	for (int c=0;c<nT;c++) {
	747	printf("%3d ",coeffs[c+r*nT]);
	748	}
	749	*/
	750
	751	for (int c=0;c<nT;c++) {
	752
	753	/*
	754	logtrace(LogTransform,"DCT-V: ");
	755	for (int i=0;i<nT;i++) {
	756	logtrace(LogTransform,"%d ",coeffs[c+inT]);
	757	}
	758	logtrace(LogTransform,"* -> ");
	759	*/
	760
	761
	762	// find last non-zero coefficient to reduce computations carried out in DCT
	763
	764	int lastCol = nT-1;
	765	for (;lastCol>=0;lastCol--) {
	766	if (coeffs[c+lastCol*nT]) { break; }
	767	}
	768
	769	for (int i=0;i<nT;i++) {
	770	int sum=0;
	771
	772	/*
	773	printf("input: ");
	774	for (int j=0;j<nT;j++) {
	775	printf("%3d ",coeffs[c+j*nT]);
	776	}
	777	printf("\n");
	778
	779	printf("mat: ");
	780	for (int j=0;j<nT;j++) {
	781	printf("%3d ",mat_dct[fact*j][i]);
	782	}
	783	printf("\n");
	784	*/
	785
	786	for (int j=0;j<=lastCol /nT/;j++) {
	787	sum += mat_dct[factj][i] coeffs[c+j*nT];
	788	}
	789
	790	g[c+i*nT] = Clip3(CoeffMin,CoeffMax, (sum+rnd1)>>7);
	791
	792	logtrace(LogTransform,"%d ",g[c+inT]);
	793	}
	794	logtrace(LogTransform,"*\n");
	795	}
	796
	797	/*
	798	printf("--- temp\n");
	799	for (int r=0;r<nT;r++, printf("\n"))
	800	for (int c=0;c<nT;c++) {
	801	printf("%3d ",g[c+r*nT]);
	802	}
	803	*/
	804
	805	for (int y=0;y<nT;y++) {
	806	/*
	807	logtrace(LogTransform,"DCT-H: ");
	808	for (int i=0;i<nT;i++) {
	809	logtrace(LogTransform,"%d ",g[i+ynT]);
	810	}
	811	logtrace(LogTransform,"* -> ");
	812	*/
	813
	814
	815	// find last non-zero coefficient to reduce computations carried out in DCT
	816
	817	int lastCol = nT-1;
	818	for (;lastCol>=0;lastCol--) {
	819	if (g[y*nT+lastCol]) { break; }
	820	}
	821
	822
	823	for (int i=0;i<nT;i++) {
	824	int sum=0;
	825
	826	for (int j=0;j<=lastCol /nT/;j++) {
	827	sum += mat_dct[factj][i] g[y*nT+j];
	828	}
	829
	830	dst[y*nT+i] = (sum + rnd2)>>bdShift;
	831
	832	logtrace(LogTransform,"*%d ",sum);
	833	}
	834	logtrace(LogTransform,"*\n");
	835	}
	836	}
	837
	838
	839	void transform_idct_4x4_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits)
	840	{
	841	transform_idct_fallback(dst,4,coeffs,bdShift,max_coeff_bits);
	842	}
	843
	844	void transform_idct_8x8_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits)
	845	{
	846	transform_idct_fallback(dst,8,coeffs,bdShift,max_coeff_bits);
	847	}
	848
	849	void transform_idct_16x16_fallback(int32_t dst, const int16_t coeffs,
	850	int bdShift, int max_coeff_bits)
	851	{
	852	transform_idct_fallback(dst,16,coeffs,bdShift,max_coeff_bits);
	853	}
	854
	855	void transform_idct_32x32_fallback(int32_t dst, const int16_t coeffs,
	856	int bdShift, int max_coeff_bits)
	857	{
	858	transform_idct_fallback(dst,32,coeffs,bdShift,max_coeff_bits);
	859	}
	860
	861
	862
	863
	864	void transform_4x4_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride)
	865	{
	866	transform_idct_add<uint8_t>(dst,stride, 4, coeffs, 8);
	867	}
	868
	869	void transform_8x8_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride)
	870	{
	871	transform_idct_add<uint8_t>(dst,stride, 8, coeffs, 8);
	872	}
	873
	874	void transform_16x16_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride)
	875	{
	876	transform_idct_add<uint8_t>(dst,stride, 16, coeffs, 8);
	877	}
	878
	879	void transform_32x32_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride)
	880	{
	881	transform_idct_add<uint8_t>(dst,stride, 32, coeffs, 8);
	882	}
	883
	884
	885	void transform_4x4_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth)
	886	{
	887	transform_idct_add<uint16_t>(dst,stride, 4, coeffs, bit_depth);
	888	}
	889
	890	void transform_8x8_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth)
	891	{
	892	transform_idct_add<uint16_t>(dst,stride, 8, coeffs, bit_depth);
	893	}
	894
	895	void transform_16x16_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth)
	896	{
	897	transform_idct_add<uint16_t>(dst,stride, 16, coeffs, bit_depth);
	898	}
	899
	900	void transform_32x32_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth)
	901	{
	902	transform_idct_add<uint16_t>(dst,stride, 32, coeffs, bit_depth);
	903	}
	904
	905
	906	static void transform_fdct_8(int16_t* coeffs, int nT,
	907	const int16_t *input, ptrdiff_t stride)
	908	{
	909	/*
	910	Each sum over a basis vector sums nT elements, which is compensated by
	911	shifting right by Log2(nT), effectively dividing by 2^Log2(nT) = nT.
	912	Do this in each of the H/V passes.
	913
	914	Each multiplication with the table includes a left shift of 6 bits.
	915	Hence, we have in total 2* 6 bits = 12 bits left shift because of the
	916	multiplications.
	917
	918	We carry out shifts after each pass:
	919	First (V) pass has BitDepth-9 bit right shift,
	920	Second (H) pass has fixed 6 bit right shift.
	921
	922	For bit-depth 8, the total shift is 7 bits left.
	923	For bit-depth 9, the total shift is 6 bits left.
	924	For bit-depth 10, the total shift is 5 bits left.
	925
	926	I.e.: a constant residual value 1 gives DC-coeff (1<<s).
	927
	928	For 8-bit images in a 32x32 block, the input are 8 bits + 1 sign bit.
	929	After the first pass, we need 9+5+6=20 bits for the intermediate sum
	930	(9 bit input, 5 bit because we sum 2^5 elements, 6 bit because of multiplication with 64).
	931	The first pass shift is Log2(32) - 1 -> 4 bits and we are down to 16 bits again.
	932	After the second pass, we need 16+5+6=27 bits for the intermediate sum
	933	(16 bit input, 5 bit because we sum 2^5 elements, 6 bit because of coefficient multiplication).
	934	The second pass shift is Log2(32)+6 = 11 and we are down again to 16 bits.
	935
	936	For larger input bit-depths, the intermediate result after the first pass
	937	will be wider accordingly, but the widths after the shifts are the same.
	938	*/
	939
	940	int BitDepth = 8;
	941
	942	// / compensate everything \| / effective word length \|
	943	int shift1 = Log2(nT) + 6 + BitDepth - 15;
	944	int shift2 = Log2(nT) + 6;
	945
	946	int rnd1 = 1<<(shift1-1);
	947	int rnd2 = 1<<(shift2-1);
	948	int fact = (1<<(5-Log2(nT)));
	949
	950	int16_t g[3232]; // actually, only [nTnT] used
	951
	952	for (int c=0;c<nT;c++) {
	953
	954	for (int i=0;i<nT;i++) {
	955	int sum=0;
	956
	957	for (int j=0;j<nT;j++) {
	958	sum += mat_dct[facti][j] input[c+j*stride];
	959	}
	960
	961	//assert((sum+rnd1)>>shift1 <= 32767);
	962	//assert((sum+rnd1)>>shift1 >= -32768);
	963	g[c+i*nT] = (sum+rnd1)>>shift1; // clipping to -32768;32767 unnecessary
	964	}
	965	}
	966
	967
	968	for (int y=0;y<nT;y++) {
	969	for (int i=0;i<nT;i++) {
	970	int sum=0;
	971
	972	for (int j=0;j<nT;j++) {
	973	sum += mat_dct[facti][j] g[y*nT+j];
	974	}
	975
	976	// no clipping to -32768;32767 required
	977	int out = (sum+rnd2)>>shift2;
	978
	979	coeffs[y*nT+i] = out;
	980	}
	981	}
	982	}
	983
	984
	985	void fdct_4x4_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	986	{
	987	transform_fdct_8(coeffs, 4, input,stride);
	988	}
	989
	990	void fdct_8x8_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	991	{
	992	transform_fdct_8(coeffs, 8, input,stride);
	993	}
	994
	995	void fdct_16x16_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	996	{
	997	transform_fdct_8(coeffs, 16, input,stride);
	998	}
	999
	1000	void fdct_32x32_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	1001	{
	1002	transform_fdct_8(coeffs, 32, input,stride);
	1003	}
	1004
	1005
	1006
	1007
	1008	void hadamard_transform_8(int16_t coeffs, int n, const int16_t input, ptrdiff_t stride)
	1009	{
	1010	int16_t tmp[32*32];
	1011
	1012	// row transforms
	1013
	1014	//printMatrix("input",input,n);
	1015
	1016	int16_t am[32],bm[32];
	1017	int16_t a = am, b = bm;
	1018	for (int row=0;row<n;row++) {
	1019	int rs = row*stride;
	1020	for (int i=0;i<(n>>1);i++) {
	1021	a[ i] = input[i+rs] + input[i+(n>>1)+rs];
	1022	a[(n>>1)+i] = input[i+rs] - input[i+(n>>1)+rs];
	1023	}
	1024
	1025	int iOuter=(n>>1);
	1026	int nInner=(n>>2);
	1027
	1028	while (nInner>=2) {
	1029	std::swap(a,b);
	1030
	1031	for (int k=0;k<n;k+=iOuter) {
	1032	for (int i=0;i<nInner;i++) {
	1033	a[k+i ] = b[k+i] + b[k+i+nInner];
	1034	a[k+i+nInner] = b[k+i] - b[k+i+nInner];
	1035	}
	1036	}
	1037
	1038	iOuter>>=1;
	1039	nInner>>=1;
	1040	}
	1041
	1042	for (int k=0;k<n;k+=2) {
	1043	tmp[k +n*row] = a[k] + a[k+1];
	1044	tmp[k+1+n*row] = a[k] - a[k+1];
	1045	}
	1046	}
	1047
	1048	//printMatrix("tmp",tmp,n);
	1049
	1050	// column transforms
	1051
	1052	for (int col=0;col<n;col++) {
	1053	for (int i=0;i<(n>>1);i++) {
	1054	a[ i] = tmp[in+col] + tmp[(i+(n>>1))n+col];
	1055	a[(n>>1)+i] = tmp[in+col] - tmp[(i+(n>>1))n+col];
	1056	}
	1057
	1058	int iOuter=(n>>1);
	1059	int nInner=(n>>2);
	1060
	1061	while (nInner>=2) {
	1062	std::swap(a,b);
	1063
	1064	for (int k=0;k<n;k+=iOuter) {
	1065	for (int i=0;i<nInner;i++) {
	1066	a[k+i ] = b[k+i] + b[k+i+nInner];
	1067	a[k+i+nInner] = b[k+i] - b[k+i+nInner];
	1068	}
	1069	}
	1070
	1071	iOuter>>=1;
	1072	nInner>>=1;
	1073	}
	1074
	1075	for (int k=0;k<n;k+=2) {
	1076	coeffs[col+(k )*n] = a[k] + a[k+1];
	1077	coeffs[col+(k+1)*n] = a[k] - a[k+1];
	1078	}
	1079	}
	1080
	1081	//printMatrix("coeffs",coeffs,n);
	1082	}
	1083
	1084
	1085	void hadamard_4x4_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	1086	{
	1087	int16_t tmp[4*4];
	1088
	1089	// row transforms
	1090
	1091	//printMatrix("input",input,4);
	1092
	1093	int16_t a[4];
	1094	for (int row=0;row<4;row++) {
	1095	int rs = row*stride;
	1096	a[0] = input[0+rs] + input[2+rs];
	1097	a[1] = input[1+rs] + input[3+rs];
	1098	a[2] = input[0+rs] - input[2+rs];
	1099	a[3] = input[1+rs] - input[3+rs];
	1100
	1101	tmp[0+4*row] = a[0]+a[1];
	1102	tmp[1+4*row] = a[0]-a[1];
	1103	tmp[2+4*row] = a[2]+a[3];
	1104	tmp[3+4*row] = a[2]-a[3];
	1105	}
	1106
	1107	//printMatrix("tmp",tmp,4);
	1108
	1109	// column transforms
	1110
	1111	for (int col=0;col<4;col++) {
	1112	a[0] = tmp[col+04] + tmp[col+24];
	1113	a[1] = tmp[col+14] + tmp[col+34];
	1114	a[2] = tmp[col+04] - tmp[col+24];
	1115	a[3] = tmp[col+14] - tmp[col+34];
	1116
	1117	coeffs[col+0*4] = a[0]+a[1];
	1118	coeffs[col+1*4] = a[0]-a[1];
	1119	coeffs[col+2*4] = a[2]+a[3];
	1120	coeffs[col+3*4] = a[2]-a[3];
	1121	}
	1122
	1123	//printMatrix("coeffs",coeffs,4);
	1124	}
	1125
	1126
	1127	void hadamard_8x8_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	1128	{
	1129	int16_t tmp[8*8];
	1130
	1131	// row transforms
	1132
	1133	//printMatrix("input",input,8);
	1134
	1135	int16_t a[8],b[8];
	1136	for (int row=0;row<8;row++) {
	1137	int rs = row*stride;
	1138	a[0] = input[0+rs] + input[4+rs];
	1139	a[1] = input[1+rs] + input[5+rs];
	1140	a[2] = input[2+rs] + input[6+rs];
	1141	a[3] = input[3+rs] + input[7+rs];
	1142	a[4] = input[0+rs] - input[4+rs];
	1143	a[5] = input[1+rs] - input[5+rs];
	1144	a[6] = input[2+rs] - input[6+rs];
	1145	a[7] = input[3+rs] - input[7+rs];
	1146
	1147	b[0] = a[0]+a[2];
	1148	b[1] = a[1]+a[3];
	1149	b[2] = a[0]-a[2];
	1150	b[3] = a[1]-a[3];
	1151	b[4] = a[4]+a[6];
	1152	b[5] = a[5]+a[7];
	1153	b[6] = a[4]-a[6];
	1154	b[7] = a[5]-a[7];
	1155
	1156	tmp[0+8*row] = b[0]+b[1];
	1157	tmp[1+8*row] = b[0]-b[1];
	1158	tmp[2+8*row] = b[2]+b[3];
	1159	tmp[3+8*row] = b[2]-b[3];
	1160	tmp[4+8*row] = b[4]+b[5];
	1161	tmp[5+8*row] = b[4]-b[5];
	1162	tmp[6+8*row] = b[6]+b[7];
	1163	tmp[7+8*row] = b[6]-b[7];
	1164	}
	1165
	1166	//printMatrix("tmp",tmp,8);
	1167
	1168	// column transforms
	1169
	1170	for (int col=0;col<8;col++) {
	1171	a[0] = tmp[col+08] + tmp[col+48];
	1172	a[1] = tmp[col+18] + tmp[col+58];
	1173	a[2] = tmp[col+28] + tmp[col+68];
	1174	a[3] = tmp[col+38] + tmp[col+78];
	1175	a[4] = tmp[col+08] - tmp[col+48];
	1176	a[5] = tmp[col+18] - tmp[col+58];
	1177	a[6] = tmp[col+28] - tmp[col+68];
	1178	a[7] = tmp[col+38] - tmp[col+78];
	1179
	1180	b[0] = a[0]+a[2];
	1181	b[1] = a[1]+a[3];
	1182	b[2] = a[0]-a[2];
	1183	b[3] = a[1]-a[3];
	1184	b[4] = a[4]+a[6];
	1185	b[5] = a[5]+a[7];
	1186	b[6] = a[4]-a[6];
	1187	b[7] = a[5]-a[7];
	1188
	1189	coeffs[col+0*8] = b[0]+b[1];
	1190	coeffs[col+1*8] = b[0]-b[1];
	1191	coeffs[col+2*8] = b[2]+b[3];
	1192	coeffs[col+3*8] = b[2]-b[3];
	1193	coeffs[col+4*8] = b[4]+b[5];
	1194	coeffs[col+5*8] = b[4]-b[5];
	1195	coeffs[col+6*8] = b[6]+b[7];
	1196	coeffs[col+7*8] = b[6]-b[7];
	1197	}
	1198
	1199	//printMatrix("coeffs",coeffs,8);
	1200	}
	1201
	1202
	1203	void hadamard_16x16_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	1204	{
	1205	hadamard_transform_8(coeffs,16, input,stride);
	1206	}
	1207
	1208	void hadamard_32x32_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride)
	1209	{
	1210	hadamard_transform_8(coeffs,32, input,stride);
	1211	}

+66

-7

libde265/fallback-dct.h less more

23	23	#include <stddef.h>
24	24	#include <stdint.h>
25	25
	26	#include "util.h"
26	27
27		void transform_skip_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
28		void transform_bypass_8_fallback(uint8_t dst, int16_t coeffs, int nT, ptrdiff_t stride);
29	28
30		void transform_4x4_luma_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
31		void transform_4x4_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
32		void transform_8x8_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
33		void transform_16x16_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
34		void transform_32x32_add_8_fallback(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
	29	// --- decoding ---
	30
	31	void transform_skip_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	32	void transform_bypass_fallback(int32_t r, const int16_t coeffs, int nT);
	33
	34	void transform_skip_rdpcm_v_8_fallback(uint8_t dst, const int16_t coeffs, int nT, ptrdiff_t stride);
	35	void transform_skip_rdpcm_h_8_fallback(uint8_t dst, const int16_t coeffs, int nT, ptrdiff_t stride);
	36	void transform_bypass_rdpcm_v_fallback(int32_t r, const int16_t coeffs,int nT);
	37	void transform_bypass_rdpcm_h_fallback(int32_t r, const int16_t coeffs,int nT);
	38
	39	void transform_4x4_luma_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	40	void transform_4x4_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	41	void transform_8x8_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	42	void transform_16x16_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	43	void transform_32x32_add_8_fallback(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	44
	45
	46	void transform_skip_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth);
	47	void transform_bypass_16_fallback(uint16_t dst, const int16_t coeffs, int nT, ptrdiff_t stride, int bit_depth);
	48
	49	void transform_4x4_luma_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth);
	50	void transform_4x4_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth);
	51	void transform_8x8_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth);
	52	void transform_16x16_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth);
	53	void transform_32x32_add_16_fallback(uint16_t dst, const int16_t coeffs, ptrdiff_t stride, int bit_depth);
	54
	55	void rotate_coefficients_fallback(int16_t *coeff, int nT);
	56
	57
	58	void transform_idst_4x4_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits);
	59	void transform_idct_4x4_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits);
	60	void transform_idct_8x8_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits);
	61	void transform_idct_16x16_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits);
	62	void transform_idct_32x32_fallback(int32_t dst, const int16_t coeffs, int bdShift, int max_coeff_bits);
	63
	64	template <class pixel_t>
	65	void add_residual_fallback(pixel_t *dst, ptrdiff_t stride,
	66	const int32_t* r, int nT, int bit_depth)
	67	{
	68	for (int y=0;y<nT;y++)
	69	for (int x=0;x<nT;x++) {
	70	dst[ystride+x] = Clip_BitDepth(dst[ystride+x] + r[y*nT+x], bit_depth);
	71	}
	72	}
	73
	74
	75	void rdpcm_v_fallback(int32_t* residual, const int16_t* coeffs, int nT, int tsShift,int bdShift);
	76	void rdpcm_h_fallback(int32_t* residual, const int16_t* coeffs, int nT, int tsShift,int bdShift);
	77
	78	void transform_skip_residual_fallback(int32_t residual, const int16_t coeffs, int nT,
	79	int tsShift,int bdShift);
	80
	81
	82	// --- encoding ---
	83
	84	void fdst_4x4_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	85	void fdct_4x4_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	86	void fdct_8x8_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	87	void fdct_16x16_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	88	void fdct_32x32_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	89
	90	void hadamard_4x4_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	91	void hadamard_8x8_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	92	void hadamard_16x16_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
	93	void hadamard_32x32_8_fallback(int16_t coeffs, const int16_t input, ptrdiff_t stride);
35	94
36	95	#endif

+204

-47

libde265/fallback-motion.cc less more

30	30
31	31
32	32	void put_unweighted_pred_8_fallback(uint8_t *dst, ptrdiff_t dststride,
33		int16_t *src, ptrdiff_t srcstride,
	33	const int16_t *src, ptrdiff_t srcstride,
34	34	int width, int height)
35	35	{
36	36	int offset8bit = 32;

39	39	assert((width&1)==0);
40	40
41	41	for (int y=0;y<height;y++) {
42		int16_t* in = &src[y*srcstride];
	42	const int16_t* in = &src[y*srcstride];
43	43	uint8_t* out = &dst[y*dststride];
44	44
45	45	for (int x=0;x<width;x+=2) {

52	52
53	53
54	54	void put_weighted_pred_8_fallback(uint8_t *dst, ptrdiff_t dststride,
55		int16_t *src, ptrdiff_t srcstride,
	55	const int16_t *src, ptrdiff_t srcstride,
56	56	int width, int height,
57	57	int w,int o,int log2WD)
58	58	{

61	61	const int rnd = (1<<(log2WD-1));
62	62
63	63	for (int y=0;y<height;y++) {
64		int16_t* in = &src[y*srcstride];
	64	const int16_t* in = &src[y*srcstride];
65	65	uint8_t* out = &dst[y*dststride];
66	66
67	67	for (int x=0;x<width;x++) {

72	72	}
73	73
74	74	void put_weighted_bipred_8_fallback(uint8_t *dst, ptrdiff_t dststride,
75		int16_t src1, int16_t src2, ptrdiff_t srcstride,
	75	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
76	76	int width, int height,
77	77	int w1,int o1, int w2,int o2, int log2WD)
78	78	{

81	81	const int rnd = ((o1+o2+1) << log2WD);
82	82
83	83	for (int y=0;y<height;y++) {
84		int16_t* in1 = &src1[y*srcstride];
85		int16_t* in2 = &src2[y*srcstride];
	84	const int16_t* in1 = &src1[y*srcstride];
	85	const int16_t* in2 = &src2[y*srcstride];
86	86	uint8_t* out = &dst[y*dststride];
87	87
88	88	for (int x=0;x<width;x++) {

94	94
95	95
96	96	void put_weighted_pred_avg_8_fallback(uint8_t *dst, ptrdiff_t dststride,
97		int16_t src1, int16_t src2,
	97	const int16_t src1, const int16_t src2,
98	98	ptrdiff_t srcstride, int width,
99	99	int height)
100	100	{

143	143	#endif
144	144	{
145	145	for (int y=0;y<height;y++) {
146		int16_t* in1 = &src1[y*srcstride];
147		int16_t* in2 = &src2[y*srcstride];
	146	const int16_t* in1 = &src1[y*srcstride];
	147	const int16_t* in2 = &src2[y*srcstride];
148	148	uint8_t* out = &dst[y*dststride];
149	149
150	150	for (int x=0;x<width;x+=2) {

158	158
159	159
160	160
	161
	162
	163	void put_unweighted_pred_16_fallback(uint16_t *dst, ptrdiff_t dststride,
	164	const int16_t *src, ptrdiff_t srcstride,
	165	int width, int height, int bit_depth)
	166	{
	167	int shift1 = 14-bit_depth;
	168	int offset1 = 0;
	169	if (shift1>0) { offset1 = 1<<(shift1-1); }
	170
	171	assert((width&1)==0);
	172
	173	for (int y=0;y<height;y++) {
	174	const int16_t* in = &src[y*srcstride];
	175	uint16_t* out = &dst[y*dststride];
	176
	177	for (int x=0;x<width;x+=2) {
	178	out[0] = Clip_BitDepth((in[0] + offset1)>>shift1, bit_depth);
	179	out[1] = Clip_BitDepth((in[1] + offset1)>>shift1, bit_depth);
	180	out+=2; in+=2;
	181	}
	182	}
	183	}
	184
	185	#include <stdlib.h>
	186
	187	void put_weighted_pred_16_fallback(uint16_t *dst, ptrdiff_t dststride,
	188	const int16_t *src, ptrdiff_t srcstride,
	189	int width, int height,
	190	int w,int o,int log2WD, int bit_depth)
	191	{
	192	assert(log2WD>=1); // TODO
	193
	194	const int rnd = (1<<(log2WD-1));
	195
	196	for (int y=0;y<height;y++) {
	197	const int16_t* in = &src[y*srcstride];
	198	uint16_t* out = &dst[y*dststride];
	199
	200	for (int x=0;x<width;x++) {
	201	out[0] = Clip_BitDepth(((in[0]*w + rnd)>>log2WD) + o, bit_depth);
	202	out++; in++;
	203	}
	204	}
	205	}
	206
	207	void put_weighted_bipred_16_fallback(uint16_t *dst, ptrdiff_t dststride,
	208	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	209	int width, int height,
	210	int w1,int o1, int w2,int o2, int log2WD, int bit_depth)
	211	{
	212	assert(log2WD>=1); // TODO
	213
	214	const int rnd = ((o1+o2+1) << log2WD);
	215
	216	for (int y=0;y<height;y++) {
	217	const int16_t* in1 = &src1[y*srcstride];
	218	const int16_t* in2 = &src2[y*srcstride];
	219	uint16_t* out = &dst[y*dststride];
	220
	221	for (int x=0;x<width;x++) {
	222	out[0] = Clip_BitDepth((in1[0]w1 + in2[0]w2 + rnd)>>(log2WD+1), bit_depth);
	223	out++; in1++; in2++;
	224	}
	225	}
	226	}
	227
	228
	229	void put_weighted_pred_avg_16_fallback(uint16_t *dst, ptrdiff_t dststride,
	230	const int16_t src1, const int16_t src2,
	231	ptrdiff_t srcstride, int width,
	232	int height, int bit_depth)
	233	{
	234	int shift2 = 15-bit_depth;
	235	int offset2 = 1<<(shift2-1);
	236
	237	assert((width&1)==0);
	238
	239	for (int y=0;y<height;y++) {
	240	const int16_t* in1 = &src1[y*srcstride];
	241	const int16_t* in2 = &src2[y*srcstride];
	242	uint16_t* out = &dst[y*dststride];
	243
	244	for (int x=0;x<width;x+=2) {
	245	out[0] = Clip_BitDepth((in1[0] + in2[0] + offset2)>>shift2, bit_depth);
	246	out[1] = Clip_BitDepth((in1[1] + in2[1] + offset2)>>shift2, bit_depth);
	247	out+=2; in1+=2; in2+=2;
	248	}
	249	}
	250	}
	251
	252
	253
	254
	255
161	256	void put_epel_8_fallback(int16_t *out, ptrdiff_t out_stride,
162		uint8_t *src, ptrdiff_t src_stride,
	257	const uint8_t *src, ptrdiff_t src_stride,
163	258	int width, int height,
164	259	int mx, int my, int16_t* mcbuffer)
165	260	{

167	262
168	263	for (int y=0;y<height;y++) {
169	264	int16_t* o = &out[y*out_stride];
170		uint8_t* i = &src[y*src_stride];
	265	const uint8_t* i = &src[y*src_stride];
171	266
172	267	for (int x=0;x<width;x++) {
173	268	o = i << shift3;

178	273	}
179	274
180	275
181		void put_epel_hv_8_fallback(int16_t *dst, ptrdiff_t dst_stride,
182		uint8_t *src, ptrdiff_t src_stride,
183		int nPbWC, int nPbHC,
184		int xFracC, int yFracC, int16_t* mcbuffer)
185		{
186		const int shift1 = 0;
	276	void put_epel_16_fallback(int16_t *out, ptrdiff_t out_stride,
	277	const uint16_t *src, ptrdiff_t src_stride,
	278	int width, int height,
	279	int mx, int my, int16_t* mcbuffer, int bit_depth)
	280	{
	281	int shift3 = 14 - bit_depth;
	282
	283	for (int y=0;y<height;y++) {
	284	int16_t* o = &out[y*out_stride];
	285	const uint16_t* i = &src[y*src_stride];
	286
	287	for (int x=0;x<width;x++) {
	288	o = i << shift3;
	289	o++;
	290	i++;
	291	}
	292	}
	293	}
	294
	295
	296	template <class pixel_t>
	297	void put_epel_hv_fallback(int16_t *dst, ptrdiff_t dst_stride,
	298	const pixel_t *src, ptrdiff_t src_stride,
	299	int nPbWC, int nPbHC,
	300	int xFracC, int yFracC, int16_t* mcbuffer, int bit_depth)
	301	{
	302	const int shift1 = bit_depth-8;
187	303	const int shift2 = 6;
188	304	//const int shift3 = 6;
189	305

223	339	//printf("---H---(%d)\n",xFracC);
224	340
225	341	for (int y=-extra_top;y<nPbHC+extra_bottom;y++) {
226		uint8_t* p = &src[y*src_stride - extra_left];
	342	const pixel_t* p = &src[y*src_stride - extra_left];
227	343
228	344	for (int x=0;x<nPbWC;x++) {
229	345	int16_t v;

240	356	}
241	357
242	358	//printf("%d %d %d %d -> %d\n",p[0],p[1],p[2],p[3],v);
243
	359
244	360	tmp2buf[y+extra_top + x*nPbH_extra] = v;
245	361	p++;
246	362

271	387	default:
272	388	case 7: v = (-2p[0]+10p[1]+58p[2]-2p[3])>>vshift; break;
273	389	}
274
	390
275	391	dst[x + y*dst_stride] = v;
276	392	p++;
277	393	}

290	406	}
291	407
292	408
	409	template
	410	void put_epel_hv_fallback<uint8_t>(int16_t *dst, ptrdiff_t dst_stride,
	411	const uint8_t *src, ptrdiff_t src_stride,
	412	int nPbWC, int nPbHC,
	413	int xFracC, int yFracC, int16_t* mcbuffer, int bit_depth);
	414	template
	415	void put_epel_hv_fallback<uint16_t>(int16_t *dst, ptrdiff_t dst_stride,
	416	const uint16_t *src, ptrdiff_t src_stride,
	417	int nPbWC, int nPbHC,
	418	int xFracC, int yFracC, int16_t* mcbuffer, int bit_depth);
	419
293	420
294	421
295	422	void put_qpel_0_0_fallback(int16_t *out, ptrdiff_t out_stride,
296		uint8_t *src, ptrdiff_t srcstride,
	423	const uint8_t *src, ptrdiff_t srcstride,
297	424	int nPbW, int nPbH, int16_t* mcbuffer)
298	425	{
299	426	//const int shift1 = 0; // sps->BitDepth_Y-8;

302	429	// straight copy
303	430
304	431	for (int y=0;y<nPbH;y++) {
305		uint8_t* p = src + srcstride*y;
	432	const uint8_t* p = src + srcstride*y;
306	433	int16_t* o = out + out_stride*y;
307	434
308	435	for (int x=0;x<nPbW;x+=4) {
309		#if 0
310		o = p << shift2;
311		o++; p++;
312		#else
	436
313	437	// does not seem to be faster...
314	438	int16_t o0,o1,o2,o3;
315	439	o0 = p[0] << shift2;

323	447
324	448	o+=4;
325	449	p+=4;
326		#endif
327		}
	450	}
	451	}
	452	}
	453
	454
	455	void put_qpel_0_0_fallback_16(int16_t *out, ptrdiff_t out_stride,
	456	const uint16_t *src, ptrdiff_t srcstride,
	457	int nPbW, int nPbH, int16_t* mcbuffer, int bit_depth)
	458	{
	459	//const int shift1 = bit_depth-8;
	460	//const int shift2 = 6;
	461	const int shift3 = 14-bit_depth;
	462
	463	// straight copy
	464
	465	for (int y=0;y<nPbH;y++) {
	466	const uint16_t* p = src + srcstride*y;
	467	int16_t* o = out + out_stride*y;
	468
	469	for (int x=0;x<nPbW;x++) {
	470	o++ = p++ << shift3;
	471	}
328	472	}
329	473	}
330	474

333	477	static int extra_before[4] = { 0,3,3,2 };
334	478	static int extra_after [4] = { 0,3,4,4 };
335	479
	480	template <class pixel_t>
336	481	void put_qpel_fallback(int16_t *out, ptrdiff_t out_stride,
337		uint8_t *src, ptrdiff_t srcstride,
	482	const pixel_t *src, ptrdiff_t srcstride,
338	483	int nPbW, int nPbH, int16_t* mcbuffer,
339		int xFracL, int yFracL)
	484	int xFracL, int yFracL, int bit_depth)
340	485	{
341	486	int extra_left = extra_before[xFracL];
342	487	//int extra_right = extra_after [xFracL];

346	491	//int nPbW_extra = extra_left + nPbW + extra_right;
347	492	int nPbH_extra = extra_top + nPbH + extra_bottom;
348	493
349		const int shift1 = 0; // sps->BitDepth_Y-8;
	494	const int shift1 = bit_depth-8;
350	495	const int shift2 = 6;
351	496
352	497

355	500	switch (xFracL) {
356	501	case 0:
357	502	for (int y=-extra_top;y<nPbH+extra_bottom;y++) {
358		uint8_t* p = src + srcstride*y - extra_left;
	503	const pixel_t* p = src + srcstride*y - extra_left;
359	504	int16_t* o = &mcbuffer[y+extra_top];
360	505
361	506	for (int x=0;x<nPbW;x++) {

367	512	break;
368	513	case 1:
369	514	for (int y=-extra_top;y<nPbH+extra_bottom;y++) {
370		uint8_t* p = src + srcstride*y - extra_left;
	515	const pixel_t* p = src + srcstride*y - extra_left;
371	516	int16_t* o = &mcbuffer[y+extra_top];
372	517
373	518	for (int x=0;x<nPbW;x++) {

379	524	break;
380	525	case 2:
381	526	for (int y=-extra_top;y<nPbH+extra_bottom;y++) {
382		uint8_t* p = src + srcstride*y - extra_left;
	527	const pixel_t* p = src + srcstride*y - extra_left;
383	528	int16_t* o = &mcbuffer[y+extra_top];
384	529
385	530	for (int x=0;x<nPbW;x++) {

391	536	break;
392	537	case 3:
393	538	for (int y=-extra_top;y<nPbH+extra_bottom;y++) {
394		uint8_t* p = src + srcstride*y - extra_left;
	539	const pixel_t* p = src + srcstride*y - extra_left;
395	540	int16_t* o = &mcbuffer[y+extra_top];
396	541
397	542	for (int x=0;x<nPbW;x++) {

420	565	switch (yFracL) {
421	566	case 0:
422	567	for (int x=0;x<nPbW;x++) {
423		int16_t* p = &mcbuffer[x*nPbH_extra];
	568	const int16_t* p = &mcbuffer[x*nPbH_extra];
424	569	int16_t* o = &out[x];
425
	570
426	571	for (int y=0;y<nPbH;y++) {
427	572	o = p;
428	573	o+=out_stride;

432	577	break;
433	578	case 1:
434	579	for (int x=0;x<nPbW;x++) {
435		int16_t* p = &mcbuffer[x*nPbH_extra];
	580	const int16_t* p = &mcbuffer[x*nPbH_extra];
436	581	int16_t* o = &out[x];
437
	582
438	583	for (int y=0;y<nPbH;y++) {
439	584	o = (-p[0]+4p[1]-10p[2]+58p[3]+17p[4] -5p[5] +p[6])>>vshift;
440	585	o+=out_stride;

444	589	break;
445	590	case 2:
446	591	for (int x=0;x<nPbW;x++) {
447		int16_t* p = &mcbuffer[x*nPbH_extra];
	592	const int16_t* p = &mcbuffer[x*nPbH_extra];
448	593	int16_t* o = &out[x];
449
	594
450	595	for (int y=0;y<nPbH;y++) {
451	596	o = (-p[0]+4p[1]-11p[2]+40p[3]+40p[4]-11p[5]+4*p[6]-p[7])>>vshift;
452	597	o+=out_stride;

456	601	break;
457	602	case 3:
458	603	for (int x=0;x<nPbW;x++) {
459		int16_t* p = &mcbuffer[x*nPbH_extra];
	604	const int16_t* p = &mcbuffer[x*nPbH_extra];
460	605	int16_t* o = &out[x];
461
	606
462	607	for (int y=0;y<nPbH;y++) {
463	608	o = ( p[0]-5p[1]+17p[2]+58p[3]-10p[4] +4p[5] -p[6])>>vshift;
464	609	o+=out_stride;

479	624	}
480	625
481	626
	627
482	628	#define QPEL(x,y) void put_qpel_ ## x ## _ ## y ## _fallback(int16_t *out, ptrdiff_t out_stride, \
483		uint8_t *src, ptrdiff_t srcstride, \
484		int nPbW, int nPbH, int16_t* mcbuffer) \
485		{ put_qpel_fallback(out,out_stride, src,srcstride, nPbW,nPbH,mcbuffer,x,y ); }
	629	const uint8_t *src, ptrdiff_t srcstride, \
	630	int nPbW, int nPbH, int16_t* mcbuffer) \
	631	{ put_qpel_fallback(out,out_stride, src,srcstride, nPbW,nPbH,mcbuffer,x,y, 8 ); }
	632
	633
	634	#define QPEL16(x,y) void put_qpel_ ## x ## _ ## y ## _fallback_16(int16_t *out, ptrdiff_t out_stride, \
	635	const uint16_t *src, ptrdiff_t srcstride, \
	636	int nPbW, int nPbH, int16_t* mcbuffer, int bit_depth) \
	637	{ put_qpel_fallback(out,out_stride, src,srcstride, nPbW,nPbH,mcbuffer,x,y, bit_depth ); }
486	638
487	639	/* */ QPEL(0,1) QPEL(0,2) QPEL(0,3)
488	640	QPEL(1,0) QPEL(1,1) QPEL(1,2) QPEL(1,3)
489	641	QPEL(2,0) QPEL(2,1) QPEL(2,2) QPEL(2,3)
490	642	QPEL(3,0) QPEL(3,1) QPEL(3,2) QPEL(3,3)
	643
	644	/* */ QPEL16(0,1) QPEL16(0,2) QPEL16(0,3)
	645	QPEL16(1,0) QPEL16(1,1) QPEL16(1,2) QPEL16(1,3)
	646	QPEL16(2,0) QPEL16(2,1) QPEL16(2,2) QPEL16(2,3)
	647	QPEL16(3,0) QPEL16(3,1) QPEL16(3,2) QPEL16(3,3)

+59

-59

libde265/fallback-motion.h less more

25	25
26	26
27	27	void put_weighted_pred_avg_8_fallback(uint8_t *dst, ptrdiff_t dststride,
28		int16_t src1, int16_t src2,
	28	const int16_t src1, const int16_t src2,
29	29	ptrdiff_t srcstride, int width,
30	30	int height);
31	31
32	32	void put_unweighted_pred_8_fallback(uint8_t *_dst, ptrdiff_t dststride,
33		int16_t *src, ptrdiff_t srcstride,
	33	const int16_t *src, ptrdiff_t srcstride,
34	34	int width, int height);
35	35
36	36	void put_weighted_pred_8_fallback(uint8_t *_dst, ptrdiff_t dststride,
37		int16_t *src, ptrdiff_t srcstride,
	37	const int16_t *src, ptrdiff_t srcstride,
38	38	int width, int height,
39	39	int w,int o,int log2WD);
40	40	void put_weighted_bipred_8_fallback(uint8_t *_dst, ptrdiff_t dststride,
41		int16_t src1, int16_t src2, ptrdiff_t srcstride,
	41	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
42	42	int width, int height,
43	43	int w1,int o1, int w2,int o2, int log2WD);
44	44
	45	void put_weighted_pred_avg_16_fallback(uint16_t *dst, ptrdiff_t dststride,
	46	const int16_t src1, const int16_t src2,
	47	ptrdiff_t srcstride, int width,
	48	int height, int bit_depth);
	49
	50	void put_unweighted_pred_16_fallback(uint16_t *_dst, ptrdiff_t dststride,
	51	const int16_t *src, ptrdiff_t srcstride,
	52	int width, int height, int bit_depth);
	53
	54	void put_weighted_pred_16_fallback(uint16_t *_dst, ptrdiff_t dststride,
	55	const int16_t *src, ptrdiff_t srcstride,
	56	int width, int height,
	57	int w,int o,int log2WD, int bit_depth);
	58	void put_weighted_bipred_16_fallback(uint16_t *_dst, ptrdiff_t dststride,
	59	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	60	int width, int height,
	61	int w1,int o1, int w2,int o2, int log2WD, int bit_depth);
	62
	63
	64
45	65	void put_epel_8_fallback(int16_t *dst, ptrdiff_t dststride,
46		uint8_t *_src, ptrdiff_t srcstride,
47		int width, int height,
48		int mx, int my, int16_t* mcbuffer);
49		void put_epel_hv_8_fallback(int16_t *dst, ptrdiff_t dststride,
50		uint8_t *_src, ptrdiff_t srcstride,
51		int width, int height,
52		int mx, int my, int16_t* mcbuffer);
	66	const uint8_t *_src, ptrdiff_t srcstride,
	67	int width, int height,
	68	int mx, int my, int16_t* mcbuffer);
53	69
54		void put_qpel_0_0_fallback(int16_t *out, ptrdiff_t out_stride,
55		uint8_t *src, ptrdiff_t srcstride,
56		int nPbW, int nPbH, int16_t* mcbuffer);
57		void put_qpel_0_1_fallback(int16_t *out, ptrdiff_t out_stride,
58		uint8_t *src, ptrdiff_t srcstride,
59		int nPbW, int nPbH, int16_t* mcbuffer);
60		void put_qpel_0_2_fallback(int16_t *out, ptrdiff_t out_stride,
61		uint8_t *src, ptrdiff_t srcstride,
62		int nPbW, int nPbH, int16_t* mcbuffer);
63		void put_qpel_0_3_fallback(int16_t *out, ptrdiff_t out_stride,
64		uint8_t *src, ptrdiff_t srcstride,
65		int nPbW, int nPbH, int16_t* mcbuffer);
66		void put_qpel_1_0_fallback(int16_t *out, ptrdiff_t out_stride,
67		uint8_t *src, ptrdiff_t srcstride,
68		int nPbW, int nPbH, int16_t* mcbuffer);
69		void put_qpel_1_1_fallback(int16_t *out, ptrdiff_t out_stride,
70		uint8_t *src, ptrdiff_t srcstride,
71		int nPbW, int nPbH, int16_t* mcbuffer);
72		void put_qpel_1_2_fallback(int16_t *out, ptrdiff_t out_stride,
73		uint8_t *src, ptrdiff_t srcstride,
74		int nPbW, int nPbH, int16_t* mcbuffer);
75		void put_qpel_1_3_fallback(int16_t *out, ptrdiff_t out_stride,
76		uint8_t *src, ptrdiff_t srcstride,
77		int nPbW, int nPbH, int16_t* mcbuffer);
78		void put_qpel_2_0_fallback(int16_t *out, ptrdiff_t out_stride,
79		uint8_t *src, ptrdiff_t srcstride,
80		int nPbW, int nPbH, int16_t* mcbuffer);
81		void put_qpel_2_1_fallback(int16_t *out, ptrdiff_t out_stride,
82		uint8_t *src, ptrdiff_t srcstride,
83		int nPbW, int nPbH, int16_t* mcbuffer);
84		void put_qpel_2_2_fallback(int16_t *out, ptrdiff_t out_stride,
85		uint8_t *src, ptrdiff_t srcstride,
86		int nPbW, int nPbH, int16_t* mcbuffer);
87		void put_qpel_2_3_fallback(int16_t *out, ptrdiff_t out_stride,
88		uint8_t *src, ptrdiff_t srcstride,
89		int nPbW, int nPbH, int16_t* mcbuffer);
90		void put_qpel_3_0_fallback(int16_t *out, ptrdiff_t out_stride,
91		uint8_t *src, ptrdiff_t srcstride,
92		int nPbW, int nPbH, int16_t* mcbuffer);
93		void put_qpel_3_1_fallback(int16_t *out, ptrdiff_t out_stride,
94		uint8_t *src, ptrdiff_t srcstride,
95		int nPbW, int nPbH, int16_t* mcbuffer);
96		void put_qpel_3_2_fallback(int16_t *out, ptrdiff_t out_stride,
97		uint8_t *src, ptrdiff_t srcstride,
98		int nPbW, int nPbH, int16_t* mcbuffer);
99		void put_qpel_3_3_fallback(int16_t *out, ptrdiff_t out_stride,
100		uint8_t *src, ptrdiff_t srcstride,
101		int nPbW, int nPbH, int16_t* mcbuffer);
	70	void put_epel_16_fallback(int16_t *out, ptrdiff_t out_stride,
	71	const uint16_t *src, ptrdiff_t src_stride,
	72	int width, int height,
	73	int mx, int my, int16_t* mcbuffer, int bit_depth);
	74
	75	template <class pixel_t>
	76	void put_epel_hv_fallback(int16_t *dst, ptrdiff_t dststride,
	77	const pixel_t *_src, ptrdiff_t srcstride,
	78	int width, int height,
	79	int mx, int my, int16_t* mcbuffer, int bit_depth);
	80
	81
	82	#define QPEL(x,y) void put_qpel_ ## x ## _ ## y ## _fallback(int16_t *out, ptrdiff_t out_stride, \
	83	const uint8_t *src, ptrdiff_t srcstride, \
	84	int nPbW, int nPbH, int16_t* mcbuffer)
	85	QPEL(0,0); QPEL(0,1); QPEL(0,2); QPEL(0,3);
	86	QPEL(1,0); QPEL(1,1); QPEL(1,2); QPEL(1,3);
	87	QPEL(2,0); QPEL(2,1); QPEL(2,2); QPEL(2,3);
	88	QPEL(3,0); QPEL(3,1); QPEL(3,2); QPEL(3,3);
	89
	90	#undef QPEL
	91
	92
	93	#define QPEL(x,y) void put_qpel_ ## x ## _ ## y ## _fallback_16(int16_t *out, ptrdiff_t out_stride, \
	94	const uint16_t *src, ptrdiff_t srcstride, \
	95	int nPbW, int nPbH, int16_t* mcbuffer, int bit_depth)
	96	QPEL(0,0); QPEL(0,1); QPEL(0,2); QPEL(0,3);
	97	QPEL(1,0); QPEL(1,1); QPEL(1,2); QPEL(1,3);
	98	QPEL(2,0); QPEL(2,1); QPEL(2,2); QPEL(2,3);
	99	QPEL(3,0); QPEL(3,1); QPEL(3,2); QPEL(3,3);
	100
	101	#undef QPEL
102	102
103	103	#endif

+74

-10

libde265/fallback.cc less more

26	26	{
27	27	accel->put_weighted_pred_avg_8 = put_weighted_pred_avg_8_fallback;
28	28	accel->put_unweighted_pred_8 = put_unweighted_pred_8_fallback;
29
30	29	accel->put_weighted_pred_8 = put_weighted_pred_8_fallback;
31	30	accel->put_weighted_bipred_8 = put_weighted_bipred_8_fallback;
32	31
	32	accel->put_weighted_pred_avg_16 = put_weighted_pred_avg_16_fallback;
	33	accel->put_unweighted_pred_16 = put_unweighted_pred_16_fallback;
	34	accel->put_weighted_pred_16 = put_weighted_pred_16_fallback;
	35	accel->put_weighted_bipred_16 = put_weighted_bipred_16_fallback;
	36
	37
33	38	accel->put_hevc_epel_8 = put_epel_8_fallback;
34		accel->put_hevc_epel_h_8 = put_epel_hv_8_fallback;
35		accel->put_hevc_epel_v_8 = put_epel_hv_8_fallback;
36		accel->put_hevc_epel_hv_8 = put_epel_hv_8_fallback;
	39	accel->put_hevc_epel_h_8 = put_epel_hv_fallback<uint8_t>;
	40	accel->put_hevc_epel_v_8 = put_epel_hv_fallback<uint8_t>;
	41	accel->put_hevc_epel_hv_8 = put_epel_hv_fallback<uint8_t>;
37	42
38	43	accel->put_hevc_qpel_8[0][0] = put_qpel_0_0_fallback;
39	44	accel->put_hevc_qpel_8[0][1] = put_qpel_0_1_fallback;

52	57	accel->put_hevc_qpel_8[3][2] = put_qpel_3_2_fallback;
53	58	accel->put_hevc_qpel_8[3][3] = put_qpel_3_3_fallback;
54	59
	60	accel->put_hevc_epel_16 = put_epel_16_fallback;
	61	accel->put_hevc_epel_h_16 = put_epel_hv_fallback<uint16_t>;
	62	accel->put_hevc_epel_v_16 = put_epel_hv_fallback<uint16_t>;
	63	accel->put_hevc_epel_hv_16 = put_epel_hv_fallback<uint16_t>;
	64
	65	accel->put_hevc_qpel_16[0][0] = put_qpel_0_0_fallback_16;
	66	accel->put_hevc_qpel_16[0][1] = put_qpel_0_1_fallback_16;
	67	accel->put_hevc_qpel_16[0][2] = put_qpel_0_2_fallback_16;
	68	accel->put_hevc_qpel_16[0][3] = put_qpel_0_3_fallback_16;
	69	accel->put_hevc_qpel_16[1][0] = put_qpel_1_0_fallback_16;
	70	accel->put_hevc_qpel_16[1][1] = put_qpel_1_1_fallback_16;
	71	accel->put_hevc_qpel_16[1][2] = put_qpel_1_2_fallback_16;
	72	accel->put_hevc_qpel_16[1][3] = put_qpel_1_3_fallback_16;
	73	accel->put_hevc_qpel_16[2][0] = put_qpel_2_0_fallback_16;
	74	accel->put_hevc_qpel_16[2][1] = put_qpel_2_1_fallback_16;
	75	accel->put_hevc_qpel_16[2][2] = put_qpel_2_2_fallback_16;
	76	accel->put_hevc_qpel_16[2][3] = put_qpel_2_3_fallback_16;
	77	accel->put_hevc_qpel_16[3][0] = put_qpel_3_0_fallback_16;
	78	accel->put_hevc_qpel_16[3][1] = put_qpel_3_1_fallback_16;
	79	accel->put_hevc_qpel_16[3][2] = put_qpel_3_2_fallback_16;
	80	accel->put_hevc_qpel_16[3][3] = put_qpel_3_3_fallback_16;
	81
	82
	83
55	84	accel->transform_skip_8 = transform_skip_8_fallback;
56		accel->transform_bypass_8 = transform_bypass_8_fallback;
57		accel->transform_4x4_luma_add_8 = transform_4x4_luma_add_8_fallback;
58		accel->transform_4x4_add_8 = transform_4x4_add_8_fallback;
59		accel->transform_8x8_add_8 = transform_8x8_add_8_fallback;
60		accel->transform_16x16_add_8 = transform_16x16_add_8_fallback;
61		accel->transform_32x32_add_8 = transform_32x32_add_8_fallback;
	85	accel->transform_skip_rdpcm_h_8 = transform_skip_rdpcm_h_8_fallback;
	86	accel->transform_skip_rdpcm_v_8 = transform_skip_rdpcm_v_8_fallback;
	87	accel->transform_bypass = transform_bypass_fallback;
	88	accel->transform_bypass_rdpcm_h = transform_bypass_rdpcm_h_fallback;
	89	accel->transform_bypass_rdpcm_v = transform_bypass_rdpcm_v_fallback;
	90	accel->transform_4x4_dst_add_8 = transform_4x4_luma_add_8_fallback;
	91	accel->transform_add_8[0] = transform_4x4_add_8_fallback;
	92	accel->transform_add_8[1] = transform_8x8_add_8_fallback;
	93	accel->transform_add_8[2] = transform_16x16_add_8_fallback;
	94	accel->transform_add_8[3] = transform_32x32_add_8_fallback;
	95
	96	accel->transform_skip_16 = transform_skip_16_fallback;
	97	accel->transform_4x4_dst_add_16 = transform_4x4_luma_add_16_fallback;
	98	accel->transform_add_16[0] = transform_4x4_add_16_fallback;
	99	accel->transform_add_16[1] = transform_8x8_add_16_fallback;
	100	accel->transform_add_16[2] = transform_16x16_add_16_fallback;
	101	accel->transform_add_16[3] = transform_32x32_add_16_fallback;
	102
	103	accel->rotate_coefficients = rotate_coefficients_fallback;
	104	accel->add_residual_8 = add_residual_fallback<uint8_t>;
	105	accel->add_residual_16 = add_residual_fallback<uint16_t>;
	106	accel->rdpcm_h = rdpcm_h_fallback;
	107	accel->rdpcm_v = rdpcm_v_fallback;
	108	accel->transform_skip_residual = transform_skip_residual_fallback;
	109
	110	accel->transform_idst_4x4 = transform_idst_4x4_fallback;
	111	accel->transform_idct_4x4 = transform_idct_4x4_fallback;
	112	accel->transform_idct_8x8 = transform_idct_8x8_fallback;
	113	accel->transform_idct_16x16 = transform_idct_16x16_fallback;
	114	accel->transform_idct_32x32 = transform_idct_32x32_fallback;
	115
	116	accel->fwd_transform_4x4_dst_8 = fdst_4x4_8_fallback;
	117	accel->fwd_transform_8[0] = fdct_4x4_8_fallback;
	118	accel->fwd_transform_8[1] = fdct_8x8_8_fallback;
	119	accel->fwd_transform_8[2] = fdct_16x16_8_fallback;
	120	accel->fwd_transform_8[3] = fdct_32x32_8_fallback;
	121
	122	accel->hadamard_transform_8[0] = hadamard_4x4_8_fallback;
	123	accel->hadamard_transform_8[1] = hadamard_8x8_8_fallback;
	124	accel->hadamard_transform_8[2] = hadamard_16x16_8_fallback;
	125	accel->hadamard_transform_8[3] = hadamard_32x32_8_fallback;
62	126	}

+194

-0

libde265/image-io.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#include "libde265/image-io.h"
	23	#include <assert.h>
	24
	25
	26	ImageSource_YUV::~ImageSource_YUV()
	27	{
	28	if (mFH) {
	29	fclose(mFH);
	30	}
	31	}
	32
	33
	34	bool ImageSource_YUV::set_input_file(const char* filename, int w,int h)
	35	{
	36	assert(mFH==NULL);
	37
	38	mFH = fopen(filename,"rb");
	39	if (mFH==NULL) {
	40	return false;
	41	}
	42
	43	width =w;
	44	height=h;
	45	mReachedEndOfFile = false;
	46
	47	return true;
	48	}
	49
	50
	51	de265_image* ImageSource_YUV::read_next_image()
	52	{
	53	if (mReachedEndOfFile) return NULL;
	54
	55	de265_image* img = new de265_image;
	56	img->alloc_image(width,height,de265_chroma_420, NULL, false,
	57	NULL, NULL, 0, NULL, false);
	58	assert(img); // TODO: error handling
	59
	60	// --- load image ---
	61
	62	uint8_t* p;
	63	int stride;
	64
	65	p = img->get_image_plane(0); stride = img->get_image_stride(0);
	66	for (int y=0;y<height;y++) {
	67	if (fread(p+y*stride,1,width,mFH) != width) {
	68	goto check_eof;
	69	}
	70	}
	71
	72	p = img->get_image_plane(1); stride = img->get_image_stride(1);
	73	for (int y=0;y<height/2;y++) {
	74	if (fread(p+y*stride,1,width/2,mFH) != width/2) {
	75	goto check_eof;
	76	}
	77	}
	78
	79	p = img->get_image_plane(2); stride = img->get_image_stride(2);
	80	for (int y=0;y<height/2;y++) {
	81	if (fread(p+y*stride,1,width/2,mFH) != width/2) {
	82	goto check_eof;
	83	}
	84	}
	85
	86	// --- check for EOF ---
	87
	88	check_eof:
	89	if (feof(mFH)) {
	90	mReachedEndOfFile = true;
	91	delete img;
	92	return NULL;
	93	}
	94	else {
	95	return img;
	96	}
	97	}
	98
	99
	100	/*
	101	ImageSource::ImageStatus ImageSource_YUV::get_status()
	102	{
	103	return Available;
	104	}
	105	*/
	106
	107	de265_image* ImageSource_YUV::get_image(bool block)
	108	{
	109	de265_image* img = read_next_image();
	110	return img;
	111	}
	112
	113
	114	void ImageSource_YUV::skip_frames(int n)
	115	{
	116	int imageSize = widthheight3/2;
	117	fseek(mFH,n * imageSize, SEEK_CUR);
	118	}
	119
	120
	121
	122
	123	ImageSink_YUV::~ImageSink_YUV()
	124	{
	125	if (mFH) {
	126	fclose(mFH);
	127	}
	128	}
	129
	130	bool ImageSink_YUV::set_filename(const char* filename)
	131	{
	132	assert(mFH==NULL);
	133
	134	mFH = fopen(filename,"wb");
	135
	136	return true;
	137	}
	138
	139	void ImageSink_YUV::send_image(const de265_image* img)
	140	{
	141	// --- write image ---
	142
	143	const uint8_t* p;
	144	int stride;
	145
	146	int width = img->get_width();
	147	int height= img->get_height();
	148
	149	p = img->get_image_plane(0); stride = img->get_image_stride(0);
	150	for (int y=0;y<height;y++) {
	151	fwrite(p+y*stride,1,width,mFH);
	152	}
	153
	154	p = img->get_image_plane(1); stride = img->get_image_stride(1);
	155	for (int y=0;y<height/2;y++) {
	156	fwrite(p+y*stride,1,width/2,mFH);
	157	}
	158
	159	p = img->get_image_plane(2); stride = img->get_image_stride(2);
	160	for (int y=0;y<height/2;y++) {
	161	fwrite(p+y*stride,1,width/2,mFH);
	162	}
	163	}
	164
	165
	166
	167	LIBDE265_API PacketSink_File::~PacketSink_File()
	168	{
	169	if (mFH) {
	170	fclose(mFH);
	171	}
	172	}
	173
	174
	175	LIBDE265_API void PacketSink_File::set_filename(const char* filename)
	176	{
	177	assert(mFH==NULL);
	178
	179	mFH = fopen(filename,"wb");
	180	}
	181
	182
	183	LIBDE265_API void PacketSink_File::send_packet(const uint8_t* data, int n)
	184	{
	185	uint8_t startCode[3];
	186	startCode[0] = 0;
	187	startCode[1] = 0;
	188	startCode[2] = 1;
	189
	190	fwrite(startCode,1,3,mFH);
	191	fwrite(data,1,n,mFH);
	192	fflush(mFH);
	193	}

+122

-0

libde265/image-io.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* Authors: struktur AG, Dirk Farin <farin@struktur.de>
	5	*
	6	* This file is part of libde265.
	7	*
	8	* libde265 is free software: you can redistribute it and/or modify
	9	* it under the terms of the GNU Lesser General Public License as
	10	* published by the Free Software Foundation, either version 3 of
	11	* the License, or (at your option) any later version.
	12	*
	13	* libde265 is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU Lesser General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU Lesser General Public License
	19	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	20	*/
	21
	22	#ifndef IMAGE_IO_H
	23	#define IMAGE_IO_H
	24
	25	#include "libde265/image.h"
	26	#include <deque>
	27
	28
	29	class ImageSource
	30	{
	31	public:
	32	LIBDE265_API ImageSource() { }
	33	virtual LIBDE265_API ~ImageSource() { }
	34
	35	//enum ImageStatus { Available, Waiting, EndOfVideo };
	36
	37	//virtual ImageStatus get_status() = 0;
	38	virtual LIBDE265_API de265_image* get_image(bool block=true) = 0;
	39	virtual LIBDE265_API void skip_frames(int n) = 0;
	40
	41	virtual LIBDE265_API int get_width() const = 0;
	42	virtual LIBDE265_API int get_height() const = 0;
	43	};
	44
	45
	46
	47	class ImageSource_YUV : public ImageSource
	48	{
	49	public:
	50	LIBDE265_API ImageSource_YUV() : mFH(NULL) { }
	51	virtual LIBDE265_API ~ImageSource_YUV();
	52
	53	bool LIBDE265_API set_input_file(const char* filename, int w,int h);
	54
	55	//virtual ImageStatus get_status();
	56	virtual LIBDE265_API de265_image* get_image(bool block=true);
	57	virtual LIBDE265_API void skip_frames(int n);
	58
	59	virtual LIBDE265_API int get_width() const { return width; }
	60	virtual LIBDE265_API int get_height() const { return height; }
	61
	62	private:
	63	FILE* mFH;
	64	bool mReachedEndOfFile;
	65
	66	int width,height;
	67
	68	de265_image* read_next_image();
	69	};
	70
	71
	72
	73
	74	class ImageSink
	75	{
	76	public:
	77	virtual LIBDE265_API ~ImageSink() { }
	78
	79	virtual LIBDE265_API void send_image(const de265_image* img) = 0;
	80	};
	81
	82	class ImageSink_YUV : public ImageSink
	83	{
	84	public:
	85	LIBDE265_API ImageSink_YUV() : mFH(NULL) { }
	86	LIBDE265_API ~ImageSink_YUV();
	87
	88	bool LIBDE265_API set_filename(const char* filename);
	89
	90	virtual LIBDE265_API void send_image(const de265_image* img);
	91
	92	private:
	93	FILE* mFH;
	94	};
	95
	96
	97
	98	class PacketSink
	99	{
	100	public:
	101	virtual LIBDE265_API ~PacketSink() { }
	102
	103	virtual LIBDE265_API void send_packet(const uint8_t* data, int n) = 0;
	104	};
	105
	106
	107	class PacketSink_File : public PacketSink
	108	{
	109	public:
	110	LIBDE265_API PacketSink_File() : mFH(NULL) { }
	111	virtual LIBDE265_API ~PacketSink_File();
	112
	113	LIBDE265_API void set_filename(const char* filename);
	114
	115	virtual LIBDE265_API void send_packet(const uint8_t* data, int n);
	116
	117	private:
	118	FILE* mFH;
	119	};
	120
	121	#endif

+206

-65

libde265/image.cc less more

19	19
20	20	#include "image.h"
21	21	#include "decctx.h"
	22	#include "encoder/encoder-context.h"
22	23
23	24	#include <stdlib.h>
24	25	#include <string.h>

36	37	#else
37	38	#define MEMORY_PADDING 0
38	39	#endif
	40
	41	#define STANDARD_ALIGNMENT 16
39	42
40	43	#ifdef HAVE___MINGW_ALIGNED_MALLOC
41	44	#define ALLOC_ALIGNED(alignment, size) __mingw_aligned_malloc((size), (alignment))

61	64
62	65	static const int alignment = 16;
63	66
	67	LIBDE265_API void* de265_alloc_image_plane(struct de265_image* img, int cIdx,
	68	void* inputdata, int inputstride, void *userdata)
	69	{
	70	int alignment = STANDARD_ALIGNMENT;
	71	int stride = (img->get_width(cIdx) + alignment-1) / alignment * alignment;
	72	int height = img->get_height(cIdx);
	73
	74	uint8_t* p = (uint8_t )ALLOC_ALIGNED_16(stride height + MEMORY_PADDING);
	75
	76	if (p==NULL) { return NULL; }
	77
	78	img->set_image_plane(cIdx, p, stride, userdata);
	79
	80	// copy input data if provided
	81
	82	if (inputdata != NULL) {
	83	if (inputstride == stride) {
	84	memcpy(p, inputdata, stride*height);
	85	}
	86	else {
	87	for (int y=0;y<height;y++) {
	88	memcpy(p+ystride, ((char)inputdata) + inputstride*y, inputstride);
	89	}
	90	}
	91	}
	92
	93	return p;
	94	}
	95
	96
	97	LIBDE265_API void de265_free_image_plane(struct de265_image* img, int cIdx)
	98	{
	99	uint8_t* p = (uint8_t*)img->get_image_plane(cIdx);
	100	assert(p);
	101	FREE_ALIGNED(p);
	102	}
	103
64	104
65	105	static int de265_image_get_buffer(de265_decoder_context* ctx,
66	106	de265_image_spec* spec, de265_image* img, void* userdata)
67	107	{
68		int luma_stride = (spec->width + spec->alignment-1) / spec->alignment * spec->alignment;
69		int chroma_stride = (spec->width/2 + spec->alignment-1) / spec->alignment * spec->alignment;
	108	const int rawChromaWidth = spec->width / img->sps.SubWidthC;
	109	const int rawChromaHeight = spec->height / img->sps.SubHeightC;
	110
	111	int luma_stride = (spec->width + spec->alignment-1) / spec->alignment * spec->alignment;
	112	int chroma_stride = (rawChromaWidth + spec->alignment-1) / spec->alignment * spec->alignment;
	113
	114	assert(img->sps.BitDepth_Y >= 8 && img->sps.BitDepth_Y <= 16);
	115	assert(img->sps.BitDepth_C >= 8 && img->sps.BitDepth_C <= 16);
	116
	117	int luma_bpl = luma_stride * ((img->sps.BitDepth_Y+7)/8);
	118	int chroma_bpl = chroma_stride * ((img->sps.BitDepth_C+7)/8);
70	119
71	120	int luma_height = spec->height;
72		int chroma_height = (spec->height+1)/2;
	121	int chroma_height = rawChromaHeight;
	122
	123	bool alloc_failed = false;
73	124
74	125	uint8_t* p[3] = { 0,0,0 };
75		p[0] = (uint8_t )ALLOC_ALIGNED_16(luma_stride luma_height + MEMORY_PADDING);
76		p[1] = (uint8_t )ALLOC_ALIGNED_16(chroma_stride chroma_height + MEMORY_PADDING);
77		p[2] = (uint8_t )ALLOC_ALIGNED_16(chroma_stride chroma_height + MEMORY_PADDING);
78
79		if (p[0]==NULL \|\| p[1]==NULL \|\| p[2]==NULL) {
	126	p[0] = (uint8_t )ALLOC_ALIGNED_16(luma_height luma_bpl + MEMORY_PADDING);
	127	if (p[0]==NULL) { alloc_failed=true; }
	128
	129	if (img->get_chroma_format() != de265_chroma_mono) {
	130	p[1] = (uint8_t )ALLOC_ALIGNED_16(chroma_height chroma_bpl + MEMORY_PADDING);
	131	p[2] = (uint8_t )ALLOC_ALIGNED_16(chroma_height chroma_bpl + MEMORY_PADDING);
	132
	133	if (p[1]==NULL \|\| p[2]==NULL) { alloc_failed=true; }
	134	}
	135	else {
	136	p[1] = NULL;
	137	p[2] = NULL;
	138	chroma_stride = 0;
	139	}
	140
	141	if (alloc_failed) {
80	142	for (int i=0;i<3;i++)
81	143	if (p[i]) {
82	144	FREE_ALIGNED(p[i]);

97	159	{
98	160	for (int i=0;i<3;i++) {
99	161	uint8_t* p = (uint8_t*)img->get_image_plane(i);
100		assert(p);
101		FREE_ALIGNED(p);
	162	if (p) {
	163	FREE_ALIGNED(p);
	164	}
102	165	}
103	166	}
104	167

127	190	removed_at_picture_id = 0; // picture not used, so we can assume it has been removed
128	191
129	192	decctx = NULL;
	193	encctx = NULL;
	194
	195	encoder_image_release_func = NULL;
130	196
131	197	//alloc_functions.get_buffer = NULL;
132	198	//alloc_functions.release_buffer = NULL;

164	230
165	231	de265_error de265_image::alloc_image(int w,int h, enum de265_chroma c,
166	232	const seq_parameter_set* sps, bool allocMetadata,
167		decoder_context* ctx, de265_PTS pts, void* user_data,
168		bool isOutputImage)
169		{
170		if (allocMetadata) { assert(sps); }
	233	decoder_context* dctx,
	234	encoder_context* ectx,
	235	de265_PTS pts, void* user_data,
	236	bool useCustomAllocFunc)
	237	{
	238	//if (allocMetadata) { assert(sps); }
	239	assert(sps);
	240
	241	this->sps = *sps;
	242
	243	release(); /* TODO: review code for efficient allocation when arrays are already
	244	allocated to the requested size. Without the release, the old image-data
	245	will not be freed. */
171	246
172	247	ID = s_next_image_ID++;
173	248	removed_at_picture_id = std::numeric_limits<int32_t>::max();
174	249
175		decctx = ctx;
	250	decctx = dctx;
	251	encctx = ectx;
176	252
177	253	// --- allocate image buffer ---
178	254

208	284
209	285	case de265_chroma_422:
210	286	spec.format = de265_image_format_YUV422P8;
211		chroma_height = (chroma_height+1)/2;
	287	chroma_width = (chroma_width+1)/2;
212	288	break;
213	289
	290	case de265_chroma_444:
	291	spec.format = de265_image_format_YUV444P8;
	292	break;
	293
	294	case de265_chroma_mono:
	295	spec.format = de265_image_format_mono8;
	296	chroma_width = 0;
	297	chroma_height= 0;
	298	break;
	299
214	300	default:
215		assert(false); // TODO: not implemented yet
	301	assert(false);
216	302	break;
217	303	}
218	304
219	305	spec.width = w;
220	306	spec.height = h;
221		spec.alignment = 16;
	307	spec.alignment = STANDARD_ALIGNMENT;
222	308
223	309
224	310	// conformance window cropping

242	328	spec.visible_height= height_confwin;
243	329
244	330
	331	bpp_shift[0] = (sps->BitDepth_Y > 8) ? 1 : 0;
	332	bpp_shift[1] = (sps->BitDepth_C > 8) ? 1 : 0;
	333	bpp_shift[2] = bpp_shift[1];
	334
	335
245	336	// allocate memory and set conformance window pointers
246	337
247		void* alloc_userdata = decctx->param_image_allocation_userdata;
248		if (isOutputImage) {
	338	void* alloc_userdata = NULL;
	339	if (decctx) alloc_userdata = decctx->param_image_allocation_userdata;
	340	if (encctx) alloc_userdata = encctx->param_image_allocation_userdata; // actually not needed
	341
	342	if (encctx && useCustomAllocFunc) {
	343	encoder_image_release_func = encctx->release_func;
	344
	345	// if we do not provide a release function, use our own
	346
	347	if (encoder_image_release_func == NULL) {
	348	image_allocation_functions = de265_image::default_image_allocation;
	349	}
	350	else {
	351	image_allocation_functions.get_buffer = NULL;
	352	image_allocation_functions.release_buffer = NULL;
	353	}
	354	}
	355	else if (decctx && useCustomAllocFunc) {
249	356	image_allocation_functions = decctx->param_image_allocation_functions;
250	357	}
251	358	else {
252	359	image_allocation_functions = de265_image::default_image_allocation;
253	360	}
254		bool mem_alloc_success = image_allocation_functions.get_buffer(decctx, &spec, this,
255		alloc_userdata);
256
257		pixels_confwin[0] = pixels[0] + leftWinUnitX + topWinUnitY*stride;
258		pixels_confwin[1] = pixels[1] + left + top*chroma_stride;
259		pixels_confwin[2] = pixels[2] + left + top*chroma_stride;
260
261
262		// check for memory shortage
263
264		if (!mem_alloc_success)
265		{
266		return DE265_ERROR_OUT_OF_MEMORY;
267		}
	361
	362	bool mem_alloc_success = true;
	363
	364	if (image_allocation_functions.get_buffer != NULL) {
	365	mem_alloc_success = image_allocation_functions.get_buffer(decctx, &spec, this,
	366	alloc_userdata);
	367
	368	pixels_confwin[0] = pixels[0] + leftWinUnitX + topWinUnitY*stride;
	369	pixels_confwin[1] = pixels[1] + left + top*chroma_stride;
	370	pixels_confwin[2] = pixels[2] + left + top*chroma_stride;
	371
	372
	373	// check for memory shortage
	374
	375	if (!mem_alloc_success)
	376	{
	377	return DE265_ERROR_OUT_OF_MEMORY;
	378	}
	379	}
268	380
269	381	//alloc_functions = *allocfunc;
270	382	//alloc_userdata = userdata;

276	388
277	389	mem_alloc_success &= intraPredMode.alloc(sps->PicWidthInMinPUs, sps->PicHeightInMinPUs,
278	390	sps->Log2MinPUSize);
	391
	392	mem_alloc_success &= intraPredModeC.alloc(sps->PicWidthInMinPUs, sps->PicHeightInMinPUs,
	393	sps->Log2MinPUSize);
279	394
280	395	// cb info
281	396

348	463
349	464	if (pixels[0])
350	465	{
351		image_allocation_functions.release_buffer(decctx, this,
352		decctx ? decctx->param_image_allocation_userdata : NULL);
353
	466	if (encoder_image_release_func != NULL) {
	467	encoder_image_release_func(encctx, this,
	468	encctx->param_image_allocation_userdata);
	469	}
	470	else {
	471	image_allocation_functions.release_buffer(decctx, this,
	472	decctx ?
	473	decctx->param_image_allocation_userdata :
	474	NULL);
	475	}
	476
354	477	for (int i=0;i<3;i++)
355	478	{
356	479	pixels[i] = NULL;

392	515	*/
393	516
394	517	de265_error err = alloc_image(src->width, src->height, src->chroma_format, &src->sps, false,
395		src->decctx, src->pts, src->user_data, false);
	518	src->decctx, src->encctx, src->pts, src->user_data, false);
396	519	if (err != DE265_OK) {
397	520	return err;
398	521	}

411	534	assert(first % 2 == 0);
412	535	assert(end % 2 == 0);
413	536
	537	int luma_bpp = (sps.BitDepth_Y+7)/8;
	538	int chroma_bpp = (sps.BitDepth_C+7)/8;
	539
414	540	if (src->stride == stride) {
415		memcpy(pixels[0] + first*stride,
416		src->pixels[0] + first*src->stride,
417		(end-first)*stride);
	541	memcpy(pixels[0] + firststride luma_bpp,
	542	src->pixels[0] + firstsrc->stride luma_bpp,
	543	(end-first)stride luma_bpp);
418	544	}
419	545	else {
420	546	for (int yp=first;yp<end;yp++) {
421		memcpy(pixels[0]+ypstride, src->pixels[0]+ypsrc->stride, src->width);
422		}
423		}
424
425		int first_chroma = first>>1;
426		int end_chroma = end>>1;
	547	memcpy(pixels[0]+ypstride luma_bpp,
	548	src->pixels[0]+ypsrc->stride luma_bpp,
	549	src->width * luma_bpp);
	550	}
	551	}
	552
	553	int first_chroma = first / src->sps.SubHeightC;
	554	int end_chroma = end / src->sps.SubHeightC;
427	555
428	556	if (src->chroma_format != de265_chroma_mono) {
429	557	if (src->chroma_stride == chroma_stride) {
430		memcpy(pixels[1] + first_chroma*chroma_stride,
431		src->pixels[1] + first_chroma*chroma_stride,
432		(end_chroma-first_chroma) * chroma_stride);
433		memcpy(pixels[2] + first_chroma*chroma_stride,
434		src->pixels[2] + first_chroma*chroma_stride,
435		(end_chroma-first_chroma) * chroma_stride);
	558	memcpy(pixels[1] + first_chromachroma_stride chroma_bpp,
	559	src->pixels[1] + first_chromachroma_stride chroma_bpp,
	560	(end_chroma-first_chroma) * chroma_stride * chroma_bpp);
	561	memcpy(pixels[2] + first_chromachroma_stride chroma_bpp,
	562	src->pixels[2] + first_chromachroma_stride chroma_bpp,
	563	(end_chroma-first_chroma) * chroma_stride * chroma_bpp);
436	564	}
437	565	else {
438	566	for (int y=first_chroma;y<end_chroma;y++) {
439		memcpy(pixels[1]+ychroma_stride, src->pixels[1]+ysrc->chroma_stride, src->chroma_width);
440		memcpy(pixels[2]+ychroma_stride, src->pixels[2]+ysrc->chroma_stride, src->chroma_width);
	567	memcpy(pixels[1]+ychroma_stride chroma_bpp,
	568	src->pixels[1]+ysrc->chroma_stride chroma_bpp,
	569	src->chroma_width * chroma_bpp);
	570	memcpy(pixels[2]+ychroma_stride chroma_bpp,
	571	src->pixels[2]+ysrc->chroma_stride chroma_bpp,
	572	src->chroma_width * chroma_bpp);
441	573	}
442	574	}
443	575	}

462	594	{
463	595	de265_mutex_lock(&mutex);
464	596
	597	//printf("nThreads before: %d %d\n",nThreadsQueued, nThreadsTotal);
	598
465	599	nThreadsQueued += nThreads;
466	600	nThreadsTotal += nThreads;
467	601
	602	//printf("nThreads after: %d %d\n",nThreadsQueued, nThreadsTotal);
	603
468	604	de265_mutex_unlock(&mutex);
469	605	}
470	606
471		void de265_image::thread_run()
472		{
	607	void de265_image::thread_run(const thread_task* task)
	608	{
	609	//printf("run thread %s\n", task->name().c_str());
	610
473	611	de265_mutex_lock(&mutex);
474	612	nThreadsQueued--;
475	613	nThreadsRunning++;

492	630	de265_mutex_unlock(&mutex);
493	631	}
494	632
495		void de265_image::thread_finishes()
496		{
	633	void de265_image::thread_finishes(const thread_task* task)
	634	{
	635	//printf("finish thread %s\n", task->name().c_str());
	636
497	637	de265_mutex_lock(&mutex);
498	638
499	639	nThreadsRunning--;

516	656
517	657	void de265_image::wait_for_progress(thread_task* task, int ctbAddrRS, int progress)
518	658	{
	659	if (task==NULL) { return; }
	660
519	661	de265_progress_lock* progresslock = &ctb_progress[ctbAddrRS];
520	662	if (progresslock->get_progress() < progress) {
521	663	thread_blocks();

557	699	// during decoding (especially log2CbSize), but it is unlikely to be faster than the memset.
558	700
559	701	cb_info.clear();
560		tu_info.clear();
	702	//tu_info.clear(); // done on the fly
561	703	ctb_info.clear();
562	704	deblk_info.clear();
563	705

569	711	}
570	712
571	713
572		void de265_image::set_mv_info(int x,int y, int nPbW,int nPbH, const PredVectorInfo* mv)
	714	void de265_image::set_mv_info(int x,int y, int nPbW,int nPbH, const MotionVectorSpec& mv)
573	715	{
574	716	int log2PuSize = 2;
575	717

583	725	for (int pby=0;pby<hPu;pby++)
584	726	for (int pbx=0;pbx<wPu;pbx++)
585	727	{
586		pb_info[ xPu+pbx + (yPu+pby)stride ].mvi = mv;
	728	pb_info[ xPu+pbx + (yPu+pby)*stride ].mv = mv;
587	729	}
588	730	}
589	731

634	776	availableN = available_zscan(xP,yP,xN,yN);
635	777	}
636	778	else {
637		availableN = !(nPbW<<1 == nCbS && nPbH<<1 == nCbS &&
	779	availableN = !(nPbW<<1 == nCbS && nPbH<<1 == nCbS && // NxN
638	780	partIdx==1 &&
639		yN >= yC+nPbH && xN < xC+nPbW);
	781	yN >= yC+nPbH && xN < xC+nPbW); // xN/yN inside partIdx 2
640	782	}
641	783
642	784	if (availableN && get_pred_mode(xN,yN) == MODE_INTRA) {

645	787
646	788	return availableN;
647	789	}
648

+242

-62

libde265/image.h less more

32	32	#include <stdbool.h>
33	33	#endif
34	34	#include "libde265/de265.h"
	35	#include "libde265/en265.h"
35	36	#include "libde265/sps.h"
36	37	#include "libde265/pps.h"
37	38	#include "libde265/motion.h"
38	39	#include "libde265/threads.h"
39	40	#include "libde265/slice.h"
40	41	#include "libde265/nal.h"
41
42	42
43	43	enum PictureState {
44	44	UnusedForReference,

77	77	#define CTB_PROGRESS_DEBLK_H 3
78	78	#define CTB_PROGRESS_SAO 4
79	79
	80	class decoder_context;
	81
80	82	template <class DataUnit> class MetaDataArray
81	83	{
82	84	public:
83	85	MetaDataArray() { data=NULL; data_size=0; log2unitSize=0; width_in_units=0; height_in_units=0; }
84	86	~MetaDataArray() { free(data); }
85	87
86		bool alloc(int w,int h, int _log2unitSize) {
	88	LIBDE265_CHECK_RESULT bool alloc(int w,int h, int _log2unitSize) {
87	89	int size = w*h;
88	90
89	91	if (size != data_size) {
90	92	free(data);
91	93	data = (DataUnit)malloc(size sizeof(DataUnit));
	94	if (data == NULL) {
	95	data_size = 0;
	96	return false;
	97	}
92	98	data_size = size;
93		width_in_units = w;
94		height_in_units = h;
95	99	}
	100
	101	width_in_units = w;
	102	height_in_units = h;
96	103
97	104	log2unitSize = _log2unitSize;
98	105

107	114	int unitX = x>>log2unitSize;
108	115	int unitY = y>>log2unitSize;
109	116
	117	assert(unitX >= 0 && unitX < width_in_units);
	118	assert(unitY >= 0 && unitY < height_in_units);
	119
110	120	return data[ unitX + unitY*width_in_units ];
111	121	}
112	122

114	124	int unitX = x>>log2unitSize;
115	125	int unitY = y>>log2unitSize;
116	126
	127	assert(unitX >= 0 && unitX < width_in_units);
	128	assert(unitY >= 0 && unitY < height_in_units);
	129
117	130	return data[ unitX + unitY*width_in_units ];
118	131	}
119	132

121	134	int unitX = x>>log2unitSize;
122	135	int unitY = y>>log2unitSize;
123	136
	137	assert(unitX >= 0 && unitX < width_in_units);
	138	assert(unitY >= 0 && unitY < height_in_units);
	139
124	140	data[ unitX + unitY*width_in_units ] = d;
125	141	}
126	142
127	143	DataUnit& operator[](int idx) { return data[idx]; }
128	144	const DataUnit& operator[](int idx) const { return data[idx]; }
	145
	146	int size() const { return data_size; }
129	147
130	148	// private:
131	149	DataUnit* data;

145	163	cb_info[ cbx + cby*cb_info.width_in_units ].Field = value; \
146	164	}
147	165
	166	#define CLEAR_TB_BLK(x,y,log2BlkWidth) \
	167	int tuX = x >> tu_info.log2unitSize; \
	168	int tuY = y >> tu_info.log2unitSize; \
	169	int width = 1 << (log2BlkWidth - tu_info.log2unitSize); \
	170	for (int tuy=tuY;tuy<tuY+width;tuy++) \
	171	for (int tux=tuX;tux<tuX+width;tux++) \
	172	{ \
	173	tu_info[ tux + tuy*tu_info.width_in_units ] = 0; \
	174	}
	175
148	176
149	177	typedef struct {
150	178	uint16_t SliceAddrRS;

152	180
153	181	sao_info saoInfo;
154	182	bool deblock; // this CTB has to be deblocked
155		bool has_pcm; // pcm is used in this CTB
156		bool has_cu_transquant_bypass; // transquant_bypass is used in this CTB
	183
	184	// The following flag helps to quickly check whether we have to
	185	// check all conditions in the SAO filter or whether we can skip them.
	186	bool has_pcm_or_cu_transquant_bypass; // pcm or transquant_bypass is used in this CTB
157	187	} CTB_info;
158	188
159	189
160	190	typedef struct {
161		uint8_t log2CbSize : 3; // [0;6] (1<<log2CbSize) = 64
	191	uint8_t log2CbSize : 3; /* [0;6] (1<<log2CbSize) = 64
	192	This is only set in the top-left corner of the CB.
	193	The other values should be zero.
	194	TODO: in the encoder, we have to clear to zero.
	195	Used in deblocking and QP-scale decoding */
162	196	uint8_t PartMode : 3; // (enum PartMode) [0;7] set only in top-left of CB
163		// TODO: could be removed if prediction-block-boundaries would be
164		// set during decoding
165		uint8_t ctDepth : 2; // [0:3]? (0:64, 1:32, 2:16, 3:8)
	197	// Used for spatial merging candidates in current frame
	198	// and for deriving interSplitFlag in decoding.
	199
	200	uint8_t ctDepth : 2; // [0:3]? (for CTB size 64: 0:64, 1:32, 2:16, 3:8)
	201	// Used for decoding/encoding split_cu flag.
	202
	203	// --- byte boundary ---
166	204	uint8_t PredMode : 2; // (enum PredMode) [0;2] must be saved for past images
167		uint8_t pcm_flag : 1; //
168		uint8_t cu_transquant_bypass : 1;
169
170		int8_t QP_Y;
171
172		// uint8_t pcm_flag; // TODO
	205	// Used in motion decoding.
	206	uint8_t pcm_flag : 1; // Stored for intra-prediction / SAO
	207	uint8_t cu_transquant_bypass : 1; // Stored for SAO
	208	// note: 4 bits left
	209
	210	// --- byte boundary ---
	211	int8_t QP_Y; // Stored for QP prediction
	212
173	213	} CB_ref_info;
174	214
175	215
176	216	typedef struct {
177		PredVectorInfo mvi; // TODO: this can be done in 16x16 grid
	217	MotionVectorSpec mv; // TODO: this can be done in 16x16 grid
178	218	} PB_ref_info;
	219
	220	// intraPredMode: Used for determining scanIdx when decoding/encoding coefficients.
179	221
180	222
181	223

184	226	~de265_image();
185	227
186	228
187		de265_error alloc_image(int w,int h, enum de265_chroma c, const seq_parameter_set* sps,
188		bool allocMetadata, decoder_context* ctx, de265_PTS pts, void* user_data,
189		bool isOutputImage);
	229	de265_error alloc_image(int w,int h, enum de265_chroma c,
	230	const seq_parameter_set* sps,
	231	bool allocMetadata,
	232	decoder_context* dctx,
	233	class encoder_context* ectx,
	234	de265_PTS pts, void* user_data,
	235	bool useCustomAllocFunctions);
	236
	237	//de265_error alloc_encoder_data(const seq_parameter_set* sps);
190	238
191	239	bool is_allocated() const { return pixels[0] != NULL; }
192	240

211	259	return pixels[cIdx] + xpos + ypos*stride;
212	260	}
213	261
	262
	263	/// xpos;ypos in actual plane resolution
	264	template <class pixel_t>
	265	pixel_t* get_image_plane_at_pos_NEW(int cIdx, int xpos,int ypos)
	266	{
	267	int stride = get_image_stride(cIdx);
	268	return (pixel_t)(pixels[cIdx] + (xpos + yposstride)*sizeof(pixel_t));
	269	}
	270
214	271	const uint8_t* get_image_plane_at_pos(int cIdx, int xpos,int ypos) const
215	272	{
216	273	int stride = get_image_stride(cIdx);
217	274	return pixels[cIdx] + xpos + ypos*stride;
218	275	}
219	276
	277	void* get_image_plane_at_pos_any_depth(int cIdx, int xpos,int ypos)
	278	{
	279	int stride = get_image_stride(cIdx);
	280	return pixels[cIdx] + ((xpos + ypos*stride) << bpp_shift[cIdx]);
	281	}
	282
	283	const void* get_image_plane_at_pos_any_depth(int cIdx, int xpos,int ypos) const
	284	{
	285	int stride = get_image_stride(cIdx);
	286	return pixels[cIdx] + ((xpos + ypos*stride) << bpp_shift[cIdx]);
	287	}
	288
	289	/* Number of pixels in one row (not number of bytes).
	290	*/
220	291	int get_image_stride(int cIdx) const
221	292	{
222	293	if (cIdx==0) return stride;

231	302
232	303	enum de265_chroma get_chroma_format() const { return chroma_format; }
233	304
	305	int get_bit_depth(int cIdx) const {
	306	if (cIdx==0) return sps.BitDepth_Y;
	307	else return sps.BitDepth_C;
	308	}
	309
	310	int get_bytes_per_pixel(int cIdx) const {
	311	return (get_bit_depth(cIdx)+7)/8;
	312	}
	313
	314	bool high_bit_depth(int cIdx) const {
	315	return get_bit_depth(cIdx)>8;
	316	}
234	317
235	318	bool can_be_released() const { return PicOutputFlag==false && PicState==UnusedForReference; }
236	319

250	333
251	334	static de265_image_allocation default_image_allocation;
252	335
	336	void printBlk(const char* title, int x0,int y0,int blkSize,int cIdx) const {
	337	::printBlk(title, get_image_plane_at_pos(cIdx,x0,y0),
	338	blkSize, get_image_stride(cIdx));
	339	}
	340
253	341	private:
254	342	uint32_t ID;
255	343	static uint32_t s_next_image_ID;
256	344
257	345	uint8_t* pixels[3];
	346	uint8_t bpp_shift[3]; // 0 for 8 bit, 1 for 16 bit
258	347
259	348	enum de265_chroma chroma_format;
260	349

290	379	seq_parameter_set sps; // the SPS used for decoding this image
291	380	pic_parameter_set pps; // the PPS used for decoding this image
292	381	decoder_context* decctx;
	382	class encoder_context* encctx;
	383
	384	int number_of_ctbs() const { return ctb_info.size(); }
293	385
294	386	private:
295	387	MetaDataArray<CTB_info> ctb_info;
296	388	MetaDataArray<CB_ref_info> cb_info;
297	389	MetaDataArray<PB_ref_info> pb_info;
298	390	MetaDataArray<uint8_t> intraPredMode;
	391	MetaDataArray<uint8_t> intraPredModeC;
299	392	MetaDataArray<uint8_t> tu_info;
300	393	MetaDataArray<uint8_t> deblk_info;
301	394

306	399	void* user_data;
307	400	void* plane_user_data[3]; // this is logically attached to the pixel data pointers
308	401	de265_image_allocation image_allocation_functions; // the functions used for memory allocation
	402	void (encoder_image_release_func)(en265_encoder_context,
	403	de265_image*,
	404	void* userdata);
309	405
310	406	uint8_t integrity; /* Whether an error occured while the image was decoded.
311	407	When generated, this is initialized to INTEGRITY_CORRECT,

327	423
328	424
329	425	void thread_start(int nThreads);
330		void thread_run();
	426	void thread_run(const thread_task*);
331	427	void thread_blocks();
332	428	void thread_unblocks();
333		void thread_finishes(); /* NOTE: you should not access any data in the thread_task after
334		calling this, as this function may unlock other threads that
335		will push this image to the output queue and free all decoder data. */
	429	/* NOTE: you should not access any data in the thread_task after
	430	calling this, as this function may unlock other threads that
	431	will push this image to the output queue and free all decoder data. */
	432	void thread_finishes(const thread_task*);
336	433
337	434	void wait_for_progress(thread_task* task, int ctbx,int ctby, int progress);
338	435	void wait_for_progress(thread_task* task, int ctbAddrRS, int progress);

383	480	return get_pred_mode(x,y)==MODE_SKIP;
384	481	}
385	482
386		void set_pcm_flag(int x,int y, int log2BlkWidth)
387		{
388		SET_CB_BLK(x,y,log2BlkWidth, pcm_flag, 1);
389		ctb_info.get(x,y).has_pcm = true;
	483	void set_pcm_flag(int x,int y, int log2BlkWidth, uint8_t value=1)
	484	{
	485	SET_CB_BLK(x,y,log2BlkWidth, pcm_flag, value);
	486
	487	// TODO: in the encoder, we somewhere have to clear this
	488	ctb_info.get(x,y).has_pcm_or_cu_transquant_bypass = true;
390	489	}
391	490
392	491	int get_pcm_flag(int x,int y) const

394	493	return cb_info.get(x,y).pcm_flag;
395	494	}
396	495
397		void set_cu_transquant_bypass(int x,int y, int log2BlkWidth)
398		{
399		SET_CB_BLK(x,y,log2BlkWidth, cu_transquant_bypass, 1);
400		ctb_info.get(x,y).has_cu_transquant_bypass = true;
	496	void set_cu_transquant_bypass(int x,int y, int log2BlkWidth, uint8_t value=1)
	497	{
	498	SET_CB_BLK(x,y,log2BlkWidth, cu_transquant_bypass, value);
	499
	500	// TODO: in the encoder, we somewhere have to clear this
	501	ctb_info.get(x,y).has_pcm_or_cu_transquant_bypass = true;
401	502	}
402	503
403	504	int get_cu_transquant_bypass(int x,int y) const

405	506	return cb_info.get(x,y).cu_transquant_bypass;
406	507	}
407	508
408		void set_log2CbSize(int x0, int y0, int log2CbSize)
409		{
	509	void set_log2CbSize(int x0, int y0, int log2CbSize, bool fill)
	510	{
	511	// In theory, we could assume that remaining cb_info blocks are initialized to zero.
	512	// But in corrupted streams, slices may overlap and set contradicting log2CbSizes.
	513	// We also need this for encoding.
	514	if (fill) {
	515	SET_CB_BLK(x0,y0,log2CbSize, log2CbSize, 0);
	516	}
	517
410	518	cb_info.get(x0,y0).log2CbSize = log2CbSize;
411
412		// assume that remaining cb_info blocks are initialized to zero
413	519	}
414	520
415	521	int get_log2CbSize(int x0, int y0) const

458	564	void set_split_transform_flag(int x0,int y0,int trafoDepth)
459	565	{
460	566	tu_info.get(x0,y0) \|= (1<<trafoDepth);
	567	}
	568
	569	void clear_split_transform_flags(int x0,int y0,int log2CbSize)
	570	{
	571	CLEAR_TB_BLK (x0,y0, log2CbSize);
461	572	}
462	573
463	574	int get_split_transform_flag(int x0,int y0,int trafoDepth) const

499	610	void set_IntraPredMode(int PUidx,int log2blkSize, enum IntraPredMode mode)
500	611	{
501	612	int pbSize = 1<<(log2blkSize - intraPredMode.log2unitSize);
502
	613
503	614	for (int y=0;y<pbSize;y++)
504	615	for (int x=0;x<pbSize;x++)
505	616	intraPredMode[PUidx + x + y*intraPredMode.width_in_units] = mode;
506	617	}
507	618
	619	void set_IntraPredMode(int x0,int y0,int log2blkSize,
	620	enum IntraPredMode mode)
	621	{
	622	int pbSize = 1<<(log2blkSize - intraPredMode.log2unitSize);
	623	int PUidx = (x0>>sps.Log2MinPUSize) + (y0>>sps.Log2MinPUSize)*sps.PicWidthInMinPUs;
	624
	625	for (int y=0;y<pbSize;y++)
	626	for (int x=0;x<pbSize;x++) {
	627	assert(x<sps.PicWidthInMinPUs);
	628	assert(y<sps.PicHeightInMinPUs);
	629
	630	int idx = PUidx + x + y*intraPredMode.width_in_units;
	631	assert(idx<intraPredMode.data_size);
	632	intraPredMode[idx] = mode;
	633	}
	634	}
	635
	636
	637	enum IntraPredMode get_IntraPredModeC(int x,int y) const
	638	{
	639	return (enum IntraPredMode)(intraPredModeC.get(x,y) & 0x3f);
	640	}
	641
	642	bool is_IntraPredModeC_Mode4(int x,int y) const
	643	{
	644	return intraPredModeC.get(x,y) & 0x80;
	645	}
	646
	647	void set_IntraPredModeC(int x0,int y0,int log2blkSize, enum IntraPredMode mode,
	648	bool is_mode4)
	649	{
	650	uint8_t combinedValue = mode;
	651	if (is_mode4) combinedValue \|= 0x80;
	652
	653	int pbSize = 1<<(log2blkSize - intraPredMode.log2unitSize);
	654	int PUidx = (x0>>sps.Log2MinPUSize) + (y0>>sps.Log2MinPUSize)*sps.PicWidthInMinPUs;
	655
	656	for (int y=0;y<pbSize;y++)
	657	for (int x=0;x<pbSize;x++) {
	658	assert(x<sps.PicWidthInMinPUs);
	659	assert(y<sps.PicHeightInMinPUs);
	660
	661	int idx = PUidx + x + y*intraPredModeC.width_in_units;
	662	assert(idx<intraPredModeC.data_size);
	663	intraPredModeC[idx] = combinedValue;
	664	}
	665	}
	666
	667
	668	/*
	669	// NOTE: encoder only
	670	void set_ChromaIntraPredMode(int x,int y,int log2BlkWidth, enum IntraChromaPredMode mode)
	671	{
	672	SET_CB_BLK (x, y, log2BlkWidth, intra_chroma_pred_mode, mode);
	673	}
	674
	675	// NOTE: encoder only
	676	enum IntraChromaPredMode get_ChromaIntraPredMode(int x,int y) const
	677	{
	678	return (enum IntraChromaPredMode)(cb_info.get(x,y).intra_chroma_pred_mode);
	679	}
	680	*/
508	681
509	682	// --- CTB metadata access ---
510	683

546	719	return ctb_info[ctb].SliceHeaderIndex;
547	720	}
548	721
	722	bool is_SliceHeader_available(int x,int y) const
	723	{
	724	int idx = ctb_info.get(x,y).SliceHeaderIndex;
	725	return idx >= 0 && idx < slices.size();
	726	}
	727
549	728	slice_segment_header* get_SliceHeader(int x, int y)
550	729	{
551		return slices[ get_SliceHeaderIndex(x,y) ];
	730	int idx = get_SliceHeaderIndex(x,y);
	731	if (idx >= slices.size()) { return NULL; }
	732	return slices[idx];
552	733	}
553	734
554	735	slice_segment_header* get_SliceHeaderCtb(int ctbX, int ctbY)
555	736	{
556		return slices[ get_SliceHeaderIndexCtb(ctbX,ctbY) ];
	737	int idx = get_SliceHeaderIndexCtb(ctbX,ctbY);
	738	if (idx >= slices.size()) { return NULL; }
	739	return slices[idx];
557	740	}
558	741
559	742	const slice_segment_header* get_SliceHeaderCtb(int ctbX, int ctbY) const
560	743	{
561		return slices[ get_SliceHeaderIndexCtb(ctbX,ctbY) ];
562		}
563
	744	int idx = get_SliceHeaderIndexCtb(ctbX,ctbY);
	745	if (idx >= slices.size()) { return NULL; }
	746	return slices[idx];
	747	}
	748
564	749	void set_sao_info(int ctbX,int ctbY,const sao_info* saoinfo)
565	750	{
566	751	sao_info* sao = &ctb_info[ctbX + ctbY*ctb_info.width_in_units].saoInfo;
567
	752
568	753	memcpy(sao,
569	754	saoinfo,
570	755	sizeof(sao_info));
571	756	}
572
	757
573	758	const sao_info* get_sao_info(int ctbX,int ctbY) const
574	759	{
575	760	return &ctb_info[ctbX + ctbY*ctb_info.width_in_units].saoInfo;

588	773	}
589	774
590	775
591		bool get_CTB_has_pcm(int ctbX,int ctbY) const
	776	bool get_CTB_has_pcm_or_cu_transquant_bypass(int ctbX,int ctbY) const
592	777	{
593	778	int idx = ctbX + ctbY*ctb_info.width_in_units;
594		return ctb_info[idx].has_pcm;
595		}
596
597		bool get_CTB_has_cu_transquant_bypass(int ctbX,int ctbY) const
598		{
599		int idx = ctbX + ctbY*ctb_info.width_in_units;
600		return ctb_info[idx].has_cu_transquant_bypass;
	779	return ctb_info[idx].has_pcm_or_cu_transquant_bypass;
601	780	}
602	781
603	782

611	790	{
612	791	const int xd = x0/4;
613	792	const int yd = y0/4;
614
	793
615	794	if (xd<deblk_info.width_in_units &&
616	795	yd<deblk_info.height_in_units) {
617	796	deblk_info[xd + yd*deblk_info.width_in_units] \|= flags;

641	820
642	821	// --- PB metadata access ---
643	822
644		const PredVectorInfo* get_mv_info(int x,int y) const
645		{
646		return &pb_info.get(x,y).mvi;
647		}
648
649		void set_mv_info(int x,int y, int nPbW,int nPbH, const PredVectorInfo* mv);
650
651		// --- value logging ---
652
	823	const MotionVectorSpec* get_mv_info(int x,int y) const
	824	{
	825	return &pb_info.get(x,y).mv;
	826	}
	827
	828	void set_mv_info(int x,int y, int nPbW,int nPbH, const MotionVectorSpec& mv);
	829
	830	// --- value logging ---
	831
	832	void printBlk(int x0,int y0, int cIdx, int log2BlkSize);
653	833	};
654	834
655	835

+346

-95

libde265/intrapred.cc less more

28	28
29	29
30	30	#ifdef DE265_LOG_TRACE
31		void print_border(uint8_t* data, uint8_t* available, int nT)
	31	template <class pixel_t>
	32	void print_border(pixel_t* data, uint8_t* available, int nT)
32	33	{
33	34	for (int i=-2nT ; i<=2nT ; i++) {
34	35	if (i==0 \|\| i==1 \|\| i==-nT \|\| i==nT+1) {

50	51	#endif
51	52
52	53
	54	void fillIntraPredModeCandidates(int candModeList[3], int x,int y, int PUidx,
	55	bool availableA, // left
	56	bool availableB, // top
	57	const de265_image* img)
	58	{
	59	const seq_parameter_set* sps = &img->sps;
	60
	61	// block on left side
	62
	63	enum IntraPredMode candIntraPredModeA, candIntraPredModeB;
	64	if (availableA==false) {
	65	candIntraPredModeA=INTRA_DC;
	66	}
	67	else if (img->get_pred_mode(x-1,y) != MODE_INTRA \|\|
	68	img->get_pcm_flag (x-1,y)) {
	69	candIntraPredModeA=INTRA_DC;
	70	}
	71	else {
	72	candIntraPredModeA = img->get_IntraPredMode_atIndex(PUidx-1);
	73	}
	74
	75	// block above
	76
	77	if (availableB==false) {
	78	candIntraPredModeB=INTRA_DC;
	79	}
	80	else if (img->get_pred_mode(x,y-1) != MODE_INTRA \|\|
	81	img->get_pcm_flag (x,y-1)) {
	82	candIntraPredModeB=INTRA_DC;
	83	}
	84	else if (y-1 < ((y >> sps->Log2CtbSizeY) << sps->Log2CtbSizeY)) {
	85	candIntraPredModeB=INTRA_DC;
	86	}
	87	else {
	88	candIntraPredModeB = img->get_IntraPredMode_atIndex(PUidx-sps->PicWidthInMinPUs);
	89	}
	90
	91
	92	// build candidate list
	93
	94	logtrace(LogSlice,"%d;%d availableA:%d candA:%d & availableB:%d candB:%d\n", x,y,
	95	availableA, candIntraPredModeA,
	96	availableB, candIntraPredModeB);
	97
	98	if (candIntraPredModeA == candIntraPredModeB) {
	99	if (candIntraPredModeA < 2) {
	100	candModeList[0] = INTRA_PLANAR;
	101	candModeList[1] = INTRA_DC;
	102	candModeList[2] = INTRA_ANGULAR_26;
	103	}
	104	else {
	105	candModeList[0] = candIntraPredModeA;
	106	candModeList[1] = 2 + ((candIntraPredModeA-2 -1 +32) % 32);
	107	candModeList[2] = 2 + ((candIntraPredModeA-2 +1 ) % 32);
	108	}
	109	}
	110	else {
	111	candModeList[0] = candIntraPredModeA;
	112	candModeList[1] = candIntraPredModeB;
	113
	114	if (candIntraPredModeA != INTRA_PLANAR &&
	115	candIntraPredModeB != INTRA_PLANAR) {
	116	candModeList[2] = INTRA_PLANAR;
	117	}
	118	else if (candIntraPredModeA != INTRA_DC &&
	119	candIntraPredModeB != INTRA_DC) {
	120	candModeList[2] = INTRA_DC;
	121	}
	122	else {
	123	candModeList[2] = INTRA_ANGULAR_26;
	124	}
	125	}
	126
	127	/*
	128	printf("candModeList: %d %d %d\n",
	129	candModeList[0],
	130	candModeList[1],
	131	candModeList[2]
	132	);
	133	*/
	134	}
	135
	136
	137	int find_intra_pred_mode(enum IntraPredMode mode,
	138	int candModeList[3])
	139	{
	140	// check whether the mode is in the candidate list
	141
	142	for (int i=0;i<3;i++) {
	143	if (candModeList[i] == mode) {
	144	return i;
	145	}
	146	}
	147
	148	// sort candModeList
	149
	150	if (candModeList[0] > candModeList[1]) {
	151	std::swap(candModeList[0],candModeList[1]);
	152	}
	153	if (candModeList[0] > candModeList[2]) {
	154	std::swap(candModeList[0],candModeList[2]);
	155	}
	156	if (candModeList[1] > candModeList[2]) {
	157	std::swap(candModeList[1],candModeList[2]);
	158	}
	159
	160	// skip modes already in the candidate list
	161
	162	int intraMode = mode;
	163
	164	for (int i=2;i>=0;i--) {
	165	if (intraMode >= candModeList[i]) { intraMode--; }
	166	}
	167
	168	return -intraMode-1;
	169	}
	170
	171
	172	void list_chroma_pred_candidates(enum IntraPredMode chroma_mode[5],
	173	enum IntraPredMode luma_mode)
	174	{
	175	enum IntraPredMode chroma_cand[5];
	176	chroma_cand[0] = INTRA_PLANAR;
	177	chroma_cand[1] = INTRA_ANGULAR_26;
	178	chroma_cand[2] = INTRA_ANGULAR_10;
	179	chroma_cand[3] = INTRA_DC;
	180	chroma_cand[4] = luma_mode;
	181
	182	switch (luma_mode) {
	183	case INTRA_PLANAR: chroma_cand[0] = INTRA_ANGULAR_34; break;
	184	case INTRA_ANGULAR_26: chroma_cand[1] = INTRA_ANGULAR_34; break;
	185	case INTRA_ANGULAR_10: chroma_cand[2] = INTRA_ANGULAR_34; break;
	186	case INTRA_DC: chroma_cand[3] = INTRA_ANGULAR_34; break;
	187	default:
	188	// use defaults from above
	189	break;
	190	}
	191	}
	192
	193
	194	int get_intra_scan_idx(int log2TrafoSize, enum IntraPredMode intraPredMode, int cIdx,
	195	const seq_parameter_set* sps)
	196	{
	197	if (log2TrafoSize==2 \|\|
	198	(log2TrafoSize==3 && (cIdx==0 \|\|
	199	sps->ChromaArrayType==CHROMA_444))) {
	200	/**/ if (intraPredMode >= 6 && intraPredMode <= 14) return 2;
	201	else if (intraPredMode >= 22 && intraPredMode <= 30) return 1;
	202	else return 0;
	203	}
	204	else { return 0; }
	205	}
	206
	207
	208	int get_intra_scan_idx_luma(int log2TrafoSize, enum IntraPredMode intraPredMode)
	209	{
	210	if (log2TrafoSize==2 \|\| log2TrafoSize==3) {
	211	/**/ if (intraPredMode >= 6 && intraPredMode <= 14) return 2;
	212	else if (intraPredMode >= 22 && intraPredMode <= 30) return 1;
	213	else return 0;
	214	}
	215	else { return 0; }
	216	}
	217
	218	int get_intra_scan_idx_chroma(int log2TrafoSize, enum IntraPredMode intraPredMode)
	219	{
	220	if (log2TrafoSize==1 \|\| log2TrafoSize==2) {
	221	/**/ if (intraPredMode >= 6 && intraPredMode <= 14) return 2;
	222	else if (intraPredMode >= 22 && intraPredMode <= 30) return 1;
	223	else return 0;
	224	}
	225	else { return 0; }
	226	}
	227
	228
	229	enum IntraPredMode lumaPredMode_to_chromaPredMode(enum IntraPredMode luma,
	230	enum IntraChromaPredMode chroma)
	231	{
	232	switch (chroma) {
	233	case INTRA_CHROMA_LIKE_LUMA:
	234	return luma;
	235
	236	case INTRA_CHROMA_PLANAR_OR_34:
	237	if (luma==INTRA_PLANAR) return INTRA_ANGULAR_34;
	238	else return INTRA_PLANAR;
	239
	240	case INTRA_CHROMA_ANGULAR_26_OR_34:
	241	if (luma==INTRA_ANGULAR_26) return INTRA_ANGULAR_34;
	242	else return INTRA_ANGULAR_26;
	243
	244	case INTRA_CHROMA_ANGULAR_10_OR_34:
	245	if (luma==INTRA_ANGULAR_10) return INTRA_ANGULAR_34;
	246	else return INTRA_ANGULAR_10;
	247
	248	case INTRA_CHROMA_DC_OR_34:
	249	if (luma==INTRA_DC) return INTRA_ANGULAR_34;
	250	else return INTRA_DC;
	251	}
	252
	253
	254	assert(false);
	255	return INTRA_DC;
	256	}
	257
	258
53	259	// (8.4.4.2.2)
54		void fill_border_samples(de265_image* img, int xB,int yB,
	260	template <class pixel_t>
	261	void fill_border_samples(de265_image* img,
	262	int xB,int yB, // in component specific resolution
55	263	int nT, int cIdx,
56		uint8_t* out_border)
	264	pixel_t* out_border)
57	265	{
58	266	const seq_parameter_set* sps = &img->sps;
59	267	const pic_parameter_set* pps = &img->pps;

61	269	uint8_t available_data[2*64 + 1];
62	270	uint8_t* available = &available_data[64];
63	271
64		uint8_t* image;
	272	pixel_t* image;
65	273	int stride;
66		image = img->get_image_plane(cIdx);
	274	image = (pixel_t*)img->get_image_plane(cIdx);
67	275	stride = img->get_image_stride(cIdx);
68	276
69		const int chromaShift = (cIdx==0) ? 0 : 1;
70		const int TUShift = (cIdx==0) ? sps->Log2MinTrafoSize : sps->Log2MinTrafoSize-1;
71
	277	const int SubWidth = (cIdx==0) ? 1 : sps->SubWidthC;
	278	const int SubHeight = (cIdx==0) ? 1 : sps->SubHeightC;
	279
	280	const int bit_depth = img->get_bit_depth(cIdx);
72	281
73	282	// --- check for CTB boundaries ---
74	283
75		int xBLuma = (cIdx==0) ? xB : 2*xB;
76		int yBLuma = (cIdx==0) ? yB : 2*yB;
77		int nTLuma = (cIdx==0) ? nT : 2*nT;
	284	int xBLuma = xB * SubWidth;
	285	int yBLuma = yB * SubHeight;
78	286
79	287	int log2CtbSize = sps->Log2CtbSizeY;
80	288	int picWidthInCtbs = sps->PicWidthInCtbsY;

84	292	bool availableTopRight=true; // is CTB at top-right side available?
85	293	bool availableTopLeft=true; // if CTB at top-left pixel available?
86	294
	295
	296	//printf("xB/yB: %d %d\n",xB,yB);
87	297
88	298	// are we at left image border
89	299

103	313	yBLuma = 0; // fake value, available flags are already set to false
104	314	}
105	315
106		if (xBLuma+nTLuma >= sps->pic_width_in_luma_samples) {
	316	if (xBLuma+nT*SubWidth >= sps->pic_width_in_luma_samples) {
107	317	availableTopRight=false;
108	318	}
109
	319
110	320	// check for tile and slice boundaries
111	321
112	322	int xCurrCtb = xBLuma >> log2CtbSize;
113	323	int yCurrCtb = yBLuma >> log2CtbSize;
114	324	int xLeftCtb = (xBLuma-1) >> log2CtbSize;
115		int xRightCtb = (xBLuma+nTLuma) >> log2CtbSize;
	325	int xRightCtb = (xBLuma+nT*SubWidth) >> log2CtbSize;
116	326	int yTopCtb = (yBLuma-1) >> log2CtbSize;
117	327
118	328	int currCTBSlice = img->get_SliceAddrRS(xCurrCtb,yCurrCtb);

121	331	int toprightCTBSlice = availableTopRight ? img->get_SliceAddrRS(xRightCtb, yTopCtb) : -1;
122	332	int topleftCTBSlice = availableTopLeft ? img->get_SliceAddrRS(xLeftCtb, yTopCtb) : -1;
123	333
	334	/*
	335	printf("size: %d\n",pps->TileIdRS.size());
	336	printf("curr: %d left: %d top: %d\n",
	337	xCurrCtb+yCurrCtb*picWidthInCtbs,
	338	availableLeft ? xLeftCtb+yCurrCtb*picWidthInCtbs : 9999,
	339	availableTop ? xCurrCtb+yTopCtb*picWidthInCtbs : 9999);
	340	*/
124	341	int currCTBTileID = pps->TileIdRS[xCurrCtb+yCurrCtb*picWidthInCtbs];
125	342	int leftCTBTileID = availableLeft ? pps->TileIdRS[xLeftCtb+yCurrCtb*picWidthInCtbs] : -1;
126	343	int topCTBTileID = availableTop ? pps->TileIdRS[xCurrCtb+yTopCtb*picWidthInCtbs] : -1;

138	355
139	356	// number of pixels that are in the valid image area to the right and to the bottom
140	357
141		int nBottom = sps->pic_height_in_luma_samples - (cIdx==0 ? yB : 2*yB);
142		if (cIdx) nBottom=(nBottom+1)/2;
	358	int nBottom = sps->pic_height_in_luma_samples - yB*SubHeight;
	359	nBottom=(nBottom+SubHeight-1)/SubHeight;
143	360	if (nBottom>2nT) nBottom=2nT;
144		int nRight = sps->pic_width_in_luma_samples - (cIdx==0 ? xB : 2*xB);
145		if (cIdx) nRight =(nRight +1)/2;
	361
	362	int nRight = sps->pic_width_in_luma_samples - xB*SubWidth;
	363	nRight =(nRight +SubWidth-1)/SubWidth;
146	364	if (nRight >2nT) nRight=2nT;
147	365
148	366	int nAvail=0;
149	367
150		uint8_t firstValue;
	368	pixel_t firstValue;
151	369
152	370	memset(available-2nT, 0, 4nT+1);
153	371

157	375	for (int y=nBottom-1 ; y>=0 ; y-=4)
158	376	if (availableLeft)
159	377	{
160		int NBlockAddr = pps->MinTbAddrZS[ ((xB-1)>>TUShift) +
161		((yB+y)>>TUShift) * sps->PicWidthInTbsY ];
162
163		bool availableN = NBlockAddr < currBlockAddr;
	378	int NBlockAddr = pps->MinTbAddrZS[ (((xB-1)*SubWidth )>>sps->Log2MinTrafoSize) +
	379	(((yB+y)*SubHeight)>>sps->Log2MinTrafoSize)
	380	* sps->PicWidthInTbsY ];
	381
	382	bool availableN = NBlockAddr <= currBlockAddr;
164	383
165	384	if (pps->constrained_intra_pred_flag) {
166		if (img->get_pred_mode((xB-1)<<chromaShift,(yB+y)<<chromaShift)!=MODE_INTRA)
	385	if (img->get_pred_mode((xB-1)SubWidth,(yB+y)SubHeight)!=MODE_INTRA)
167	386	availableN = false;
168	387	}
169	388

183	402
184	403	if (availableTopLeft)
185	404	{
186		int NBlockAddr = pps->MinTbAddrZS[ ((xB-1)>>TUShift) +
187		((yB-1)>>TUShift) * sps->PicWidthInTbsY ];
188
189		bool availableN = NBlockAddr < currBlockAddr;
	405	int NBlockAddr = pps->MinTbAddrZS[ (((xB-1)*SubWidth )>>sps->Log2MinTrafoSize) +
	406	(((yB-1)*SubHeight)>>sps->Log2MinTrafoSize)
	407	* sps->PicWidthInTbsY ];
	408
	409	bool availableN = NBlockAddr <= currBlockAddr;
190	410
191	411	if (pps->constrained_intra_pred_flag) {
192		if (img->get_pred_mode((xB-1)<<chromaShift,(yB-1)<<chromaShift)!=MODE_INTRA) {
	412	if (img->get_pred_mode((xB-1)SubWidth,(yB-1)SubHeight)!=MODE_INTRA) {
193	413	availableN = false;
194	414	}
195	415	}

212	432
213	433	if (borderAvailable)
214	434	{
215		int NBlockAddr = pps->MinTbAddrZS[ ((xB+x)>>TUShift) +
216		((yB-1)>>TUShift) * sps->PicWidthInTbsY ];
217
218		bool availableN = NBlockAddr < currBlockAddr;
	435	int NBlockAddr = pps->MinTbAddrZS[ (((xB+x)*SubWidth )>>sps->Log2MinTrafoSize) +
	436	(((yB-1)*SubHeight)>>sps->Log2MinTrafoSize)
	437	* sps->PicWidthInTbsY ];
	438
	439	bool availableN = NBlockAddr <= currBlockAddr;
219	440
220	441	if (pps->constrained_intra_pred_flag) {
221		if (img->get_pred_mode((xB+x)<<chromaShift,(yB-1)<<chromaShift)!=MODE_INTRA) {
	442	if (img->get_pred_mode((xB+x)SubWidth,(yB-1)SubHeight)!=MODE_INTRA) {
222	443	availableN = false;
223	444	}
224	445	}

242	463
243	464	if (nAvail!=4*nT+1) {
244	465	if (nAvail==0) {
245		memset(out_border-2nT, 1<<(sps->bit_depth_luma-1), 4nT+1);
	466	if (sizeof(pixel_t)==1) {
	467	memset(out_border-2nT, 1<<(bit_depth-1), 4nT+1);
	468	}
	469	else {
	470	for (int i = -2nT; i <= 2nT ; i++) {
	471	out_border[i] = 1<<(bit_depth-1);
	472	}
	473	}
246	474	}
247	475	else {
248	476	if (!available[-2*nT]) {

268	496
269	497
270	498	// (8.4.4.2.3)
	499	template <class pixel_t>
271	500	void intra_prediction_sample_filtering(de265_image* img,
272		uint8_t* p,
273		int nT,
	501	pixel_t* p,
	502	int nT, int cIdx,
274	503	enum IntraPredMode intraPredMode)
275	504	{
276	505	int filterFlag;

292	521
293	522	if (filterFlag) {
294	523	int biIntFlag = (img->sps.strong_intra_smoothing_enable_flag &&
	524	cIdx==0 &&
295	525	nT==32 &&
296	526	abs_value(p[0]+p[ 64]-2*p[ 32]) < (1<<(img->sps.bit_depth_luma-5)) &&
297	527	abs_value(p[0]+p[-64]-2*p[-32]) < (1<<(img->sps.bit_depth_luma-5)))
298	528	? 1 : 0;
299	529
300		uint8_t pF_mem[2*64+1];
301		uint8_t* pF = &pF_mem[64];
	530	pixel_t pF_mem[2*64+1];
	531	pixel_t* pF = &pF_mem[64];
302	532
303	533	if (biIntFlag) {
304	534	pF[-2nT] = p[-2nT];

322	552
323	553	// copy back to original array
324	554
325		memcpy(p-2nT, pF-2nT, 4*nT+1);
	555	memcpy(p-2nT, pF-2nT, (4nT+1) sizeof(pixel_t));
326	556	}
327	557	else {
328	558	// do nothing ?

344	574	-315,-390,-482,-630,-910,-1638,-4096 };
345	575
346	576
347		// TODO: clip to read BitDepthY
348		LIBDE265_INLINE static int Clip1Y(int x) { if (x<0) return 0; else if (x>255) return 255; else return x; }
349
350
351	577	// (8.4.4.2.6)
	578	template <class pixel_t>
352	579	void intra_prediction_angular(de265_image* img,
353	580	int xB0,int yB0,
354	581	enum IntraPredMode intraPredMode,
355	582	int nT,int cIdx,
356		uint8_t* border)
357		{
358		uint8_t ref_mem[2*64+1];
359		uint8_t* ref=&ref_mem[64];
360
361		uint8_t* pred;
	583	pixel_t* border)
	584	{
	585	pixel_t ref_mem[2*64+1];
	586	pixel_t* ref=&ref_mem[64];
	587
	588	pixel_t* pred;
362	589	int stride;
363		pred = img->get_image_plane_at_pos(cIdx,xB0,yB0);
	590	pred = img->get_image_plane_at_pos_NEW<pixel_t>(cIdx,xB0,yB0);
364	591	stride = img->get_image_stride(cIdx);
365	592
	593	int bit_depth = img->get_bit_depth(cIdx);
	594
	595	assert(intraPredMode<35);
	596	assert(intraPredMode>=2);
	597
366	598	int intraPredAngle = intraPredAngle_table[intraPredMode];
	599
	600	bool disableIntraBoundaryFilter =
	601	(img->sps.range_extension.implicit_rdpcm_enabled_flag &&
	602	img->get_cu_transquant_bypass(xB0,yB0));
367	603
368	604	if (intraPredMode >= 18) {
369	605

397	633	}
398	634	}
399	635
400		if (intraPredMode==26 && cIdx==0 && nT<32) {
	636	if (intraPredMode==26 && cIdx==0 && nT<32 && !disableIntraBoundaryFilter) {
401	637	for (int y=0;y<nT;y++) {
402		pred[0+y*stride] = Clip1Y(border[1] + ((border[-1-y] - border[0])>>1));
	638	pred[0+y*stride] = Clip_BitDepth(border[1] + ((border[-1-y] - border[0])>>1), bit_depth);
403	639	}
404	640	}
405	641	}

435	671	}
436	672	}
437	673
438		if (intraPredMode==10 && cIdx==0 && nT<32) { // DIFF 26->10
	674	if (intraPredMode==10 && cIdx==0 && nT<32 && !disableIntraBoundaryFilter) { // DIFF 26->10
439	675	for (int x=0;x<nT;x++) { // DIFF (x<->y)
440		pred[x] = Clip1Y(border[-1] + ((border[1+x] - border[0])>>1)); // DIFF (x<->y && neg)
	676	pred[x] = Clip_BitDepth(border[-1] + ((border[1+x] - border[0])>>1), bit_depth); // DIFF (x<->y && neg)
441	677	}
442	678	}
443	679	}

455	691	}
456	692
457	693
	694	template <class pixel_t>
458	695	void intra_prediction_planar(de265_image* img,int xB0,int yB0,int nT,int cIdx,
459		uint8_t* border)
460		{
461		uint8_t* pred;
	696	pixel_t* border)
	697	{
	698	pixel_t* pred;
462	699	int stride;
463		pred = img->get_image_plane_at_pos(cIdx,xB0,yB0);
	700	pred = img->get_image_plane_at_pos_NEW<pixel_t>(cIdx,xB0,yB0);
464	701	stride = img->get_image_stride(cIdx);
465	702
466	703	int Log2_nT = Log2(nT);

485	722	}
486	723
487	724
	725	template <class pixel_t>
488	726	void intra_prediction_DC(de265_image* img,int xB0,int yB0,int nT,int cIdx,
489		uint8_t* border)
490		{
491		uint8_t* pred;
	727	pixel_t* border)
	728	{
	729	pixel_t* pred;
492	730	int stride;
493		pred = img->get_image_plane_at_pos(cIdx,xB0,yB0);
	731	pred = img->get_image_plane_at_pos_NEW<pixel_t>(cIdx,xB0,yB0);
494	732	stride = img->get_image_stride(cIdx);
495	733
496	734	int Log2_nT = Log2(nT);

524	762	}
525	763
526	764
527		/*
528		printf("INTRAPRED DC\n");
529		for (int y=0;y<nT;y++) {
530		for (int x=0;x<nT;x++)
531		{
532		printf("%d ",pred[x+y*stride]);
533		}
534		printf("\n");
535		}
536		*/
537		}
538
539
540
541		// (8.4.4.2.1)
542		void decode_intra_prediction(de265_image* img,
543		int xB0,int yB0,
544		enum IntraPredMode intraPredMode,
545		int nT, int cIdx)
546		{
547		logtrace(LogIntraPred,"decode_intra_prediction xy0:%d/%d mode=%d nT=%d, cIdx=%d\n",
548		xB0,yB0, intraPredMode, nT,cIdx);
549		/*
550		printf("decode_intra_prediction xy0:%d/%d mode=%d nT=%d, cIdx=%d\n",
551		xB0,yB0, intraPredMode, nT,cIdx);
552		*/
553
554		uint8_t border_pixels_mem[2*64+1];
555		uint8_t* border_pixels = &border_pixels_mem[64];
	765	logtrace(LogIntraPred,"INTRAPRED DC\n");
	766	for (int y=0;y<nT;y++)
	767	{
	768	for (int x=0;x<nT;x++)
	769	logtrace(LogIntraPred,"%02x ", pred[x+y*stride]);
	770
	771	logtrace(LogIntraPred,"\n");
	772	}
	773	}
	774
	775
	776
	777	template <class pixel_t>
	778	void decode_intra_prediction_internal(de265_image* img,
	779	int xB0,int yB0,
	780	enum IntraPredMode intraPredMode,
	781	int nT, int cIdx)
	782	{
	783	pixel_t border_pixels_mem[2*64+1];
	784	pixel_t* border_pixels = &border_pixels_mem[64];
556	785
557	786	fill_border_samples(img, xB0,yB0, nT, cIdx, border_pixels);
558	787
559		if (cIdx==0) {
560		intra_prediction_sample_filtering(img, border_pixels, nT, intraPredMode);
561		}
	788	if (img->sps.range_extension.intra_smoothing_disabled_flag == 0 &&
	789	(cIdx==0 \|\| img->sps.ChromaArrayType==CHROMA_444))
	790	{
	791	intra_prediction_sample_filtering(img, border_pixels, nT, cIdx, intraPredMode);
	792	}
562	793
563	794
564	795	switch (intraPredMode) {

575	806	}
576	807
577	808
	809	// (8.4.4.2.1)
	810	void decode_intra_prediction(de265_image* img,
	811	int xB0,int yB0,
	812	enum IntraPredMode intraPredMode,
	813	int nT, int cIdx)
	814	{
	815	logtrace(LogIntraPred,"decode_intra_prediction xy0:%d/%d mode=%d nT=%d, cIdx=%d\n",
	816	xB0,yB0, intraPredMode, nT,cIdx);
	817	/*
	818	printf("decode_intra_prediction xy0:%d/%d mode=%d nT=%d, cIdx=%d\n",
	819	xB0,yB0, intraPredMode, nT,cIdx);
	820	*/
	821
	822	if (img->high_bit_depth(cIdx)) {
	823	decode_intra_prediction_internal<uint16_t>(img,xB0,yB0, intraPredMode,nT,cIdx);
	824	}
	825	else {
	826	decode_intra_prediction_internal<uint8_t>(img,xB0,yB0, intraPredMode,nT,cIdx);
	827	}
	828	}

+46

-2

libde265/intrapred.h less more

24	24
25	25	extern const int intraPredAngle_table[1+34];
26	26
	27
	28	/* Fill the three intra-pred-mode candidates into candModeList.
	29	Block position is (x,y) and you also have to give the PUidx for this
	30	block (which is (x>>Log2MinPUSize) + (y>>Log2MinPUSize)*PicWidthInMinPUs).
	31	availableA/B is the output of check_CTB_available().
	32	*/
	33	void fillIntraPredModeCandidates(int candModeList[3],
	34	int x,int y, int PUidx,
	35	bool availableA, // left
	36	bool availableB, // top
	37	const de265_image* img);
	38
	39
	40	inline void fillIntraPredModeCandidates(int candModeList[3], int x,int y,
	41	bool availableA, // left
	42	bool availableB, // top
	43	const de265_image* img)
	44	{
	45	int PUidx = img->sps.getPUIndexRS(x,y);
	46	fillIntraPredModeCandidates(candModeList, x,y, PUidx, availableA,availableB, img);
	47	}
	48
	49
	50
	51	/* Return value >= 0 -> use mpm_idx(return value)
	52	else -> use rem_intra(-return value-1)
	53
	54	This function may modify the candModeList !
	55	*/
	56	int find_intra_pred_mode(enum IntraPredMode mode,
	57	int candModeList[3]);
	58
	59	void list_chroma_pred_candidates(enum IntraPredMode chroma_mode[5],
	60	enum IntraPredMode luma_mode);
	61
	62	int get_intra_scan_idx(int log2TrafoSize, enum IntraPredMode intraPredMode, int cIdx,
	63	const seq_parameter_set* sps);
	64
	65	int get_intra_scan_idx_luma (int log2TrafoSize, enum IntraPredMode intraPredMode); // DEPRECATED
	66	int get_intra_scan_idx_chroma(int log2TrafoSize, enum IntraPredMode intraPredMode); // DEPRECATED
	67
	68	enum IntraPredMode lumaPredMode_to_chromaPredMode(enum IntraPredMode luma,
	69	enum IntraChromaPredMode chroma);
	70
27	71	void decode_intra_block(decoder_context* ctx,
28	72	thread_context* tctx,
29	73	int cIdx,

33	77	enum IntraPredMode intraPredMode,
34	78	bool transform_skip_flag);
35	79
36		void fill_border_samples(decoder_context* ctx, int xB,int yB,
37		int nT, int cIdx, uint8_t* out_border);
	80	//void fill_border_samples(decoder_context* ctx, int xB,int yB,
	81	// int nT, int cIdx, uint8_t* out_border);
38	82
39	83	void decode_intra_prediction(de265_image* img,
40	84	int xB0,int yB0,

+885

-653

libde265/motion.cc less more

38	38	#define MAX_CU_SIZE 64
39	39
40	40
41		enum {
42		// important! order like shown in 8.5.3.1.1
43		PRED_A1 = 0,
44		PRED_B1 = 1,
45		PRED_B0 = 2,
46		PRED_A0 = 3,
47		PRED_B2 = 4,
48		PRED_COL = 5,
49		PRED_ZERO= 6
50		};
51
52
53		typedef struct
54		{
55		uint8_t available[7];
56		PredVectorInfo pred_vector[7];
57		} MergingCandidates;
58
59
60		void reset_pred_vector(PredVectorInfo* pvec)
61		{
62		for (int X=0;X<2;X++) {
63		pvec->mv[X].x = 0;
64		pvec->mv[X].y = 0;
65		pvec->refIdx[X] = -1;
66		pvec->predFlag[X] = 0;
67		}
68		}
69
70
71	41	static int extra_before[4] = { 0,3,3,2 };
72	42	static int extra_after [4] = { 0,3,4,4 };
73	43
74	44
75	45
76		void mc_luma(const decoder_context* ctx,
77		const de265_image* img, int mv_x, int mv_y,
	46	template <class pixel_t>
	47	void mc_luma(const base_context* ctx,
	48	const seq_parameter_set* sps, int mv_x, int mv_y,
78	49	int xP,int yP,
79	50	int16_t* out, int out_stride,
80		uint8_t* ref, int ref_stride,
81		int nPbW, int nPbH)
	51	const pixel_t* ref, int ref_stride,
	52	int nPbW, int nPbH, int bitDepth_L)
82	53	{
83		const seq_parameter_set* sps = &img->sps;
84
85	54	int xFracL = mv_x & 3;
86	55	int yFracL = mv_y & 3;
87	56

100	69	ALIGNED_16(int16_t) mcbuffer[MAX_CU_SIZE * (MAX_CU_SIZE+7)];
101	70
102	71	if (xFracL==0 && yFracL==0) {
	72
103	73	if (xIntOffsL >= 0 && yIntOffsL >= 0 &&
104	74	nPbW+xIntOffsL <= w && nPbH+yIntOffsL <= h) {
105		}
106
107		if (xIntOffsL >= 0 && yIntOffsL >= 0 &&
108		nPbW+xIntOffsL <= w && nPbH+yIntOffsL <= h) {
109
110		ctx->acceleration.put_hevc_qpel_8[0][0](out, out_stride,
111		&ref[yIntOffsL*ref_stride + xIntOffsL],
112		ref_stride,
113		nPbW,nPbH, mcbuffer);
	75
	76	ctx->acceleration.put_hevc_qpel(out, out_stride,
	77	&ref[yIntOffsL*ref_stride + xIntOffsL],
	78	ref_stride /* sizeof(pixel_t)*/,
	79	nPbW,nPbH, mcbuffer, 0,0, bitDepth_L);
114	80	}
115	81	else {
116	82	for (int y=0;y<nPbH;y++)
117	83	for (int x=0;x<nPbW;x++) {
118
	84
119	85	int xA = Clip3(0,w-1,x + xIntOffsL);
120	86	int yA = Clip3(0,h-1,y + yIntOffsL);
121
	87
122	88	out[yout_stride+x] = ref[ xA + yAref_stride ] << shift3;
123	89	}
124	90	}

128	94
129	95	for (int y=0;y<nPbH;y++) {
130	96	for (int x=0;x<nPbW;x++) {
131
	97
132	98	int xA = Clip3(0,w-1,x + xIntOffsL);
133	99	int yA = Clip3(0,h-1,y + yIntOffsL);
134
	100
135	101	logtrace(LogMotion,"%02x ", ref[ xA + yA*ref_stride ]);
136	102	}
137	103	logtrace(LogMotion,"\n");

141	107
142	108	for (int y=0;y<nPbH;y++) {
143	109	for (int x=0;x<nPbW;x++) {
144
	110
145	111	logtrace(LogMotion,"%02x ",out[y*out_stride+x] >> 6); // 6 will be used when summing predictions
146	112	}
147	113	logtrace(LogMotion,"\n");

158	124	//int nPbH_extra = extra_top + nPbH + extra_bottom;
159	125
160	126
161		uint8_t padbuf[(MAX_CU_SIZE+16)*(MAX_CU_SIZE+7)];
162
163		uint8_t* src_ptr;
	127	pixel_t padbuf[(MAX_CU_SIZE+16)*(MAX_CU_SIZE+7)];
	128
	129	const pixel_t* src_ptr;
164	130	int src_stride;
165	131
166	132	if (-extra_left + xIntOffsL >= 0 &&

173	139	else {
174	140	for (int y=-extra_top;y<nPbH+extra_bottom;y++) {
175	141	for (int x=-extra_left;x<nPbW+extra_right;x++) {
176
	142
177	143	int xA = Clip3(0,w-1,x + xIntOffsL);
178	144	int yA = Clip3(0,h-1,y + yIntOffsL);
179
	145
180	146	padbuf[x+extra_left + (y+extra_top)(MAX_CU_SIZE+16)] = ref[ xA + yAref_stride ];
181	147	}
182	148	}

185	151	src_stride = MAX_CU_SIZE+16;
186	152	}
187	153
188		ctx->acceleration.put_hevc_qpel_8[xFracL][yFracL](out, out_stride,
189		src_ptr, src_stride,
190		nPbW,nPbH, mcbuffer);
	154	ctx->acceleration.put_hevc_qpel(out, out_stride,
	155	src_ptr, src_stride /* sizeof(pixel_t) */,
	156	nPbW,nPbH, mcbuffer, xFracL,yFracL, bitDepth_L);
191	157
192	158
193	159	logtrace(LogMotion,"---V---\n");

202	168
203	169
204	170
205		void mc_chroma(const decoder_context* ctx,
206		const de265_image* img,
	171	template <class pixel_t>
	172	void mc_chroma(const base_context* ctx,
	173	const seq_parameter_set* sps,
207	174	int mv_x, int mv_y,
208	175	int xP,int yP,
209	176	int16_t* out, int out_stride,
210		uint8_t* ref, int ref_stride,
211		int nPbWC, int nPbHC)
	177	const pixel_t* ref, int ref_stride,
	178	int nPbWC, int nPbHC, int bit_depth_C)
212	179	{
213		const seq_parameter_set* sps = &img->sps;
214
215	180	// chroma sample interpolation process (8.5.3.2.2.2)
216	181
217	182	//const int shift1 = sps->BitDepth_C-8;

221	186	int wC = sps->pic_width_in_luma_samples /sps->SubWidthC;
222	187	int hC = sps->pic_height_in_luma_samples/sps->SubHeightC;
223	188
	189	mv_x *= 2 / sps->SubWidthC;
	190	mv_y *= 2 / sps->SubHeightC;
	191
224	192	int xFracC = mv_x & 7;
225	193	int yFracC = mv_y & 7;
226	194
227		int xIntOffsC = xP/2 + (mv_x>>3);
228		int yIntOffsC = yP/2 + (mv_y>>3);
	195	int xIntOffsC = xP/sps->SubWidthC + (mv_x>>3);
	196	int yIntOffsC = yP/sps->SubHeightC + (mv_y>>3);
229	197
230	198	ALIGNED_32(int16_t mcbuffer[MAX_CU_SIZE*(MAX_CU_SIZE+7)]);
231	199
232	200	if (xFracC == 0 && yFracC == 0) {
233	201	if (xIntOffsC>=0 && nPbWC+xIntOffsC<=wC &&
234	202	yIntOffsC>=0 && nPbHC+yIntOffsC<=hC) {
235		ctx->acceleration.put_hevc_epel_8(out, out_stride,
236		&ref[xIntOffsC + yIntOffsC*ref_stride], ref_stride,
237		nPbWC,nPbHC, 0,0, NULL);
	203	ctx->acceleration.put_hevc_epel(out, out_stride,
	204	&ref[xIntOffsC + yIntOffsC*ref_stride], ref_stride,
	205	nPbWC,nPbHC, 0,0, NULL, bit_depth_C);
238	206	}
239	207	else
240	208	{

249	217	}
250	218	}
251	219	else {
252		uint8_t padbuf[(MAX_CU_SIZE+16)*(MAX_CU_SIZE+3)];
253
254		uint8_t* src_ptr;
	220	pixel_t padbuf[(MAX_CU_SIZE+16)*(MAX_CU_SIZE+3)];
	221
	222	const pixel_t* src_ptr;
255	223	int src_stride;
256	224
257	225	int extra_top = 1;

267	235	else {
268	236	for (int y=-extra_top;y<nPbHC+extra_bottom;y++) {
269	237	for (int x=-extra_left;x<nPbWC+extra_right;x++) {
270
	238
271	239	int xA = Clip3(0,wC-1,x + xIntOffsC);
272	240	int yA = Clip3(0,hC-1,y + yIntOffsC);
273
	241
274	242	padbuf[x+extra_left + (y+extra_top)(MAX_CU_SIZE+16)] = ref[ xA + yAref_stride ];
275	243	}
276	244	}

281	249
282	250
283	251	if (xFracC && yFracC) {
284		ctx->acceleration.put_hevc_epel_hv_8(out, out_stride,
285		src_ptr, src_stride,
286		nPbWC,nPbHC, xFracC,yFracC, mcbuffer);
	252	ctx->acceleration.put_hevc_epel_hv(out, out_stride,
	253	src_ptr, src_stride,
	254	nPbWC,nPbHC, xFracC,yFracC, mcbuffer, bit_depth_C);
287	255	}
288	256	else if (xFracC) {
289		ctx->acceleration.put_hevc_epel_h_8(out, out_stride,
290		src_ptr, src_stride,
291		nPbWC,nPbHC, xFracC,yFracC, mcbuffer);
	257	ctx->acceleration.put_hevc_epel_h(out, out_stride,
	258	src_ptr, src_stride,
	259	nPbWC,nPbHC, xFracC,yFracC, mcbuffer, bit_depth_C);
292	260	}
293	261	else if (yFracC) {
294		ctx->acceleration.put_hevc_epel_v_8(out, out_stride,
295		src_ptr, src_stride,
296		nPbWC,nPbHC, xFracC,yFracC, mcbuffer);
	262	ctx->acceleration.put_hevc_epel_v(out, out_stride,
	263	src_ptr, src_stride,
	264	nPbWC,nPbHC, xFracC,yFracC, mcbuffer, bit_depth_C);
297	265	}
298	266	else {
299	267	assert(false); // full-pel shifts are handled above

305	273
306	274	// 8.5.3.2
307	275	// NOTE: for full-pel shifts, we can introduce a fast path, simply copying without shifts
308		void generate_inter_prediction_samples(decoder_context* ctx,
	276	void generate_inter_prediction_samples(base_context* ctx,
	277	const slice_segment_header* shdr,
309	278	de265_image* img,
310		slice_segment_header* shdr,
311	279	int xC,int yC,
312	280	int xB,int yB,
313	281	int nCS, int nPbW,int nPbH,
314		const VectorInfo* vi)
	282	const MotionVectorSpec* vi)
315	283	{
	284	int xP = xC+xB;
	285	int yP = yC+yB;
	286
	287	void* pixels[3];
	288	int stride[3];
	289
	290	const int SubWidthC = img->sps.SubWidthC;
	291	const int SubHeightC = img->sps.SubHeightC;
	292
	293	pixels[0] = img->get_image_plane_at_pos_any_depth(0,xP,yP);
	294	stride[0] = img->get_image_stride(0);
	295
	296	pixels[1] = img->get_image_plane_at_pos_any_depth(1,xP/SubWidthC,yP/SubHeightC);
	297	stride[1] = img->get_image_stride(1);
	298
	299	pixels[2] = img->get_image_plane_at_pos_any_depth(2,xP/SubWidthC,yP/SubHeightC);
	300	stride[2] = img->get_image_stride(2);
	301
	302
316	303	ALIGNED_16(int16_t) predSamplesL [2 /* LX /][MAX_CU_SIZE MAX_CU_SIZE];
317	304	ALIGNED_16(int16_t) predSamplesC[2 /* chroma / ][2 / LX /][MAX_CU_SIZE MAX_CU_SIZE];
318	305
319		int xP = xC+xB;
320		int yP = yC+yB;
	306	//int xP = xC+xB;
	307	//int yP = yC+yB;
321	308
322	309	int predFlag[2];
323		predFlag[0] = vi->lum.predFlag[0];
324		predFlag[1] = vi->lum.predFlag[1];
325
	310	predFlag[0] = vi->predFlag[0];
	311	predFlag[1] = vi->predFlag[1];
	312
	313	const int bit_depth_L = img->sps.BitDepth_Y;
	314	const int bit_depth_C = img->sps.BitDepth_C;
326	315
327	316	// Some encoders use bi-prediction with two similar MVs.
328	317	// Identify this case and use only one MV.

330	319	// do this only without weighted prediction, because the weights/offsets may be different
331	320	if (img->pps.weighted_pred_flag==0) {
332	321	if (predFlag[0] && predFlag[1]) {
333		if (vi->lum.mv[0].x == vi->lum.mv[1].x &&
334		vi->lum.mv[0].y == vi->lum.mv[1].y &&
335		shdr->RefPicList[0][vi->lum.refIdx[0]] ==
336		shdr->RefPicList[1][vi->lum.refIdx[1]]) {
	322	if (vi->mv[0].x == vi->mv[1].x &&
	323	vi->mv[0].y == vi->mv[1].y &&
	324	shdr->RefPicList[0][vi->refIdx[0]] ==
	325	shdr->RefPicList[1][vi->refIdx[1]]) {
337	326	predFlag[1] = 0;
338	327	}
339	328	}

344	333	if (predFlag[l]) {
345	334	// 8.5.3.2.1
346	335
347		if (vi->lum.refIdx[l] >= MAX_NUM_REF_PICS) {
	336	if (vi->refIdx[l] >= MAX_NUM_REF_PICS) {
348	337	img->integrity = INTEGRITY_DECODING_ERRORS;
349	338	ctx->add_warning(DE265_WARNING_NONEXISTING_REFERENCE_PICTURE_ACCESSED, false);
350	339	return;
351	340	}
352	341
353		de265_image* refPic;
354		refPic = ctx->get_image(shdr->RefPicList[l][vi->lum.refIdx[l]]);
355
356		logtrace(LogMotion, "refIdx: %d -> dpb[%d]\n", vi->lum.refIdx[l], shdr->RefPicList[l][vi->lum.refIdx[l]]);
	342	const de265_image* refPic = ctx->get_image(shdr->RefPicList[l][vi->refIdx[l]]);
	343
	344	logtrace(LogMotion, "refIdx: %d -> dpb[%d]\n", vi->refIdx[l], shdr->RefPicList[l][vi->refIdx[l]]);
357	345
358	346	if (refPic->PicState == UnusedForReference) {
359	347	img->integrity = INTEGRITY_DECODING_ERRORS;
360	348	ctx->add_warning(DE265_WARNING_NONEXISTING_REFERENCE_PICTURE_ACCESSED, false);
	349
	350	// TODO: fill predSamplesC with black or grey
361	351	}
362	352	else {
363	353	// 8.5.3.2.2
364	354
365	355	logtrace(LogMotion,"do MC: L%d,MV=%d;%d RefPOC=%d\n",
366		l,vi->lum.mv[l].x,vi->lum.mv[l].y,refPic->PicOrderCntVal);
	356	l,vi->mv[l].x,vi->mv[l].y,refPic->PicOrderCntVal);
367	357
368	358
369	359	// TODO: must predSamples stride really be nCS or can it be somthing smaller like nPbW?
370		mc_luma(ctx, img, vi->lum.mv[l].x, vi->lum.mv[l].y, xP,yP,
371		predSamplesL[l],nCS,
372		refPic->get_image_plane(0),refPic->get_luma_stride(), nPbW,nPbH);
373
374
375		mc_chroma(ctx, img, vi->lum.mv[l].x, vi->lum.mv[l].y, xP,yP,
376		predSamplesC[0][l],nCS, refPic->get_image_plane(1),
377		refPic->get_chroma_stride(), nPbW/2,nPbH/2);
378		mc_chroma(ctx, img, vi->lum.mv[l].x, vi->lum.mv[l].y, xP,yP,
379		predSamplesC[1][l],nCS, refPic->get_image_plane(2),
380		refPic->get_chroma_stride(), nPbW/2,nPbH/2);
	360
	361	if (img->high_bit_depth(0)) {
	362	mc_luma(ctx, &img->sps, vi->mv[l].x, vi->mv[l].y, xP,yP,
	363	predSamplesL[l],nCS,
	364	(const uint16_t*)refPic->get_image_plane(0),
	365	refPic->get_luma_stride(), nPbW,nPbH, bit_depth_L);
	366	}
	367	else {
	368	mc_luma(ctx, &img->sps, vi->mv[l].x, vi->mv[l].y, xP,yP,
	369	predSamplesL[l],nCS,
	370	(const uint8_t*)refPic->get_image_plane(0),
	371	refPic->get_luma_stride(), nPbW,nPbH, bit_depth_L);
	372	}
	373
	374	if (img->high_bit_depth(0)) {
	375	mc_chroma(ctx, &img->sps, vi->mv[l].x, vi->mv[l].y, xP,yP,
	376	predSamplesC[0][l],nCS, (const uint16_t*)refPic->get_image_plane(1),
	377	refPic->get_chroma_stride(), nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
	378	mc_chroma(ctx, &img->sps, vi->mv[l].x, vi->mv[l].y, xP,yP,
	379	predSamplesC[1][l],nCS, (const uint16_t*)refPic->get_image_plane(2),
	380	refPic->get_chroma_stride(), nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
	381	}
	382	else {
	383	mc_chroma(ctx, &img->sps, vi->mv[l].x, vi->mv[l].y, xP,yP,
	384	predSamplesC[0][l],nCS, (const uint8_t*)refPic->get_image_plane(1),
	385	refPic->get_chroma_stride(), nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
	386	mc_chroma(ctx, &img->sps, vi->mv[l].x, vi->mv[l].y, xP,yP,
	387	predSamplesC[1][l],nCS, (const uint8_t*)refPic->get_image_plane(2),
	388	refPic->get_chroma_stride(), nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
	389	}
381	390	}
382	391	}
383	392	}

385	394
386	395	// weighted sample prediction (8.5.3.2.3)
387	396
388		//const int shift1 = 6; // TODO
389		//const int offset1= 1<<(shift1-1);
	397	const int shift1_L = libde265_max(2,14-img->sps.BitDepth_Y);
	398	const int offset_shift1_L = img->sps.WpOffsetBdShiftY;
	399	const int shift1_C = libde265_max(2,14-img->sps.BitDepth_C);
	400	const int offset_shift1_C = img->sps.WpOffsetBdShiftC;
	401
	402	/*
	403	const int shift1_L = 14-img->sps.BitDepth_Y;
	404	const int offset_shift1_L = img->sps.BitDepth_Y-8;
	405	const int shift1_C = 14-img->sps.BitDepth_C;
	406	const int offset_shift1_C = img->sps.BitDepth_C-8;
	407	*/
	408
	409	/*
	410	if (0)
	411	printf("%d/%d %d/%d %d/%d %d/%d\n",
	412	shift1_L,
	413	Nshift1_L,
	414	offset_shift1_L,
	415	Noffset_shift1_L,
	416	shift1_C,
	417	Nshift1_C,
	418	offset_shift1_C,
	419	Noffset_shift1_C);
	420
	421	assert(shift1_L==
	422	Nshift1_L);
	423	assert(offset_shift1_L==
	424	Noffset_shift1_L);
	425	assert(shift1_C==
	426	Nshift1_C);
	427	assert(offset_shift1_C==
	428	Noffset_shift1_C);
	429	*/
	430
390	431
391	432	logtrace(LogMotion,"predFlags (modified): %d %d\n", predFlag[0], predFlag[1]);
392	433
393	434	if (shdr->slice_type == SLICE_TYPE_P) {
394	435	if (img->pps.weighted_pred_flag==0) {
395	436	if (predFlag[0]==1 && predFlag[1]==0) {
396		ctx->acceleration.put_unweighted_pred_8(img->get_image_plane_at_pos(0,xP,yP),
397		img->get_image_stride(0),
398		predSamplesL[0],nCS, nPbW,nPbH);
399		ctx->acceleration.put_unweighted_pred_8(img->get_image_plane_at_pos(1,xP/2,yP/2),
400		img->get_image_stride(1),
401		predSamplesC[0][0],nCS, nPbW/2,nPbH/2);
402		ctx->acceleration.put_unweighted_pred_8(img->get_image_plane_at_pos(2,xP/2,yP/2),
403		img->get_image_stride(2),
404		predSamplesC[1][0],nCS, nPbW/2,nPbH/2);
	437	ctx->acceleration.put_unweighted_pred(pixels[0], stride[0],
	438	predSamplesL[0],nCS, nPbW,nPbH, bit_depth_L);
	439	ctx->acceleration.put_unweighted_pred(pixels[1], stride[1],
	440	predSamplesC[0][0],nCS,
	441	nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
	442	ctx->acceleration.put_unweighted_pred(pixels[2], stride[2],
	443	predSamplesC[1][0],nCS,
	444	nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
405	445	}
406	446	else {
407	447	ctx->add_warning(DE265_WARNING_BOTH_PREDFLAGS_ZERO, false);

413	453
414	454	if (predFlag[0]==1 && predFlag[1]==0) {
415	455
416		int refIdx0 = vi->lum.refIdx[0];
417
418		int luma_log2WD = shdr->luma_log2_weight_denom + (14-8); // TODO: bitDepth
419		int chroma_log2WD = shdr->ChromaLog2WeightDenom + (14-8); // TODO: bitDepth
	456	int refIdx0 = vi->refIdx[0];
	457
	458	int luma_log2WD = shdr->luma_log2_weight_denom + shift1_L;
	459	int chroma_log2WD = shdr->ChromaLog2WeightDenom + shift1_C;
420	460
421	461	int luma_w0 = shdr->LumaWeight[0][refIdx0];
422		int luma_o0 = shdr->luma_offset[0][refIdx0] * (1<<(8-8)); // TODO: bitDepth
	462	int luma_o0 = shdr->luma_offset[0][refIdx0] * (1<<(offset_shift1_L));
423	463
424	464	int chroma0_w0 = shdr->ChromaWeight[0][refIdx0][0];
425		int chroma0_o0 = shdr->ChromaOffset[0][refIdx0][0] * (1<<(8-8)); // TODO: bitDepth
	465	int chroma0_o0 = shdr->ChromaOffset[0][refIdx0][0] * (1<<(offset_shift1_C));
426	466	int chroma1_w0 = shdr->ChromaWeight[0][refIdx0][1];
427		int chroma1_o0 = shdr->ChromaOffset[0][refIdx0][1] * (1<<(8-8)); // TODO: bitDepth
	467	int chroma1_o0 = shdr->ChromaOffset[0][refIdx0][1] * (1<<(offset_shift1_C));
428	468
429	469	logtrace(LogMotion,"weighted-0 [%d] %d %d %d %dx%d\n", refIdx0, luma_log2WD-6,luma_w0,luma_o0,nPbW,nPbH);
430	470
431		ctx->acceleration.put_weighted_pred_8(img->get_image_plane_at_pos(0,xP,yP),
432		img->get_image_stride(0),
433		predSamplesL[0],nCS, nPbW,nPbH,
434		luma_w0, luma_o0, luma_log2WD);
435		ctx->acceleration.put_weighted_pred_8(img->get_image_plane_at_pos(1,xP/2,yP/2),
436		img->get_image_stride(1),
437		predSamplesC[0][0],nCS, nPbW/2,nPbH/2,
438		chroma0_w0, chroma0_o0, chroma_log2WD);
439		ctx->acceleration.put_weighted_pred_8(img->get_image_plane_at_pos(2,xP/2,yP/2),
440		img->get_image_stride(2),
441		predSamplesC[1][0],nCS, nPbW/2,nPbH/2,
442		chroma1_w0, chroma1_o0, chroma_log2WD);
	471	ctx->acceleration.put_weighted_pred(pixels[0], stride[0],
	472	predSamplesL[0],nCS, nPbW,nPbH,
	473	luma_w0, luma_o0, luma_log2WD, bit_depth_L);
	474	ctx->acceleration.put_weighted_pred(pixels[1], stride[1],
	475	predSamplesC[0][0],nCS, nPbW/SubWidthC,nPbH/SubHeightC,
	476	chroma0_w0, chroma0_o0, chroma_log2WD, bit_depth_C);
	477	ctx->acceleration.put_weighted_pred(pixels[2], stride[2],
	478	predSamplesC[1][0],nCS, nPbW/SubWidthC,nPbH/SubHeightC,
	479	chroma1_w0, chroma1_o0, chroma_log2WD, bit_depth_C);
443	480	}
444	481	else {
445	482	ctx->add_warning(DE265_WARNING_BOTH_PREDFLAGS_ZERO, false);

457	494
458	495	int16_t* in0 = predSamplesL[0];
459	496	int16_t* in1 = predSamplesL[1];
460		uint8_t* out = img->get_image_plane_at_pos(0, xP,yP);
461
462		ctx->acceleration.put_weighted_pred_avg_8(out, img->get_luma_stride(),
463		in0,in1, nCS, nPbW, nPbH);
	497
	498	ctx->acceleration.put_weighted_pred_avg(pixels[0], stride[0],
	499	in0,in1, nCS, nPbW, nPbH, bit_depth_L);
464	500
465	501	int16_t* in00 = predSamplesC[0][0];
466	502	int16_t* in01 = predSamplesC[0][1];
467	503	int16_t* in10 = predSamplesC[1][0];
468	504	int16_t* in11 = predSamplesC[1][1];
469		uint8_t* out0 = img->get_image_plane_at_pos(1,xP/2,yP/2);
470		uint8_t* out1 = img->get_image_plane_at_pos(2,xP/2,yP/2);
471
472		ctx->acceleration.put_weighted_pred_avg_8(out0, img->get_chroma_stride(),
473		in00,in01, nCS, nPbW/2, nPbH/2);
474		ctx->acceleration.put_weighted_pred_avg_8(out1, img->get_chroma_stride(),
475		in10,in11, nCS, nPbW/2, nPbH/2);
	505
	506	ctx->acceleration.put_weighted_pred_avg(pixels[1], stride[1],
	507	in00,in01, nCS,
	508	nPbW/SubWidthC, nPbH/SubHeightC, bit_depth_C);
	509	ctx->acceleration.put_weighted_pred_avg(pixels[2], stride[2],
	510	in10,in11, nCS,
	511	nPbW/SubWidthC, nPbH/SubHeightC, bit_depth_C);
476	512	}
477	513	else {
478	514	// weighted prediction
479	515
480		int refIdx0 = vi->lum.refIdx[0];
481		int refIdx1 = vi->lum.refIdx[1];
482
483		int luma_log2WD = shdr->luma_log2_weight_denom + (14-8); // TODO: bitDepth
484		int chroma_log2WD = shdr->ChromaLog2WeightDenom + (14-8); // TODO: bitDepth
	516	int refIdx0 = vi->refIdx[0];
	517	int refIdx1 = vi->refIdx[1];
	518
	519	int luma_log2WD = shdr->luma_log2_weight_denom + shift1_L;
	520	int chroma_log2WD = shdr->ChromaLog2WeightDenom + shift1_C;
485	521
486	522	int luma_w0 = shdr->LumaWeight[0][refIdx0];
487		int luma_o0 = shdr->luma_offset[0][refIdx0] * (1<<(8-8)); // TODO: bitDepth
	523	int luma_o0 = shdr->luma_offset[0][refIdx0] * (1<<(offset_shift1_L));
488	524	int luma_w1 = shdr->LumaWeight[1][refIdx1];
489		int luma_o1 = shdr->luma_offset[1][refIdx1] * (1<<(8-8)); // TODO: bitDepth
	525	int luma_o1 = shdr->luma_offset[1][refIdx1] * (1<<(offset_shift1_L));
490	526
491	527	int chroma0_w0 = shdr->ChromaWeight[0][refIdx0][0];
492		int chroma0_o0 = shdr->ChromaOffset[0][refIdx0][0] * (1<<(8-8)); // TODO: bitDepth
	528	int chroma0_o0 = shdr->ChromaOffset[0][refIdx0][0] * (1<<(offset_shift1_C));
493	529	int chroma1_w0 = shdr->ChromaWeight[0][refIdx0][1];
494		int chroma1_o0 = shdr->ChromaOffset[0][refIdx0][1] * (1<<(8-8)); // TODO: bitDepth
	530	int chroma1_o0 = shdr->ChromaOffset[0][refIdx0][1] * (1<<(offset_shift1_C));
495	531	int chroma0_w1 = shdr->ChromaWeight[1][refIdx1][0];
496		int chroma0_o1 = shdr->ChromaOffset[1][refIdx1][0] * (1<<(8-8)); // TODO: bitDepth
	532	int chroma0_o1 = shdr->ChromaOffset[1][refIdx1][0] * (1<<(offset_shift1_C));
497	533	int chroma1_w1 = shdr->ChromaWeight[1][refIdx1][1];
498		int chroma1_o1 = shdr->ChromaOffset[1][refIdx1][1] * (1<<(8-8)); // TODO: bitDepth
	534	int chroma1_o1 = shdr->ChromaOffset[1][refIdx1][1] * (1<<(offset_shift1_C));
499	535
500	536	logtrace(LogMotion,"weighted-BI-0 [%d] %d %d %d %dx%d\n", refIdx0, luma_log2WD-6,luma_w0,luma_o0,nPbW,nPbH);
501	537	logtrace(LogMotion,"weighted-BI-1 [%d] %d %d %d %dx%d\n", refIdx1, luma_log2WD-6,luma_w1,luma_o1,nPbW,nPbH);
502	538
503	539	int16_t* in0 = predSamplesL[0];
504	540	int16_t* in1 = predSamplesL[1];
505		uint8_t* out = img->get_image_plane_at_pos(0, xP,yP);
506
507		ctx->acceleration.put_weighted_bipred_8(out, img->get_luma_stride(),
508		in0,in1, nCS, nPbW, nPbH,
509		luma_w0,luma_o0,
510		luma_w1,luma_o1,
511		luma_log2WD);
	541
	542	ctx->acceleration.put_weighted_bipred(pixels[0], stride[0],
	543	in0,in1, nCS, nPbW, nPbH,
	544	luma_w0,luma_o0,
	545	luma_w1,luma_o1,
	546	luma_log2WD, bit_depth_L);
512	547
513	548	int16_t* in00 = predSamplesC[0][0];
514	549	int16_t* in01 = predSamplesC[0][1];
515	550	int16_t* in10 = predSamplesC[1][0];
516	551	int16_t* in11 = predSamplesC[1][1];
517		uint8_t* out0 = img->get_image_plane_at_pos(1,xP/2,yP/2);
518		uint8_t* out1 = img->get_image_plane_at_pos(2,xP/2,yP/2);
519
520		ctx->acceleration.put_weighted_bipred_8(out0, img->get_chroma_stride(),
521		in00,in01, nCS, nPbW/2, nPbH/2,
522		chroma0_w0,chroma0_o0,
523		chroma0_w1,chroma0_o1,
524		chroma_log2WD);
525		ctx->acceleration.put_weighted_bipred_8(out1, img->get_chroma_stride(),
526		in10,in11, nCS, nPbW/2, nPbH/2,
527		chroma1_w0,chroma1_o0,
528		chroma1_w1,chroma1_o1,
529		chroma_log2WD);
	552
	553	ctx->acceleration.put_weighted_bipred(pixels[1], stride[1],
	554	in00,in01, nCS, nPbW/SubWidthC, nPbH/SubHeightC,
	555	chroma0_w0,chroma0_o0,
	556	chroma0_w1,chroma0_o1,
	557	chroma_log2WD, bit_depth_C);
	558	ctx->acceleration.put_weighted_bipred(pixels[2], stride[2],
	559	in10,in11, nCS, nPbW/SubWidthC, nPbH/SubHeightC,
	560	chroma1_w0,chroma1_o0,
	561	chroma1_w1,chroma1_o1,
	562	chroma_log2WD, bit_depth_C);
530	563	}
531	564	}
532	565	else if (predFlag[0]==1 \|\| predFlag[1]==1) {
533	566	int l = predFlag[0] ? 0 : 1;
534	567
535	568	if (img->pps.weighted_bipred_flag==0) {
536		ctx->acceleration.put_unweighted_pred_8(img->get_image_plane_at_pos(0,xP,yP),
537		img->get_image_stride(0),
538		predSamplesL[l],nCS, nPbW,nPbH);
539		ctx->acceleration.put_unweighted_pred_8(img->get_image_plane_at_pos(1,xP/2,yP/2),
540		img->get_image_stride(1),
541		predSamplesC[0][l],nCS, nPbW/2,nPbH/2);
542		ctx->acceleration.put_unweighted_pred_8(img->get_image_plane_at_pos(2,xP/2,yP/2),
543		img->get_image_stride(2),
544		predSamplesC[1][l],nCS, nPbW/2,nPbH/2);
	569	ctx->acceleration.put_unweighted_pred(pixels[0], stride[0],
	570	predSamplesL[l],nCS, nPbW,nPbH, bit_depth_L);
	571	ctx->acceleration.put_unweighted_pred(pixels[1], stride[1],
	572	predSamplesC[0][l],nCS,
	573	nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
	574	ctx->acceleration.put_unweighted_pred(pixels[2], stride[2],
	575	predSamplesC[1][l],nCS,
	576	nPbW/SubWidthC,nPbH/SubHeightC, bit_depth_C);
545	577	}
546	578	else {
547		int refIdx = vi->lum.refIdx[l];
548
549		int luma_log2WD = shdr->luma_log2_weight_denom + (14-8); // TODO: bitDepth
550		int chroma_log2WD = shdr->ChromaLog2WeightDenom + (14-8); // TODO: bitDepth
	579	int refIdx = vi->refIdx[l];
	580
	581	int luma_log2WD = shdr->luma_log2_weight_denom + shift1_L;
	582	int chroma_log2WD = shdr->ChromaLog2WeightDenom + shift1_C;
551	583
552	584	int luma_w = shdr->LumaWeight[l][refIdx];
553		int luma_o = shdr->luma_offset[l][refIdx] * (1<<(8-8)); // TODO: bitDepth
	585	int luma_o = shdr->luma_offset[l][refIdx] * (1<<(offset_shift1_L));
554	586
555	587	int chroma0_w = shdr->ChromaWeight[l][refIdx][0];
556		int chroma0_o = shdr->ChromaOffset[l][refIdx][0] * (1<<(8-8)); // TODO: bitDepth
	588	int chroma0_o = shdr->ChromaOffset[l][refIdx][0] * (1<<(offset_shift1_C));
557	589	int chroma1_w = shdr->ChromaWeight[l][refIdx][1];
558		int chroma1_o = shdr->ChromaOffset[l][refIdx][1] * (1<<(8-8)); // TODO: bitDepth
	590	int chroma1_o = shdr->ChromaOffset[l][refIdx][1] * (1<<(offset_shift1_C));
559	591
560	592	logtrace(LogMotion,"weighted-B-L%d [%d] %d %d %d %dx%d\n", l, refIdx, luma_log2WD-6,luma_w,luma_o,nPbW,nPbH);
561	593
562		ctx->acceleration.put_weighted_pred_8(img->get_image_plane_at_pos(0,xP,yP),
563		img->get_image_stride(0),
564		predSamplesL[l],nCS, nPbW,nPbH,
565		luma_w, luma_o, luma_log2WD);
566		ctx->acceleration.put_weighted_pred_8(img->get_image_plane_at_pos(1,xP/2,yP/2),
567		img->get_image_stride(1),
568		predSamplesC[0][l],nCS, nPbW/2,nPbH/2,
569		chroma0_w, chroma0_o, chroma_log2WD);
570		ctx->acceleration.put_weighted_pred_8(img->get_image_plane_at_pos(2,xP/2,yP/2),
571		img->get_image_stride(2),
572		predSamplesC[1][l],nCS, nPbW/2,nPbH/2,
573		chroma1_w, chroma1_o, chroma_log2WD);
	594	ctx->acceleration.put_weighted_pred(pixels[0], stride[0],
	595	predSamplesL[l],nCS, nPbW,nPbH,
	596	luma_w, luma_o, luma_log2WD, bit_depth_L);
	597	ctx->acceleration.put_weighted_pred(pixels[1], stride[1],
	598	predSamplesC[0][l],nCS,
	599	nPbW/SubWidthC,nPbH/SubHeightC,
	600	chroma0_w, chroma0_o, chroma_log2WD, bit_depth_C);
	601	ctx->acceleration.put_weighted_pred(pixels[2], stride[2],
	602	predSamplesC[1][l],nCS,
	603	nPbW/SubWidthC,nPbH/SubHeightC,
	604	chroma1_w, chroma1_o, chroma_log2WD, bit_depth_C);
574	605	}
575	606	}
576	607	else {

589	620	logtrace(LogTransform,"MC-y-%d-%d ",xP,yP+y);
590	621
591	622	for (int x=0;x<nPbW;x++) {
592		logtrace(LogTransform,"%02x ", img->y[xP+x+(yP+y)img->stride]);
	623	logtrace(LogTransform,"%02x ", pixels[0][x+ystride[0]]);
593	624	}
594	625
595	626	logtrace(LogTransform,"*\n");
596		}
	627	}
597	628
598	629
599	630	logtrace(LogTransform,"MC pixels (chroma cb), position %d %d:\n", xP/2,yP/2);

602	633	logtrace(LogTransform,"MC-cb-%d-%d ",xP/2,yP/2+y);
603	634
604	635	for (int x=0;x<nPbW/2;x++) {
605		logtrace(LogTransform,"%02x ", img->cb[xP/2+x+(yP/2+y)img->chroma_stride]);
	636	logtrace(LogTransform,"%02x ", pixels[1][x+ystride[1]]);
606	637	}
607	638
608	639	logtrace(LogTransform,"*\n");
609		}
	640	}
610	641
611	642
612	643	logtrace(LogTransform,"MC pixels (chroma cr), position %d %d:\n", xP/2,yP/2);

615	646	logtrace(LogTransform,"MC-cr-%d-%d ",xP/2,yP/2+y);
616	647
617	648	for (int x=0;x<nPbW/2;x++) {
618		logtrace(LogTransform,"%02x ", img->cr[xP/2+x+(yP/2+y)img->chroma_stride]);
	649	logtrace(LogTransform,"%02x ", pixels[2][x+ystride[2]]);
619	650	}
620	651
621	652	logtrace(LogTransform,"*\n");
622		}
	653	}
623	654	#endif
624	655	}
625	656
626	657
627	658	#ifdef DE265_LOG_TRACE
628		void logmvcand(PredVectorInfo p)
	659	void logmvcand(const MotionVectorSpec& p)
629	660	{
630	661	for (int v=0;v<2;v++) {
631	662	if (p.predFlag[v]) {

641	672	#endif
642	673
643	674
644		LIBDE265_INLINE static bool equal_cand_MV(const PredVectorInfo* a, const PredVectorInfo* b)
	675	LIBDE265_INLINE static bool equal_cand_MV(const MotionVectorSpec* a, const MotionVectorSpec* b)
645	676	{
646	677	// TODO: is this really correct? no check for predFlag? Standard says so... (p.127)
647	678

667	698	\| \|
668	699	\| \|
669	700	\| \|
670		\| \|
	701	\| PB \|
671	702	\| \|
672	703	\| \|
673	704	+--+ \|

680	711
681	712	// 8.5.3.1.2
682	713	// TODO: check: can we fill the candidate list directly in this function and omit to copy later
683		void derive_spatial_merging_candidates(const de265_image* img,
684		int xC, int yC, int nCS, int xP, int yP,
685		uint8_t singleMCLFlag,
686		int nPbW, int nPbH,
687		int partIdx,
688		MergingCandidates* out_cand)
	714	/*
	715	xC/yC: CB position
	716	nCS: CB size (probably modified because of singleMCLFlag)
	717	xP/yP: PB position (absolute) (probably modified because of singleMCLFlag)
	718	singleMCLFlag
	719	nPbW/nPbH: PB size
	720	partIdx
	721	out_cand: merging candidate vectors
	722
	723	Add these candidates:
	724	- A1
	725	- B1 (if != A1)
	726	- B0 (if != B1)
	727	- A0 (if != A1)
	728	- B2 (if != A1 and != B1)
	729
	730	A maximum of 4 candidates are generated.
	731
	732	Note 1: For a CB splitted into two PBs, it does not make sense to merge the
	733	second part to the parameters of the first part, since then, we could use 2Nx2N
	734	right away. -> Exclude this candidate.
	735	*/
	736	int derive_spatial_merging_candidates(const de265_image* img,
	737	int xC, int yC, int nCS, int xP, int yP,
	738	uint8_t singleMCLFlag,
	739	int nPbW, int nPbH,
	740	int partIdx,
	741	MotionVectorSpec* out_cand,
	742	int maxCandidates)
689	743	{
690	744	const pic_parameter_set* pps = &img->pps;
691		int log2_parallel_merge_level = pps->log2_parallel_merge_level;
	745	const int log2_parallel_merge_level = pps->log2_parallel_merge_level;
692	746
693	747	enum PartMode PartMode = img->get_PartMode(xC,yC);
694	748
	749	/*
	750	const int A0 = SpatialMergingCandidates::PRED_A0;
	751	const int A1 = SpatialMergingCandidates::PRED_A1;
	752	const int B0 = SpatialMergingCandidates::PRED_B0;
	753	const int B1 = SpatialMergingCandidates::PRED_B1;
	754	const int B2 = SpatialMergingCandidates::PRED_B2;
	755	*/
	756
695	757	// --- A1 ---
696	758
697		// a pixel within A1
	759	// a pixel within A1 (bottom right of A1)
698	760	int xA1 = xP-1;
699	761	int yA1 = yP+nPbH-1;
700	762
701	763	bool availableA1;
702
	764	int idxA1;
	765
	766	int computed_candidates = 0;
	767
	768	// check if candidate is in same motion-estimation region (MER) -> discard
703	769	if ((xP>>log2_parallel_merge_level) == (xA1>>log2_parallel_merge_level) &&
704	770	(yP>>log2_parallel_merge_level) == (yA1>>log2_parallel_merge_level)) {
705	771	availableA1 = false;
706	772	logtrace(LogMotion,"spatial merging candidate A1: below parallel merge level\n");
707	773	}
708		else if (!singleMCLFlag &&
	774	// redundant candidate? (Note 1) -> discard
	775	else if (// !singleMCLFlag && automatically true when partIdx==1
709	776	partIdx==1 &&
710	777	(PartMode==PART_Nx2N \|\|
711	778	PartMode==PART_nLx2N \|\|

713	780	availableA1 = false;
714	781	logtrace(LogMotion,"spatial merging candidate A1: second part ignore\n");
715	782	}
	783	// MV available in A1
716	784	else {
717	785	availableA1 = img->available_pred_blk(xC,yC, nCS, xP,yP, nPbW,nPbH,partIdx, xA1,yA1);
718	786	if (!availableA1) logtrace(LogMotion,"spatial merging candidate A1: unavailable\n");
719	787	}
720	788
721		if (!availableA1) {
722		out_cand->available[PRED_A1] = 0;
723		reset_pred_vector(&out_cand->pred_vector[PRED_A1]);
724		}
725		else {
726		out_cand->available[PRED_A1] = 1;
727		out_cand->pred_vector[PRED_A1] = *img->get_mv_info(xA1,yA1);
	789	if (availableA1) {
	790	idxA1 = computed_candidates++;
	791	out_cand[idxA1] = *img->get_mv_info(xA1,yA1);
728	792
729	793	logtrace(LogMotion,"spatial merging candidate A1:\n");
730		logmvcand(out_cand->pred_vector[PRED_A1]);
731		}
	794	logmvcand(out_cand[idxA1]);
	795	}
	796
	797	if (computed_candidates>=maxCandidates) return computed_candidates;
732	798
733	799
734	800	// --- B1 ---

737	803	int yB1 = yP-1;
738	804
739	805	bool availableB1;
740
	806	int idxB1;
	807
	808	// same MER -> discard
741	809	if ((xP>>log2_parallel_merge_level) == (xB1>>log2_parallel_merge_level) &&
742	810	(yP>>log2_parallel_merge_level) == (yB1>>log2_parallel_merge_level)) {
743	811	availableB1 = false;
744	812	logtrace(LogMotion,"spatial merging candidate B1: below parallel merge level\n");
745	813	}
746		else if (!singleMCLFlag &&
	814	// redundant candidate (Note 1) -> discard
	815	else if (// !singleMCLFlag && automatically true when partIdx==1
747	816	partIdx==1 &&
748	817	(PartMode==PART_2NxN \|\|
749	818	PartMode==PART_2NxnU \|\|

751	820	availableB1 = false;
752	821	logtrace(LogMotion,"spatial merging candidate B1: second part ignore\n");
753	822	}
	823	// MV available in B1
754	824	else {
755	825	availableB1 = img->available_pred_blk(xC,yC, nCS, xP,yP, nPbW,nPbH,partIdx, xB1,yB1);
756	826	if (!availableB1) logtrace(LogMotion,"spatial merging candidate B1: unavailable\n");
757	827	}
758	828
759		if (!availableB1) {
760		out_cand->available[PRED_B1] = 0;
761		reset_pred_vector(&out_cand->pred_vector[PRED_B1]);
762		}
763		else {
764		out_cand->available[PRED_B1] = 1;
765		out_cand->pred_vector[PRED_B1] = *img->get_mv_info(xB1,yB1);
766
	829	if (availableB1) {
	830	const MotionVectorSpec* b1 = img->get_mv_info(xB1,yB1);
	831
	832	// B1 == A1 -> discard B1
767	833	if (availableA1 &&
768		equal_cand_MV(&out_cand->pred_vector[PRED_A1],
769		&out_cand->pred_vector[PRED_B1])) {
770		out_cand->available[PRED_B1] = 0;
	834	equal_cand_MV(&out_cand[idxA1], b1)) {
	835	idxB1 = idxA1;
771	836	logtrace(LogMotion,"spatial merging candidate B1: redundant to A1\n");
772	837	}
773	838	else {
	839	idxB1 = computed_candidates++;
	840	out_cand[idxB1] = *b1;
	841
774	842	logtrace(LogMotion,"spatial merging candidate B1:\n");
775		logmvcand(out_cand->pred_vector[PRED_B1]);
776		}
777		}
	843	logmvcand(out_cand[idxB1]);
	844	}
	845	}
	846
	847	if (computed_candidates>=maxCandidates) return computed_candidates;
778	848
779	849
780	850	// --- B0 ---

783	853	int yB0 = yP-1;
784	854
785	855	bool availableB0;
	856	int idxB0;
786	857
787	858	if ((xP>>log2_parallel_merge_level) == (xB0>>log2_parallel_merge_level) &&
788	859	(yP>>log2_parallel_merge_level) == (yB0>>log2_parallel_merge_level)) {

794	865	if (!availableB0) logtrace(LogMotion,"spatial merging candidate B0: unavailable\n");
795	866	}
796	867
797		if (!availableB0) {
798		out_cand->available[PRED_B0] = 0;
799		reset_pred_vector(&out_cand->pred_vector[PRED_B0]);
800		}
801		else {
802		out_cand->available[PRED_B0] = 1;
803		out_cand->pred_vector[PRED_B0] = *img->get_mv_info(xB0,yB0);
804
	868	if (availableB0) {
	869	const MotionVectorSpec* b0 = img->get_mv_info(xB0,yB0);
	870
	871	// B0 == B1 -> discard B0
805	872	if (availableB1 &&
806		equal_cand_MV(&out_cand->pred_vector[PRED_B1],
807		&out_cand->pred_vector[PRED_B0])) {
808		out_cand->available[PRED_B0] = 0;
	873	equal_cand_MV(&out_cand[idxB1], b0)) {
	874	idxB0 = idxB1;
809	875	logtrace(LogMotion,"spatial merging candidate B0: redundant to B1\n");
810	876	}
811	877	else {
	878	idxB0 = computed_candidates++;
	879	out_cand[idxB0] = *b0;
812	880	logtrace(LogMotion,"spatial merging candidate B0:\n");
813		logmvcand(out_cand->pred_vector[PRED_B0]);
814		}
815		}
	881	logmvcand(out_cand[idxB0]);
	882	}
	883	}
	884
	885	if (computed_candidates>=maxCandidates) return computed_candidates;
816	886
817	887
818	888	// --- A0 ---

821	891	int yA0 = yP+nPbH;
822	892
823	893	bool availableA0;
	894	int idxA0;
824	895
825	896	if ((xP>>log2_parallel_merge_level) == (xA0>>log2_parallel_merge_level) &&
826	897	(yP>>log2_parallel_merge_level) == (yA0>>log2_parallel_merge_level)) {

832	903	if (!availableA0) logtrace(LogMotion,"spatial merging candidate A0: unavailable\n");
833	904	}
834	905
835		if (!availableA0) {
836		out_cand->available[PRED_A0] = 0;
837		reset_pred_vector(&out_cand->pred_vector[PRED_A0]);
838		}
839		else {
840		out_cand->available[PRED_A0] = 1;
841		out_cand->pred_vector[PRED_A0] = *img->get_mv_info(xA0,yA0);
842
	906	if (availableA0) {
	907	const MotionVectorSpec* a0 = img->get_mv_info(xA0,yA0);
	908
	909	// A0 == A1 -> discard A0
843	910	if (availableA1 &&
844		equal_cand_MV(&out_cand->pred_vector[PRED_A1],
845		&out_cand->pred_vector[PRED_A0])) {
846		out_cand->available[PRED_A0] = 0;
	911	equal_cand_MV(&out_cand[idxA1], a0)) {
	912	idxA0 = idxA1;
847	913	logtrace(LogMotion,"spatial merging candidate A0: redundant to A1\n");
848	914	}
849	915	else {
	916	idxA0 = computed_candidates++;
	917	out_cand[idxA0] = *a0;
850	918	logtrace(LogMotion,"spatial merging candidate A0:\n");
851		logmvcand(out_cand->pred_vector[PRED_A0]);
852		}
853		}
	919	logmvcand(out_cand[idxA0]);
	920	}
	921	}
	922
	923	if (computed_candidates>=maxCandidates) return computed_candidates;
854	924
855	925
856	926	// --- B2 ---

859	929	int yB2 = yP-1;
860	930
861	931	bool availableB2;
862
863		if (out_cand->available[PRED_A0] && out_cand->available[PRED_A1] &&
864		out_cand->available[PRED_B0] && out_cand->available[PRED_B1]) {
	932	int idxB2;
	933
	934	// if we already have four candidates, do not consider B2 anymore
	935	if (computed_candidates==4) {
865	936	availableB2 = false;
866	937	logtrace(LogMotion,"spatial merging candidate B2: ignore\n");
867	938	}

875	946	if (!availableB2) logtrace(LogMotion,"spatial merging candidate B2: unavailable\n");
876	947	}
877	948
878		if (!availableB2) {
879		out_cand->available[PRED_B2] = 0;
880		reset_pred_vector(&out_cand->pred_vector[PRED_B2]);
881		}
882		else {
883		out_cand->available[PRED_B2] = 1;
884		out_cand->pred_vector[PRED_B2] = *img->get_mv_info(xB2,yB2);
885
	949	if (availableB2) {
	950	const MotionVectorSpec* b2 = img->get_mv_info(xB2,yB2);
	951
	952	// B2 == B1 -> discard B2
886	953	if (availableB1 &&
887		equal_cand_MV(&out_cand->pred_vector[PRED_B1],
888		&out_cand->pred_vector[PRED_B2])) {
889		out_cand->available[PRED_B2] = 0;
	954	equal_cand_MV(&out_cand[idxB1], b2)) {
	955	idxB2 = idxB1;
890	956	logtrace(LogMotion,"spatial merging candidate B2: redundant to B1\n");
891	957	}
	958	// B2 == A1 -> discard B2
892	959	else if (availableA1 &&
893		equal_cand_MV(&out_cand->pred_vector[PRED_A1],
894		&out_cand->pred_vector[PRED_B2])) {
895		out_cand->available[PRED_B2] = 0;
	960	equal_cand_MV(&out_cand[idxA1], b2)) {
	961	idxB2 = idxA1;
896	962	logtrace(LogMotion,"spatial merging candidate B2: redundant to A1\n");
897	963	}
898	964	else {
899		logtrace(LogMotion,"spatial merging candidate B0:\n");
900		logmvcand(out_cand->pred_vector[PRED_B0]);
901		}
902		}
	965	idxB2 = computed_candidates++;
	966	out_cand[idxB2] = *b2;
	967	logtrace(LogMotion,"spatial merging candidate B2:\n");
	968	logmvcand(out_cand[idxB2]);
	969	}
	970	}
	971
	972	return computed_candidates;
903	973	}
904	974
905	975
906	976	// 8.5.3.1.4
907		void derive_zero_motion_vector_candidates(slice_segment_header* shdr,
908		PredVectorInfo* inout_mergeCandList,
909		int* inout_numCurrMergeCand)
	977	void derive_zero_motion_vector_candidates(const slice_segment_header* shdr,
	978	MotionVectorSpec* out_mergeCandList,
	979	int* inout_numCurrMergeCand,
	980	int maxCandidates)
910	981	{
911	982	logtrace(LogMotion,"derive_zero_motion_vector_candidates\n");
912	983

924	995	//int numInputMergeCand = *inout_numMergeCand;
925	996	int zeroIdx = 0;
926	997
927		while (*inout_numCurrMergeCand < shdr->MaxNumMergeCand) {
	998	while (*inout_numCurrMergeCand < maxCandidates) {
928	999	// 1.
929	1000
930	1001	logtrace(LogMotion,"zeroIdx:%d numRefIdx:%d\n", zeroIdx, numRefIdx);
931	1002
932		PredVectorInfo* newCand = &inout_mergeCandList[*inout_numCurrMergeCand];
	1003	MotionVectorSpec* newCand = &out_mergeCandList[*inout_numCurrMergeCand];
	1004
	1005	const int refIdx = (zeroIdx < numRefIdx) ? zeroIdx : 0;
933	1006
934	1007	if (shdr->slice_type==SLICE_TYPE_P) {
935		newCand->refIdx[0] = (zeroIdx < numRefIdx) ? zeroIdx : 0;
	1008	newCand->refIdx[0] = refIdx;
936	1009	newCand->refIdx[1] = -1;
937	1010	newCand->predFlag[0] = 1;
938	1011	newCand->predFlag[1] = 0;
939	1012	}
940	1013	else {
941		newCand->refIdx[0] = (zeroIdx < numRefIdx) ? zeroIdx : 0;
942		newCand->refIdx[1] = (zeroIdx < numRefIdx) ? zeroIdx : 0;
	1014	newCand->refIdx[0] = refIdx;
	1015	newCand->refIdx[1] = refIdx;
943	1016	newCand->predFlag[0] = 1;
944	1017	newCand->predFlag[1] = 1;
945	1018	}

950	1023	newCand->mv[1].y = 0;
951	1024
952	1025	(*inout_numCurrMergeCand)++;
953
	1026
954	1027	// 2.
955	1028
956	1029	zeroIdx++;

981	1054
982	1055	// (L1003) 8.5.3.2.8
983	1056
984		void derive_collocated_motion_vectors(decoder_context* ctx,
	1057	void derive_collocated_motion_vectors(base_context* ctx,
985	1058	de265_image* img,
986	1059	const slice_segment_header* shdr,
987	1060	int xP,int yP,
988	1061	int colPic,
989	1062	int xColPb,int yColPb,
990		int refIdxLX, int X,
	1063	int refIdxLX, // (always 0 for merge mode)
	1064	int X,
991	1065	MotionVector* out_mvLXCol,
992	1066	uint8_t* out_availableFlagLXCol)
993	1067	{
994	1068	logtrace(LogMotion,"derive_collocated_motion_vectors %d;%d\n",xP,yP);
995	1069
	1070
	1071	// get collocated image and the prediction mode at the collocated position
	1072
996	1073	assert(ctx->has_image(colPic));
997	1074	const de265_image* colImg = ctx->get_image(colPic);
	1075
	1076	// check for access outside image area
	1077
	1078	if (xColPb >= colImg->get_width() \|\|
	1079	yColPb >= colImg->get_height()) {
	1080	ctx->add_warning(DE265_WARNING_COLLOCATED_MOTION_VECTOR_OUTSIDE_IMAGE_AREA, false);
	1081	*out_availableFlagLXCol = 0;
	1082	return;
	1083	}
	1084
998	1085	enum PredMode predMode = colImg->get_pred_mode(xColPb,yColPb);
	1086
	1087
	1088	// collocated block is Intra -> no collocated MV
999	1089
1000	1090	if (predMode == MODE_INTRA) {
1001	1091	out_mvLXCol->x = 0;

1003	1093	*out_availableFlagLXCol = 0;
1004	1094	return;
1005	1095	}
	1096
	1097
	1098	logtrace(LogMotion,"colPic:%d (POC=%d) X:%d refIdxLX:%d refpiclist:%d\n",
	1099	colPic,
	1100	colImg->PicOrderCntVal,
	1101	X,refIdxLX,shdr->RefPicList[X][refIdxLX]);
	1102
	1103
	1104	// collocated reference image is unavailable -> no collocated MV
	1105
	1106	if (colImg->integrity == INTEGRITY_UNAVAILABLE_REFERENCE) {
	1107	out_mvLXCol->x = 0;
	1108	out_mvLXCol->y = 0;
	1109	*out_availableFlagLXCol = 0;
	1110	return;
	1111	}
	1112
	1113
	1114	// get the collocated MV
	1115
	1116	const MotionVectorSpec* mvi = colImg->get_mv_info(xColPb,yColPb);
	1117	int listCol;
	1118	int refIdxCol;
	1119	MotionVector mvCol;
	1120
	1121	logtrace(LogMotion,"read MVI %d;%d:\n",xColPb,yColPb);
	1122	logmvcand(*mvi);
	1123
	1124
	1125	// collocated MV uses only L1 -> use L1
	1126	if (mvi->predFlag[0]==0) {
	1127	mvCol = mvi->mv[1];
	1128	refIdxCol = mvi->refIdx[1];
	1129	listCol = 1;
	1130	}
	1131	// collocated MV uses only L0 -> use L0
	1132	else if (mvi->predFlag[1]==0) {
	1133	mvCol = mvi->mv[0];
	1134	refIdxCol = mvi->refIdx[0];
	1135	listCol = 0;
	1136	}
	1137	// collocated MV uses L0 and L1
1006	1138	else {
1007		logtrace(LogMotion,"colPic:%d (POC=%d) X:%d refIdxLX:%d refpiclist:%d\n",
1008		colPic,
1009		colImg->PicOrderCntVal,
1010		X,refIdxLX,shdr->RefPicList[X][refIdxLX]);
1011
1012		if (colImg->integrity == INTEGRITY_UNAVAILABLE_REFERENCE) {
1013		out_mvLXCol->x = 0;
1014		out_mvLXCol->y = 0;
1015		*out_availableFlagLXCol = 0;
1016		return;
1017		}
1018
1019		const PredVectorInfo* mvi = colImg->get_mv_info(xColPb,yColPb);
1020		int listCol;
1021		int refIdxCol;
1022		MotionVector mvCol;
1023
1024		logtrace(LogMotion,"read MVI %d;%d:\n",xColPb,yColPb);
1025		logmvcand(*mvi);
1026
1027		if (mvi->predFlag[0]==0) {
1028		mvCol = mvi->mv[1];
1029		refIdxCol = mvi->refIdx[1];
1030		listCol = 1;
	1139	bool allRefFramesBeforeCurrentFrame = true;
	1140
	1141	const int currentPOC = img->PicOrderCntVal;
	1142
	1143	// all reference POCs earlier than current POC (list 1)
	1144	// Test L1 first, because there is a higher change to find a future reference frame.
	1145
	1146	for (int rIdx=0; rIdx<shdr->num_ref_idx_l1_active && allRefFramesBeforeCurrentFrame; rIdx++)
	1147	{
	1148	const de265_image* refimg = ctx->get_image(shdr->RefPicList[1][rIdx]);
	1149	int refPOC = refimg->PicOrderCntVal;
	1150
	1151	if (refPOC > currentPOC) {
	1152	allRefFramesBeforeCurrentFrame = false;
	1153	}
	1154	}
	1155
	1156	// all reference POCs earlier than current POC (list 0)
	1157
	1158	for (int rIdx=0; rIdx<shdr->num_ref_idx_l0_active && allRefFramesBeforeCurrentFrame; rIdx++)
	1159	{
	1160	const de265_image* refimg = ctx->get_image(shdr->RefPicList[0][rIdx]);
	1161	int refPOC = refimg->PicOrderCntVal;
	1162
	1163	if (refPOC > currentPOC) {
	1164	allRefFramesBeforeCurrentFrame = false;
	1165	}
	1166	}
	1167
	1168
	1169	/* TODO: What is the rationale behind this ???
	1170
	1171	My guess:
	1172	when there are images before the current frame (most probably in L0) and images after
	1173	the current frame (most probably in L1), we take the reference in the opposite
	1174	direction than where the collocated frame is positioned in the hope that the distance
	1175	to the current frame will be smaller and thus give a better prediction.
	1176
	1177	If all references point into the past, we cannot say much about the temporal order or
	1178	L0,L1 and thus take over both parts.
	1179	*/
	1180
	1181	if (allRefFramesBeforeCurrentFrame) {
	1182	mvCol = mvi->mv[X];
	1183	refIdxCol = mvi->refIdx[X];
	1184	listCol = X;
1031	1185	}
1032	1186	else {
1033		if (mvi->predFlag[1]==0) {
1034		mvCol = mvi->mv[0];
1035		refIdxCol = mvi->refIdx[0];
1036		listCol = 0;
1037		}
1038		else {
1039		int AllDiffPicOrderCntLEZero = true;
1040
1041		const int PicOrderCntVal = img->PicOrderCntVal;
1042
1043		for (int rIdx=0; rIdx<shdr->num_ref_idx_l0_active && AllDiffPicOrderCntLEZero; rIdx++)
1044		{
1045		const de265_image* imgA = ctx->get_image(shdr->RefPicList[0][rIdx]);
1046		int aPOC = imgA->PicOrderCntVal;
1047
1048		if (aPOC > PicOrderCntVal) {
1049		AllDiffPicOrderCntLEZero = false;
1050		}
1051		}
1052
1053		for (int rIdx=0; rIdx<shdr->num_ref_idx_l1_active && AllDiffPicOrderCntLEZero; rIdx++)
1054		{
1055		const de265_image* imgA = ctx->get_image(shdr->RefPicList[1][rIdx]);
1056		int aPOC = imgA->PicOrderCntVal;
1057
1058		if (aPOC > PicOrderCntVal) {
1059		AllDiffPicOrderCntLEZero = false;
1060		}
1061		}
1062
1063		if (AllDiffPicOrderCntLEZero) {
1064		mvCol = mvi->mv[X];
1065		refIdxCol = mvi->refIdx[X];
1066		listCol = X;
1067		}
1068		else {
1069		int N = shdr->collocated_from_l0_flag;
1070		mvCol = mvi->mv[N];
1071		refIdxCol = mvi->refIdx[N];
1072		listCol = N;
1073		}
1074		}
1075		}
1076
1077
1078
1079		const slice_segment_header* colShdr = colImg->slices[ colImg->get_SliceHeaderIndex(xColPb,yColPb) ];
1080
1081		if (shdr->LongTermRefPic[X][refIdxLX] !=
1082		colShdr->LongTermRefPic[listCol][refIdxCol]) {
1083		*out_availableFlagLXCol = 0;
1084		out_mvLXCol->x = 0;
1085		out_mvLXCol->y = 0;
	1187	int N = shdr->collocated_from_l0_flag;
	1188	mvCol = mvi->mv[N];
	1189	refIdxCol = mvi->refIdx[N];
	1190	listCol = N;
	1191	}
	1192	}
	1193
	1194
	1195
	1196	const slice_segment_header* colShdr = colImg->slices[ colImg->get_SliceHeaderIndex(xColPb,yColPb) ];
	1197
	1198	if (shdr->LongTermRefPic[X][refIdxLX] !=
	1199	colShdr->LongTermRefPic[listCol][refIdxCol]) {
	1200	*out_availableFlagLXCol = 0;
	1201	out_mvLXCol->x = 0;
	1202	out_mvLXCol->y = 0;
	1203	}
	1204	else {
	1205	*out_availableFlagLXCol = 1;
	1206
	1207	const bool isLongTerm = shdr->LongTermRefPic[X][refIdxLX];
	1208
	1209	int colDist = colImg->PicOrderCntVal - colShdr->RefPicList_POC[listCol][refIdxCol];
	1210	int currDist = img->PicOrderCntVal - shdr->RefPicList_POC[X][refIdxLX];
	1211
	1212	logtrace(LogMotion,"COLPOCDIFF %d %d [%d %d / %d %d]\n",colDist, currDist,
	1213	colImg->PicOrderCntVal, colShdr->RefPicList_POC[listCol][refIdxCol],
	1214	img->PicOrderCntVal, shdr->RefPicList_POC[X][refIdxLX]
	1215	);
	1216
	1217	if (isLongTerm \|\| colDist == currDist) {
	1218	*out_mvLXCol = mvCol;
1086	1219	}
1087	1220	else {
1088		*out_availableFlagLXCol = 1;
1089
1090		const bool isLongTerm = shdr->LongTermRefPic[X][refIdxLX];
1091
1092		int colDist = colImg->PicOrderCntVal - colShdr->RefPicList_POC[listCol][refIdxCol];
1093		int currDist = img->PicOrderCntVal - shdr->RefPicList_POC[X][refIdxLX];
1094
1095		logtrace(LogMotion,"COLPOCDIFF %d %d [%d %d / %d %d]\n",colDist, currDist,
1096		colImg->PicOrderCntVal, colShdr->RefPicList_POC[listCol][refIdxCol],
1097		img->PicOrderCntVal, shdr->RefPicList_POC[X][refIdxLX]
1098		);
1099
1100		if (isLongTerm \|\| colDist == currDist) {
1101		*out_mvLXCol = mvCol;
1102		}
1103		else {
1104		if (!scale_mv(out_mvLXCol, mvCol, colDist, currDist)) {
1105		ctx->add_warning(DE265_WARNING_INCORRECT_MOTION_VECTOR_SCALING, false);
1106		img->integrity = INTEGRITY_DECODING_ERRORS;
1107		}
1108
1109		logtrace(LogMotion,"scale: %d;%d to %d;%d\n",
1110		mvCol.x,mvCol.y, out_mvLXCol->x,out_mvLXCol->y);
1111		}
	1221	if (!scale_mv(out_mvLXCol, mvCol, colDist, currDist)) {
	1222	ctx->add_warning(DE265_WARNING_INCORRECT_MOTION_VECTOR_SCALING, false);
	1223	img->integrity = INTEGRITY_DECODING_ERRORS;
	1224	}
	1225
	1226	logtrace(LogMotion,"scale: %d;%d to %d;%d\n",
	1227	mvCol.x,mvCol.y, out_mvLXCol->x,out_mvLXCol->y);
1112	1228	}
1113	1229	}
1114	1230	}
1115	1231
1116	1232
1117	1233	// 8.5.3.1.7
1118		void derive_temporal_luma_vector_prediction(decoder_context* ctx,
	1234	void derive_temporal_luma_vector_prediction(base_context* ctx,
1119	1235	de265_image* img,
1120	1236	const slice_segment_header* shdr,
1121	1237	int xP,int yP,
1122	1238	int nPbW,int nPbH,
1123		int refIdxL, int X,
	1239	int refIdxL,
	1240	int X, // which MV (L0/L1) to get
1124	1241	MotionVector* out_mvLXCol,
1125	1242	uint8_t* out_availableFlagLXCol)
1126	1243	{
	1244	// --- no temporal MVP -> exit ---
1127	1245
1128	1246	if (shdr->slice_temporal_mvp_enabled_flag == 0) {
1129	1247	out_mvLXCol->x = 0;

1132	1250	return;
1133	1251	}
1134	1252
	1253
	1254	// --- find collocated reference image ---
	1255
1135	1256	int Log2CtbSizeY = img->sps.Log2CtbSizeY;
1136	1257
1137	1258	int colPic; // TODO: this is the same for the whole slice. We can precompute it.

1141	1262	{
1142	1263	logtrace(LogMotion,"collocated L1 ref_idx=%d\n",shdr->collocated_ref_idx);
1143	1264
1144		// TODO: make sure that shdr->collocated_ref_idx is a valid index
1145	1265	colPic = shdr->RefPicList[1][ shdr->collocated_ref_idx ];
1146	1266	}
1147	1267	else
1148	1268	{
1149	1269	logtrace(LogMotion,"collocated L0 ref_idx=%d\n",shdr->collocated_ref_idx);
1150	1270
1151		// TODO: make sure that shdr->collocated_ref_idx is a valid index
1152	1271	colPic = shdr->RefPicList[0][ shdr->collocated_ref_idx ];
1153	1272	}
1154	1273
1155		//logtrace(LogMotion,"collocated reference POC=%d\n",ctx->dpb[colPic].PicOrderCntVal);
1156
	1274
	1275	// check whether collocated reference picture exists
1157	1276
1158	1277	if (!ctx->has_image(colPic)) {
1159	1278	out_mvLXCol->x = 0;

1165	1284	}
1166	1285
1167	1286
	1287	// --- get collocated MV either at bottom-right corner or from center of PB ---
	1288
1168	1289	int xColPb,yColPb;
1169	1290	int yColBr = yP + nPbH; // bottom right collocated motion vector position
1170	1291	int xColBr = xP + nPbW;
1171	1292
	1293	/* If neighboring pixel at bottom-right corner is in the same CTB-row and inside the image,
	1294	use this (reduced down to 16 pixels resolution) as collocated MV position.
	1295
	1296	Note: see 2014, Sze, Sect. 5.2.1.2 why candidate C0 is excluded when on another CTB-row.
	1297	This is to reduce the memory bandwidth requirements.
	1298	*/
1172	1299	if ((yP>>Log2CtbSizeY) == (yColBr>>Log2CtbSizeY) &&
1173	1300	xColBr < img->sps.pic_width_in_luma_samples &&
1174	1301	yColBr < img->sps.pic_height_in_luma_samples)

1207	1334	};
1208	1335
1209	1336	// 8.5.3.1.3
1210		void derive_combined_bipredictive_merging_candidates(const decoder_context* ctx,
1211		slice_segment_header* shdr,
1212		PredVectorInfo* inout_mergeCandList,
	1337	/* Note (TODO): during decoding, we know which of the candidates we will select.
	1338	+ Hence, we do not really have to generate the other ones...
	1339	+ */
	1340	void derive_combined_bipredictive_merging_candidates(const base_context* ctx,
	1341	const slice_segment_header* shdr,
	1342	MotionVectorSpec* inout_mergeCandList,
1213	1343	int* inout_numMergeCand,
1214		int numOrigMergeCand)
	1344	int maxCandidates)
1215	1345	{
1216		if (inout_numMergeCand>1 && inout_numMergeCand < shdr->MaxNumMergeCand) {
	1346	if (inout_numMergeCand>1 && inout_numMergeCand < maxCandidates) {
	1347	int numOrigMergeCand = *inout_numMergeCand;
	1348
1217	1349	int numInputMergeCand = *inout_numMergeCand;
1218	1350	int combIdx = 0;
1219	1351	uint8_t combStop = false;

1227	1359	assert(false); // bitstream error -> TODO: conceal error
1228	1360	}
1229	1361
1230		PredVectorInfo* l0Cand = &inout_mergeCandList[l0CandIdx];
1231		PredVectorInfo* l1Cand = &inout_mergeCandList[l1CandIdx];
	1362	MotionVectorSpec* l0Cand = &inout_mergeCandList[l0CandIdx];
	1363	MotionVectorSpec* l1Cand = &inout_mergeCandList[l1CandIdx];
1232	1364
1233	1365	logtrace(LogMotion,"add bipredictive merging candidate (combIdx:%d)\n",combIdx);
1234	1366	logtrace(LogMotion,"l0Cand:\n"); logmvcand(*l0Cand);

1236	1368
1237	1369	const de265_image* img0 = l0Cand->predFlag[0] ? ctx->get_image(shdr->RefPicList[0][l0Cand->refIdx[0]]) : NULL;
1238	1370	const de265_image* img1 = l1Cand->predFlag[1] ? ctx->get_image(shdr->RefPicList[1][l1Cand->refIdx[1]]) : NULL;
	1371
	1372	if (l0Cand->predFlag[0] && !img0) {
	1373	return; // TODO error
	1374	}
	1375
	1376	if (l1Cand->predFlag[1] && !img1) {
	1377	return; // TODO error
	1378	}
1239	1379
1240	1380	if (l0Cand->predFlag[0] && l1Cand->predFlag[1] &&
1241	1381	(img0->PicOrderCntVal != img1->PicOrderCntVal \|\|
1242	1382	l0Cand->mv[0].x != l1Cand->mv[1].x \|\|
1243	1383	l0Cand->mv[0].y != l1Cand->mv[1].y)) {
1244		PredVectorInfo* p = &inout_mergeCandList[ *inout_numMergeCand ];
	1384	MotionVectorSpec* p = &inout_mergeCandList[ *inout_numMergeCand ];
1245	1385	p->refIdx[0] = l0Cand->refIdx[0];
1246	1386	p->refIdx[1] = l1Cand->refIdx[1];
1247	1387	p->predFlag[0] = l0Cand->predFlag[0];

1256	1396
1257	1397	combIdx++;
1258	1398	if (combIdx == numOrigMergeCand*(numOrigMergeCand-1) \|\|
1259		*inout_numMergeCand == shdr->MaxNumMergeCand) {
	1399	*inout_numMergeCand == maxCandidates) {
1260	1400	combStop = true;
1261	1401	}
1262	1402	}

1265	1405
1266	1406
1267	1407	// 8.5.3.1.1
1268		void derive_luma_motion_merge_mode(decoder_context* ctx,
1269		thread_context* tctx,
1270		int xC,int yC, int xP,int yP,
1271		int nCS, int nPbW,int nPbH, int partIdx,
1272		VectorInfo* out_vi)
	1408	static void get_merge_candidate_list_without_step_9(base_context* ctx,
	1409	const slice_segment_header* shdr,
	1410	de265_image* img,
	1411	int xC,int yC, int xP,int yP,
	1412	int nCS, int nPbW,int nPbH, int partIdx,
	1413	int max_merge_idx,
	1414	MotionVectorSpec* mergeCandList)
1273	1415	{
1274		slice_segment_header* shdr = tctx->shdr;
1275	1416
1276	1417	//int xOrigP = xP;
1277	1418	//int yOrigP = yP;
1278	1419	int nOrigPbW = nPbW;
1279	1420	int nOrigPbH = nPbH;
1280	1421
1281		int singleMCLFlag;
1282		singleMCLFlag = (tctx->img->pps.log2_parallel_merge_level > 2 && nCS==8);
	1422	int singleMCLFlag; // single merge-candidate-list (MCL) flag
	1423
	1424	/* Use single MCL for CBs of size 8x8, except when parallel-merge-level is at 4x4.
	1425	Without this flag, PBs smaller than 8x8 would not receive as much merging candidates.
	1426	Having additional candidates might have these advantages:
	1427	- coding MVs for these small PBs is expensive, and
	1428	- since the PBs are not far away from a proper (neighboring) merging candidate,
	1429	the quality of the candidates will still be good.
	1430	*/
	1431	singleMCLFlag = (img->pps.log2_parallel_merge_level > 2 && nCS==8);
1283	1432
1284	1433	if (singleMCLFlag) {
1285	1434	xP=xC;

1289	1438	partIdx=0;
1290	1439	}
1291	1440
1292		MergingCandidates mergeCand;
1293		derive_spatial_merging_candidates(tctx->img, xC,yC, nCS, xP,yP, singleMCLFlag,
1294		nPbW,nPbH,partIdx, &mergeCand);
1295
1296		int refIdxCol[2] = { 0,0 };
1297
1298		MotionVector mvCol[2];
1299		uint8_t predFlagLCol[2];
1300		derive_temporal_luma_vector_prediction(ctx,tctx->img,shdr, xP,yP,nPbW,nPbH,
1301		refIdxCol[0],0, &mvCol[0],
1302		&predFlagLCol[0]);
1303
1304		uint8_t availableFlagCol = predFlagLCol[0];
1305		predFlagLCol[1] = 0;
1306
1307		if (shdr->slice_type == SLICE_TYPE_B) {
1308		derive_temporal_luma_vector_prediction(ctx,tctx->img,shdr,
1309		xP,yP,nPbW,nPbH, refIdxCol[1],1, &mvCol[1],
1310		&predFlagLCol[1]);
1311		availableFlagCol \|= predFlagLCol[1];
1312		}
1313
1314
1315		// 4.
1316
1317		PredVectorInfo mergeCandList[5];
	1441	int maxCandidates = max_merge_idx+1;
	1442	//MotionVectorSpec mergeCandList[5];
1318	1443	int numMergeCand=0;
1319	1444
1320		for (int i=0;i<5;i++) {
1321		if (mergeCand.available[i]) {
1322		mergeCandList[numMergeCand++] = mergeCand.pred_vector[i];
1323		}
1324		}
1325
1326		if (availableFlagCol) {
1327		// TODO: save in mergeCand directly...
1328		mergeCand.available[PRED_COL] = availableFlagCol;
1329		mergeCand.pred_vector[PRED_COL].mv[0] = mvCol[0];
1330		mergeCand.pred_vector[PRED_COL].mv[1] = mvCol[1];
1331		mergeCand.pred_vector[PRED_COL].predFlag[0] = predFlagLCol[0];
1332		mergeCand.pred_vector[PRED_COL].predFlag[1] = predFlagLCol[1];
1333		mergeCand.pred_vector[PRED_COL].refIdx[0] = refIdxCol[0];
1334		mergeCand.pred_vector[PRED_COL].refIdx[1] = refIdxCol[1];
1335
1336		mergeCandList[numMergeCand++] = mergeCand.pred_vector[PRED_COL];
1337		}
1338
1339		// 5.
1340
1341		//int numOrigMergeCand = numMergeCand;
1342
1343		// 6.
1344
1345		//int numCombMergeCand = 0;
	1445	// --- spatial merge candidates
	1446
	1447	numMergeCand = derive_spatial_merging_candidates(img, xC,yC, nCS, xP,yP, singleMCLFlag,
	1448	nPbW,nPbH,partIdx, mergeCandList,
	1449	maxCandidates);
	1450
	1451	// --- collocated merge candidate
	1452	if (numMergeCand < maxCandidates) {
	1453	int refIdxCol[2] = { 0,0 };
	1454
	1455	MotionVector mvCol[2];
	1456	uint8_t predFlagLCol[2];
	1457	derive_temporal_luma_vector_prediction(ctx,img,shdr, xP,yP,nPbW,nPbH,
	1458	refIdxCol[0],0, &mvCol[0],
	1459	&predFlagLCol[0]);
	1460
	1461	uint8_t availableFlagCol = predFlagLCol[0];
	1462	predFlagLCol[1] = 0;
	1463
	1464	if (shdr->slice_type == SLICE_TYPE_B) {
	1465	derive_temporal_luma_vector_prediction(ctx,img,shdr,
	1466	xP,yP,nPbW,nPbH, refIdxCol[1],1, &mvCol[1],
	1467	&predFlagLCol[1]);
	1468	availableFlagCol \|= predFlagLCol[1];
	1469	}
	1470
	1471
	1472	if (availableFlagCol) {
	1473	MotionVectorSpec* colVec = &mergeCandList[numMergeCand++];
	1474
	1475	colVec->mv[0] = mvCol[0];
	1476	colVec->mv[1] = mvCol[1];
	1477	colVec->predFlag[0] = predFlagLCol[0];
	1478	colVec->predFlag[1] = predFlagLCol[1];
	1479	colVec->refIdx[0] = refIdxCol[0];
	1480	colVec->refIdx[1] = refIdxCol[1];
	1481	}
	1482	}
	1483
	1484
	1485	// --- bipredictive merge candidates ---
1346	1486
1347	1487	if (shdr->slice_type == SLICE_TYPE_B) {
1348	1488	derive_combined_bipredictive_merging_candidates(ctx, shdr,
1349		mergeCandList, &numMergeCand, numMergeCand);
1350
1351		//numCombMergeCand = numMergeCand - numOrigMergeCand;
1352		}
1353
1354
1355		// 7.
1356
1357		derive_zero_motion_vector_candidates(shdr, mergeCandList, &numMergeCand);
1358
1359		// 8.
1360
1361		int merge_idx = tctx->merge_idx; // get_merge_idx(ctx,xP,yP);
1362		out_vi->lum = mergeCandList[merge_idx];
	1489	mergeCandList, &numMergeCand, maxCandidates);
	1490	}
	1491
	1492
	1493	// --- zero-vector merge candidates ---
	1494
	1495	derive_zero_motion_vector_candidates(shdr, mergeCandList, &numMergeCand, maxCandidates);
1363	1496
1364	1497
1365	1498	logtrace(LogMotion,"mergeCandList:\n");
1366	1499	for (int i=0;i<shdr->MaxNumMergeCand;i++)
1367	1500	{
1368		logtrace(LogMotion, " %d:%s\n", i, i==merge_idx ? " SELECTED":"");
	1501	//logtrace(LogMotion, " %d:%s\n", i, i==merge_idx ? " SELECTED":"");
1369	1502	logmvcand(mergeCandList[i]);
1370	1503	}
1371
1372		// 9.
1373
1374		if (out_vi->lum.predFlag[0] && out_vi->lum.predFlag[1] && nOrigPbW+nOrigPbH==12) {
1375		out_vi->lum.refIdx[1] = -1;
1376		out_vi->lum.predFlag[1] = 0;
1377		}
1378	1504	}
1379	1505
1380	1506
	1507
	1508	void get_merge_candidate_list(base_context* ctx,
	1509	const slice_segment_header* shdr,
	1510	de265_image* img,
	1511	int xC,int yC, int xP,int yP,
	1512	int nCS, int nPbW,int nPbH, int partIdx,
	1513	MotionVectorSpec* mergeCandList)
	1514	{
	1515	int max_merge_idx = 5-shdr->five_minus_max_num_merge_cand -1;
	1516
	1517	get_merge_candidate_list_without_step_9(ctx, shdr, img,
	1518	xC,yC,xP,yP,nCS,nPbW,nPbH, partIdx,
	1519	max_merge_idx, mergeCandList);
	1520
	1521	// 9. for encoder: modify all merge candidates
	1522
	1523	for (int i=0;i<=max_merge_idx;i++) {
	1524	if (mergeCandList[i].predFlag[0] &&
	1525	mergeCandList[i].predFlag[1] &&
	1526	nPbW+nPbH==12)
	1527	{
	1528	mergeCandList[i].refIdx[1] = -1;
	1529	mergeCandList[i].predFlag[1] = 0;
	1530	}
	1531	}
	1532	}
	1533
	1534
	1535
	1536	void derive_luma_motion_merge_mode(base_context* ctx,
	1537	const slice_segment_header* shdr,
	1538	de265_image* img,
	1539	int xC,int yC, int xP,int yP,
	1540	int nCS, int nPbW,int nPbH, int partIdx,
	1541	int merge_idx,
	1542	MotionVectorSpec* out_vi)
	1543	{
	1544	MotionVectorSpec mergeCandList[5];
	1545
	1546	get_merge_candidate_list_without_step_9(ctx, shdr, img,
	1547	xC,yC,xP,yP,nCS,nPbW,nPbH, partIdx,
	1548	merge_idx, mergeCandList);
	1549
	1550
	1551	*out_vi = mergeCandList[merge_idx];
	1552
	1553	// 8.5.3.1.1 / 9.
	1554
	1555	if (out_vi->predFlag[0] && out_vi->predFlag[1] && nPbW+nPbH==12) {
	1556	out_vi->refIdx[1] = -1;
	1557	out_vi->predFlag[1] = 0;
	1558	}
	1559	}
	1560
	1561
1381	1562	// 8.5.3.1.6
1382		void derive_spatial_luma_vector_prediction(de265_image* img,
	1563	void derive_spatial_luma_vector_prediction(base_context* ctx,
	1564	de265_image* img,
1383	1565	const slice_segment_header* shdr,
1384	1566	int xC,int yC,int nCS,int xP,int yP,
1385	1567	int nPbW,int nPbH, int X,

1387	1569	uint8_t out_availableFlagLXN[2],
1388	1570	MotionVector out_mvLXN[2])
1389	1571	{
1390		const decoder_context* ctx = img->decctx;
1391
1392	1572	int isScaledFlagLX = 0;
1393	1573
1394	1574	const int A=0;
1395	1575	const int B=1;
	1576
	1577	out_availableFlagLXN[A] = 0;
	1578	out_availableFlagLXN[B] = 0;
	1579
1396	1580
1397	1581	// --- A ---
1398	1582

1427	1611	int refIdxA=-1;
1428	1612
1429	1613	// the POC we want to reference in this PB
1430		const int referenced_POC = ctx->get_image(shdr->RefPicList[X][ refIdxLX ])->PicOrderCntVal;
1431
1432		const int referenced_refIdx = refIdxLX;
	1614	const de265_image* tmpimg = ctx->get_image(shdr->RefPicList[X][ refIdxLX ]);
	1615	if (tmpimg==NULL) { return; }
	1616	const int referenced_POC = tmpimg->PicOrderCntVal;
1433	1617
1434	1618	for (int k=0;k<=1;k++) {
1435	1619	if (availableA[k] &&

1437	1621	img->get_pred_mode(xA[k],yA[k]) != MODE_INTRA) {
1438	1622
1439	1623	int Y=1-X;
1440
1441		const PredVectorInfo* vi = img->get_mv_info(xA[k],yA[k]);
	1624
	1625	const MotionVectorSpec* vi = img->get_mv_info(xA[k],yA[k]);
1442	1626	logtrace(LogMotion,"MVP A%d=\n",k);
1443	1627	logmvcand(*vi);
1444	1628
	1629	const de265_image* imgX = NULL;
	1630	if (vi->predFlag[X]) imgX = ctx->get_image(shdr->RefPicList[X][ vi->refIdx[X] ]);
	1631	const de265_image* imgY = NULL;
	1632	if (vi->predFlag[Y]) imgY = ctx->get_image(shdr->RefPicList[Y][ vi->refIdx[Y] ]);
	1633
1445	1634	// check whether the predictor X is available and references the same POC
1446		if (vi->predFlag[X] &&
1447		ctx->get_image(shdr->RefPicList[X][ vi->refIdx[X] ])->PicOrderCntVal == referenced_POC) {
1448		//vi->refIdx[X] == referenced_refIdx) {
	1635	if (vi->predFlag[X] && imgX && imgX->PicOrderCntVal == referenced_POC) {
1449	1636
1450	1637	logtrace(LogMotion,"take A%d/L%d as A candidate with same POC\n",k,X);
1451	1638

1454	1641	refIdxA = vi->refIdx[X];
1455	1642	}
1456	1643	// check whether the other predictor (Y) is available and references the same POC
1457		else if (vi->predFlag[Y] &&
1458		ctx->get_image(shdr->RefPicList[Y][ vi->refIdx[Y] ])->PicOrderCntVal == referenced_POC) {
1459		//vi->refIdx[Y] == referenced_refIdx) {
	1644	else if (vi->predFlag[Y] && imgY && imgY->PicOrderCntVal == referenced_POC) {
1460	1645
1461	1646	logtrace(LogMotion,"take A%d/L%d as A candidate with same POC\n",k,Y);
1462	1647

1478	1663	img->get_pred_mode(xA[k],yA[k]) != MODE_INTRA) {
1479	1664
1480	1665	int Y=1-X;
1481
1482		const PredVectorInfo* vi = img->get_mv_info(xA[k],yA[k]);
	1666
	1667	const MotionVectorSpec* vi = img->get_mv_info(xA[k],yA[k]);
1483	1668	if (vi->predFlag[X]==1 &&
1484	1669	shdr->LongTermRefPic[X][refIdxLX] == shdr->LongTermRefPic[X][ vi->refIdx[X] ]) {
1485	1670

1503	1688	}
1504	1689
1505	1690	if (out_availableFlagLXN[A]==1) {
	1691	if (refIdxA<0) {
	1692	out_availableFlagLXN[0] = out_availableFlagLXN[1] = false;
	1693	return; // error
	1694	}
	1695
1506	1696	assert(refIdxA>=0);
1507	1697	assert(refPicList>=0);
1508	1698
1509	1699	const de265_image* refPicA = ctx->get_image(shdr->RefPicList[refPicList][refIdxA ]);
1510	1700	const de265_image* refPicX = ctx->get_image(shdr->RefPicList[X ][refIdxLX]);
1511	1701
1512		int picStateA = shdr->RefPicList_PicState[refPicList][refIdxA ];
1513		int picStateX = shdr->RefPicList_PicState[X ][refIdxLX];
	1702	//int picStateA = shdr->RefPicList_PicState[refPicList][refIdxA ];
	1703	//int picStateX = shdr->RefPicList_PicState[X ][refIdxLX];
1514	1704
1515	1705	int isLongTermA = shdr->LongTermRefPic[refPicList][refIdxA ];
1516	1706	int isLongTermX = shdr->LongTermRefPic[X ][refIdxLX];

1526	1716	{
1527	1717	int distA = img->PicOrderCntVal - refPicA->PicOrderCntVal;
1528	1718	int distX = img->PicOrderCntVal - referenced_POC;
1529
	1719
1530	1720	if (!scale_mv(&out_mvLXN[A], out_mvLXN[A], distA, distX)) {
1531		img->decctx->add_warning(DE265_WARNING_INCORRECT_MOTION_VECTOR_SCALING, false);
	1721	ctx->add_warning(DE265_WARNING_INCORRECT_MOTION_VECTOR_SCALING, false);
1532	1722	img->integrity = INTEGRITY_DECODING_ERRORS;
1533	1723	}
1534	1724	}

1563	1753	availableB[k] = img->available_pred_blk(xC,yC, nCS, xP,yP, nPbW,nPbH,partIdx, xB[k],yB[k]);
1564	1754
1565	1755	if (availableB[k] && out_availableFlagLXN[B]==0) {
1566
	1756
1567	1757	int Y=1-X;
1568
1569		const PredVectorInfo* vi = img->get_mv_info(xB[k],yB[k]);
	1758
	1759	const MotionVectorSpec* vi = img->get_mv_info(xB[k],yB[k]);
1570	1760	logtrace(LogMotion,"MVP B%d=\n",k);
1571	1761	logmvcand(*vi);
1572	1762
1573	1763
1574		if (vi->predFlag[X] &&
1575		ctx->get_image(shdr->RefPicList[X][ vi->refIdx[X] ])->PicOrderCntVal == referenced_POC) {
	1764	const de265_image* imgX = NULL;
	1765	if (vi->predFlag[X]) imgX = ctx->get_image(shdr->RefPicList[X][ vi->refIdx[X] ]);
	1766	const de265_image* imgY = NULL;
	1767	if (vi->predFlag[Y]) imgY = ctx->get_image(shdr->RefPicList[Y][ vi->refIdx[Y] ]);
	1768
	1769	if (vi->predFlag[X] && imgX && imgX->PicOrderCntVal == referenced_POC) {
1576	1770	logtrace(LogMotion,"a) take B%d/L%d as B candidate with same POC\n",k,X);
1577	1771
1578	1772	out_availableFlagLXN[B]=1;
1579	1773	out_mvLXN[B] = vi->mv[X];
1580	1774	refIdxB = vi->refIdx[X];
1581	1775	}
1582		else if (vi->predFlag[Y] &&
1583		ctx->get_image(shdr->RefPicList[Y][ vi->refIdx[Y] ])->PicOrderCntVal == referenced_POC) {
	1776	else if (vi->predFlag[Y] && imgY && imgY->PicOrderCntVal == referenced_POC) {
1584	1777	logtrace(LogMotion,"b) take B%d/L%d as B candidate with same POC\n",k,Y);
1585	1778
1586	1779	out_availableFlagLXN[B]=1;

1617	1810
1618	1811	if (availableB[k]) {
1619	1812	int Y=1-X;
1620
1621		const PredVectorInfo* vi = img->get_mv_info(xB[k],yB[k]);
	1813
	1814	const MotionVectorSpec* vi = img->get_mv_info(xB[k],yB[k]);
1622	1815
1623	1816	if (vi->predFlag[X]==1 &&
1624	1817	shdr->LongTermRefPic[X][refIdxLX] == shdr->LongTermRefPic[X][ vi->refIdx[X] ]) {

1637	1830	}
1638	1831
1639	1832	if (out_availableFlagLXN[B]==1) {
	1833	if (refIdxB<0) {
	1834	out_availableFlagLXN[0] = out_availableFlagLXN[1] = false;
	1835	return; // error
	1836	}
	1837
1640	1838	assert(refPicList>=0);
1641	1839	assert(refIdxB>=0);
1642	1840
1643		const de265_image* refPicB=img->decctx->get_image(shdr->RefPicList[refPicList][refIdxB ]);
1644		const de265_image* refPicX=img->decctx->get_image(shdr->RefPicList[X ][refIdxLX]);
	1841	const de265_image* refPicB=ctx->get_image(shdr->RefPicList[refPicList][refIdxB ]);
	1842	const de265_image* refPicX=ctx->get_image(shdr->RefPicList[X ][refIdxLX]);
1645	1843
1646	1844	int isLongTermB = shdr->LongTermRefPic[refPicList][refIdxB ];
1647	1845	int isLongTermX = shdr->LongTermRefPic[X ][refIdxLX];
1648	1846
1649		if (refPicB->PicOrderCntVal != refPicX->PicOrderCntVal &&
1650		!isLongTermB && !isLongTermX) {
	1847	if (refPicB==NULL \|\| refPicX==NULL) {
	1848	img->decctx->add_warning(DE265_WARNING_NONEXISTING_REFERENCE_PICTURE_ACCESSED,false);
	1849	img->integrity = INTEGRITY_DECODING_ERRORS;
	1850	}
	1851	else if (refPicB->PicOrderCntVal != refPicX->PicOrderCntVal &&
	1852	!isLongTermB && !isLongTermX) {
1651	1853	int distB = img->PicOrderCntVal - refPicB->PicOrderCntVal;
1652	1854	int distX = img->PicOrderCntVal - referenced_POC;
1653	1855
1654	1856	logtrace(LogMotion,"scale MVP B: B-POC:%d X-POC:%d\n",refPicB->PicOrderCntVal,refPicX->PicOrderCntVal);
1655	1857
1656	1858	if (!scale_mv(&out_mvLXN[B], out_mvLXN[B], distB, distX)) {
1657		img->decctx->add_warning(DE265_WARNING_INCORRECT_MOTION_VECTOR_SCALING, false);
	1859	ctx->add_warning(DE265_WARNING_INCORRECT_MOTION_VECTOR_SCALING, false);
1658	1860	img->integrity = INTEGRITY_DECODING_ERRORS;
1659	1861	}
1660	1862	}

1663	1865	}
1664	1866	}
1665	1867
	1868
1666	1869	// 8.5.3.1.5
1667		MotionVector luma_motion_vector_prediction(decoder_context* ctx,
1668		thread_context* tctx,
	1870	void fill_luma_motion_vector_predictors(base_context* ctx,
	1871	const slice_segment_header* shdr,
	1872	de265_image* img,
	1873	int xC,int yC,int nCS,int xP,int yP,
	1874	int nPbW,int nPbH, int l,
	1875	int refIdx, int partIdx,
	1876	MotionVector out_mvpList[2])
	1877	{
	1878	// 8.5.3.1.6: derive two spatial vector predictors A (0) and B (1)
	1879
	1880	uint8_t availableFlagLXN[2];
	1881	MotionVector mvLXN[2];
	1882
	1883	derive_spatial_luma_vector_prediction(ctx, img, shdr, xC,yC, nCS, xP,yP,
	1884	nPbW,nPbH, l, refIdx, partIdx,
	1885	availableFlagLXN, mvLXN);
	1886
	1887	// 8.5.3.1.7: if we only have one spatial vector or both spatial vectors are the same,
	1888	// derive a temporal predictor
	1889
	1890	uint8_t availableFlagLXCol;
	1891	MotionVector mvLXCol;
	1892
	1893
	1894	if (availableFlagLXN[0] &&
	1895	availableFlagLXN[1] &&
	1896	(mvLXN[0].x != mvLXN[1].x \|\| mvLXN[0].y != mvLXN[1].y)) {
	1897	availableFlagLXCol = 0;
	1898	}
	1899	else {
	1900	derive_temporal_luma_vector_prediction(ctx, img, shdr,
	1901	xP,yP, nPbW,nPbH, refIdx,l,
	1902	&mvLXCol, &availableFlagLXCol);
	1903	}
	1904
	1905
	1906	// --- build candidate vector list with exactly two entries ---
	1907
	1908	int numMVPCandLX=0;
	1909
	1910	// spatial predictor A
	1911
	1912	if (availableFlagLXN[0])
	1913	{
	1914	out_mvpList[numMVPCandLX++] = mvLXN[0];
	1915	}
	1916
	1917	// spatial predictor B (if not same as A)
	1918
	1919	if (availableFlagLXN[1] &&
	1920	(!availableFlagLXN[0] \|\| // in case A in not available, but mvLXA initialized to same as mvLXB
	1921	(mvLXN[0].x != mvLXN[1].x \|\| mvLXN[0].y != mvLXN[1].y)))
	1922	{
	1923	out_mvpList[numMVPCandLX++] = mvLXN[1];
	1924	}
	1925
	1926	// temporal predictor
	1927
	1928	if (availableFlagLXCol)
	1929	{
	1930	out_mvpList[numMVPCandLX++] = mvLXCol;
	1931	}
	1932
	1933	// fill with zero predictors
	1934
	1935	while (numMVPCandLX<2) {
	1936	out_mvpList[numMVPCandLX].x = 0;
	1937	out_mvpList[numMVPCandLX].y = 0;
	1938	numMVPCandLX++;
	1939	}
	1940
	1941
	1942	assert(numMVPCandLX==2);
	1943	}
	1944
	1945
	1946	MotionVector luma_motion_vector_prediction(base_context* ctx,
	1947	const slice_segment_header* shdr,
	1948	de265_image* img,
	1949	const motion_spec& motion,
1669	1950	int xC,int yC,int nCS,int xP,int yP,
1670	1951	int nPbW,int nPbH, int l,
1671	1952	int refIdx, int partIdx)
1672	1953	{
1673		const slice_segment_header* shdr = tctx->shdr;
1674
1675
1676		// 8.5.3.1.6: derive two spatial vector predictors A (0) and B (1)
1677
1678		uint8_t availableFlagLXN[2];
1679		MotionVector mvLXN[2];
1680
1681		derive_spatial_luma_vector_prediction(tctx->img, shdr, xC,yC, nCS, xP,yP, nPbW,nPbH, l, refIdx, partIdx,
1682		availableFlagLXN, mvLXN);
1683
1684		// 8.5.3.1.7: if we only have one spatial vector or both spatial vectors are the same,
1685		// derive a temporal predictor
1686
1687		uint8_t availableFlagLXCol;
1688		MotionVector mvLXCol;
1689
1690
1691		if (availableFlagLXN[0] &&
1692		availableFlagLXN[1] &&
1693		(mvLXN[0].x != mvLXN[1].x \|\| mvLXN[0].y != mvLXN[1].y)) {
1694		availableFlagLXCol = 0;
1695		}
1696		else {
1697		derive_temporal_luma_vector_prediction(ctx, tctx->img, shdr, xP,yP, nPbW,nPbH, refIdx,l,
1698		&mvLXCol, &availableFlagLXCol);
1699		}
1700
1701
1702		// --- build candidate vector list with exactly two entries ---
1703
1704		int numMVPCandLX=0;
1705
1706		// spatial predictor A
1707
1708		MotionVector mvpList[3];
1709		if (availableFlagLXN[0])
1710		{
1711		mvpList[numMVPCandLX++] = mvLXN[0];
1712		}
1713
1714		// spatial predictor B (if not same as A)
1715
1716		if (availableFlagLXN[1] &&
1717		(!availableFlagLXN[0] \|\| // in case A in not available, but mvLXA initialized to same as mvLXB
1718		(mvLXN[0].x != mvLXN[1].x \|\| mvLXN[0].y != mvLXN[1].y)))
1719		{
1720		mvpList[numMVPCandLX++] = mvLXN[1];
1721		}
1722
1723		// temporal predictor
1724
1725		if (availableFlagLXCol)
1726		{
1727		mvpList[numMVPCandLX++] = mvLXCol;
1728		}
1729
1730		// fill with zero predictors
1731
1732		while (numMVPCandLX<2) {
1733		mvpList[numMVPCandLX].x = 0;
1734		mvpList[numMVPCandLX].y = 0;
1735		numMVPCandLX++;
1736		}
1737
	1954	MotionVector mvpList[2];
	1955
	1956	fill_luma_motion_vector_predictors(ctx, shdr, img,
	1957	xC,yC,nCS,xP,yP,
	1958	nPbW, nPbH, l, refIdx, partIdx,
	1959	mvpList);
1738	1960
1739	1961	// select predictor according to mvp_lX_flag
1740	1962
1741		return mvpList[ tctx->mvp_lX_flag[l] ];
	1963	return mvpList[ l ? motion.mvp_l1_flag : motion.mvp_l0_flag ];
1742	1964	}
1743	1965
	1966
1744	1967	#if DE265_LOG_TRACE
1745		void logMV(int x0,int y0,int nPbW,int nPbH, const char* mode,const VectorInfo* mv)
	1968	void logMV(int x0,int y0,int nPbW,int nPbH, const char* mode,const MotionVectorSpec* mv)
1746	1969	{
1747		int pred0 = mv->lum.predFlag[0];
1748		int pred1 = mv->lum.predFlag[1];
	1970	int pred0 = mv->predFlag[0];
	1971	int pred1 = mv->predFlag[1];
1749	1972
1750	1973	logtrace(LogMotion,
1751	1974	"*MV %d;%d [%d;%d] %s: (%d) %d;%d @%d (%d) %d;%d @%d\n", x0,y0,nPbW,nPbH,mode,
1752	1975	pred0,
1753		pred0 ? mv->lum.mv[0].x : 0,pred0 ? mv->lum.mv[0].y : 0, pred0 ? mv->lum.refIdx[0] : 0,
	1976	pred0 ? mv->mv[0].x : 0,pred0 ? mv->mv[0].y : 0, pred0 ? mv->refIdx[0] : 0,
1754	1977	pred1,
1755		pred1 ? mv->lum.mv[1].x : 0,pred1 ? mv->lum.mv[1].y : 0, pred1 ? mv->lum.refIdx[1] : 0);
	1978	pred1 ? mv->mv[1].x : 0,pred1 ? mv->mv[1].y : 0, pred1 ? mv->refIdx[1] : 0);
1756	1979	}
1757	1980	#else
1758	1981	#define logMV(x0,y0,nPbW,nPbH,mode,mv)

1761	1984
1762	1985
1763	1986	// 8.5.3.1
1764		void motion_vectors_and_ref_indices(decoder_context* ctx,
1765		thread_context* tctx,
1766		int xC,int yC, int xB,int yB, int nCS, int nPbW,int nPbH, int partIdx,
1767		VectorInfo* out_vi)
	1987	void motion_vectors_and_ref_indices(base_context* ctx,
	1988	const slice_segment_header* shdr,
	1989	de265_image* img,
	1990	const motion_spec& motion,
	1991	int xC,int yC, int xB,int yB, int nCS, int nPbW,int nPbH,
	1992	int partIdx,
	1993	MotionVectorSpec* out_vi)
1768	1994	{
1769	1995	//slice_segment_header* shdr = tctx->shdr;
1770	1996
1771	1997	int xP = xC+xB;
1772	1998	int yP = yC+yB;
1773	1999
1774		enum PredMode predMode = tctx->img->get_pred_mode(xC,yC);
	2000	enum PredMode predMode = img->get_pred_mode(xC,yC);
1775	2001
1776	2002	if (predMode == MODE_SKIP \|\|
1777		(predMode == MODE_INTER && tctx->merge_flag))
	2003	(predMode == MODE_INTER && motion.merge_flag))
1778	2004	{
1779		derive_luma_motion_merge_mode(ctx,tctx, xC,yC, xP,yP, nCS,nPbW,nPbH, partIdx, out_vi);
	2005	derive_luma_motion_merge_mode(ctx,shdr,img,
	2006	xC,yC, xP,yP, nCS,nPbW,nPbH, partIdx,
	2007	motion.merge_idx, out_vi);
1780	2008
1781	2009	logMV(xP,yP,nPbW,nPbH, "merge_mode", out_vi);
1782	2010	}

1787	2015	for (int l=0;l<2;l++) {
1788	2016	// 1.
1789	2017
1790		enum InterPredIdc inter_pred_idc = (enum InterPredIdc)tctx->inter_pred_idc;
	2018	enum InterPredIdc inter_pred_idc = (enum InterPredIdc)motion.inter_pred_idc;
1791	2019
1792	2020	if (inter_pred_idc == PRED_BI \|\|
1793	2021	(inter_pred_idc == PRED_L0 && l==0) \|\|
1794	2022	(inter_pred_idc == PRED_L1 && l==1)) {
1795		out_vi->lum.refIdx[l] = tctx->refIdx[l];
1796		out_vi->lum.predFlag[l] = 1;
	2023	out_vi->refIdx[l] = motion.refIdx[l];
	2024	out_vi->predFlag[l] = 1;
1797	2025	}
1798	2026	else {
1799		out_vi->lum.refIdx[l] = -1;
1800		out_vi->lum.predFlag[l] = 0;
	2027	out_vi->refIdx[l] = -1;
	2028	out_vi->predFlag[l] = 0;
1801	2029	}
1802	2030
1803	2031	// 2.
1804	2032
1805		mvdL[l][0] = tctx->mvd[l][0];
1806		mvdL[l][1] = tctx->mvd[l][1];
1807
1808
1809		if (out_vi->lum.predFlag[l]) {
	2033	mvdL[l][0] = motion.mvd[l][0];
	2034	mvdL[l][1] = motion.mvd[l][1];
	2035
	2036
	2037	if (out_vi->predFlag[l]) {
1810	2038	// 3.
1811	2039
1812		mvpL[l] = luma_motion_vector_prediction(ctx,tctx,xC,yC,nCS,xP,yP, nPbW,nPbH, l,
1813		out_vi->lum.refIdx[l], partIdx);
	2040	mvpL[l] = luma_motion_vector_prediction(ctx,shdr,img,motion,
	2041	xC,yC,nCS,xP,yP, nPbW,nPbH, l,
	2042	out_vi->refIdx[l], partIdx);
1814	2043
1815	2044	// 4.
1816	2045
1817	2046	int32_t x = (mvpL[l].x + mvdL[l][0] + 0x10000) & 0xFFFF;
1818	2047	int32_t y = (mvpL[l].y + mvdL[l][1] + 0x10000) & 0xFFFF;
1819	2048
1820		out_vi->lum.mv[l].x = (x>=0x8000) ? x-0x10000 : x;
1821		out_vi->lum.mv[l].y = (y>=0x8000) ? y-0x10000 : y;
	2049	out_vi->mv[l].x = (x>=0x8000) ? x-0x10000 : x;
	2050	out_vi->mv[l].y = (y>=0x8000) ? y-0x10000 : y;
1822	2051	}
1823	2052	}
1824	2053

1828	2057
1829	2058
1830	2059	// 8.5.3
1831		void decode_prediction_unit(thread_context* tctx,
	2060	void decode_prediction_unit(base_context* ctx,
	2061	const slice_segment_header* shdr,
	2062	de265_image* img,
	2063	const motion_spec& motion,
1832	2064	int xC,int yC, int xB,int yB, int nCS, int nPbW,int nPbH, int partIdx)
1833	2065	{
1834	2066	logtrace(LogMotion,"decode_prediction_unit POC=%d %d;%d %dx%d\n",
1835		tctx->img->PicOrderCntVal, xC+xB,yC+yB, nPbW,nPbH);
1836
1837		slice_segment_header* shdr = tctx->shdr;
	2067	img->PicOrderCntVal, xC+xB,yC+yB, nPbW,nPbH);
	2068
	2069	//slice_segment_header* shdr = tctx->shdr;
1838	2070
1839	2071	// 1.
1840	2072
1841		VectorInfo vi;
1842		motion_vectors_and_ref_indices(tctx->decctx,tctx, xC,yC, xB,yB, nCS, nPbW,nPbH, partIdx, &vi);
	2073	MotionVectorSpec vi;
	2074	motion_vectors_and_ref_indices(ctx, shdr, img, motion,
	2075	xC,yC, xB,yB, nCS, nPbW,nPbH, partIdx, &vi);
1843	2076
1844	2077	// 2.
1845	2078
1846		generate_inter_prediction_samples(tctx->decctx,tctx->img, shdr, xC,yC, xB,yB, nCS, nPbW,nPbH, &vi);
1847
1848
1849		tctx->img->set_mv_info(xC+xB,yC+yB,nPbW,nPbH, &vi.lum);
	2079	generate_inter_prediction_samples(ctx,shdr, img, xC,yC, xB,yB, nCS, nPbW,nPbH, &vi);
	2080
	2081
	2082	img->set_mv_info(xC+xB,yC+yB,nPbW,nPbH, vi);
1850	2083	}
1851

+55

-13

libde265/motion.h less more

22	22
23	23	#include <stdint.h>
24	24
	25	class base_context;
	26	class slice_segment_header;
25	27
26	28	typedef struct
27	29	{

31	33
32	34	typedef struct
33	35	{
34		int8_t refIdx[2];
35		uint8_t predFlag[2];
36		MotionVector mv[2];
37		} PredVectorInfo;
	36	uint8_t predFlag[2]; // which of the two vectors is actually used
	37	int8_t refIdx[2];
	38	MotionVector mv[2];
	39	} MotionVectorSpec;
38	40
39	41
40		typedef struct
41		{
42		PredVectorInfo lum;
43		MotionVector mvC[2];
44		} VectorInfo;
	42	typedef struct {
	43	int8_t refIdx[2];
	44	int16_t mvd[2][2]; // [L0/L1][x/y] (only in top left position - ???)
	45
	46	uint8_t inter_pred_idc : 2; // enum InterPredIdc
	47	uint8_t mvp_l0_flag : 1;
	48	uint8_t mvp_l1_flag : 1;
	49	uint8_t merge_flag : 1;
	50	uint8_t merge_idx : 3;
	51	} motion_spec;
45	52
46	53
47		void decode_prediction_unit(struct thread_context* shdr,
	54	void get_merge_candidate_list(base_context* ctx,
	55	const slice_segment_header* shdr,
	56	struct de265_image* img,
	57	int xC,int yC, int xP,int yP,
	58	int nCS, int nPbW,int nPbH, int partIdx,
	59	MotionVectorSpec* mergeCandList);
	60
	61	/*
	62	int derive_spatial_merging_candidates(const struct de265_image* img,
	63	int xC, int yC, int nCS, int xP, int yP,
	64	uint8_t singleMCLFlag,
	65	int nPbW, int nPbH,
	66	int partIdx,
	67	MotionVectorSpec* out_cand,
	68	int maxCandidates);
	69	*/
	70
	71	void generate_inter_prediction_samples(base_context* ctx,
	72	const slice_segment_header* shdr,
	73	struct de265_image* img,
	74	int xC,int yC,
	75	int xB,int yB,
	76	int nCS, int nPbW,int nPbH,
	77	const MotionVectorSpec* vi);
	78
	79
	80	/* Fill list (two entries) of motion-vector predictors for MVD coding.
	81	*/
	82	void fill_luma_motion_vector_predictors(base_context* ctx,
	83	const slice_segment_header* shdr,
	84	de265_image* img,
	85	int xC,int yC,int nCS,int xP,int yP,
	86	int nPbW,int nPbH, int l,
	87	int refIdx, int partIdx,
	88	MotionVector out_mvpList[2]);
	89
	90
	91	void decode_prediction_unit(base_context* ctx,const slice_segment_header* shdr,
	92	de265_image* img, const motion_spec& motion,
48	93	int xC,int yC, int xB,int yB, int nCS, int nPbW,int nPbH, int partIdx);
49	94
50		void inter_prediction(struct decoder_context* ctx,struct slice_segment_header* shdr,
51		int xC,int yC, int log2CbSize);
52
53	95	#endif

+53

-17

libde265/nal-parser.cc less more

57	57	skipped_bytes.clear();
58	58	}
59	59
60		void NAL_unit::resize(int new_size)
	60	LIBDE265_CHECK_RESULT bool NAL_unit::resize(int new_size)
61	61	{
62	62	if (capacity < new_size) {
63	63	unsigned char* newbuffer = (unsigned char*)malloc(new_size);
	64	if (newbuffer == NULL) {
	65	return false;
	66	}
64	67
65	68	if (nal_data != NULL) {
66	69	memcpy(newbuffer, nal_data, data_size);

70	73	nal_data = newbuffer;
71	74	capacity = new_size;
72	75	}
73		}
74
75		void NAL_unit::append(const unsigned char* in_data, int n)
76		{
77		resize(data_size + n);
	76	return true;
	77	}
	78
	79	LIBDE265_CHECK_RESULT bool NAL_unit::append(const unsigned char* in_data, int n)
	80	{
	81	if (!resize(data_size + n)) {
	82	return false;
	83	}
78	84	memcpy(nal_data + data_size, in_data, n);
79	85	data_size += n;
80		}
81
82		void NAL_unit::set_data(const unsigned char* in_data, int n)
83		{
84		resize(n);
	86	return true;
	87	}
	88
	89	bool LIBDE265_CHECK_RESULT NAL_unit::set_data(const unsigned char* in_data, int n)
	90	{
	91	if (!resize(n)) {
	92	return false;
	93	}
85	94	memcpy(nal_data, in_data, n);
86	95	data_size = n;
	96	return true;
87	97	}
88	98
89	99	void NAL_unit::insert_skipped_byte(int pos)

176	186	}
177	187
178	188
179		NAL_unit* NAL_Parser::alloc_NAL_unit(int size)
	189	LIBDE265_CHECK_RESULT NAL_unit* NAL_Parser::alloc_NAL_unit(int size)
180	190	{
181	191	NAL_unit* nal;
182	192

191	201	}
192	202
193	203	nal->clear();
194		nal->resize(size);
	204	if (!nal->resize(size)) {
	205	free_NAL_unit(nal);
	206	return NULL;
	207	}
195	208
196	209	return nal;
197	210	}
198	211
199	212	void NAL_Parser::free_NAL_unit(NAL_unit* nal)
200	213	{
	214	if (nal == NULL) {
	215	// Allow calling with NULL just like regular "free()"
	216	return;
	217	}
201	218	if (NAL_free_list.size() < DE265_NAL_FREE_LIST_SIZE) {
202	219	NAL_free_list.push_back(nal);
203	220	}

234	251
235	252	if (pending_input_NAL == NULL) {
236	253	pending_input_NAL = alloc_NAL_unit(len+3);
	254	if (pending_input_NAL == NULL) {
	255	return DE265_ERROR_OUT_OF_MEMORY;
	256	}
237	257	pending_input_NAL->pts = pts;
238	258	pending_input_NAL->user_data = user_data;
239	259	}

242	262
243	263	// Resize output buffer so that complete input would fit.
244	264	// We add 3, because in the worst case 3 extra bytes are created for an input byte.
245		nal->resize(nal->size() + len + 3);
	265	if (!nal->resize(nal->size() + len + 3)) {
	266	return DE265_ERROR_OUT_OF_MEMORY;
	267	}
246	268
247	269	unsigned char* out = nal->data() + nal->size();
248	270

315	337	// initialize new, empty NAL unit
316	338
317	339	pending_input_NAL = alloc_NAL_unit(len+3);
	340	if (pending_input_NAL == NULL) {
	341	return DE265_ERROR_OUT_OF_MEMORY;
	342	}
318	343	pending_input_NAL->pts = pts;
319	344	pending_input_NAL->user_data = user_data;
320	345	nal = pending_input_NAL;

351	376	end_of_frame = false;
352	377
353	378	NAL_unit* nal = alloc_NAL_unit(len);
354		nal->set_data(data, len);
	379	if (nal == NULL \|\| !nal->set_data(data, len)) {
	380	free_NAL_unit(nal);
	381	return DE265_ERROR_OUT_OF_MEMORY;
	382	}
355	383	nal->pts = pts;
356	384	nal->user_data = user_data;
357	385

371	399
372	400	// append bytes that are implied by the push state
373	401
374		if (input_push_state==6) { nal->append(null,1); }
375		if (input_push_state==7) { nal->append(null,2); }
	402	if (input_push_state==6) {
	403	if (!nal->append(null,1)) {
	404	return DE265_ERROR_OUT_OF_MEMORY;
	405	}
	406	}
	407	if (input_push_state==7) {
	408	if (!nal->append(null,2)) {
	409	return DE265_ERROR_OUT_OF_MEMORY;
	410	}
	411	}
376	412
377	413
378	414	// only push the NAL if it contains at least the NAL header

+13

-7

libde265/nal-parser.h less more

23	23	#include "libde265/sps.h"
24	24	#include "libde265/pps.h"
25	25	#include "libde265/nal.h"
	26	#include "libde265/util.h"
26	27
27	28	#include <vector>
28	29	#include <queue>

46	47
47	48	// --- rbsp data ---
48	49
49		void resize(int new_size);
50		void append(const unsigned char* data, int n);
51		void set_data(const unsigned char* data, int n);
	50	LIBDE265_CHECK_RESULT bool resize(int new_size);
	51	LIBDE265_CHECK_RESULT bool append(const unsigned char* data, int n);
	52	LIBDE265_CHECK_RESULT bool set_data(const unsigned char* data, int n);
52	53
53	54	int size() const { return data_size; }
54	55	void set_size(int s) { data_size=s; }

89	90	~NAL_Parser();
90	91
91	92	de265_error push_data(const unsigned char* data, int len,
92		de265_PTS pts, void* user_data);
	93	de265_PTS pts, void* user_data = NULL);
93	94
94	95	de265_error push_NAL(const unsigned char* data, int len,
95		de265_PTS pts, void* user_data);
	96	de265_PTS pts, void* user_data = NULL);
96	97
97	98	NAL_unit* pop_from_NAL_queue();
98		void push_to_NAL_queue(NAL_unit*);
99	99	de265_error flush_data();
100	100	void mark_end_of_stream() { end_of_stream=true; }
101	101	void mark_end_of_frame() { end_of_frame=true; }

111	111	int size = NAL_queue.size();
112	112	if (pending_input_NAL) { size++; }
113	113	return size;
	114	}
	115
	116	int number_of_complete_NAL_units_pending() const {
	117	return NAL_queue.size();
114	118	}
115	119
116	120	void free_NAL_unit(NAL_unit*);

135	139	std::queue<NAL_unit*> NAL_queue; // enqueued NALs have suffing bytes removed
136	140	int nBytes_in_NAL_queue; // data bytes currently in NAL_queue
137	141
	142	void push_to_NAL_queue(NAL_unit*);
	143
138	144
139	145	// pool of unused NAL memory
140	146
141	147	std::vector<NAL_unit*> NAL_free_list; // maximum size: DE265_NAL_FREE_LIST_SIZE
142	148
143		NAL_unit* alloc_NAL_unit(int size);
	149	LIBDE265_CHECK_RESULT NAL_unit* alloc_NAL_unit(int size);
144	150	};
145	151
146	152

+32

-6

libde265/nal.cc less more

18	18	*/
19	19
20	20	#include "nal.h"
	21	#include "cabac.h"
21	22	#include <assert.h>
22	23
23	24
24		void nal_read_header(bitreader* reader, nal_header* hdr)
	25	void nal_header::read(bitreader* reader)
25	26	{
26	27	skip_bits(reader,1);
27		hdr->nal_unit_type = get_bits(reader,6);
28		hdr->nuh_layer_id = get_bits(reader,6);
29		hdr->nuh_temporal_id = get_bits(reader,3) -1;
	28	nal_unit_type = get_bits(reader,6);
	29	nuh_layer_id = get_bits(reader,6);
	30	nuh_temporal_id = get_bits(reader,3) -1;
	31	}
	32
	33
	34	void nal_header::write(CABAC_encoder& out) const
	35	{
	36	out.skip_bits(1);
	37	out.write_bits(nal_unit_type,6);
	38	out.write_bits(nuh_layer_id ,6);
	39	out.write_bits(nuh_temporal_id+1,3);
30	40	}
31	41
32	42

79	89	(unit_type <= NAL_UNIT_RESERVED_IRAP_VCL23)) );
80	90	}
81	91
	92	bool isSublayerNonReference(uint8_t unit_type)
	93	{
	94	switch (unit_type) {
	95	case NAL_UNIT_TRAIL_N:
	96	case NAL_UNIT_TSA_N:
	97	case NAL_UNIT_STSA_N:
	98	case NAL_UNIT_RADL_N:
	99	case NAL_UNIT_RASL_N:
	100	case NAL_UNIT_RESERVED_VCL_N10:
	101	case NAL_UNIT_RESERVED_VCL_N12:
	102	case NAL_UNIT_RESERVED_VCL_N14:
	103	return true;
	104
	105	default:
	106	return false;
	107	}
	108	}
82	109
83	110	static const char* NAL_unit_name[] = {
84	111	"TRAIL_N", // 0

133	160
134	161	const char* get_NAL_name(uint8_t unit_type)
135	162	{
136		assert(unit_type <= 47);
	163	if (unit_type >= 48) { return "INVALID NAL >= 48"; }
137	164	return NAL_unit_name[unit_type];
138	165	}
139

+23

-5

libde265/nal.h less more

30	30	#endif
31	31
32	32	#include "libde265/bitstream.h"
	33	#include "libde265/cabac.h"
33	34
34	35	struct nal_header {
35	36	nal_header() {

38	39	nuh_temporal_id = 0;
39	40	}
40	41
41		int nal_unit_type;
42		int nuh_layer_id;
43		int nuh_temporal_id;
	42	void read(bitreader* reader);
	43	void write(CABAC_encoder& writer) const;
	44
	45	void set(int unit_type, int layer_id=0, int temporal_id=0) {
	46	nal_unit_type =unit_type;
	47	nuh_layer_id =layer_id;
	48	nuh_temporal_id=temporal_id;
	49	}
	50
	51	uint8_t nal_unit_type;
	52	uint8_t nuh_layer_id;
	53	uint8_t nuh_temporal_id;
44	54	};
45	55
46	56	#define NAL_UNIT_TRAIL_N 0

94	104
95	105	#define NAL_UNIT_UNDEFINED 255
96	106
97		void nal_read_header(bitreader* reader, nal_header*);
98
99	107	bool isIDR(uint8_t unit_type);
100	108	bool isBLA(uint8_t unit_type);
101	109	bool isCRA(uint8_t unit_type);

104	112	bool isIRAP(uint8_t unit_type);
105	113	bool isRADL(uint8_t unit_type);
106	114	bool isReferenceNALU(uint8_t unit_type);
	115	bool isSublayerNonReference(uint8_t unit_type);
107	116
108	117	const char* get_NAL_name(uint8_t unit_type);
109	118
	119	inline bool isIdrPic(uint8_t nal_unit_type) {
	120	return (nal_unit_type == NAL_UNIT_IDR_W_RADL \|\|
	121	nal_unit_type == NAL_UNIT_IDR_N_LP);
	122	}
	123
	124	inline bool isRapPic(uint8_t nal_unit_type) {
	125	return nal_unit_type >= 16 && nal_unit_type <= 23;
	126	}
	127
110	128	#endif

+598

-188

libde265/pps.cc less more

31	31	#endif
32	32
33	33
	34	void pps_range_extension::reset()
	35	{
	36	log2_max_transform_skip_block_size = 2;
	37	cross_component_prediction_enabled_flag = false;
	38	chroma_qp_offset_list_enabled_flag = false;
	39	diff_cu_chroma_qp_offset_depth = 0;
	40	chroma_qp_offset_list_len = 0;
	41	log2_sao_offset_scale_luma = 0;
	42	log2_sao_offset_scale_chroma = 0;
	43	}
	44
	45
	46	bool pps_range_extension::read(bitreader* br, decoder_context* ctx, const pic_parameter_set* pps)
	47	{
	48	const seq_parameter_set* sps = ctx->get_sps(pps->seq_parameter_set_id);
	49
	50	int uvlc;
	51
	52	if (pps->transform_skip_enabled_flag) {
	53	uvlc = get_uvlc(br);
	54	if (uvlc == UVLC_ERROR \|\|
	55	uvlc+2 > sps->Log2MaxTrafoSize) {
	56
	57	// Note: this is out of spec, but the conformance stream
	58	// PERSIST_RPARAM_A_RExt_Sony_2 codes a too large value.
	59
	60	//ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	61	//return false;
	62	}
	63
	64	log2_max_transform_skip_block_size = uvlc+2;
	65	}
	66
	67	cross_component_prediction_enabled_flag = get_bits(br,1);
	68	if (sps->ChromaArrayType != CHROMA_444 &&
	69	cross_component_prediction_enabled_flag) {
	70	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	71	}
	72
	73	chroma_qp_offset_list_enabled_flag = get_bits(br,1);
	74	if (sps->ChromaArrayType == CHROMA_MONO &&
	75	chroma_qp_offset_list_enabled_flag) {
	76	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	77	}
	78
	79	if (chroma_qp_offset_list_enabled_flag) {
	80	uvlc = get_uvlc(br);
	81	if (uvlc == UVLC_ERROR \|\|
	82	uvlc > sps->log2_diff_max_min_luma_coding_block_size) {
	83	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	84	return false;
	85	}
	86
	87	diff_cu_chroma_qp_offset_depth = uvlc;
	88
	89
	90	uvlc = get_uvlc(br);
	91	if (uvlc == UVLC_ERROR \|\|
	92	uvlc > 5) {
	93	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	94	return false;
	95	}
	96
	97	chroma_qp_offset_list_len = uvlc+1;
	98
	99	for (int i=0;i<chroma_qp_offset_list_len;i++) {
	100	int svlc;
	101	svlc = get_svlc(br);
	102	if (svlc == UVLC_ERROR \|\|
	103	svlc < -12 \|\| svlc > 12) {
	104	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	105	return false;
	106	}
	107
	108	cb_qp_offset_list[i] = svlc;
	109
	110	svlc = get_svlc(br);
	111	if (svlc == UVLC_ERROR \|\|
	112	svlc < -12 \|\| svlc > 12) {
	113	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	114	return false;
	115	}
	116
	117	cr_qp_offset_list[i] = svlc;
	118	}
	119	}
	120
	121
	122	uvlc = get_uvlc(br);
	123	if (uvlc == UVLC_ERROR \|\|
	124	uvlc > libde265_max(0, sps->BitDepth_Y-10)) {
	125	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	126	return false;
	127	}
	128
	129	log2_sao_offset_scale_luma = uvlc;
	130
	131	uvlc = get_uvlc(br);
	132	if (uvlc == UVLC_ERROR \|\|
	133	uvlc > libde265_max(0, sps->BitDepth_C-10)) {
	134	ctx->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	135	return false;
	136	}
	137
	138	log2_sao_offset_scale_chroma = uvlc;
	139
	140	return true;
	141	}
	142
	143
	144	void pps_range_extension::dump(int fd) const
	145	{
	146	FILE* fh;
	147	if (fd==1) fh=stdout;
	148	else if (fd==2) fh=stderr;
	149	else { return; }
	150
	151	#define LOG0(t) log2fh(fh, t)
	152	#define LOG1(t,d) log2fh(fh, t,d)
	153	#define LOG2(t,d,e) log2fh(fh, t,d,e)
	154
	155	LOG0("---------- PPS range-extension ----------\n");
	156	LOG1("log2_max_transform_skip_block_size : %d\n", log2_max_transform_skip_block_size);
	157	LOG1("cross_component_prediction_enabled_flag : %d\n", cross_component_prediction_enabled_flag);
	158	LOG1("chroma_qp_offset_list_enabled_flag : %d\n", chroma_qp_offset_list_enabled_flag);
	159	if (chroma_qp_offset_list_enabled_flag) {
	160	LOG1("diff_cu_chroma_qp_offset_depth : %d\n", diff_cu_chroma_qp_offset_depth);
	161	LOG1("chroma_qp_offset_list_len : %d\n", chroma_qp_offset_list_len);
	162	for (int i=0;i<chroma_qp_offset_list_len;i++) {
	163	LOG2("cb_qp_offset_list[%d] : %d\n", i,cb_qp_offset_list[i]);
	164	LOG2("cr_qp_offset_list[%d] : %d\n", i,cr_qp_offset_list[i]);
	165	}
	166	}
	167
	168	LOG1("log2_sao_offset_scale_luma : %d\n", log2_sao_offset_scale_luma);
	169	LOG1("log2_sao_offset_scale_chroma : %d\n", log2_sao_offset_scale_chroma);
	170	#undef LOG2
	171	#undef LOG1
	172	#undef LOG0
	173	}
	174
	175
	176
	177
	178
34	179	pic_parameter_set::pic_parameter_set()
35	180	{
	181	reset();
	182	}
	183
	184
	185	pic_parameter_set::~pic_parameter_set()
	186	{
	187	}
	188
	189
	190	void pic_parameter_set::set_defaults(enum PresetSet)
	191	{
36	192	pps_read = false;
37		}
38
39
40		pic_parameter_set::~pic_parameter_set()
41		{
	193
	194	pic_parameter_set_id = 0;
	195	seq_parameter_set_id = 0;
	196	dependent_slice_segments_enabled_flag = 0;
	197	sign_data_hiding_flag = 0;
	198	cabac_init_present_flag = 0;
	199	num_ref_idx_l0_default_active = 1;
	200	num_ref_idx_l1_default_active = 1;
	201
	202	pic_init_qp = 27;
	203	constrained_intra_pred_flag = 0;
	204	transform_skip_enabled_flag = 0;
	205
	206	cu_qp_delta_enabled_flag = 0;
	207	diff_cu_qp_delta_depth = 0;
	208
	209	pic_cb_qp_offset = 0;
	210	pic_cr_qp_offset = 0;
	211	pps_slice_chroma_qp_offsets_present_flag = 0;
	212	weighted_pred_flag = 0;
	213	weighted_bipred_flag= 0;
	214	output_flag_present_flag = 0;
	215	transquant_bypass_enable_flag = 0;
	216	entropy_coding_sync_enabled_flag = 0;
	217
	218	// --- tiles ---
	219
	220	tiles_enabled_flag = 0;
	221	num_tile_columns = 1;
	222	num_tile_rows = 1;
	223	uniform_spacing_flag = 1;
	224
	225
	226	// --- ---
	227
	228	loop_filter_across_tiles_enabled_flag = 1;
	229	pps_loop_filter_across_slices_enabled_flag = 1;
	230
	231	for (int i=0;i<DE265_MAX_TILE_COLUMNS;i++) { colWidth[i]=0; }
	232	for (int i=0;i<DE265_MAX_TILE_ROWS;i++) { rowHeight[i]=0; }
	233	for (int i=0;i<=DE265_MAX_TILE_COLUMNS;i++) { colBd[i]=0; }
	234	for (int i=0;i<=DE265_MAX_TILE_ROWS;i++) { rowBd[i]=0; }
	235
	236	CtbAddrRStoTS.clear();
	237	CtbAddrTStoRS.clear();
	238	TileId.clear();
	239	TileIdRS.clear();
	240	MinTbAddrZS.clear();
	241
	242
	243	Log2MinCuQpDeltaSize = 0;
	244
	245	deblocking_filter_control_present_flag = 0;
	246	deblocking_filter_override_enabled_flag = 0;
	247	pic_disable_deblocking_filter_flag = 0;
	248
	249	beta_offset = 0;
	250	tc_offset = 0;
	251
	252	pic_scaling_list_data_present_flag = 0;
	253	// TODO struct scaling_list_data scaling_list;
	254
	255	lists_modification_present_flag = 0;
	256	log2_parallel_merge_level = 2;
	257
	258	num_extra_slice_header_bits = 0;
	259	slice_segment_header_extension_present_flag = 0;
	260	pps_extension_flag = 0;
	261
	262	pps_range_extension_flag = 0;
	263	pps_multilayer_extension_flag = 0;
	264	pps_extension_6bits = 0;
42	265	}
43	266
44	267
45	268	bool pic_parameter_set::read(bitreader* br, decoder_context* ctx)
46	269	{
47		pps_read = false; // incomplete pps
	270	reset();
	271
48	272
49	273	int uvlc;
50	274	pic_parameter_set_id = uvlc = get_uvlc(br);

160	384
161	385	lastColumnWidth -= colWidth[i];
162	386	}
	387
	388	if (lastColumnWidth <= 0) {
	389	return false;
	390	}
163	391
164	392	colWidth[num_tile_columns-1] = lastColumnWidth;
165	393

174	402	lastRowHeight -= rowHeight[i];
175	403	}
176	404
	405	if (lastRowHeight <= 0) {
	406	return false;
	407	}
	408
	409
177	410	rowHeight[num_tile_rows-1] = lastRowHeight;
178	411	}
179	412

188	421
189	422
190	423
191		if (uniform_spacing_flag) {
192
193		// set columns widths
194
195		int const colPos = (int )alloca((num_tile_columns+1) * sizeof(int));
196
197		for (int i=0;i<=num_tile_columns;i++) {
198		colPos[i] = i*sps->PicWidthInCtbsY / num_tile_columns;
199		}
200		for (int i=0;i<num_tile_columns;i++) {
201		colWidth[i] = colPos[i+1] - colPos[i];
202		}
203
204		// set row heights
205
206		int const rowPos = (int )alloca((num_tile_rows+1) * sizeof(int));
207
208		for (int i=0;i<=num_tile_rows;i++) {
209		rowPos[i] = i*sps->PicHeightInCtbsY / num_tile_rows;
210		}
211		for (int i=0;i<num_tile_rows;i++) {
212		rowHeight[i] = rowPos[i+1] - rowPos[i];
213		}
214		}
215
216
217		// set tile boundaries
218
219		colBd[0]=0;
220		for (int i=0;i<num_tile_columns;i++) {
221		colBd[i+1] = colBd[i] + colWidth[i];
222		}
223
224		rowBd[0]=0;
225		for (int i=0;i<num_tile_rows;i++) {
226		rowBd[i+1] = rowBd[i] + rowHeight[i];
227		}
228
229
230
231		// alloc raster scan arrays
232
233		CtbAddrRStoTS.resize(sps->PicSizeInCtbsY);
234		CtbAddrTStoRS.resize(sps->PicSizeInCtbsY);
235		TileId .resize(sps->PicSizeInCtbsY);
236		TileIdRS .resize(sps->PicSizeInCtbsY);
237		MinTbAddrZS .resize(sps->PicSizeInTbsY );
238
239
240		// raster scan (RS) <-> tile scan (TS) conversion
241
242		for (int ctbAddrRS=0 ; ctbAddrRS < sps->PicSizeInCtbsY ; ctbAddrRS++)
243		{
244		int tbX = ctbAddrRS % sps->PicWidthInCtbsY;
245		int tbY = ctbAddrRS / sps->PicWidthInCtbsY;
246		int tileX=-1,tileY=-1;
247
248		for (int i=0;i<num_tile_columns;i++)
249		if (tbX >= colBd[i])
250		tileX=i;
251
252		for (int j=0;j<num_tile_rows;j++)
253		if (tbY >= rowBd[j])
254		tileY=j;
255
256		CtbAddrRStoTS[ctbAddrRS] = 0;
257		for (int i=0;i<tileX;i++)
258		CtbAddrRStoTS[ctbAddrRS] += rowHeight[tileY]*colWidth[i];
259
260		for (int j=0;j<tileY;j++)
261		{
262		//pps->CtbAddrRStoTS[ctbAddrRS] += (tbY - pps->rowBd[tileY])*pps->colWidth[tileX];
263		//pps->CtbAddrRStoTS[ctbAddrRS] += tbX - pps->colBd[tileX];
264
265		CtbAddrRStoTS[ctbAddrRS] += sps->PicWidthInCtbsY * rowHeight[j];
266		}
267
268		assert(tileX>=0 && tileY>=0);
269
270		CtbAddrRStoTS[ctbAddrRS] += (tbY-rowBd[tileY])*colWidth[tileX];
271		CtbAddrRStoTS[ctbAddrRS] += tbX - colBd[tileX];
272
273
274		// inverse mapping
275
276		CtbAddrTStoRS[ CtbAddrRStoTS[ctbAddrRS] ] = ctbAddrRS;
277		}
278
279
280		logtrace(LogHeaders,"6.5.1 CtbAddrRSToTS\n");
281		for (int y=0;y<sps->PicHeightInCtbsY;y++)
282		{
283		for (int x=0;x<sps->PicWidthInCtbsY;x++)
284		{
285		logtrace(LogHeaders,"%3d ", CtbAddrRStoTS[x + y*sps->PicWidthInCtbsY]);
286		}
287
288		logtrace(LogHeaders,"\n");
289		}
290
291
292		// tile id
293
294		for (int j=0, tIdx=0 ; j<num_tile_rows ; j++)
295		for (int i=0 ; i<num_tile_columns;i++)
296		{
297		for (int y=rowBd[j] ; y<rowBd[j+1] ; y++)
298		for (int x=colBd[i] ; x<colBd[i+1] ; x++) {
299		TileId [ CtbAddrRStoTS[y*sps->PicWidthInCtbsY + x] ] = tIdx;
300		TileIdRS[ y*sps->PicWidthInCtbsY + x ] = tIdx;
301
302		//logtrace(LogHeaders,"tileID[%d,%d] = %d\n",x,y,pps->TileIdRS[ y*sps->PicWidthInCtbsY + x ]);
303		}
304
305		tIdx++;
306		}
307
308		logtrace(LogHeaders,"Tile IDs RS:\n");
309		for (int y=0;y<sps->PicHeightInCtbsY;y++) {
310		for (int x=0;x<sps->PicWidthInCtbsY;x++) {
311		logtrace(LogHeaders,"%2d ",TileIdRS[y*sps->PicWidthInCtbsY+x]);
312		}
313		logtrace(LogHeaders,"\n");
314		}
315
316		// 6.5.2 Z-scan order array initialization process
317
318		for (int y=0;y<sps->PicHeightInTbsY;y++)
319		for (int x=0;x<sps->PicWidthInTbsY;x++)
320		{
321		int tbX = (x<<sps->Log2MinTrafoSize)>>sps->Log2CtbSizeY;
322		int tbY = (y<<sps->Log2MinTrafoSize)>>sps->Log2CtbSizeY;
323		int ctbAddrRS = sps->PicWidthInCtbsY*tbY + tbX;
324
325		MinTbAddrZS[x + y*sps->PicWidthInTbsY] = CtbAddrRStoTS[ctbAddrRS]
326		<< ((sps->Log2CtbSizeY-sps->Log2MinTrafoSize)*2);
327
328		int p=0;
329		for (int i=0 ; i<(sps->Log2CtbSizeY - sps->Log2MinTrafoSize) ; i++) {
330		int m=1<<i;
331		p += (m & x ? mm : 0) + (m & y ? 2m*m : 0);
332		}
333
334		MinTbAddrZS[x + y*sps->PicWidthInTbsY] += p;
335		}
336
337
338		// --- debug logging ---
339
340		/*
341		logtrace(LogHeaders,"6.5.2 Z-scan order array\n");
342		for (int y=0;y<sps->PicHeightInTbsY;y++)
343		{
344		for (int x=0;x<sps->PicWidthInTbsY;x++)
345		{
346		logtrace(LogHeaders,"%4d ", pps->MinTbAddrZS[x + y*sps->PicWidthInTbsY]);
347		}
348
349		logtrace(LogHeaders,"\n");
350		}
351
352		for (int i=0;i<sps->PicSizeInTbsY;i++)
353		{
354		for (int y=0;y<sps->PicHeightInTbsY;y++)
355		{
356		for (int x=0;x<sps->PicWidthInTbsY;x++)
357		{
358		if (pps->MinTbAddrZS[x + y*sps->PicWidthInTbsY] == i) {
359		logtrace(LogHeaders,"%d %d\n",x,y);
360		}
361		}
362		}
363		}
364		*/
365
366	424	// END tiles
367	425
368
369		Log2MinCuQpDeltaSize = sps->Log2CtbSizeY - diff_cu_qp_delta_depth;
370	426
371	427
372	428	beta_offset = 0; // default value

433	489	}
434	490	log2_parallel_merge_level += 2;
435	491
	492	if (log2_parallel_merge_level-2 > sps->log2_min_luma_coding_block_size-3 +1 +
	493	sps->log2_diff_max_min_luma_coding_block_size) {
	494	return false;
	495	}
	496
436	497	slice_segment_header_extension_present_flag = get_bits(br,1);
437	498	pps_extension_flag = get_bits(br,1);
	499
	500	if (pps_extension_flag) {
	501	pps_range_extension_flag = get_bits(br,1);
	502	pps_multilayer_extension_flag = get_bits(br,1);
	503	pps_extension_6bits = get_bits(br,6);
	504
	505	if (pps_range_extension_flag) {
	506	bool success = range_extension.read(br, ctx, this);
	507	if (!success) {
	508	return false;
	509	}
	510	}
	511
	512	//assert(false);
	513	/*
	514	while( more_rbsp_data() )
	515
	516	pps_extension_data_flag
	517	u(1)
	518	rbsp_trailing_bits()
	519
	520	}
	521	*/
	522	}
	523
	524
	525	set_derived_values(sps);
	526
	527	pps_read = true;
	528
	529	return true;
	530	}
	531
	532
	533	void pic_parameter_set::set_derived_values(const seq_parameter_set* sps)
	534	{
	535	Log2MinCuQpDeltaSize = sps->Log2CtbSizeY - diff_cu_qp_delta_depth;
	536
	537	Log2MinCuChromaQpOffsetSize = sps->Log2CtbSizeY - range_extension.diff_cu_chroma_qp_offset_depth;
	538	Log2MaxTransformSkipSize = range_extension.log2_max_transform_skip_block_size;
	539
	540	if (uniform_spacing_flag) {
	541
	542	// set columns widths
	543
	544	int const colPos = (int )alloca((num_tile_columns+1) * sizeof(int));
	545
	546	for (int i=0;i<=num_tile_columns;i++) {
	547	colPos[i] = i*sps->PicWidthInCtbsY / num_tile_columns;
	548	}
	549	for (int i=0;i<num_tile_columns;i++) {
	550	colWidth[i] = colPos[i+1] - colPos[i];
	551	}
	552
	553	// set row heights
	554
	555	int const rowPos = (int )alloca((num_tile_rows+1) * sizeof(int));
	556
	557	for (int i=0;i<=num_tile_rows;i++) {
	558	rowPos[i] = i*sps->PicHeightInCtbsY / num_tile_rows;
	559	}
	560	for (int i=0;i<num_tile_rows;i++) {
	561	rowHeight[i] = rowPos[i+1] - rowPos[i];
	562	}
	563	}
	564
	565
	566	// set tile boundaries
	567
	568	colBd[0]=0;
	569	for (int i=0;i<num_tile_columns;i++) {
	570	colBd[i+1] = colBd[i] + colWidth[i];
	571	}
	572
	573	rowBd[0]=0;
	574	for (int i=0;i<num_tile_rows;i++) {
	575	rowBd[i+1] = rowBd[i] + rowHeight[i];
	576	}
	577
	578
	579
	580	// alloc raster scan arrays
	581
	582	CtbAddrRStoTS.resize(sps->PicSizeInCtbsY);
	583	CtbAddrTStoRS.resize(sps->PicSizeInCtbsY);
	584	TileId .resize(sps->PicSizeInCtbsY);
	585	TileIdRS .resize(sps->PicSizeInCtbsY);
	586	MinTbAddrZS .resize(sps->PicSizeInTbsY );
	587
	588
	589	// raster scan (RS) <-> tile scan (TS) conversion
	590
	591	for (int ctbAddrRS=0 ; ctbAddrRS < sps->PicSizeInCtbsY ; ctbAddrRS++)
	592	{
	593	int tbX = ctbAddrRS % sps->PicWidthInCtbsY;
	594	int tbY = ctbAddrRS / sps->PicWidthInCtbsY;
	595	int tileX=-1,tileY=-1;
	596
	597	for (int i=0;i<num_tile_columns;i++)
	598	if (tbX >= colBd[i])
	599	tileX=i;
	600
	601	for (int j=0;j<num_tile_rows;j++)
	602	if (tbY >= rowBd[j])
	603	tileY=j;
	604
	605	CtbAddrRStoTS[ctbAddrRS] = 0;
	606	for (int i=0;i<tileX;i++)
	607	CtbAddrRStoTS[ctbAddrRS] += rowHeight[tileY]*colWidth[i];
	608
	609	for (int j=0;j<tileY;j++)
	610	{
	611	//pps->CtbAddrRStoTS[ctbAddrRS] += (tbY - pps->rowBd[tileY])*pps->colWidth[tileX];
	612	//pps->CtbAddrRStoTS[ctbAddrRS] += tbX - pps->colBd[tileX];
	613
	614	CtbAddrRStoTS[ctbAddrRS] += sps->PicWidthInCtbsY * rowHeight[j];
	615	}
	616
	617	assert(tileX>=0 && tileY>=0);
	618
	619	CtbAddrRStoTS[ctbAddrRS] += (tbY-rowBd[tileY])*colWidth[tileX];
	620	CtbAddrRStoTS[ctbAddrRS] += tbX - colBd[tileX];
	621
	622
	623	// inverse mapping
	624
	625	CtbAddrTStoRS[ CtbAddrRStoTS[ctbAddrRS] ] = ctbAddrRS;
	626	}
	627
	628
	629	#if 0
	630	logtrace(LogHeaders,"6.5.1 CtbAddrRSToTS\n");
	631	for (int y=0;y<sps->PicHeightInCtbsY;y++)
	632	{
	633	for (int x=0;x<sps->PicWidthInCtbsY;x++)
	634	{
	635	logtrace(LogHeaders,"%3d ", CtbAddrRStoTS[x + y*sps->PicWidthInCtbsY]);
	636	}
	637
	638	logtrace(LogHeaders,"\n");
	639	}
	640	#endif
	641
	642	// tile id
	643
	644	for (int j=0, tIdx=0 ; j<num_tile_rows ; j++)
	645	for (int i=0 ; i<num_tile_columns;i++)
	646	{
	647	for (int y=rowBd[j] ; y<rowBd[j+1] ; y++)
	648	for (int x=colBd[i] ; x<colBd[i+1] ; x++) {
	649	TileId [ CtbAddrRStoTS[y*sps->PicWidthInCtbsY + x] ] = tIdx;
	650	TileIdRS[ y*sps->PicWidthInCtbsY + x ] = tIdx;
	651
	652	//logtrace(LogHeaders,"tileID[%d,%d] = %d\n",x,y,pps->TileIdRS[ y*sps->PicWidthInCtbsY + x ]);
	653	}
	654
	655	tIdx++;
	656	}
	657
	658	#if 0
	659	logtrace(LogHeaders,"Tile IDs RS:\n");
	660	for (int y=0;y<sps->PicHeightInCtbsY;y++) {
	661	for (int x=0;x<sps->PicWidthInCtbsY;x++) {
	662	logtrace(LogHeaders,"%2d ",TileIdRS[y*sps->PicWidthInCtbsY+x]);
	663	}
	664	logtrace(LogHeaders,"\n");
	665	}
	666	#endif
	667
	668	// 6.5.2 Z-scan order array initialization process
	669
	670	for (int y=0;y<sps->PicHeightInTbsY;y++)
	671	for (int x=0;x<sps->PicWidthInTbsY;x++)
	672	{
	673	int tbX = (x<<sps->Log2MinTrafoSize)>>sps->Log2CtbSizeY;
	674	int tbY = (y<<sps->Log2MinTrafoSize)>>sps->Log2CtbSizeY;
	675	int ctbAddrRS = sps->PicWidthInCtbsY*tbY + tbX;
	676
	677	MinTbAddrZS[x + y*sps->PicWidthInTbsY] = CtbAddrRStoTS[ctbAddrRS]
	678	<< ((sps->Log2CtbSizeY-sps->Log2MinTrafoSize)*2);
	679
	680	int p=0;
	681	for (int i=0 ; i<(sps->Log2CtbSizeY - sps->Log2MinTrafoSize) ; i++) {
	682	int m=1<<i;
	683	p += (m & x ? mm : 0) + (m & y ? 2m*m : 0);
	684	}
	685
	686	MinTbAddrZS[x + y*sps->PicWidthInTbsY] += p;
	687	}
	688
	689
	690	// --- debug logging ---
	691
	692	/*
	693	logtrace(LogHeaders,"6.5.2 Z-scan order array\n");
	694	for (int y=0;y<sps->PicHeightInTbsY;y++)
	695	{
	696	for (int x=0;x<sps->PicWidthInTbsY;x++)
	697	{
	698	logtrace(LogHeaders,"%4d ", pps->MinTbAddrZS[x + y*sps->PicWidthInTbsY]);
	699	}
	700
	701	logtrace(LogHeaders,"\n");
	702	}
	703
	704	for (int i=0;i<sps->PicSizeInTbsY;i++)
	705	{
	706	for (int y=0;y<sps->PicHeightInTbsY;y++)
	707	{
	708	for (int x=0;x<sps->PicWidthInTbsY;x++)
	709	{
	710	if (pps->MinTbAddrZS[x + y*sps->PicWidthInTbsY] == i) {
	711	logtrace(LogHeaders,"%d %d\n",x,y);
	712	}
	713	}
	714	}
	715	}
	716	*/
	717	}
	718
	719
	720	bool pic_parameter_set::write(error_queue* errqueue, CABAC_encoder& out,
	721	const seq_parameter_set* sps)
	722	{
	723	if (pic_parameter_set_id >= DE265_MAX_PPS_SETS) {
	724	errqueue->add_warning(DE265_WARNING_NONEXISTING_PPS_REFERENCED, false);
	725	return false;
	726	}
	727	out.write_uvlc(pic_parameter_set_id);
	728
	729	if (seq_parameter_set_id >= DE265_MAX_PPS_SETS) {
	730	errqueue->add_warning(DE265_WARNING_NONEXISTING_SPS_REFERENCED, false);
	731	return false;
	732	}
	733	out.write_uvlc(seq_parameter_set_id);
	734
	735	out.write_bit(dependent_slice_segments_enabled_flag);
	736	out.write_bit(output_flag_present_flag);
	737	out.write_bits(num_extra_slice_header_bits,3);
	738	out.write_bit(sign_data_hiding_flag);
	739	out.write_bit(cabac_init_present_flag);
	740	out.write_uvlc(num_ref_idx_l0_default_active-1);
	741	out.write_uvlc(num_ref_idx_l1_default_active-1);
	742
	743	out.write_svlc(pic_init_qp-26);
	744
	745	out.write_bit(constrained_intra_pred_flag);
	746	out.write_bit(transform_skip_enabled_flag);
	747	out.write_bit(cu_qp_delta_enabled_flag);
	748
	749	if (cu_qp_delta_enabled_flag) {
	750	out.write_uvlc(diff_cu_qp_delta_depth);
	751	}
	752
	753	out.write_svlc(pic_cb_qp_offset);
	754	out.write_svlc(pic_cr_qp_offset);
	755
	756	out.write_bit(pps_slice_chroma_qp_offsets_present_flag);
	757	out.write_bit(weighted_pred_flag);
	758	out.write_bit(weighted_bipred_flag);
	759	out.write_bit(transquant_bypass_enable_flag);
	760	out.write_bit(tiles_enabled_flag);
	761	out.write_bit(entropy_coding_sync_enabled_flag);
	762
	763
	764	// --- tiles ---
	765
	766	if (tiles_enabled_flag) {
	767	if (num_tile_columns > DE265_MAX_TILE_COLUMNS) {
	768	errqueue->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	769	return false;
	770	}
	771	out.write_uvlc(num_tile_columns-1);
	772
	773	if (num_tile_rows > DE265_MAX_TILE_ROWS) {
	774	errqueue->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	775	return false;
	776	}
	777	out.write_uvlc(num_tile_rows-1);
	778
	779	out.write_bit(uniform_spacing_flag);
	780
	781	if (uniform_spacing_flag==false) {
	782	for (int i=0; i<num_tile_columns-1; i++)
	783	{
	784	out.write_uvlc(colWidth[i]-1);
	785	}
	786
	787	for (int i=0; i<num_tile_rows-1; i++)
	788	{
	789	out.write_uvlc(rowHeight[i]-1);
	790	}
	791	}
	792
	793	out.write_bit(loop_filter_across_tiles_enabled_flag);
	794	}
	795
	796
	797	out.write_bit(pps_loop_filter_across_slices_enabled_flag);
	798	out.write_bit(deblocking_filter_control_present_flag);
	799
	800	if (deblocking_filter_control_present_flag) {
	801	out.write_bit(deblocking_filter_override_enabled_flag);
	802	out.write_bit(pic_disable_deblocking_filter_flag);
	803
	804	if (!pic_disable_deblocking_filter_flag) {
	805	out.write_svlc(beta_offset/2);
	806	out.write_svlc(tc_offset /2);
	807	}
	808	}
	809
	810
	811	// --- scaling list ---
	812
	813	out.write_bit(pic_scaling_list_data_present_flag);
	814
	815	// check consistency: if scaling-lists are not enabled, pic_scalign_list_data_present_flag
	816	// must be FALSE
	817	if (sps->scaling_list_enable_flag==0 &&
	818	pic_scaling_list_data_present_flag != 0) {
	819	errqueue->add_warning(DE265_WARNING_PPS_HEADER_INVALID, false);
	820	return false;
	821	}
	822
	823	if (pic_scaling_list_data_present_flag) {
	824	de265_error err = write_scaling_list(out, sps, &scaling_list, true);
	825	if (err != DE265_OK) {
	826	errqueue->add_warning(err, false);
	827	return false;
	828	}
	829	}
	830
	831
	832
	833	out.write_bit(lists_modification_present_flag);
	834	out.write_uvlc(log2_parallel_merge_level-2);
	835
	836	out.write_bit(slice_segment_header_extension_present_flag);
	837	out.write_bit(pps_extension_flag);
438	838
439	839	if (pps_extension_flag) {
440	840	//assert(false);

456	856	}
457	857
458	858
459		void pic_parameter_set::dump_pps(int fd) const
	859	void pic_parameter_set::dump(int fd) const
460	860	{
461	861	FILE* fh;
462	862	if (fd==1) fh=stdout;

535	935	LOG1("pic_disable_deblocking_filter_flag: %d\n", pic_disable_deblocking_filter_flag);
536	936
537	937	LOG1("beta_offset: %d\n", beta_offset);
538		LOG1("tc_offset: %d\n", tc_offset);
	938	LOG1("tc_offset: %d\n", tc_offset);
539	939	}
540	940
541	941	LOG1("pic_scaling_list_data_present_flag: %d\n", pic_scaling_list_data_present_flag);

547	947	LOG1("log2_parallel_merge_level : %d\n", log2_parallel_merge_level);
548	948	LOG1("num_extra_slice_header_bits : %d\n", num_extra_slice_header_bits);
549	949	LOG1("slice_segment_header_extension_present_flag : %d\n", slice_segment_header_extension_present_flag);
550		LOG1("pps_extension_flag : %d\n", pps_extension_flag);
551
552		LOG1("Log2MinCuQpDeltaSize : %d\n", Log2MinCuQpDeltaSize);
	950	LOG1("pps_extension_flag : %d\n", pps_extension_flag);
	951	LOG1("pps_range_extension_flag : %d\n", pps_range_extension_flag);
	952	LOG1("pps_multilayer_extension_flag : %d\n", pps_multilayer_extension_flag);
	953	LOG1("pps_extension_6bits : %d\n", pps_extension_6bits);
	954
	955	LOG1("Log2MinCuQpDeltaSize : %d\n", Log2MinCuQpDeltaSize);
	956	LOG1("Log2MinCuChromaQpOffsetSize (RExt) : %d\n", Log2MinCuChromaQpOffsetSize);
	957	LOG1("Log2MaxTransformSkipSize (RExt) : %d\n", Log2MaxTransformSkipSize);
553	958
554	959	#undef LOG0
555	960	#undef LOG1
	961
	962
	963	if (pps_range_extension_flag) {
	964	range_extension.dump(fd);
	965	}
556	966	}
557	967
558	968

+68

-28

libde265/pps.h less more

28	28	#define DE265_MAX_TILE_COLUMNS 10
29	29	#define DE265_MAX_TILE_ROWS 10
30	30
	31	class decoder_context;
	32	class pic_parameter_set;
31	33
32		struct pic_parameter_set {
	34
	35	class pps_range_extension
	36	{
	37	public:
	38	pps_range_extension() { reset(); }
	39
	40	void reset();
	41
	42	bool read(bitreader, decoder_context, const pic_parameter_set*);
	43	void dump(int fd) const;
	44
	45	uint8_t log2_max_transform_skip_block_size;
	46	bool cross_component_prediction_enabled_flag;
	47	bool chroma_qp_offset_list_enabled_flag;
	48	uint8_t diff_cu_chroma_qp_offset_depth;
	49	uint8_t chroma_qp_offset_list_len;
	50	int8_t cb_qp_offset_list[6];
	51	int8_t cr_qp_offset_list[6];
	52	uint8_t log2_sao_offset_scale_luma;
	53	uint8_t log2_sao_offset_scale_chroma;
	54	};
	55
	56
	57	class pic_parameter_set {
	58	public:
33	59	pic_parameter_set();
34	60	~pic_parameter_set();
35	61
36		bool read(bitreader, struct decoder_context);
	62	void reset() { set_defaults(); }
	63	bool read(bitreader, decoder_context);
	64	bool write(error_queue*, CABAC_encoder&,
	65	const seq_parameter_set* sps);
37	66
38	67	bool is_tile_start_CTB(int ctbX,int ctbY) const;
39		void dump_pps(int fd) const;
	68	void dump(int fd) const;
40	69
	70
	71	void set_defaults(enum PresetSet = Preset_Default);
41	72
42	73	bool pps_read; // whether this pps has been read from bitstream
43	74

46	77	char dependent_slice_segments_enabled_flag;
47	78	char sign_data_hiding_flag;
48	79	char cabac_init_present_flag;
49		char num_ref_idx_l0_default_active;
50		char num_ref_idx_l1_default_active;
	80	char num_ref_idx_l0_default_active; // [1;16]
	81	char num_ref_idx_l1_default_active; // [1;16]
51	82
52	83	int pic_init_qp;
53	84	char constrained_intra_pred_flag;
54	85	char transform_skip_enabled_flag;
55	86
56		// if ( cu_qp_delta_enabled_flag )
	87	// --- QP ---
	88
57	89	char cu_qp_delta_enabled_flag;
58		int diff_cu_qp_delta_depth;
	90	int diff_cu_qp_delta_depth; // [ 0 ; log2_diff_max_min_luma_coding_block_size ]
59	91
60	92	int pic_cb_qp_offset;
61	93	int pic_cr_qp_offset;
62	94	char pps_slice_chroma_qp_offsets_present_flag;
	95
	96
63	97	char weighted_pred_flag;
64	98	char weighted_bipred_flag;
65	99	char output_flag_present_flag;

69	103
70	104	// --- tiles ---
71	105
72		//if( tiles_enabled_flag ) {
73	106	char tiles_enabled_flag;
74		int num_tile_columns;
75		int num_tile_rows;
	107	int num_tile_columns; // [1;PicWidthInCtbsY]
	108	int num_tile_rows; // [1;PicHeightInCtbsY]
76	109	char uniform_spacing_flag;
77	110
78		// derived values
79		int colWidth [ DE265_MAX_TILE_COLUMNS ];
80		int rowHeight[ DE265_MAX_TILE_ROWS ];
81		int colBd [ DE265_MAX_TILE_COLUMNS+1 ];
82		int rowBd [ DE265_MAX_TILE_ROWS+1 ];
83
84		std::vector<int> CtbAddrRStoTS; // #CTBs
85		std::vector<int> CtbAddrTStoRS; // #CTBs
86		std::vector<int> TileId; // #CTBs // index in tile-scan order
87		std::vector<int> TileIdRS; // #CTBs // index in raster-scan order
88		std::vector<int> MinTbAddrZS; // #TBs [x + y*PicWidthInTbsY]
89
90
91		// --- QP ---
92
93		int Log2MinCuQpDeltaSize;
94	111
95	112	// --- ---
96	113
97	114	char loop_filter_across_tiles_enabled_flag;
98
99	115	char pps_loop_filter_across_slices_enabled_flag;
100	116	char deblocking_filter_control_present_flag;
101	117

109	125	struct scaling_list_data scaling_list; // contains valid data if sps->scaling_list_enabled_flag set
110	126
111	127	char lists_modification_present_flag;
112		int log2_parallel_merge_level;
	128	int log2_parallel_merge_level; // [2 ; log2(max CB size)]
113	129	char num_extra_slice_header_bits;
114	130	char slice_segment_header_extension_present_flag;
115	131	char pps_extension_flag;
	132	char pps_range_extension_flag;
	133	char pps_multilayer_extension_flag;
	134	char pps_extension_6bits;
116	135
	136	pps_range_extension range_extension;
	137
	138
	139	// --- derived values ---
	140
	141	int Log2MinCuQpDeltaSize;
	142	int Log2MinCuChromaQpOffsetSize;
	143	int Log2MaxTransformSkipSize;
	144
	145	int colWidth [ DE265_MAX_TILE_COLUMNS ];
	146	int rowHeight[ DE265_MAX_TILE_ROWS ];
	147	int colBd [ DE265_MAX_TILE_COLUMNS+1 ];
	148	int rowBd [ DE265_MAX_TILE_ROWS+1 ];
	149
	150	std::vector<int> CtbAddrRStoTS; // #CTBs
	151	std::vector<int> CtbAddrTStoRS; // #CTBs
	152	std::vector<int> TileId; // #CTBs // index in tile-scan order
	153	std::vector<int> TileIdRS; // #CTBs // index in raster-scan order
	154	std::vector<int> MinTbAddrZS; // #TBs [x + y*PicWidthInTbsY]
	155
	156	void set_derived_values(const seq_parameter_set* sps);
117	157	};
118	158
119	159	#endif

+112

-0

libde265/quality.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "quality.h"
	21	#include <math.h>
	22
	23
	24	uint32_t SSD(const uint8_t* img, int imgStride,
	25	const uint8_t* ref, int refStride,
	26	int width, int height)
	27	{
	28	uint32_t sum=0;
	29
	30	const uint8_t* iPtr = img;
	31	const uint8_t* rPtr = ref;
	32
	33	for (int y=0;y<height;y++) {
	34	for (int x=0;x<width;x++) {
	35	int diff = iPtr[x] - rPtr[x];
	36	sum += diff*diff;
	37	}
	38
	39	iPtr += imgStride;
	40	rPtr += refStride;
	41	}
	42
	43	return sum;
	44	}
	45
	46
	47	uint32_t SAD(const uint8_t* img, int imgStride,
	48	const uint8_t* ref, int refStride,
	49	int width, int height)
	50	{
	51	uint32_t sum=0;
	52
	53	const uint8_t* iPtr = img;
	54	const uint8_t* rPtr = ref;
	55
	56	for (int y=0;y<height;y++) {
	57	for (int x=0;x<width;x++) {
	58	int diff = iPtr[x] - rPtr[x];
	59	sum += abs_value(diff);
	60	}
	61
	62	iPtr += imgStride;
	63	rPtr += refStride;
	64	}
	65
	66	return sum;
	67	}
	68
	69
	70	double MSE(const uint8_t* img, int imgStride,
	71	const uint8_t* ref, int refStride,
	72	int width, int height)
	73	{
	74	double sum=0.0;
	75
	76	const uint8_t* iPtr = img;
	77	const uint8_t* rPtr = ref;
	78
	79	for (int y=0;y<height;y++) {
	80	uint32_t lineSum=0;
	81
	82	for (int x=0;x<width;x++) {
	83	int diff = iPtr[x] - rPtr[x];
	84	lineSum += diff*diff;
	85	}
	86
	87	sum += ((double)lineSum)/width;
	88
	89	iPtr += imgStride;
	90	rPtr += refStride;
	91	}
	92
	93	return sum/height;
	94	}
	95
	96
	97	double PSNR(double mse)
	98	{
	99	if (mse==0) { return 99.99999; }
	100
	101	return 10log10(255.0255.0/mse);
	102	}
	103
	104	uint32_t compute_distortion_ssd(const de265_image* img1, const de265_image* img2,
	105	int x0, int y0, int log2size, int cIdx)
	106	{
	107	return SSD(img1->get_image_plane_at_pos(cIdx,x0,y0), img1->get_image_stride(cIdx),
	108	img2->get_image_plane_at_pos(cIdx,x0,y0), img2->get_image_stride(cIdx),
	109	1<<log2size, 1<<log2size);
	110	}
	111

+47

-0

libde265/quality.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifndef DE265_QUALITY_H
	21	#define DE265_QUALITY_H
	22
	23	#include <stdint.h>
	24	#include <libde265/de265.h>
	25	#include <libde265/image.h>
	26
	27
	28	LIBDE265_API uint32_t SSD(const uint8_t* img, int imgStride,
	29	const uint8_t* ref, int refStride,
	30	int width, int height);
	31
	32	LIBDE265_API uint32_t SAD(const uint8_t* img, int imgStride,
	33	const uint8_t* ref, int refStride,
	34	int width, int height);
	35
	36	LIBDE265_API double MSE(const uint8_t* img, int imgStride,
	37	const uint8_t* ref, int refStride,
	38	int width, int height);
	39
	40	LIBDE265_API double PSNR(double mse);
	41
	42
	43	LIBDE265_API uint32_t compute_distortion_ssd(const de265_image* img1, const de265_image* img2,
	44	int x0, int y0, int log2size, int cIdx);
	45
	46	#endif

+133

-19

libde265/refpic.cc less more

30	30	#endif
31	31
32	32
33		static void compute_NumPoc(ref_pic_set* rpset)
34		{
35		rpset->NumPocTotalCurr_shortterm_only = 0;
36
37		for (int i=0; i<rpset->NumNegativePics; i++)
38		if (rpset->UsedByCurrPicS0[i])
39		rpset->NumPocTotalCurr_shortterm_only++;
40
41		for (int i=0; i<rpset->NumPositivePics; i++)
42		if (rpset->UsedByCurrPicS1[i])
43		rpset->NumPocTotalCurr_shortterm_only++;
	33	void ref_pic_set::reset()
	34	{
	35	NumNegativePics = 0;
	36	NumPositivePics = 0;
	37	NumDeltaPocs = 0;
	38	NumPocTotalCurr_shortterm_only = 0;
	39
	40	for (int i=0;i<MAX_NUM_REF_PICS;i++) {
	41	DeltaPocS0[i] = 0;
	42	DeltaPocS1[i] = 0;
	43
	44	UsedByCurrPicS0[i] = 0;
	45	UsedByCurrPicS1[i] = 0;
	46	}
	47	}
	48
	49
	50	void ref_pic_set::compute_derived_values()
	51	{
	52	NumPocTotalCurr_shortterm_only = 0;
	53
	54	for (int i=0; i<NumNegativePics; i++)
	55	if (UsedByCurrPicS0[i])
	56	NumPocTotalCurr_shortterm_only++;
	57
	58	for (int i=0; i<NumPositivePics; i++)
	59	if (UsedByCurrPicS1[i])
	60	NumPocTotalCurr_shortterm_only++;
	61
	62	NumDeltaPocs = NumNegativePics + NumPositivePics;
	63
44	64
45	65	/*
46	66	NOTE: this is done when reading the slice header.

61	81	When coding the ref-pic-sets in the SPS, predicition is always from the previous set.
62	82	In the slice header, the ref-pic-set can use any previous set as reference.
63	83	*/
64		bool read_short_term_ref_pic_set(decoder_context* ctx,
	84	bool read_short_term_ref_pic_set(error_queue* errqueue,
65	85	const seq_parameter_set* sps,
66	86	bitreader* br,
67	87	ref_pic_set* out_set, // where to store the read set

91	111	int delta_idx;
92	112	if (sliceRefPicSet) { // idxRps == num_short_term_ref_pic_sets) {
93	113	delta_idx = vlc = get_uvlc(br);
	114	if (delta_idx==UVLC_ERROR) {
	115	return false;
	116	}
	117
	118	if (delta_idx>=idxRps) {
	119	return false;
	120	}
	121
94	122	delta_idx++;
95	123	} else {
96	124	delta_idx = 1;

101	129
102	130	int delta_rps_sign = get_bits(br,1);
103	131	int abs_delta_rps = vlc = get_uvlc(br);
	132	if (vlc==UVLC_ERROR) { return false; }
104	133	abs_delta_rps++;
105	134	int DeltaRPS = (delta_rps_sign ? -abs_delta_rps : abs_delta_rps);
106	135

141	170
142	171	// positive list
143	172	for (int j=nPositiveRIdx-1;j>=0;j--) {
	173	assert(RIdx >= 0 && RIdx < sets.size());
	174	assert(j>=0 && j < MAX_NUM_REF_PICS);
	175
144	176	int dPoc = sets[RIdx].DeltaPocS1[j] + DeltaRPS; // new delta
145	177	if (dPoc<0 && use_delta_flag[nNegativeRIdx+j]) {
	178	if (i>= MAX_NUM_REF_PICS) { return false; }
	179
146	180	out_set->DeltaPocS0[i] = dPoc;
147	181	out_set->UsedByCurrPicS0[i] = used_by_curr_pic_flag[nNegativeRIdx+j];
148	182	i++;

151	185
152	186	// frame 0
153	187	if (DeltaRPS<0 && use_delta_flag[nDeltaPocsRIdx]) {
	188	if (i>= MAX_NUM_REF_PICS) { return false; }
	189
154	190	out_set->DeltaPocS0[i] = DeltaRPS;
155	191	out_set->UsedByCurrPicS0[i] = used_by_curr_pic_flag[nDeltaPocsRIdx];
156	192	i++;

160	196	for (int j=0;j<nNegativeRIdx;j++) {
161	197	int dPoc = sets[RIdx].DeltaPocS0[j] + DeltaRPS;
162	198	if (dPoc<0 && use_delta_flag[j]) {
	199	if (i>= MAX_NUM_REF_PICS) { return false; }
	200
163	201	out_set->DeltaPocS0[i] = dPoc;
164	202	out_set->UsedByCurrPicS0[i] = used_by_curr_pic_flag[j];
165	203	i++;

178	216	for (int j=nNegativeRIdx-1;j>=0;j--) {
179	217	int dPoc = sets[RIdx].DeltaPocS0[j] + DeltaRPS;
180	218	if (dPoc>0 && use_delta_flag[j]) {
	219	if (i>= MAX_NUM_REF_PICS) { return false; }
	220
181	221	out_set->DeltaPocS1[i] = dPoc;
182	222	out_set->UsedByCurrPicS1[i] = used_by_curr_pic_flag[j];
183	223	i++;

186	226
187	227	// frame 0
188	228	if (DeltaRPS>0 && use_delta_flag[nDeltaPocsRIdx]) {
	229	if (i>= MAX_NUM_REF_PICS) { return false; }
	230
189	231	out_set->DeltaPocS1[i] = DeltaRPS;
190	232	out_set->UsedByCurrPicS1[i] = used_by_curr_pic_flag[nDeltaPocsRIdx];
191	233	i++;

195	237	for (int j=0;j<nPositiveRIdx;j++) {
196	238	int dPoc = sets[RIdx].DeltaPocS1[j] + DeltaRPS;
197	239	if (dPoc>0 && use_delta_flag[nNegativeRIdx+j]) {
	240	if (i>= MAX_NUM_REF_PICS) { return false; }
	241
198	242	out_set->DeltaPocS1[i] = dPoc;
199	243	out_set->UsedByCurrPicS1[i] = used_by_curr_pic_flag[nNegativeRIdx+j];
200	244	i++;

202	246	}
203	247
204	248	out_set->NumPositivePics = i;
205
206		out_set->NumDeltaPocs = out_set->NumNegativePics + out_set->NumPositivePics;
207	249
208	250	} else {
209	251

221	263	out_set->NumDeltaPocs = 0;
222	264	out_set->NumPocTotalCurr_shortterm_only = 0;
223	265
224		ctx->add_warning(DE265_WARNING_MAX_NUM_REF_PICS_EXCEEDED, false);
	266	errqueue->add_warning(DE265_WARNING_MAX_NUM_REF_PICS_EXCEEDED, false);
225	267	return false;
226	268	}
227	269
	270	if (num_negative_pics > MAX_NUM_REF_PICS \|\|
	271	num_positive_pics > MAX_NUM_REF_PICS) {
	272	errqueue->add_warning(DE265_WARNING_MAX_NUM_REF_PICS_EXCEEDED, false);
	273	return false;
	274	}
228	275
229	276	out_set->NumNegativePics = num_negative_pics;
230	277	out_set->NumPositivePics = num_positive_pics;
231		out_set->NumDeltaPocs = num_positive_pics + num_negative_pics;
232	278
233	279	// --- now, read the deltas between the reference frames to fill the lists ---
234	280

236	282
237	283	int lastPocS=0;
238	284	for (int i=0;i<num_negative_pics;i++) {
239		int delta_poc_s0 = get_uvlc(br)+1;
	285	int delta_poc_s0 = get_uvlc(br);
	286	if (delta_poc_s0==UVLC_ERROR) { return false; }
	287	delta_poc_s0++;
240	288	char used_by_curr_pic_s0_flag = get_bits(br,1);
241	289
242	290	out_set->DeltaPocS0[i] = lastPocS - delta_poc_s0;

248	296
249	297	lastPocS=0;
250	298	for (int i=0;i<num_positive_pics;i++) {
251		int delta_poc_s1 = get_uvlc(br)+1;
	299	int delta_poc_s1 = get_uvlc(br);
	300	if (delta_poc_s1==UVLC_ERROR) { return false; }
	301	delta_poc_s1++;
252	302	char used_by_curr_pic_s1_flag = get_bits(br,1);
253	303
254	304	out_set->DeltaPocS1[i] = lastPocS + delta_poc_s1;

258	308	}
259	309
260	310
261		compute_NumPoc(out_set);
	311	out_set->compute_derived_values();
262	312
263	313	return true;
	314	}
	315
	316
	317	bool write_short_term_ref_pic_set_nopred(error_queue* errqueue,
	318	const seq_parameter_set* sps,
	319	CABAC_encoder& out,
	320	const ref_pic_set* in_set, // which set to write
	321	int idxRps, // index of the set to be written
	322	const std::vector<ref_pic_set>& sets, // previously read sets
	323	bool sliceRefPicSet) // is this in the slice header?
	324	{
	325	if (idxRps != 0) {
	326	// inter_ref_pic_set_prediction_flag
	327	out.write_bit(0);
	328	}
	329
	330
	331	// --- first, write the number of past and future frames in this set ---
	332
	333	out.write_uvlc(in_set->NumNegativePics);
	334	out.write_uvlc(in_set->NumPositivePics);
	335
	336	// --- now, write the deltas between the reference frames to fill the lists ---
	337
	338	// past frames
	339
	340	int lastPocS=0;
	341	for (int i=0;i<in_set->NumNegativePics;i++) {
	342	int delta_poc_s0 = lastPocS - in_set->DeltaPocS0[i];
	343	char used_by_curr_pic_s0_flag = in_set->UsedByCurrPicS0[i];
	344
	345	assert(delta_poc_s0 >= 1);
	346	out.write_uvlc(delta_poc_s0-1);
	347	out.write_bit(used_by_curr_pic_s0_flag);
	348	lastPocS = in_set->DeltaPocS0[i];
	349	}
	350
	351	// future frames
	352
	353	lastPocS=0;
	354	for (int i=0;i<in_set->NumPositivePics;i++) {
	355	int delta_poc_s1 = in_set->DeltaPocS1[i] - lastPocS;
	356	char used_by_curr_pic_s1_flag = in_set->UsedByCurrPicS1[i];
	357
	358	assert(delta_poc_s1 >= 1);
	359	out.write_uvlc(delta_poc_s1-1);
	360	out.write_bit(used_by_curr_pic_s1_flag);
	361	lastPocS = in_set->DeltaPocS1[i];
	362	}
	363
	364	return true;
	365	}
	366
	367
	368	bool write_short_term_ref_pic_set(error_queue* errqueue,
	369	const seq_parameter_set* sps,
	370	CABAC_encoder& out,
	371	const ref_pic_set* in_set, // which set to write
	372	int idxRps, // index of the set to be read
	373	const std::vector<ref_pic_set>& sets, // previously read sets
	374	bool sliceRefPicSet) // is this in the slice header?
	375	{
	376	return write_short_term_ref_pic_set_nopred(errqueue, sps, out, in_set, idxRps, sets,
	377	sliceRefPicSet);
264	378	}
265	379
266	380

+17

-9

libde265/refpic.h less more

25	25	#define MAX_NUM_REF_PICS 16 // maximum defined by standard, may be lower for some Levels
26	26
27	27
28		typedef struct {
29		uint8_t NumNegativePics; // number of past reference pictures
30		uint8_t NumPositivePics; // number of future reference pictures
31		uint8_t NumDeltaPocs; // total number of reference pictures (past + future)
32
33		uint8_t NumPocTotalCurr_shortterm_only; /* Total number of reference pictures that may actually
34		be used for prediction in the current frame. */
35
	28	class ref_pic_set
	29	{
	30	public:
36	31	// Lists of pictures that have to be kept in the decoded picture buffer for future
37	32	// reference and that may optionally be used for prediction in the current frame.
38	33	// Lists contain the relative POC positions.

43	38	uint8_t UsedByCurrPicS0[MAX_NUM_REF_PICS];
44	39	uint8_t UsedByCurrPicS1[MAX_NUM_REF_PICS];
45	40
46		} ref_pic_set;
	41	uint8_t NumNegativePics; // number of past reference pictures
	42	uint8_t NumPositivePics; // number of future reference pictures
	43
	44	// --- derived values ---
	45
	46	void compute_derived_values();
	47
	48	uint8_t NumDeltaPocs; // total number of reference pictures (past + future)
	49
	50	uint8_t NumPocTotalCurr_shortterm_only; /* Total number of reference pictures that may actually
	51	be used for prediction in the current frame. */
	52
	53	void reset();
	54	};
47	55
48	56
49	57	void dump_short_term_ref_pic_set(const ref_pic_set, FILE fh);

+106

-48

libde265/sao.cc less more

24	24	#include <string.h>
25	25
26	26
27		void apply_sao(de265_image* img, int xCtb,int yCtb,
28		const slice_segment_header* shdr, int cIdx, int nS,
29		const uint8_t* in_img, int in_stride,
30		/* / uint8_t out_img, int out_stride)
	27	template <class pixel_t>
	28	void apply_sao_internal(de265_image* img, int xCtb,int yCtb,
	29	const slice_segment_header* shdr, int cIdx, int nSW,int nSH,
	30	const pixel_t* in_img, int in_stride,
	31	/* / pixel_t out_img, int out_stride)
31	32	{
32	33	const sao_info* saoinfo = img->get_sao_info(xCtb,yCtb);
33	34
34	35	int SaoTypeIdx = (saoinfo->SaoTypeIdx >> (2*cIdx)) & 0x3;
35	36
36		logtrace(LogSAO,"apply_sao CTB %d;%d cIdx:%d type=%d (%dx%d)\n",xCtb,yCtb,cIdx, SaoTypeIdx, nS,nS);
	37	logtrace(LogSAO,"apply_sao CTB %d;%d cIdx:%d type=%d (%dx%d)\n",xCtb,yCtb,cIdx, SaoTypeIdx, nSW,nSH);
37	38
38	39	if (SaoTypeIdx==0) {
39	40	return;

45	46	const int maxPixelValue = (1<<bitDepth)-1;
46	47
47	48	// top left position of CTB in pixels
48		const int xC = xCtb*nS;
49		const int yC = yCtb*nS;
	49	const int xC = xCtb*nSW;
	50	const int yC = yCtb*nSH;
50	51
51	52	const int width = img->get_width(cIdx);
52	53	const int height = img->get_height(cIdx);

54	55	const int ctbSliceAddrRS = img->get_SliceHeader(xC,yC)->SliceAddrRS;
55	56
56	57	const int picWidthInCtbs = sps->PicWidthInCtbsY;
57		const int ctbshift = sps->Log2CtbSizeY - (cIdx>0 ? 1 : 0);
58		const int chromashift = (cIdx>0 ? 1 : 0);
	58	const int chromashiftW = sps->get_chroma_shift_W(cIdx);
	59	const int chromashiftH = sps->get_chroma_shift_H(cIdx);
	60	const int ctbshiftW = sps->Log2CtbSizeY - chromashiftW;
	61	const int ctbshiftH = sps->Log2CtbSizeY - chromashiftH;
59	62
60	63
61	64	for (int i=0;i<5;i++)

65	68
66	69
67	70	// actual size of CTB to be processed (can be smaller when partially outside of image)
68		const int ctbW = (xC+nS>width) ? width -xC : nS;
69		const int ctbH = (yC+nS>height) ? height-yC : nS;
70
71
72		const bool extendedTests = (img->get_CTB_has_pcm(xCtb,yCtb) \|\|
73		img->get_CTB_has_cu_transquant_bypass(xCtb,yCtb));
	71	const int ctbW = (xC+nSW>width) ? width -xC : nSW;
	72	const int ctbH = (yC+nSH>height) ? height-yC : nSH;
	73
	74
	75	const bool extendedTests = img->get_CTB_has_pcm_or_cu_transquant_bypass(xCtb,yCtb);
74	76
75	77	if (SaoTypeIdx==2) {
76	78	int hPos[2], vPos[2];

98	100
99	101
100	102	for (int j=0;j<ctbH;j++) {
101		const uint8_t* in_ptr = &in_img [xC+(yC+j)*in_stride];
102		/* / uint8_t out_ptr = &out_img[xC+(yC+j)*out_stride];
	103	const pixel_t* in_ptr = &in_img [xC+(yC+j)*in_stride];
	104	/* / pixel_t out_ptr = &out_img[xC+(yC+j)*out_stride];
103	105
104	106	for (int i=0;i<ctbW;i++) {
105	107	int edgeIdx = -1;
106	108
107	109	logtrace(LogSAO, "pos %d,%d\n",xC+i,yC+j);
108	110
109		if (extendedTests &&
110		(sps->pcm_loop_filter_disable_flag &&
111		img->get_pcm_flag((xC+i)<<chromashift,(yC+j)<<chromashift)) \|\|
112		img->get_cu_transquant_bypass((xC+i)<<chromashift,(yC+j)<<chromashift)) {
	111	if ((extendedTests &&
	112	(sps->pcm_loop_filter_disable_flag &&
	113	img->get_pcm_flag((xC+i)<<chromashiftW,(yC+j)<<chromashiftH))) \|\|
	114	img->get_cu_transquant_bypass((xC+i)<<chromashiftW,(yC+j)<<chromashiftH)) {
113	115	continue;
114	116	}
115	117

132	134	// slice anyway) reduced computation time only by 1.3%.
133	135	// TODO: however, this may still be a big part of SAO itself.
134	136
135		int sliceAddrRS = img->get_SliceHeader(xS<<chromashift,yS<<chromashift)->SliceAddrRS;
	137	slice_segment_header* sliceHeader = img->get_SliceHeader(xS<<chromashiftW,
	138	yS<<chromashiftH);
	139	if (sliceHeader==NULL) { return; }
	140
	141	int sliceAddrRS = sliceHeader->SliceAddrRS;
136	142	if (sliceAddrRS < ctbSliceAddrRS &&
137		img->get_SliceHeader((xC+i)<<chromashift,(yC+j)<<chromashift)->slice_loop_filter_across_slices_enabled_flag==0) {
	143	img->get_SliceHeader((xC+i)<<chromashiftW,
	144	(yC+j)<<chromashiftH)->slice_loop_filter_across_slices_enabled_flag==0) {
138	145	edgeIdx=0;
139	146	break;
140	147	}
141	148
142	149	if (sliceAddrRS > ctbSliceAddrRS &&
143		img->get_SliceHeader(xS<<chromashift,yS<<chromashift)->slice_loop_filter_across_slices_enabled_flag==0) {
	150	img->get_SliceHeader(xS<<chromashiftW,
	151	yS<<chromashiftH)->slice_loop_filter_across_slices_enabled_flag==0) {
144	152	edgeIdx=0;
145	153	break;
146	154	}
147	155
148	156
149		if (pps->loop_filter_across_tiles_enabled_flag==0 &&
150		pps->TileIdRS[(xS>>ctbshift) + (yS>>ctbshift)*picWidthInCtbs] !=
151		pps->TileIdRS[(xC>>ctbshift) + (yC>>ctbshift)*picWidthInCtbs]) {
	157	if (pps->loop_filter_across_tiles_enabled_flag==0 &&
	158	pps->TileIdRS[(xS>>ctbshiftW) + (yS>>ctbshiftH)*picWidthInCtbs] !=
	159	pps->TileIdRS[(xC>>ctbshiftW) + (yC>>ctbshiftH)*picWidthInCtbs]) {
152	160	edgeIdx=0;
153	161	break;
154	162	}

197	205	for (int i=0;i<ctbW;i++) {
198	206
199	207	if ((sps->pcm_loop_filter_disable_flag &&
200		img->get_pcm_flag((xC+i)<<chromashift,(yC+j)<<chromashift)) \|\|
201		img->get_cu_transquant_bypass((xC+i)<<chromashift,(yC+j)<<chromashift)) {
	208	img->get_pcm_flag((xC+i)<<chromashiftW,(yC+j)<<chromashiftH)) \|\|
	209	img->get_cu_transquant_bypass((xC+i)<<chromashiftW,(yC+j)<<chromashiftH)) {
202	210	continue;
203	211	}
204	212
205	213	int bandIdx = bandTable[ in_img[xC+i+(yC+j)*in_stride]>>bandShift ];
	214
	215	// Shifts are a strange thing. On x86, >>x actually computes >>(x%64).
	216	// So we have to take care of large bandShifts.
	217	if (bandShift>=8) { bandIdx=0; }
206	218
207	219	if (bandIdx>0) {
208	220	int offset = saoinfo->saoOffsetVal[cIdx][bandIdx-1];

211	223	offset,
212	224	in_img[xC+i+(yC+j)*in_stride],
213	225	in_img[xC+i+(yC+j)*in_stride]+offset);
214
	226
215	227	out_img[xC+i+(yC+j)*out_stride] = Clip3(0,maxPixelValue,
216	228	in_img[xC+i+(yC+j)*in_stride] + offset);
217	229	}

226	238
227	239	int bandIdx = bandTable[ in_img[xC+i+(yC+j)*in_stride]>>bandShift ];
228	240
	241	// see above
	242	if (bandShift>=8) { bandIdx=0; }
	243
229	244	if (bandIdx>0) {
230	245	int offset = saoinfo->saoOffsetVal[cIdx][bandIdx-1];
231	246

234	249	}
235	250	}
236	251	}
	252	}
	253	}
	254
	255
	256	template <class pixel_t>
	257	void apply_sao(de265_image* img, int xCtb,int yCtb,
	258	const slice_segment_header* shdr, int cIdx, int nSW,int nSH,
	259	const pixel_t* in_img, int in_stride,
	260	/* / pixel_t out_img, int out_stride)
	261	{
	262	if (img->high_bit_depth(cIdx)) {
	263	apply_sao_internal<uint16_t>(img,xCtb,yCtb, shdr,cIdx,nSW,nSH,
	264	(uint16_t*)in_img, in_stride,
	265	(uint16_t*)out_img,out_stride);
	266	}
	267	else {
	268	apply_sao_internal<uint8_t>(img,xCtb,yCtb, shdr,cIdx,nSW,nSH,
	269	in_img, in_stride,
	270	out_img,out_stride);
237	271	}
238	272	}
239	273

257	291	const slice_segment_header* shdr = img->get_SliceHeaderCtb(xCtb,yCtb);
258	292
259	293	if (shdr->slice_sao_luma_flag) {
260		apply_sao(img, xCtb,yCtb, shdr, 0, 1<<img->sps.Log2CtbSizeY,
	294	apply_sao(img, xCtb,yCtb, shdr, 0, 1<<img->sps.Log2CtbSizeY, 1<<img->sps.Log2CtbSizeY,
261	295	inputCopy.get_image_plane(0), inputCopy.get_image_stride(0),
262	296	img->get_image_plane(0), img->get_image_stride(0));
263	297	}
264	298
265	299	if (shdr->slice_sao_chroma_flag) {
266		apply_sao(img, xCtb,yCtb, shdr, 1, 1<<(img->sps.Log2CtbSizeY-1),
	300	int nSW = (1<<img->sps.Log2CtbSizeY) / img->sps.SubWidthC;
	301	int nSH = (1<<img->sps.Log2CtbSizeY) / img->sps.SubHeightC;
	302
	303	apply_sao(img, xCtb,yCtb, shdr, 1, nSW,nSH,
267	304	inputCopy.get_image_plane(1), inputCopy.get_image_stride(1),
268	305	img->get_image_plane(1), img->get_image_stride(1));
269	306
270		apply_sao(img, xCtb,yCtb, shdr, 2, 1<<(img->sps.Log2CtbSizeY-1),
	307	apply_sao(img, xCtb,yCtb, shdr, 2, nSW,nSH,
271	308	inputCopy.get_image_plane(2), inputCopy.get_image_stride(2),
272	309	img->get_image_plane(2), img->get_image_stride(2));
273	310	}

281	318	return;
282	319	}
283	320
284
285		uint8_t* inputCopy = new uint8_t[ img->get_image_stride(0) * img->get_height(0) ];
	321	int lumaImageSize = img->get_image_stride(0) * img->get_height(0) * img->get_bytes_per_pixel(0);
	322	int chromaImageSize = img->get_image_stride(1) * img->get_height(1) * img->get_bytes_per_pixel(1);
	323
	324	uint8_t* inputCopy = new uint8_t[ libde265_max(lumaImageSize, chromaImageSize) ];
286	325	if (inputCopy == NULL) {
287	326	img->decctx->add_warning(DE265_WARNING_CANNOT_APPLY_SAO_OUT_OF_MEMORY,false);
288	327	return;
289	328	}
290	329
291	330
292		for (int cIdx=0;cIdx<3;cIdx++) {
	331	int nChannels = 3;
	332	if (img->sps.ChromaArrayType == CHROMA_MONO) { nChannels=1; }
	333
	334	for (int cIdx=0;cIdx<nChannels;cIdx++) {
293	335
294	336	int stride = img->get_image_stride(cIdx);
295	337	int height = img->get_height(cIdx);
296	338
297		memcpy(inputCopy, img->get_image_plane(cIdx), stride * height);
	339	memcpy(inputCopy, img->get_image_plane(cIdx), stride * height * img->get_bytes_per_pixel(cIdx));
298	340
299	341	for (int yCtb=0; yCtb<img->sps.PicHeightInCtbsY; yCtb++)
300	342	for (int xCtb=0; xCtb<img->sps.PicWidthInCtbsY; xCtb++)
301	343	{
302	344	const slice_segment_header* shdr = img->get_SliceHeaderCtb(xCtb,yCtb);
	345	if (shdr==NULL) { return; }
303	346
304	347	if (cIdx==0 && shdr->slice_sao_luma_flag) {
305		apply_sao(img, xCtb,yCtb, shdr, 0, 1<<img->sps.Log2CtbSizeY,
	348	apply_sao(img, xCtb,yCtb, shdr, 0, 1<<img->sps.Log2CtbSizeY, 1<<img->sps.Log2CtbSizeY,
306	349	inputCopy, stride,
307	350	img->get_image_plane(0), img->get_image_stride(0));
308	351	}
309	352
310	353	if (cIdx!=0 && shdr->slice_sao_chroma_flag) {
311		apply_sao(img, xCtb,yCtb, shdr, cIdx, 1<<(img->sps.Log2CtbSizeY-1),
	354	int nSW = (1<<img->sps.Log2CtbSizeY) / img->sps.SubWidthC;
	355	int nSH = (1<<img->sps.Log2CtbSizeY) / img->sps.SubHeightC;
	356
	357	apply_sao(img, xCtb,yCtb, shdr, cIdx, nSW,nSH,
312	358	inputCopy, stride,
313	359	img->get_image_plane(cIdx), img->get_image_stride(cIdx));
314	360	}

334	380	int inputProgress;
335	381
336	382	virtual void work();
	383	virtual std::string name() const {
	384	char buf[100];
	385	sprintf(buf,"sao-%d",ctb_y);
	386	return buf;
	387	}
337	388	};
338	389
339	390
340	391	void thread_task_sao::work()
341	392	{
342	393	state = Running;
343		img->thread_run();
	394	img->thread_run(this);
344	395
345	396	const int rightCtb = img->sps.PicWidthInCtbsY-1;
346	397	const int ctbSize = (1<<img->sps.Log2CtbSizeY);

353	404	if (ctb_y>0) {
354	405	img->wait_for_progress(this, rightCtb,ctb_y-1, inputProgress);
355	406	}
356
	407
357	408	if (ctb_y+1<img->sps.PicHeightInCtbsY) {
358	409	img->wait_for_progress(this, rightCtb,ctb_y+1, inputProgress);
359	410	}

369	420	for (int xCtb=0; xCtb<img->sps.PicWidthInCtbsY; xCtb++)
370	421	{
371	422	const slice_segment_header* shdr = img->get_SliceHeaderCtb(xCtb,ctb_y);
	423	if (shdr==NULL) {
	424	break;
	425	}
372	426
373	427	if (shdr->slice_sao_luma_flag) {
374		apply_sao(img, xCtb,ctb_y, shdr, 0, ctbSize,
	428	apply_sao(img, xCtb,ctb_y, shdr, 0, ctbSize, ctbSize,
375	429	inputImg ->get_image_plane(0), inputImg ->get_image_stride(0),
376	430	outputImg->get_image_plane(0), outputImg->get_image_stride(0));
377	431	}
378	432
379	433	if (shdr->slice_sao_chroma_flag) {
380		apply_sao(img, xCtb,ctb_y, shdr, 1, ctbSize>>1,
	434	int nSW = ctbSize / img->sps.SubWidthC;
	435	int nSH = ctbSize / img->sps.SubHeightC;
	436
	437	apply_sao(img, xCtb,ctb_y, shdr, 1, nSW,nSH,
381	438	inputImg ->get_image_plane(1), inputImg ->get_image_stride(1),
382	439	outputImg->get_image_plane(1), outputImg->get_image_stride(1));
383	440
384		apply_sao(img, xCtb,ctb_y, shdr, 2, ctbSize>>1,
	441	apply_sao(img, xCtb,ctb_y, shdr, 2, nSW,nSH,
385	442	inputImg ->get_image_plane(2), inputImg ->get_image_stride(2),
386	443	outputImg->get_image_plane(2), outputImg->get_image_stride(2));
387	444	}

397	454
398	455
399	456	state = Finished;
400		img->thread_finishes();
	457	img->thread_finishes(this);
401	458	}
402	459
403	460

414	471
415	472	de265_error err = imgunit->sao_output.alloc_image(img->get_width(), img->get_height(),
416	473	img->get_chroma_format(), &img->sps, false,
417		img->decctx, img->pts, img->user_data, true);
	474	img->decctx, img->encctx,
	475	img->pts, img->user_data, true);
418	476	if (err != DE265_OK) {
419	477	img->decctx->add_warning(DE265_WARNING_CANNOT_APPLY_SAO_OUT_OF_MEMORY,false);
420	478	return false;

425	483	int n=0;
426	484	img->thread_start(nRows);
427	485
428		for (int y=0;y<img->sps.PicHeightInCtbsY;y++)
	486	for (int y=0;y<nRows;y++)
429	487	{
430	488	thread_task_sao* task = new thread_task_sao;
431	489

436	494	task->inputProgress = saoInputProgress;
437	495
438	496	imgunit->tasks.push_back(task);
439		add_task(&ctx->thread_pool, task);
	497	add_task(&ctx->thread_pool_, task);
440	498	n++;
441	499	}
442	500

+111

-25

libde265/sei.cc less more

95	95	}
96	96
97	97
98		static uint32_t compute_checksum_8bit(uint8_t* data,int w,int h,int stride)
	98	class raw_hash_data
	99	{
	100	public:
	101	raw_hash_data(int w, int stride);
	102	~raw_hash_data();
	103
	104	struct data_chunk {
	105	const uint8_t* data;
	106	int len;
	107	};
	108
	109	data_chunk prepare_8bit(const uint8_t* data,int y);
	110	data_chunk prepare_16bit(const uint8_t* data,int y);
	111
	112	private:
	113	int mWidth, mStride;
	114
	115	uint8_t* mMem;
	116	};
	117
	118
	119	raw_hash_data::raw_hash_data(int w, int stride)
	120	{
	121	mWidth=w;
	122	mStride=stride;
	123	mMem = NULL;
	124	}
	125
	126	raw_hash_data::~raw_hash_data()
	127	{
	128	delete[] mMem;
	129	}
	130
	131	raw_hash_data::data_chunk raw_hash_data::prepare_8bit(const uint8_t* data,int y)
	132	{
	133	data_chunk chunk;
	134	chunk.data = data+y*mStride;
	135	chunk.len = mWidth;
	136	return chunk;
	137	}
	138
	139	raw_hash_data::data_chunk raw_hash_data::prepare_16bit(const uint8_t* data,int y)
	140	{
	141	if (mMem == NULL) {
	142	mMem = new uint8_t[2*mWidth];
	143	}
	144
	145	const uint16_t* data16 = (uint16_t*)data;
	146
	147	for (int x=0; x<mWidth; x++) {
	148	mMem[2x+0] = data16[ymStride+x] & 0xFF;
	149	mMem[2x+1] = data16[ymStride+x] >> 8;
	150	}
	151
	152	data_chunk chunk;
	153	chunk.data = mMem;
	154	chunk.len = 2*mWidth;
	155	return chunk;
	156	}
	157
	158
	159	static uint32_t compute_checksum_8bit(uint8_t* data,int w,int h,int stride, int bit_depth)
99	160	{
100	161	uint32_t sum = 0;
101		for (int y=0; y<h; y++)
102		for(int x=0; x<w; x++) {
103		uint8_t xorMask = ( x & 0xFF ) ^ ( y & 0xFF ) ^ ( x >> 8 ) ^ ( y >> 8 );
104		sum += data[y*stride + x] ^ xorMask;
105
106		/*
107		if (compDepth[cIdx] > 8 )
108		sum = ( sum + ( ( component[cIdx][y * compWidth[cIdx] + x] >> 8 ) ^ xorMask ) ) &
109		0xFFFFFFFF
110		}
111		*/
112		}
	162
	163	if (bit_depth<=8) {
	164	for (int y=0; y<h; y++)
	165	for(int x=0; x<w; x++) {
	166	uint8_t xorMask = ( x & 0xFF ) ^ ( y & 0xFF ) ^ ( x >> 8 ) ^ ( y >> 8 );
	167	sum += data[y*stride + x] ^ xorMask;
	168	}
	169	}
	170	else {
	171	for (int y=0; y<h; y++)
	172	for(int x=0; x<w; x++) {
	173	uint8_t xorMask = ( x & 0xFF ) ^ ( y & 0xFF ) ^ ( x >> 8 ) ^ ( y >> 8 );
	174	sum += (data[y*stride + x] & 0xFF) ^ xorMask;
	175	sum += (data[y*stride + x] >> 8) ^ xorMask;
	176	}
	177	}
113	178
114	179	return sum & 0xFFFFFFFF;
115	180	}

156	221	(t << 12)) & 0xFFFF;
157	222	}
158	223
159		static uint32_t compute_CRC_8bit_fast(const uint8_t* data,int w,int h,int stride)
160		{
	224	static uint32_t compute_CRC_8bit_fast(const uint8_t* data,int w,int h,int stride, int bit_depth)
	225	{
	226	raw_hash_data raw_data(w,stride);
	227
161	228	uint16_t crc = 0xFFFF;
162	229
163	230	crc = crc_process_byte_parallel(crc, 0);
164	231	crc = crc_process_byte_parallel(crc, 0);
165	232
166	233	for (int y=0; y<h; y++) {
167		const uint8_t* d = &data[y*stride];
168
169		for(int x=0; x<w; x++) {
170		crc = crc_process_byte_parallel(crc, *d++);
	234	raw_hash_data::data_chunk chunk;
	235
	236	if (bit_depth>8)
	237	chunk = raw_data.prepare_16bit(data, y);
	238	else
	239	chunk = raw_data.prepare_8bit(data, y);
	240
	241	for(int x=0; x<chunk.len; x++) {
	242	crc = crc_process_byte_parallel(crc, chunk.data[x]);
171	243	}
172	244	}
173	245
174	246	return crc;
175	247	}
176	248
177		static void compute_MD5_8bit(uint8_t* data,int w,int h,int stride, uint8_t* result)
	249
	250	static void compute_MD5(uint8_t* data,int w,int h,int stride, uint8_t* result, int bit_depth)
178	251	{
179	252	MD5_CTX md5;
180	253	MD5_Init(&md5);
181	254
	255	raw_hash_data raw_data(w,stride);
	256
182	257	for (int y=0; y<h; y++) {
183		MD5_Update(&md5, &data[y*stride], w);
	258	raw_hash_data::data_chunk chunk;
	259
	260	if (bit_depth>8)
	261	chunk = raw_data.prepare_16bit(data, y);
	262	else
	263	chunk = raw_data.prepare_8bit(data, y);
	264
	265	MD5_Update(&md5, (void*)chunk.data, chunk.len);
184	266	}
185	267
186	268	MD5_Final(result, &md5);

216	298	case sei_decoded_picture_hash_type_MD5:
217	299	{
218	300	uint8_t md5[16];
219		compute_MD5_8bit(data,w,h,stride,md5);
	301	compute_MD5(data,w,h,stride,md5, img->get_bit_depth(i));
220	302
221	303	/*
222	304	fprintf(stderr,"computed MD5: ");

237	319
238	320	case sei_decoded_picture_hash_type_CRC:
239	321	{
240		uint16_t crc = compute_CRC_8bit_fast(data,w,h,stride);
	322	uint16_t crc = compute_CRC_8bit_fast(data,w,h,stride, img->get_bit_depth(i));
241	323
242	324	logtrace(LogSEI,"SEI decoded picture hash: %04x <-[%d]-> decoded picture: %04x\n",
243	325	seihash->crc[i], i, crc);

252	334
253	335	case sei_decoded_picture_hash_type_checksum:
254	336	{
255		uint32_t chksum = compute_checksum_8bit(data,w,h,stride);
	337	uint32_t chksum = compute_checksum_8bit(data,w,h,stride, img->get_bit_depth(i));
256	338
257	339	if (chksum != seihash->checksum[i]) {
258	340	fprintf(stderr,"SEI decoded picture hash: %04x, decoded picture: %04x (POC=%d)\n",

281	363	if (byte != 0xFF) { break; }
282	364	}
283	365
	366	//printf("SEI payload: %d\n",payload_type);
	367
284	368	int payload_size = 0;
285	369	for (;;)
286	370	{

334	418	case sei_payload_type_decoded_picture_hash:
335	419	if (img->decctx->param_sei_check_hash) {
336	420	err = process_sei_decoded_picture_hash(sei, img);
	421	if (err==DE265_OK) {
	422	//printf("SEI check ok\n");
	423	}
337	424	}
338	425
339	426	break;

405	492	return "unknown SEI message";
406	493	}
407	494	}
408

-4

libde265/sei.h less more

68	68	} sei_decoded_picture_hash;
69	69
70	70
71		struct sei_message {
	71	typedef struct {
72	72	enum sei_payload_type payload_type;
73	73	int payload_size;
74	74
75	75	union {
76	76	sei_decoded_picture_hash decoded_picture_hash;
77	77	} data;
78		};
79
	78	} sei_message;
80	79
81	80	class seq_parameter_set;
82	81

84	83
85	84	de265_error read_sei(bitreader* reader, sei_message, bool suffix, const seq_parameter_set sps);
86	85	void dump_sei(const sei_message, const seq_parameter_set sps);
87		de265_error process_sei(const sei_message, class de265_image img);
	86	de265_error process_sei(const sei_message, struct de265_image img);
88	87
89	88	#endif

+1681

-594

libde265/slice.cc less more

37	37	#define LOCK de265_mutex_lock(&ctx->thread_pool.mutex)
38	38	#define UNLOCK de265_mutex_unlock(&ctx->thread_pool.mutex)
39	39
40		extern bool read_short_term_ref_pic_set(decoder_context* ctx,
	40	extern bool read_short_term_ref_pic_set(error_queue* errqueue,
41	41	const seq_parameter_set* sps,
42	42	bitreader* br,
43	43	ref_pic_set* out_set,

48	48
49	49	void read_coding_tree_unit(thread_context* tctx);
50	50	void read_coding_quadtree(thread_context* tctx,
51		int xCtb, int yCtb,
	51	int xCtb, int yCtb,
52	52	int Log2CtbSizeY,
53	53	int ctDepth);
54		int check_CTB_available(de265_image* img,
55		slice_segment_header* shdr,
56		int xC,int yC, int xN,int yN);
57	54	/*
58	55	void decode_inter_block(decoder_context* ctx,thread_context* tctx,
59	56	int xC, int yC, int log2CbSize);
60	57	*/
	58
	59	void slice_segment_header::set_defaults()
	60	{
	61	slice_index = 0;
	62
	63	first_slice_segment_in_pic_flag = 1;
	64	no_output_of_prior_pics_flag = 0;
	65	slice_pic_parameter_set_id = 0;
	66	dependent_slice_segment_flag = 0;
	67	slice_segment_address = 0;
	68
	69	slice_type = SLICE_TYPE_I;
	70	pic_output_flag = 1;
	71	colour_plane_id = 0;
	72	slice_pic_order_cnt_lsb = 0;
	73	short_term_ref_pic_set_sps_flag = 1;
	74	// ref_pic_set slice_ref_pic_set;
	75
	76	short_term_ref_pic_set_idx = 0;
	77	num_long_term_sps = 0;
	78	num_long_term_pics = 0;
	79
	80	//uint8_t lt_idx_sps[MAX_NUM_REF_PICS];
	81	//int poc_lsb_lt[MAX_NUM_REF_PICS];
	82	//char used_by_curr_pic_lt_flag[MAX_NUM_REF_PICS];
	83
	84	//char delta_poc_msb_present_flag[MAX_NUM_REF_PICS];
	85	//int delta_poc_msb_cycle_lt[MAX_NUM_REF_PICS];
	86
	87	slice_temporal_mvp_enabled_flag = 0;
	88	slice_sao_luma_flag = 0;
	89	slice_sao_chroma_flag = 0;
	90
	91	num_ref_idx_active_override_flag = 0;
	92	num_ref_idx_l0_active=1; // [1;16]
	93	num_ref_idx_l1_active=1; // [1;16]
	94
	95	ref_pic_list_modification_flag_l0 = 0;
	96	ref_pic_list_modification_flag_l1 = 0;
	97	//uint8_t list_entry_l0[16];
	98	//uint8_t list_entry_l1[16];
	99
	100	mvd_l1_zero_flag = 0;
	101	cabac_init_flag = 0;
	102	collocated_from_l0_flag = 0;
	103	collocated_ref_idx = 0;
	104
	105	// --- pred_weight_table ---
	106
	107	luma_log2_weight_denom=0; // [0;7]
	108	ChromaLog2WeightDenom=0; // [0;7]
	109
	110	// first index is L0/L1
	111	/*
	112	uint8_t luma_weight_flag[2][16]; // bool
	113	uint8_t chroma_weight_flag[2][16]; // bool
	114	int16_t LumaWeight[2][16];
	115	int8_t luma_offset[2][16];
	116	int16_t ChromaWeight[2][16][2];
	117	int8_t ChromaOffset[2][16][2];
	118	*/
	119
	120
	121	five_minus_max_num_merge_cand = 0;
	122	slice_qp_delta = 0;
	123
	124	slice_cb_qp_offset = 0;
	125	slice_cr_qp_offset = 0;
	126
	127	cu_chroma_qp_offset_enabled_flag = 0;
	128
	129	deblocking_filter_override_flag = 0;
	130	slice_deblocking_filter_disabled_flag = 0;
	131	slice_beta_offset=0; // = pps->beta_offset if undefined
	132	slice_tc_offset=0; // = pps->tc_offset if undefined
	133
	134	slice_loop_filter_across_slices_enabled_flag = 0;
	135
	136	num_entry_point_offsets = 0;
	137	//int offset_len;
	138	//std::vector<int> entry_point_offset;
	139
	140	slice_segment_header_extension_length = 0;
	141
	142	SliceAddrRS = slice_segment_address;
	143	}
	144
61	145
62	146	bool read_pred_weight_table(bitreader* br, slice_segment_header* shdr, decoder_context* ctx)
63	147	{

110	194	// luma_offset
111	195
112	196	vlc = get_svlc(br);
113		if (vlc < -128 \|\| vlc > 127) return false;
	197	if (vlc < -sps->WpOffsetHalfRangeY \|\| vlc > sps->WpOffsetHalfRangeY-1) return false;
114	198	shdr->luma_offset[l][i] = vlc;
115	199	}
116	200	else {

123	207	// delta_chroma_weight
124	208
125	209	vlc = get_svlc(br);
126		if (vlc < -128 \|\| vlc > 127) return false;
	210	if (vlc < -128 \|\| vlc > 127) return false;
127	211
128	212	shdr->ChromaWeight[l][i][j] = (1<<shdr->ChromaLog2WeightDenom) + vlc;
129	213
130	214	// delta_chroma_offset
131	215
132	216	vlc = get_svlc(br);
133		if (vlc < -512 \|\| vlc > 511) return false;
134
135		vlc = Clip3(-128,127, (vlc-((128*shdr->ChromaWeight[l][i][j])
136		>> shdr->ChromaLog2WeightDenom) + 128));
	217	if (vlc < -4*sps->WpOffsetHalfRangeC \|\|
	218	vlc > 4*sps->WpOffsetHalfRangeC-1) return false;
	219
	220	vlc = Clip3(-sps->WpOffsetHalfRangeC,
	221	sps->WpOffsetHalfRangeC-1,
	222	(sps->WpOffsetHalfRangeC
	223	+vlc
	224	-((sps->WpOffsetHalfRangeC*shdr->ChromaWeight[l][i][j])
	225	>> shdr->ChromaLog2WeightDenom)));
137	226
138	227	shdr->ChromaOffset[l][i][j] = vlc;
139	228	}

152	241	}
153	242
154	243
	244	void slice_segment_header::reset()
	245	{
	246	slice_index = 0;
	247
	248	first_slice_segment_in_pic_flag = 0;
	249	no_output_of_prior_pics_flag = 0;
	250	slice_pic_parameter_set_id = 0;
	251	dependent_slice_segment_flag = 0;
	252	slice_segment_address = 0;
	253
	254	slice_type = 0;
	255	pic_output_flag = 0;
	256	colour_plane_id = 0;
	257	slice_pic_order_cnt_lsb = 0;
	258	short_term_ref_pic_set_sps_flag = 0;
	259	slice_ref_pic_set.reset();
	260
	261	short_term_ref_pic_set_idx = 0;
	262	num_long_term_sps = 0;
	263	num_long_term_pics= 0;
	264
	265	for (int i=0;i<MAX_NUM_REF_PICS;i++) {
	266	lt_idx_sps[i] = 0;
	267	poc_lsb_lt[i] = 0;
	268	used_by_curr_pic_lt_flag[i] = 0;
	269	delta_poc_msb_present_flag[i] = 0;
	270	delta_poc_msb_cycle_lt[i] = 0;
	271	}
	272
	273	slice_temporal_mvp_enabled_flag = 0;
	274	slice_sao_luma_flag = 0;
	275	slice_sao_chroma_flag = 0;
	276
	277	num_ref_idx_active_override_flag = 0;
	278	num_ref_idx_l0_active = 0;
	279	num_ref_idx_l1_active = 0;
	280
	281	ref_pic_list_modification_flag_l0 = 0;
	282	ref_pic_list_modification_flag_l1 = 0;
	283	for (int i=0;i<16;i++) {
	284	list_entry_l0[i] = 0;
	285	list_entry_l1[i] = 0;
	286	}
	287
	288	mvd_l1_zero_flag = 0;
	289	cabac_init_flag = 0;
	290	collocated_from_l0_flag = 0;
	291	collocated_ref_idx = 0;
	292
	293	luma_log2_weight_denom = 0;
	294	ChromaLog2WeightDenom = 0;
	295
	296	for (int i=0;i<2;i++)
	297	for (int j=0;j<16;j++) {
	298	luma_weight_flag[i][j] = 0;
	299	chroma_weight_flag[i][j] = 0;
	300	LumaWeight[i][j] = 0;
	301	luma_offset[i][j] = 0;
	302	ChromaWeight[i][j][0] = ChromaWeight[i][j][1] = 0;
	303	ChromaOffset[i][j][0] = ChromaOffset[i][j][1] = 0;
	304	}
	305
	306	five_minus_max_num_merge_cand = 0;
	307	slice_qp_delta = 0;
	308
	309	slice_cb_qp_offset = 0;
	310	slice_cr_qp_offset = 0;
	311
	312	cu_chroma_qp_offset_enabled_flag = 0;
	313
	314	deblocking_filter_override_flag = 0;
	315	slice_deblocking_filter_disabled_flag = 0;
	316	slice_beta_offset = 0;
	317	slice_tc_offset = 0;
	318
	319	slice_loop_filter_across_slices_enabled_flag = 0;
	320
	321	num_entry_point_offsets = 0;
	322	offset_len = 0;
	323	entry_point_offset.clear();
	324
	325	slice_segment_header_extension_length = 0;
	326
	327	SliceAddrRS = 0;
	328	SliceQPY = 0;
	329
	330	initType = 0;
	331
	332	MaxNumMergeCand = 0;
	333	CurrRpsIdx = 0;
	334	CurrRps.reset();
	335	NumPocTotalCurr = 0;
	336
	337	for (int i=0;i<2;i++)
	338	for (int j=0;j<MAX_NUM_REF_PICS;j++) {
	339	RefPicList[i][j] = 0;
	340	RefPicList_POC[i][j] = 0;
	341	RefPicList_PicState[i][j] = 0;
	342	LongTermRefPic[i][j] = 0;
	343	}
	344
	345	//context_model ctx_model_storage[CONTEXT_MODEL_TABLE_LENGTH];
	346
	347	RemoveReferencesList.clear();
	348
	349	ctx_model_storage_defined = false;
	350	}
	351
	352
155	353	de265_error slice_segment_header::read(bitreader* br, decoder_context* ctx,
156	354	bool* continueDecoding)
157	355	{
158	356	*continueDecoding = false;
	357	reset();
159	358
160	359	// set defaults
161	360

206	405	return DE265_OK;
207	406	}
208	407
	408	if (ctx->previous_slice_header == NULL) {
	409	return DE265_ERROR_NO_INITIAL_SLICE_HEADER;
	410	}
	411
209	412	this = ctx->previous_slice_header;
210	413
211	414	first_slice_segment_in_pic_flag = 0;

219	422	}
220	423
221	424	if (slice_segment_address < 0 \|\|
222		slice_segment_address > sps->PicSizeInCtbsY) {
	425	slice_segment_address >= sps->PicSizeInCtbsY) {
223	426	ctx->add_warning(DE265_WARNING_SLICE_SEGMENT_ADDRESS_INVALID, false);
224	427	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
225	428	}
226	429
	430	//printf("SLICE %d (%d)\n",slice_segment_address, sps->PicSizeInCtbsY);
227	431
228	432
229	433	if (!dependent_slice_segment_flag) {
230	434	for (int i=0; i<pps->num_extra_slice_header_bits; i++) {
231	435	//slice_reserved_undetermined_flag[i]
232		skip_bits(br,1);
	436	skip_bits(br,1);
233	437	}
234	438
235	439	slice_type = get_uvlc(br);

248	452	}
249	453
250	454	if (sps->separate_colour_plane_flag == 1) {
251		colour_plane_id = get_bits(br,1);
	455	colour_plane_id = get_bits(br,2);
252	456	}
253	457
254	458

265	469	if (!short_term_ref_pic_set_sps_flag) {
266	470	read_short_term_ref_pic_set(ctx, sps,
267	471	br, &slice_ref_pic_set,
268		sps->num_short_term_ref_pic_sets,
	472	sps->num_short_term_ref_pic_sets(),
269	473	sps->ref_pic_sets,
270	474	true);
271	475
272		CurrRpsIdx = sps->num_short_term_ref_pic_sets;
	476	CurrRpsIdx = sps->num_short_term_ref_pic_sets();
273	477	CurrRps = slice_ref_pic_set;
274	478	}
275	479	else {
276		int nBits = ceil_log2(sps->num_short_term_ref_pic_sets);
	480	int nBits = ceil_log2(sps->num_short_term_ref_pic_sets());
277	481	if (nBits>0) short_term_ref_pic_set_idx = get_bits(br,nBits);
278	482	else short_term_ref_pic_set_idx = 0;
279	483
280		if (short_term_ref_pic_set_idx > sps->num_short_term_ref_pic_sets) {
	484	if (short_term_ref_pic_set_idx >= sps->num_short_term_ref_pic_sets()) {
281	485	ctx->add_warning(DE265_WARNING_SHORT_TERM_REF_PIC_SET_OUT_OF_RANGE, false);
282	486	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
283	487	}

292	496	if (sps->long_term_ref_pics_present_flag) {
293	497	if (sps->num_long_term_ref_pics_sps > 0) {
294	498	num_long_term_sps = get_uvlc(br);
	499	if (num_long_term_sps == UVLC_ERROR) {
	500	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	501	}
295	502	}
296	503	else {
297	504	num_long_term_sps = 0;
298	505	}
299	506
300	507	num_long_term_pics= get_uvlc(br);
301
	508	if (num_long_term_pics == UVLC_ERROR) {
	509	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	510	}
302	511
303	512	// check maximum number of reference frames
304	513

347	556	delta_poc_msb_present_flag[i] = get_bits(br,1);
348	557	if (delta_poc_msb_present_flag[i]) {
349	558	delta_poc_msb_cycle_lt[i] = get_uvlc(br);
	559	if (delta_poc_msb_cycle_lt[i]==UVLC_ERROR) {
	560	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	561	}
350	562	}
351	563	else {
352	564	delta_poc_msb_cycle_lt[i] = 0;

381	593
382	594
383	595	// --- SAO ---
384
	596
385	597	if (sps->sample_adaptive_offset_enabled_flag) {
386	598	slice_sao_luma_flag = get_bits(br,1);
387		slice_sao_chroma_flag = get_bits(br,1);
	599
	600	if (sps->ChromaArrayType != CHROMA_MONO) {
	601	slice_sao_chroma_flag = get_bits(br,1);
	602	}
	603	else {
	604	slice_sao_chroma_flag = 0;
	605	}
388	606	}
389	607	else {
390	608	slice_sao_luma_flag = 0;
391	609	slice_sao_chroma_flag = 0;
392	610	}
	611
	612	num_ref_idx_l0_active = 0;
	613	num_ref_idx_l1_active = 0;
393	614
394	615	if (slice_type == SLICE_TYPE_P \|\|
395	616	slice_type == SLICE_TYPE_B) {

416	637	num_ref_idx_l1_active = pps->num_ref_idx_l1_default_active;
417	638	}
418	639
	640	if (num_ref_idx_l0_active > 16) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	641	if (num_ref_idx_l1_active > 16) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	642
419	643	NumPocTotalCurr = CurrRps.NumPocTotalCurr_shortterm_only + NumLtPics;
420	644
421	645	if (pps->lists_modification_present_flag && NumPocTotalCurr > 1) {

474	698	else {
475	699	collocated_ref_idx = 0;
476	700	}
477		}
	701
	702	// check whether collocated_ref_idx points to a valid index
	703
	704	if (( collocated_from_l0_flag && collocated_ref_idx >= num_ref_idx_l0_active) \|\|
	705	(!collocated_from_l0_flag && collocated_ref_idx >= num_ref_idx_l1_active)) {
	706	ctx->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	707	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	708	}
	709	}
	710
478	711
479	712	if ((pps->weighted_pred_flag && slice_type == SLICE_TYPE_P) \|\|
480	713	(pps->weighted_bipred_flag && slice_type == SLICE_TYPE_B)) {

493	726	}
494	727	MaxNumMergeCand = 5-five_minus_max_num_merge_cand;
495	728	}
496
	729
497	730	slice_qp_delta = get_svlc(br);
498	731	if (slice_qp_delta == UVLC_ERROR) {
499	732	ctx->add_warning(DE265_WARNING_SLICEHEADER_INVALID, false);

517	750	else {
518	751	slice_cb_qp_offset = 0;
519	752	slice_cr_qp_offset = 0;
	753	}
	754
	755	if (pps->range_extension.chroma_qp_offset_list_enabled_flag) {
	756	cu_chroma_qp_offset_enabled_flag = get_bits(br,1);
520	757	}
521	758
522	759	if (pps->deblocking_filter_override_enabled_flag) {

569	806	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
570	807	}
571	808
	809	if (pps->entropy_coding_sync_enabled_flag) {
	810	// check num_entry_points for valid range
	811
	812	int firstCTBRow = slice_segment_address / sps->PicWidthInCtbsY;
	813	int lastCTBRow = firstCTBRow + num_entry_point_offsets;
	814	if (lastCTBRow >= sps->PicHeightInCtbsY) {
	815	ctx->add_warning(DE265_WARNING_SLICEHEADER_INVALID, false);
	816	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	817	}
	818	}
	819
	820	if (pps->tiles_enabled_flag) {
	821	if (num_entry_point_offsets > pps->num_tile_columns * pps->num_tile_rows) {
	822	ctx->add_warning(DE265_WARNING_SLICEHEADER_INVALID, false);
	823	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	824	}
	825	}
	826
572	827	entry_point_offset.resize( num_entry_point_offsets );
573	828
574	829	if (num_entry_point_offsets > 0) {

578	833	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
579	834	}
580	835	offset_len++;
	836
	837	if (offset_len > 32) {
	838	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	839	}
581	840
582	841	for (int i=0; i<num_entry_point_offsets; i++) {
583	842	{

601	860	ctx->add_warning(DE265_WARNING_SLICEHEADER_INVALID, false);
602	861	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
603	862	}
604
	863
605	864	for (int i=0; i<slice_segment_header_extension_length; i++) {
606	865	//slice_segment_header_extension_data_byte[i]
607	866	get_bits(br,8);

609	868	}
610	869
611	870
	871	compute_derived_values(pps);
	872
	873	*continueDecoding = true;
	874	return DE265_OK;
	875	}
	876
	877
	878	de265_error slice_segment_header::write(error_queue* errqueue, CABAC_encoder& out,
	879	const seq_parameter_set* sps,
	880	const pic_parameter_set* pps,
	881	uint8_t nal_unit_type)
	882	{
	883	out.write_bit(first_slice_segment_in_pic_flag);
	884
	885	if (isRapPic(nal_unit_type)) { // TODO: is this still correct ? Should we drop RapPicFlag ?
	886	out.write_bit(no_output_of_prior_pics_flag);
	887	}
	888
	889	if (slice_pic_parameter_set_id > DE265_MAX_PPS_SETS) {
	890	errqueue->add_warning(DE265_WARNING_NONEXISTING_PPS_REFERENCED, false);
	891	return DE265_OK;
	892	}
	893	out.write_uvlc(slice_pic_parameter_set_id);
	894
	895	if (!first_slice_segment_in_pic_flag) {
	896	if (pps->dependent_slice_segments_enabled_flag) {
	897	out.write_bit(dependent_slice_segment_flag);
	898	}
	899
	900	out.write_bits(slice_segment_address, ceil_log2(sps->PicSizeInCtbsY));
	901
	902	if (dependent_slice_segment_flag) {
	903	if (slice_segment_address == 0) {
	904	errqueue->add_warning(DE265_WARNING_DEPENDENT_SLICE_WITH_ADDRESS_ZERO, false);
	905	return DE265_OK;
	906	}
	907	}
	908	}
	909
	910	if (slice_segment_address < 0 \|\|
	911	slice_segment_address > sps->PicSizeInCtbsY) {
	912	errqueue->add_warning(DE265_WARNING_SLICE_SEGMENT_ADDRESS_INVALID, false);
	913	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	914	}
	915
	916
	917
	918	if (!dependent_slice_segment_flag) {
	919	for (int i=0; i<pps->num_extra_slice_header_bits; i++) {
	920	//slice_reserved_undetermined_flag[i]
	921	out.skip_bits(1);
	922	}
	923
	924	if (slice_type > 2) {
	925	errqueue->add_warning(DE265_WARNING_SLICEHEADER_INVALID, false);
	926	return DE265_OK;
	927	}
	928	out.write_uvlc(slice_type);
	929
	930	if (pps->output_flag_present_flag) {
	931	out.write_bit(pic_output_flag);
	932	}
	933
	934	if (sps->separate_colour_plane_flag == 1) {
	935	out.write_bits(colour_plane_id,2);
	936	}
	937
	938
	939	int NumLtPics = 0;
	940
	941	if (nal_unit_type != NAL_UNIT_IDR_W_RADL &&
	942	nal_unit_type != NAL_UNIT_IDR_N_LP) {
	943	out.write_bits(slice_pic_order_cnt_lsb, sps->log2_max_pic_order_cnt_lsb);
	944	out.write_bit(short_term_ref_pic_set_sps_flag);
	945
	946	if (!short_term_ref_pic_set_sps_flag) {
	947	/* TODO
	948	read_short_term_ref_pic_set(ctx, sps,
	949	br, &slice_ref_pic_set,
	950	sps->num_short_term_ref_pic_sets,
	951	sps->ref_pic_sets,
	952	true);
	953	*/
	954	//CurrRpsIdx = sps->num_short_term_ref_pic_sets;
	955	//CurrRps = slice_ref_pic_set;
	956	}
	957	else {
	958	int nBits = ceil_log2(sps->num_short_term_ref_pic_sets());
	959	if (nBits>0) out.write_bits(short_term_ref_pic_set_idx,nBits);
	960	else { assert(short_term_ref_pic_set_idx==0); }
	961
	962	if (short_term_ref_pic_set_idx > sps->num_short_term_ref_pic_sets()) {
	963	errqueue->add_warning(DE265_WARNING_SHORT_TERM_REF_PIC_SET_OUT_OF_RANGE, false);
	964	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	965	}
	966
	967	//CurrRpsIdx = short_term_ref_pic_set_idx;
	968	//CurrRps = sps->ref_pic_sets[CurrRpsIdx];
	969	}
	970
	971
	972	// --- long-term MC ---
	973
	974	if (sps->long_term_ref_pics_present_flag) {
	975	if (sps->num_long_term_ref_pics_sps > 0) {
	976	out.write_uvlc(num_long_term_sps);
	977	}
	978	else {
	979	assert(num_long_term_sps == 0);
	980	}
	981
	982	out.write_uvlc(num_long_term_pics);
	983
	984
	985	// check maximum number of reference frames
	986
	987	if (num_long_term_sps +
	988	num_long_term_pics +
	989	CurrRps.NumNegativePics +
	990	CurrRps.NumPositivePics
	991	> sps->sps_max_dec_pic_buffering[sps->sps_max_sub_layers-1])
	992	{
	993	errqueue->add_warning(DE265_WARNING_MAX_NUM_REF_PICS_EXCEEDED, false);
	994	return DE265_OK;
	995	}
	996
	997	for (int i=0; i<num_long_term_sps + num_long_term_pics; i++) {
	998	if (i < num_long_term_sps) {
	999	int nBits = ceil_log2(sps->num_long_term_ref_pics_sps);
	1000	out.write_bits(lt_idx_sps[i], nBits);
	1001
	1002	// check that the referenced lt-reference really exists
	1003
	1004	if (lt_idx_sps[i] >= sps->num_long_term_ref_pics_sps) {
	1005	errqueue->add_warning(DE265_NON_EXISTING_LT_REFERENCE_CANDIDATE_IN_SLICE_HEADER, false);
	1006	return DE265_OK;
	1007	}
	1008
	1009	//ctx->PocLsbLt[i] = sps->lt_ref_pic_poc_lsb_sps[ lt_idx_sps[i] ];
	1010	//ctx->UsedByCurrPicLt[i] = sps->used_by_curr_pic_lt_sps_flag[ lt_idx_sps[i] ];
	1011	}
	1012	else {
	1013	int nBits = sps->log2_max_pic_order_cnt_lsb;
	1014	out.write_bits(poc_lsb_lt[i], nBits);
	1015	out.write_bit(used_by_curr_pic_lt_flag[i]);
	1016
	1017	//ctx->PocLsbLt[i] = poc_lsb_lt[i];
	1018	//ctx->UsedByCurrPicLt[i] = used_by_curr_pic_lt_flag[i];
	1019	}
	1020
	1021	//if (ctx->UsedByCurrPicLt[i]) {
	1022	//NumLtPics++;
	1023	//}
	1024
	1025	out.write_bit(delta_poc_msb_present_flag[i]);
	1026	if (delta_poc_msb_present_flag[i]) {
	1027	out.write_uvlc(delta_poc_msb_cycle_lt[i]);
	1028	}
	1029	else {
	1030	assert(delta_poc_msb_cycle_lt[i] == 0);
	1031	}
	1032
	1033	/*
	1034	if (i==0 \|\| i==num_long_term_sps) {
	1035	ctx->DeltaPocMsbCycleLt[i] = delta_poc_msb_cycle_lt[i];
	1036	}
	1037	else {
	1038	ctx->DeltaPocMsbCycleLt[i] = (delta_poc_msb_cycle_lt[i] +
	1039	ctx->DeltaPocMsbCycleLt[i-1]);
	1040	}
	1041	*/
	1042	}
	1043	}
	1044	else {
	1045	assert(num_long_term_sps == 0);
	1046	assert(num_long_term_pics== 0);
	1047	}
	1048
	1049	if (sps->sps_temporal_mvp_enabled_flag) {
	1050	out.write_bit(slice_temporal_mvp_enabled_flag);
	1051	}
	1052	else {
	1053	assert(slice_temporal_mvp_enabled_flag == 0);
	1054	}
	1055	}
	1056	else {
	1057	assert(slice_pic_order_cnt_lsb == 0);
	1058	assert(num_long_term_sps == 0);
	1059	assert(num_long_term_pics== 0);
	1060	}
	1061
	1062
	1063	// --- SAO ---
	1064
	1065	if (sps->sample_adaptive_offset_enabled_flag) {
	1066	out.write_bit(slice_sao_luma_flag);
	1067	out.write_bit(slice_sao_chroma_flag);
	1068	}
	1069	else {
	1070	assert(slice_sao_luma_flag == 0);
	1071	assert(slice_sao_chroma_flag== 0);
	1072	}
	1073
	1074	if (slice_type == SLICE_TYPE_P \|\|
	1075	slice_type == SLICE_TYPE_B) {
	1076	out.write_bit(num_ref_idx_active_override_flag);
	1077
	1078	if (num_ref_idx_active_override_flag) {
	1079	out.write_uvlc(num_ref_idx_l0_active);
	1080	num_ref_idx_l0_active++;;
	1081
	1082	if (slice_type == SLICE_TYPE_B) {
	1083	out.write_uvlc(num_ref_idx_l1_active);
	1084	num_ref_idx_l1_active++;
	1085	}
	1086	}
	1087	else {
	1088	assert(num_ref_idx_l0_active == pps->num_ref_idx_l0_default_active);
	1089	assert(num_ref_idx_l1_active == pps->num_ref_idx_l1_default_active);
	1090	}
	1091
	1092	NumPocTotalCurr = CurrRps.NumPocTotalCurr_shortterm_only + NumLtPics;
	1093
	1094	if (pps->lists_modification_present_flag && NumPocTotalCurr > 1) {
	1095
	1096	int nBits = ceil_log2(NumPocTotalCurr);
	1097
	1098	out.write_bit(ref_pic_list_modification_flag_l0);
	1099	if (ref_pic_list_modification_flag_l0) {
	1100	for (int i=0;i<num_ref_idx_l0_active;i++) {
	1101	out.write_bits(list_entry_l0[i], nBits);
	1102	}
	1103	}
	1104
	1105	if (slice_type == SLICE_TYPE_B) {
	1106	out.write_bit(ref_pic_list_modification_flag_l1);
	1107	if (ref_pic_list_modification_flag_l1) {
	1108	for (int i=0;i<num_ref_idx_l1_active;i++) {
	1109	out.write_bits(list_entry_l1[i], nBits);
	1110	}
	1111	}
	1112	}
	1113	else {
	1114	assert(ref_pic_list_modification_flag_l1 == 0);
	1115	}
	1116	}
	1117	else {
	1118	assert(ref_pic_list_modification_flag_l0 == 0);
	1119	assert(ref_pic_list_modification_flag_l1 == 0);
	1120	}
	1121
	1122	if (slice_type == SLICE_TYPE_B) {
	1123	out.write_bit(mvd_l1_zero_flag);
	1124	}
	1125
	1126	if (pps->cabac_init_present_flag) {
	1127	out.write_bit(cabac_init_flag);
	1128	}
	1129	else {
	1130	assert(cabac_init_flag == 0);
	1131	}
	1132
	1133	if (slice_temporal_mvp_enabled_flag) {
	1134	if (slice_type == SLICE_TYPE_B)
	1135	out.write_bit(collocated_from_l0_flag);
	1136	else
	1137	{ assert(collocated_from_l0_flag == 1); }
	1138
	1139	if (( collocated_from_l0_flag && num_ref_idx_l0_active > 1) \|\|
	1140	(!collocated_from_l0_flag && num_ref_idx_l1_active > 1)) {
	1141	out.write_uvlc(collocated_ref_idx);
	1142	}
	1143	else {
	1144	assert(collocated_ref_idx == 0);
	1145	}
	1146	}
	1147
	1148	if ((pps->weighted_pred_flag && slice_type == SLICE_TYPE_P) \|\|
	1149	(pps->weighted_bipred_flag && slice_type == SLICE_TYPE_B)) {
	1150
	1151	assert(0);
	1152	/* TODO
	1153	if (!read_pred_weight_table(br,this,ctx))
	1154	{
	1155	ctx->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	1156	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	1157	}
	1158	*/
	1159	}
	1160
	1161	out.write_uvlc(five_minus_max_num_merge_cand);
	1162	//MaxNumMergeCand = 5-five_minus_max_num_merge_cand;
	1163	}
	1164
	1165	out.write_svlc(slice_qp_delta);
	1166
	1167	if (pps->pps_slice_chroma_qp_offsets_present_flag) {
	1168	out.write_svlc(slice_cb_qp_offset);
	1169	out.write_svlc(slice_cr_qp_offset);
	1170	}
	1171	else {
	1172	assert(slice_cb_qp_offset == 0);
	1173	assert(slice_cr_qp_offset == 0);
	1174	}
	1175
	1176	if (pps->deblocking_filter_override_enabled_flag) {
	1177	out.write_bit(deblocking_filter_override_flag);
	1178	}
	1179	else {
	1180	assert(deblocking_filter_override_flag == 0);
	1181	}
	1182
	1183	//slice_beta_offset = pps->beta_offset;
	1184	//slice_tc_offset = pps->tc_offset;
	1185
	1186	if (deblocking_filter_override_flag) {
	1187	out.write_bit(slice_deblocking_filter_disabled_flag);
	1188	if (!slice_deblocking_filter_disabled_flag) {
	1189	out.write_svlc(slice_beta_offset/2);
	1190	out.write_svlc(slice_tc_offset /2);
	1191	}
	1192	}
	1193	else {
	1194	assert(slice_deblocking_filter_disabled_flag == pps->pic_disable_deblocking_filter_flag);
	1195	}
	1196
	1197	if (pps->pps_loop_filter_across_slices_enabled_flag &&
	1198	(slice_sao_luma_flag \|\| slice_sao_chroma_flag \|\|
	1199	!slice_deblocking_filter_disabled_flag )) {
	1200	out.write_bit(slice_loop_filter_across_slices_enabled_flag);
	1201	}
	1202	else {
	1203	assert(slice_loop_filter_across_slices_enabled_flag ==
	1204	pps->pps_loop_filter_across_slices_enabled_flag);
	1205	}
	1206	}
	1207
	1208	if (pps->tiles_enabled_flag \|\| pps->entropy_coding_sync_enabled_flag ) {
	1209	out.write_uvlc(num_entry_point_offsets);
	1210
	1211	if (num_entry_point_offsets > 0) {
	1212	out.write_uvlc(offset_len-1);
	1213
	1214	for (int i=0; i<num_entry_point_offsets; i++) {
	1215	{
	1216	int prev=0;
	1217	if (i>0) prev = entry_point_offset[i-1];
	1218	out.write_bits(entry_point_offset[i]-prev-1, offset_len);
	1219	}
	1220	}
	1221	}
	1222	}
	1223	else {
	1224	assert(num_entry_point_offsets == 0);
	1225	}
	1226
	1227	if (pps->slice_segment_header_extension_present_flag) {
	1228	out.write_uvlc(slice_segment_header_extension_length);
	1229	if (slice_segment_header_extension_length > 1000) { // TODO: safety check against too large values
	1230	errqueue->add_warning(DE265_WARNING_SLICEHEADER_INVALID, false);
	1231	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	1232	}
	1233
	1234	for (int i=0; i<slice_segment_header_extension_length; i++) {
	1235	//slice_segment_header_extension_data_byte[i]
	1236	out.skip_bits(8);
	1237	}
	1238	}
	1239
	1240	return DE265_OK;
	1241	}
	1242
	1243	void slice_segment_header::compute_derived_values(const pic_parameter_set* pps)
	1244	{
612	1245	// --- init variables ---
613	1246
614	1247	SliceQPY = pps->pic_init_qp + slice_qp_delta;

620	1253	case SLICE_TYPE_B: initType = 2 - cabac_init_flag; break;
621	1254	}
622	1255
623		*continueDecoding = true;
624		return DE265_OK;
625		}
626
	1256	MaxNumMergeCand = 5-five_minus_max_num_merge_cand;
	1257	}
627	1258
628	1259
629	1260	//-----------------------------------------------------------------------

650	1281
651	1282
652	1283	LOG0("----------------- SLICE -----------------\n");
653		LOG1("first_slice_segment_in_pic_flag : %d\n", first_slice_segment_in_pic_flag);
	1284	LOG1("first_slice_segment_in_pic_flag : %d\n", first_slice_segment_in_pic_flag);
654	1285	if (ctx->get_nal_unit_type() >= NAL_UNIT_BLA_W_LP &&
655	1286	ctx->get_nal_unit_type() <= NAL_UNIT_RESERVED_IRAP_VCL23) {
656		LOG1("no_output_of_prior_pics_flag : %d\n", no_output_of_prior_pics_flag);
657		}
658
659		LOG1("slice_pic_parameter_set_id : %d\n", slice_pic_parameter_set_id);
	1287	LOG1("no_output_of_prior_pics_flag : %d\n", no_output_of_prior_pics_flag);
	1288	}
	1289
	1290	LOG1("slice_pic_parameter_set_id : %d\n", slice_pic_parameter_set_id);
660	1291
661	1292	if (!first_slice_segment_in_pic_flag) {
662		if (pps->dependent_slice_segments_enabled_flag) {
	1293	//if (pps->dependent_slice_segments_enabled_flag) {
663	1294	LOG1("dependent_slice_segment_flag : %d\n", dependent_slice_segment_flag);
664		}
	1295	//}
665	1296	LOG1("slice_segment_address : %d\n", slice_segment_address);
666	1297	}
667	1298

689	1320	LOG1("short_term_ref_pic_set_sps_flag : %d\n", short_term_ref_pic_set_sps_flag);
690	1321
691	1322	if (!short_term_ref_pic_set_sps_flag) {
692		LOG1("ref_pic_set[ %2d ]: ",sps->num_short_term_ref_pic_sets);
	1323	LOG1("ref_pic_set[ %2d ]: ",sps->num_short_term_ref_pic_sets());
693	1324	dump_compact_short_term_ref_pic_set(&slice_ref_pic_set, 16, fh);
694	1325	}
695		else if (sps->num_short_term_ref_pic_sets > 1) {
	1326	else if (sps->num_short_term_ref_pic_sets() > 1) {
696	1327	LOG1("short_term_ref_pic_set_idx : %d\n", short_term_ref_pic_set_idx);
697	1328	dump_compact_short_term_ref_pic_set(&sps->ref_pic_sets[short_term_ref_pic_set_idx], 16, fh);
698	1329	}

704	1335
705	1336	LOG1("num_long_term_pics : %d\n", num_long_term_pics);
706	1337
707		#if 0
	1338	#if 0
708	1339	for (int i=0; i<num_long_term_sps + num_long_term_pics; i++) {
709	1340	LOG2("PocLsbLt[%d] : %d\n", i, ctx->PocLsbLt[i]);
710	1341	LOG2("UsedByCurrPicLt[%d] : %d\n", i, ctx->UsedByCurrPicLt[i]);

717	1348	LOG1("slice_temporal_mvp_enabled_flag : %d\n", slice_temporal_mvp_enabled_flag);
718	1349	}
719	1350	}
720
	1351
721	1352
722	1353	if (sps->sample_adaptive_offset_enabled_flag) {
723	1354	LOG1("slice_sao_luma_flag : %d\n", slice_sao_luma_flag);

756	1387	if (slice_type == SLICE_TYPE_B) {
757	1388	LOG1("mvd_l1_zero_flag : %d\n", mvd_l1_zero_flag);
758	1389	}
759
	1390
760	1391	LOG1("cabac_init_flag : %d\n", cabac_init_flag);
761	1392
762	1393	if (slice_temporal_mvp_enabled_flag) {

852	1483	/*
853	1484	if( slice_segment_header_extension_present_flag ) {
854	1485	slice_segment_header_extension_length
855		for( i = 0; i < slice_segment_header_extension_length; i++)
	1486	for( i = 0; i < slice_segment_header_extension_length; i++)
856	1487	slice_segment_header_extension_data_byte[i]
857	1488	}
858	1489	byte_alignment()

869	1500
870	1501
871	1502
872
873
874		static void set_initValue(slice_segment_header* shdr,
875		context_model* model, int initValue)
876		{
877		int slopeIdx = initValue >> 4;
878		int intersecIdx = initValue & 0xF;
879		int m = slopeIdx*5 - 45;
880		int n = (intersecIdx<<3) - 16;
881		int preCtxState = Clip3(1,126, ((m*Clip3(0,51, shdr->SliceQPY))>>4)+n);
882
883		logtrace(LogSlice,"QP=%d slopeIdx=%d intersecIdx=%d m=%d n=%d\n",shdr->SliceQPY,slopeIdx,intersecIdx,m,n);
884
885		model->MPSbit=(preCtxState<=63) ? 0 : 1;
886		model->state = model->MPSbit ? (preCtxState-64) : (63-preCtxState);
887
888		// model state will always be between [0;62]
889
890		assert(model->state <= 62);
891		}
892
893
894		static const int initValue_split_cu_flag[3][3] = {
895		{ 139,141,157 },
896		{ 107,139,126 },
897		{ 107,139,126 },
898		};
899		static const int initValue_cu_skip_flag[2][3] = {
900		{ 197,185,201 },
901		{ 197,185,201 },
902		};
903		static const int initValue_part_mode[9] = { 184,154,139, 154,154,154, 139,154,154 };
904		static const int initValue_prev_intra_luma_pred_flag[3] = { 184,154,183 };
905		static const int initValue_intra_chroma_pred_mode[3] = { 63,152,152 };
906		static const int initValue_cbf_luma[4] = { 111,141,153,111 };
907		static const int initValue_cbf_chroma[12] = { 94,138,182,154,149,107,167,154,149,92,167,154 };
908		static const int initValue_split_transform_flag[9] = { 153,138,138, 124,138,94, 224,167,122 }; // FIX712
909		static const int initValue_last_significant_coefficient_prefix[54] = {
910		110,110,124,125,140,153,125,127,140,109,111,143,127,111, 79,108,123, 63,
911		125,110, 94,110, 95, 79,125,111,110, 78,110,111,111, 95, 94,108,123,108,
912		125,110,124,110, 95, 94,125,111,111, 79,125,126,111,111, 79,108,123, 93
913		};
914		static const int initValue_coded_sub_block_flag[12] = { 91,171,134,141,121,140,61,154,121,140,61,154 };
915		static const int initValue_significant_coeff_flag[3][42] = {
916		{
917		111, 111, 125, 110, 110, 94, 124, 108, 124, 107, 125, 141, 179, 153, 125, 107,
918		125, 141, 179, 153, 125, 107, 125, 141, 179, 153, 125, 140, 139, 182, 182, 152,
919		136, 152, 136, 153, 136, 139, 111, 136, 139, 111
920		},
921		{
922		155, 154, 139, 153, 139, 123, 123, 63, 153, 166, 183, 140, 136, 153, 154, 166,
923		183, 140, 136, 153, 154, 166, 183, 140, 136, 153, 154, 170, 153, 123, 123, 107,
924		121, 107, 121, 167, 151, 183, 140, 151, 183, 140,
925		},
926		{
927		170, 154, 139, 153, 139, 123, 123, 63, 124, 166, 183, 140, 136, 153, 154, 166,
928		183, 140, 136, 153, 154, 166, 183, 140, 136, 153, 154, 170, 153, 138, 138, 122,
929		121, 122, 121, 167, 151, 183, 140, 151, 183, 140
930		},
931		};
932		static const int initValue_coeff_abs_level_greater1_flag[72] = {
933		140, 92,137,138,140,152,138,139,153, 74,149, 92,139,107,122,152,
934		140,179,166,182,140,227,122,197,154,196,196,167,154,152,167,182,
935		182,134,149,136,153,121,136,137,169,194,166,167,154,167,137,182,
936		154,196,167,167,154,152,167,182,182,134,149,136,153,121,136,122,
937		169,208,166,167,154,152,167,182
938		};
939		static const int initValue_coeff_abs_level_greater2_flag[18] = {
940		138,153,136,167,152,152,107,167, 91,122,107,167,
941		107,167, 91,107,107,167
942		};
943		static const int initValue_sao_merge_leftUp_flag[3] = { 153,153,153 };
944		static const int initValue_sao_type_idx_lumaChroma_flag[3] = { 200,185,160 };
945		static const int initValue_cu_qp_delta_abs[2] = { 154,154 };
946		static const int initValue_transform_skip_flag[2] = { 139,139 };
947		static const int initValue_merge_flag[2] = { 110,154 };
948		static const int initValue_merge_idx[2] = { 122,137 };
949		static const int initValue_pred_mode_flag[2] = { 149,134 };
950		static const int initValue_abs_mvd_greater01_flag[4] = { 140,198,169,198 };
951		static const int initValue_mvp_lx_flag[1] = { 168 };
952		static const int initValue_rqt_root_cbf[1] = { 79 };
953		static const int initValue_ref_idx_lX[2] = { 153,153 };
954		static const int initValue_inter_pred_idc[5] = { 95,79,63,31,31 };
955		static const int initValue_cu_transquant_bypass_flag[3] = { 154,154,154 };
956
957
958		static void init_context(thread_context* tctx,
959		enum context_model_indices idx,
960		const int* initValues, int len)
961		{
962		for (int i=0;i<len;i++)
963		{
964		set_initValue(tctx->shdr,
965		&tctx->ctx_model[idx+i],
966		initValues[i]);
967		}
968		}
	1503	void initialize_CABAC_models(thread_context* tctx)
	1504	{
	1505	const int QPY = tctx->shdr->SliceQPY;
	1506	const int initType = tctx->shdr->initType;
	1507	assert(initType >= 0 && initType <= 2);
	1508
	1509	tctx->ctx_model.init(initType, QPY);
	1510
	1511	for (int i=0;i<4;i++) {
	1512	tctx->StatCoeff[i] = 0;
	1513	}
	1514	}
	1515
969	1516
970	1517
971	1518	static int decode_transform_skip_flag(thread_context* tctx, int cIdx)

976	1523
977	1524	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
978	1525	&tctx->ctx_model[CONTEXT_MODEL_TRANSFORM_SKIP_FLAG+context]);
	1526
	1527	logtrace(LogSymbols,"$1 transform_skip_flag=%d\n",bit);
	1528
979	1529	return bit;
980	1530	}
981	1531

985	1535	logtrace(LogSlice,"# sao_merge_left/up_flag\n");
986	1536	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
987	1537	&tctx->ctx_model[CONTEXT_MODEL_SAO_MERGE_FLAG]);
	1538
	1539	logtrace(LogSymbols,"$1 sao_merge_flag=%d\n",bit);
	1540
988	1541	return bit;
989	1542	}
990	1543

998	1551	&tctx->ctx_model[CONTEXT_MODEL_SAO_TYPE_IDX]);
999	1552
1000	1553	if (bit0==0) {
	1554	logtrace(LogSymbols,"$1 sao_type_idx=%d\n",0);
1001	1555	return 0;
1002	1556	}
1003	1557	else {
1004	1558	int bit1 = decode_CABAC_bypass(&tctx->cabac_decoder);
1005	1559	if (bit1==0) {
	1560	logtrace(LogSymbols,"$1 sao_type_idx=%d\n",1);
1006	1561	return 1;
1007	1562	}
1008	1563	else {
	1564	logtrace(LogSymbols,"$1 sao_type_idx=%d\n",2);
1009	1565	return 2;
1010	1566	}
1011	1567	}
1012	1568	}
1013	1569
1014	1570
1015		static int decode_sao_offset_abs(thread_context* tctx)
	1571	static int decode_sao_offset_abs(thread_context* tctx, int bitDepth)
1016	1572	{
1017	1573	logtrace(LogSlice,"# sao_offset_abs\n");
1018		int bitDepth = 8;
1019	1574	int cMax = (1<<(libde265_min(bitDepth,10)-5))-1;
1020	1575	int value = decode_CABAC_TU_bypass(&tctx->cabac_decoder, cMax);
	1576	logtrace(LogSymbols,"$1 sao_offset_abs=%d\n",value);
1021	1577	return value;
1022	1578	}
1023	1579

1026	1582	{
1027	1583	logtrace(LogSlice,"# sao_class\n");
1028	1584	int value = decode_CABAC_FL_bypass(&tctx->cabac_decoder, 2);
	1585	logtrace(LogSymbols,"$1 sao_class=%d\n",value);
1029	1586	return value;
1030	1587	}
1031	1588

1034	1591	{
1035	1592	logtrace(LogSlice,"# sao_offset_sign\n");
1036	1593	int value = decode_CABAC_bypass(&tctx->cabac_decoder);
	1594	logtrace(LogSymbols,"$1 sao_offset_sign=%d\n",value);
1037	1595	return value;
1038	1596	}
1039	1597

1042	1600	{
1043	1601	logtrace(LogSlice,"# sao_band_position\n");
1044	1602	int value = decode_CABAC_FL_bypass(&tctx->cabac_decoder,5);
	1603	logtrace(LogSymbols,"$1 sao_band_position=%d\n",value);
1045	1604	return value;
1046	1605	}
1047	1606

1051	1610	logtrace(LogSlice,"# cu_transquant_bypass_enable_flag\n");
1052	1611	int value = decode_CABAC_bit(&tctx->cabac_decoder,
1053	1612	&tctx->ctx_model[CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG]);
	1613	logtrace(LogSymbols,"$1 transquant_bypass_flag=%d\n",value);
1054	1614	return value;
1055	1615	}
1056	1616

1061	1621	static int decode_split_cu_flag(thread_context* tctx,
1062	1622	int x0, int y0, int ctDepth)
1063	1623	{
1064		//decoder_context* ctx = tctx->decctx;
1065
1066		if (x0==64 && y0==448) {
1067		//raise(SIGINT);
1068		}
1069
1070	1624	// check if neighbors are available
1071	1625
1072		int availableL = check_CTB_available(tctx->img,tctx->shdr, x0,y0, x0-1,y0);
1073		int availableA = check_CTB_available(tctx->img,tctx->shdr, x0,y0, x0,y0-1);
	1626	int availableL = check_CTB_available(tctx->img, x0,y0, x0-1,y0);
	1627	int availableA = check_CTB_available(tctx->img, x0,y0, x0,y0-1);
1074	1628
1075	1629	int condL = 0;
1076	1630	int condA = 0;

1089	1643
1090	1644	logtrace(LogSlice,"> split_cu_flag R=%x, ctx=%d, bit=%d\n", tctx->cabac_decoder.range,context,bit);
1091	1645
	1646	logtrace(LogSymbols,"$1 split_cu_flag=%d\n",bit);
	1647
1092	1648	return bit;
1093	1649	}
1094	1650

1100	1656
1101	1657	// check if neighbors are available
1102	1658
1103		int availableL = check_CTB_available(tctx->img,tctx->shdr, x0,y0, x0-1,y0);
1104		int availableA = check_CTB_available(tctx->img,tctx->shdr, x0,y0, x0,y0-1);
	1659	int availableL = check_CTB_available(tctx->img, x0,y0, x0-1,y0);
	1660	int availableA = check_CTB_available(tctx->img, x0,y0, x0,y0-1);
1105	1661
1106	1662	int condL = 0;
1107	1663	int condA = 0;

1120	1676
1121	1677	logtrace(LogSlice,"> cu_skip_flag R=%x, ctx=%d, bit=%d\n", tctx->cabac_decoder.range,context,bit);
1122	1678
	1679	logtrace(LogSymbols,"$1 cu_skip_flag=%d\n",bit);
	1680
1123	1681	return bit;
1124	1682	}
1125	1683

1136	1694
1137	1695	logtrace(LogSlice,"> %s\n",bit ? "2Nx2N" : "NxN");
1138	1696
	1697	logtrace(LogSymbols,"$1 part_mode=%d\n",bit ? PART_2Nx2N : PART_NxN);
	1698
1139	1699	return bit ? PART_2Nx2N : PART_NxN;
1140	1700	}
1141	1701	else {
1142	1702	int bit0 = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_PART_MODE+0]);
1143		if (bit0) { return PART_2Nx2N; }
	1703	if (bit0) { logtrace(LogSymbols,"$1 part_mode=%d\n",PART_2Nx2N); return PART_2Nx2N; }
1144	1704
1145	1705	// CHECK_ME: I optimize code and fix bug here, need more VERIFY!
1146	1706	int bit1 = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_PART_MODE+1]);
1147	1707	if (cLog2CbSize > img->sps.Log2MinCbSizeY) {
1148	1708	if (!img->sps.amp_enabled_flag) {
	1709	logtrace(LogSymbols,"$1 part_mode=%d\n",bit1 ? PART_2NxN : PART_Nx2N);
1149	1710	return bit1 ? PART_2NxN : PART_Nx2N;
1150	1711	}
1151	1712	else {
1152	1713	int bit3 = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_PART_MODE+3]);
1153	1714	if (bit3) {
	1715	logtrace(LogSymbols,"$1 part_mode=%d\n",bit1 ? PART_2NxN : PART_Nx2N);
1154	1716	return bit1 ? PART_2NxN : PART_Nx2N;
1155	1717	}
1156	1718
1157	1719	int bit4 = decode_CABAC_bypass(&tctx->cabac_decoder);
1158		if ( bit1 && bit4) return PART_2NxnD;
1159		if ( bit1 && !bit4) return PART_2NxnU;
1160		if (!bit1 && !bit4) return PART_nLx2N;
1161		if (!bit1 && bit4) return PART_nRx2N;
	1720	if ( bit1 && bit4) {
	1721	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_2NxnD);
	1722	return PART_2NxnD;
	1723	}
	1724	if ( bit1 && !bit4) {
	1725	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_2NxnU);
	1726	return PART_2NxnU;
	1727	}
	1728	if (!bit1 && !bit4) {
	1729	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_nLx2N);
	1730	return PART_nLx2N;
	1731	}
	1732	if (!bit1 && bit4) {
	1733	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_nRx2N);
	1734	return PART_nRx2N;
	1735	}
1162	1736	}
1163	1737	}
1164	1738	else {
1165	1739	// TODO, we could save one if here when first decoding the next bin and then
1166	1740	// checkcLog2CbSize==3 when it is '0'
1167	1741
1168		if (bit1) return PART_2NxN;
	1742	if (bit1) {
	1743	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_2NxN);
	1744	return PART_2NxN;
	1745	}
1169	1746
1170	1747	if (cLog2CbSize==3) {
	1748	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_Nx2N);
1171	1749	return PART_Nx2N;
1172	1750	}
1173	1751	else {
1174	1752	int bit2 = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_PART_MODE+2]);
	1753	logtrace(LogSymbols,"$1 part_mode=%d\n",PART_NxN-bit2);
1175	1754	return (enum PartMode)((int)PART_NxN - bit2)/bit2 ? PART_Nx2N : PART_NxN/;
1176	1755	}
1177	1756	}

1182	1761	}
1183	1762
1184	1763
1185		static int decode_prev_intra_luma_pred_flag(thread_context* tctx)
	1764	static inline int decode_prev_intra_luma_pred_flag(thread_context* tctx)
1186	1765	{
1187	1766	logtrace(LogSlice,"# prev_intra_luma_pred_flag\n");
1188	1767	int bit = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG]);
	1768	logtrace(LogSymbols,"$1 prev_intra_luma_pred_flag=%d\n",bit);
1189	1769	return bit;
1190	1770	}
1191	1771
1192	1772
1193		static int decode_mpm_idx(thread_context* tctx)
	1773	static inline int decode_mpm_idx(thread_context* tctx)
1194	1774	{
1195	1775	logtrace(LogSlice,"# mpm_idx (TU:2)\n");
1196	1776	int mpm = decode_CABAC_TU_bypass(&tctx->cabac_decoder, 2);
1197	1777	logtrace(LogSlice,"> mpm_idx = %d\n",mpm);
	1778	logtrace(LogSymbols,"$1 mpm_idx=%d\n",mpm);
1198	1779	return mpm;
1199	1780	}
1200	1781
1201	1782
1202		static int decode_rem_intra_luma_pred_mode(thread_context* tctx)
	1783	static inline int decode_rem_intra_luma_pred_mode(thread_context* tctx)
1203	1784	{
1204	1785	logtrace(LogSlice,"# rem_intra_luma_pred_mode (5 bits)\n");
1205		return decode_CABAC_FL_bypass(&tctx->cabac_decoder, 5);
	1786	int value = decode_CABAC_FL_bypass(&tctx->cabac_decoder, 5);
	1787	logtrace(LogSymbols,"$1 rem_intra_luma_pred_mode=%d\n",value);
	1788	return value;
1206	1789	}
1207	1790
1208	1791

1221	1804	}
1222	1805
1223	1806	logtrace(LogSlice,"> intra_chroma_pred_mode = %d\n",mode);
	1807	logtrace(LogSymbols,"$1 intra_chroma_pred_mode=%d\n",mode);
1224	1808
1225	1809	return mode;
1226	1810	}

1237	1821	logtrace(LogSlice,"# context: %d\n",context);
1238	1822
1239	1823	int bit = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG + context]);
	1824	logtrace(LogSymbols,"$1 split_transform_flag=%d\n",bit);
1240	1825	return bit;
1241	1826	}
1242	1827

1248	1833
1249	1834	int bit = decode_CABAC_bit(&tctx->cabac_decoder, &tctx->ctx_model[CONTEXT_MODEL_CBF_CHROMA + trafoDepth]);
1250	1835
	1836	logtrace(LogSymbols,"$1 cbf_chroma=%d\n",bit);
1251	1837	return bit;
1252	1838	}
1253	1839

1261	1847
1262	1848	logtrace(LogSlice,"> cbf_luma = %d\n",bit);
1263	1849
	1850	logtrace(LogSymbols,"$1 cbf_luma=%d\n",bit);
1264	1851	return bit;
1265	1852	}
1266	1853

1283	1870	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1284	1871	&tctx->ctx_model[CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG + ctxIdxInc]);
1285	1872
	1873	logtrace(LogSymbols,"$1 coded_sub_block_flag=%d\n",bit);
1286	1874	return bit;
1287	1875	}
1288	1876

1294	1882	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1295	1883	&tctx->ctx_model[CONTEXT_MODEL_CU_QP_DELTA_ABS + 0]);
1296	1884	if (bit==0) {
	1885	logtrace(LogSymbols,"$1 cu_qp_delta_abs=%d\n",0);
1297	1886	return 0;
1298	1887	}
1299	1888

1307	1896
1308	1897	if (prefix==5) {
1309	1898	int value = decode_CABAC_EGk_bypass(&tctx->cabac_decoder, 0);
	1899	logtrace(LogSymbols,"$1 cu_qp_delta_abs=%d\n",value+5);
1310	1900	return value + 5;
1311	1901	}
1312	1902	else {
	1903	logtrace(LogSymbols,"$1 cu_qp_delta_abs=%d\n",prefix);
1313	1904	return prefix;
1314	1905	}
1315	1906	}
1316	1907
1317
	1908
1318	1909	static int decode_last_significant_coeff_prefix(thread_context* tctx,
1319	1910	int log2TrafoSize,
1320	1911	int cIdx,

1404	1995	for (int scanIdx=0;scanIdx<2;scanIdx++) {
1405	1996	ctxIdxLookup[2][cIdx][scanIdx][prevCsbf] = p;
1406	1997	}
1407
	1998
1408	1999	p += 16*16;
1409	2000	}
1410	2001

1751	2342
1752	2343	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1753	2344	&tctx->ctx_model[CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG + ctxIdxInc]);
	2345
	2346	logtrace(LogSymbols,"$1 significant_coeff_flag=%d\n",bit);
	2347
1754	2348	return bit;
1755	2349	}
1756	2350

1819	2413	*lastInvocation_coeff_abs_level_greater1_flag = bit;
1820	2414	*lastInvocation_ctxSet = ctxSet;
1821	2415
	2416	//logtrace(LogSymbols,"$1 coeff_abs_level_greater1=%d\n",bit);
	2417
1822	2418	return bit;
1823	2419	}
1824	2420

1836	2432	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1837	2433	&tctx->ctx_model[CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG + ctxIdxInc]);
1838	2434
	2435	logtrace(LogSymbols,"$1 coeff_abs_level_greater2=%d\n",bit);
	2436
1839	2437	return bit;
1840	2438	}
1841	2439
	2440
	2441	#define MAX_PREFIX 64
1842	2442
1843	2443	static int decode_coeff_abs_level_remaining(thread_context* tctx,
1844	2444	int cRiceParam)

1850	2450	do {
1851	2451	prefix++;
1852	2452	codeword = decode_CABAC_bypass(&tctx->cabac_decoder);
	2453
	2454	if (prefix>MAX_PREFIX) {
	2455	return 0; // TODO: error
	2456	}
1853	2457	}
1854	2458	while (codeword);
1855	2459

1871	2475	value = (((1<<(prefix-3))+3-1)<<cRiceParam)+codeword;
1872	2476	}
1873	2477
	2478	logtrace(LogSymbols,"$1 coeff_abs_level_remaining=%d\n",value);
	2479
1874	2480	return value;
1875	2481	}
1876	2482

1882	2488	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1883	2489	&tctx->ctx_model[CONTEXT_MODEL_MERGE_FLAG]);
1884	2490
	2491	logtrace(LogSymbols,"$1 merge_flag=%d\n",bit);
	2492
1885	2493	return bit;
1886	2494	}
1887	2495

1889	2497	static int decode_merge_idx(thread_context* tctx)
1890	2498	{
1891	2499	logtrace(LogSlice,"# merge_idx\n");
	2500
	2501	if (tctx->shdr->MaxNumMergeCand <= 1) {
	2502	logtrace(LogSymbols,"$1 merge_idx=%d\n",0);
	2503	return 0;
	2504	}
1892	2505
1893	2506	// TU coding, first bin is CABAC, remaining are bypass.
1894	2507	// cMax = MaxNumMergeCand-1

1913	2526	}
1914	2527
1915	2528	logtrace(LogSlice,"> merge_idx = %d\n",idx);
	2529	logtrace(LogSymbols,"$1 merge_idx=%d\n",idx);
1916	2530
1917	2531	return idx;
1918	2532	}

1925	2539	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1926	2540	&tctx->ctx_model[CONTEXT_MODEL_PRED_MODE_FLAG]);
1927	2541
	2542	logtrace(LogSymbols,"$1 pred_mode=%d\n",bit);
1928	2543	return bit;
1929	2544	}
1930	2545

1935	2550	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1936	2551	&tctx->ctx_model[CONTEXT_MODEL_MVP_LX_FLAG]);
1937	2552
	2553	logtrace(LogSymbols,"$1 mvp_lx_flag=%d\n",bit);
1938	2554	return bit;
1939	2555	}
1940	2556

1945	2561	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
1946	2562	&tctx->ctx_model[CONTEXT_MODEL_RQT_ROOT_CBF]);
1947	2563
	2564	logtrace(LogSymbols,"$1 rqt_root_cbf=%d\n",bit);
1948	2565	return bit;
1949	2566	}
1950	2567

1979	2596
1980	2597	logtrace(LogSlice,"> ref_idx = %d\n",idx);
1981	2598
	2599	logtrace(LogSymbols,"$1 ref_idx_lX=%d\n",idx);
1982	2600	return idx;
1983	2601	}
1984	2602

2013	2631	logtrace(LogSlice,"> inter_pred_idc = %d (%s)\n",value,
2014	2632	value==0 ? "L0" : (value==1 ? "L1" : "BI"));
2015	2633
2016		return (enum InterPredIdc) value;
2017		}
2018
2019
2020
2021		void initialize_CABAC(thread_context* tctx)
2022		{
2023		const int initType = tctx->shdr->initType;
2024		assert(initType >= 0 && initType <= 2);
2025
2026		init_context(tctx, CONTEXT_MODEL_SPLIT_CU_FLAG, initValue_split_cu_flag[initType], 3);
2027		if (initType > 0) {
2028		init_context(tctx, CONTEXT_MODEL_CU_SKIP_FLAG, initValue_cu_skip_flag[initType-1], 3);
2029		init_context(tctx, CONTEXT_MODEL_PRED_MODE_FLAG, &initValue_pred_mode_flag[initType-1], 1);
2030		init_context(tctx, CONTEXT_MODEL_MERGE_FLAG, &initValue_merge_flag[initType-1],1);
2031		init_context(tctx, CONTEXT_MODEL_MERGE_IDX, &initValue_merge_idx[initType-1], 1);
2032		init_context(tctx, CONTEXT_MODEL_INTER_PRED_IDC, initValue_inter_pred_idc, 5);
2033		init_context(tctx, CONTEXT_MODEL_REF_IDX_LX, initValue_ref_idx_lX, 2);
2034		init_context(tctx, CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG, &initValue_abs_mvd_greater01_flag[initType == 1 ? 0 : 2], 2);
2035		init_context(tctx, CONTEXT_MODEL_MVP_LX_FLAG, initValue_mvp_lx_flag, 1);
2036		init_context(tctx, CONTEXT_MODEL_RQT_ROOT_CBF, initValue_rqt_root_cbf, 1);
2037		}
2038
2039		init_context(tctx, CONTEXT_MODEL_PART_MODE, &initValue_part_mode[(initType!=2 ? initType : 5)], 4);
2040		init_context(tctx, CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG, &initValue_prev_intra_luma_pred_flag[initType], 1);
2041		init_context(tctx, CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE, &initValue_intra_chroma_pred_mode[initType], 1);
2042		init_context(tctx, CONTEXT_MODEL_CBF_LUMA, &initValue_cbf_luma[initType == 0 ? 0 : 2], 2);
2043		init_context(tctx, CONTEXT_MODEL_CBF_CHROMA, &initValue_cbf_chroma[initType * 4], 4);
2044		init_context(tctx, CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG, &initValue_split_transform_flag[initType * 3], 3);
2045		init_context(tctx, CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX, &initValue_last_significant_coefficient_prefix[initType * 18], 18);
2046		init_context(tctx, CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX, &initValue_last_significant_coefficient_prefix[initType * 18], 18);
2047		init_context(tctx, CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG, &initValue_coded_sub_block_flag[initType * 4], 4);
2048		init_context(tctx, CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG, initValue_significant_coeff_flag[initType], 42);
2049		init_context(tctx, CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG, &initValue_coeff_abs_level_greater1_flag[initType * 24], 24);
2050		init_context(tctx, CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG, &initValue_coeff_abs_level_greater2_flag[initType * 6], 6);
2051		init_context(tctx, CONTEXT_MODEL_SAO_MERGE_FLAG, &initValue_sao_merge_leftUp_flag[initType], 1);
2052		init_context(tctx, CONTEXT_MODEL_SAO_TYPE_IDX, &initValue_sao_type_idx_lumaChroma_flag[initType], 1);
2053		init_context(tctx, CONTEXT_MODEL_CU_QP_DELTA_ABS, initValue_cu_qp_delta_abs, 2);
2054		init_context(tctx, CONTEXT_MODEL_TRANSFORM_SKIP_FLAG, initValue_transform_skip_flag, 2);
2055		init_context(tctx, CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG, &initValue_cu_transquant_bypass_flag[initType], 1);
2056		}
	2634	logtrace(LogSymbols,"$1 decode_inter_pred_idx=%d\n",value+1);
	2635
	2636	return (enum InterPredIdc) (value+1);
	2637	}
	2638
	2639
	2640	static int decode_explicit_rdpcm_flag(thread_context* tctx,int cIdx)
	2641	{
	2642	context_model* model = &tctx->ctx_model[CONTEXT_MODEL_RDPCM_FLAG];
	2643	int value = decode_CABAC_bit(&tctx->cabac_decoder, &model[cIdx ? 1 : 0]);
	2644	return value;
	2645	}
	2646
	2647
	2648	static int decode_explicit_rdpcm_dir(thread_context* tctx,int cIdx)
	2649	{
	2650	context_model* model = &tctx->ctx_model[CONTEXT_MODEL_RDPCM_DIR];
	2651	int value = decode_CABAC_bit(&tctx->cabac_decoder, &model[cIdx ? 1 : 0]);
	2652	return value;
	2653	}
	2654
2057	2655
2058	2656
2059	2657	/* Take CtbAddrInTS and compute

2134	2732	}
2135	2733
2136	2734	if (!sao_merge_up_flag && !sao_merge_left_flag) {
2137		for (int cIdx=0; cIdx<3; cIdx++) {
	2735	int nChroma = 3;
	2736	if (sps->ChromaArrayType == CHROMA_MONO) nChroma=1;
	2737
	2738	for (int cIdx=0; cIdx<nChroma; cIdx++) {
2138	2739	if ((shdr->slice_sao_luma_flag && cIdx==0) \|\|
2139	2740	(shdr->slice_sao_chroma_flag && cIdx>0)) {
2140	2741

2160	2761
2161	2762	if (SaoTypeIdx != 0) {
2162	2763	for (int i=0;i<4;i++) {
2163		saoinfo.saoOffsetVal[cIdx][i] = decode_sao_offset_abs(tctx);
	2764	saoinfo.saoOffsetVal[cIdx][i] = decode_sao_offset_abs(tctx, img->get_bit_depth(cIdx));
2164	2765	logtrace(LogSlice,"saoOffsetVal[%d][%d] = %d\n",cIdx,i, saoinfo.saoOffsetVal[cIdx][i]);
2165	2766	}
2166	2767

2195	2796	logtrace(LogSlice,"SaoEoClass[%d] = %d\n",cIdx,SaoEoClass);
2196	2797	}
2197	2798
2198		int bitDepth = (cIdx==0 ?
2199		sps->BitDepth_Y :
2200		sps->BitDepth_C);
2201		int shift = bitDepth-libde265_min(bitDepth,10);
	2799	int log2OffsetScale;
	2800
	2801	if (cIdx==0) {
	2802	log2OffsetScale = pps->range_extension.log2_sao_offset_scale_luma;
	2803	}
	2804	else {
	2805	log2OffsetScale = pps->range_extension.log2_sao_offset_scale_chroma;
	2806	}
2202	2807
2203	2808	for (int i=0;i<4;i++) {
2204		saoinfo.saoOffsetVal[cIdx][i] = sign[i]*(saoinfo.saoOffsetVal[cIdx][i] << shift);
	2809	saoinfo.saoOffsetVal[cIdx][i] = sign[i]*(saoinfo.saoOffsetVal[cIdx][i] << log2OffsetScale);
2205	2810	}
2206	2811	}
2207	2812	}

2251	2856	}
2252	2857
2253	2858
2254		LIBDE265_INLINE static int luma_pos_to_ctbAddrRS(seq_parameter_set* sps, int x,int y)
	2859	LIBDE265_INLINE static int luma_pos_to_ctbAddrRS(const seq_parameter_set* sps, int x,int y)
2255	2860	{
2256	2861	int ctbX = x >> sps->Log2CtbSizeY;
2257	2862	int ctbY = y >> sps->Log2CtbSizeY;

2260	2865	}
2261	2866
2262	2867
2263		int check_CTB_available(de265_image* img,
2264		slice_segment_header* shdr,
	2868	int check_CTB_available(const de265_image* img,
2265	2869	int xC,int yC, int xN,int yN)
2266	2870	{
2267	2871	// check whether neighbor is outside of frame

2294	2898
2295	2899	int residual_coding(thread_context* tctx,
2296	2900	int x0, int y0, // position of TU in frame
2297		int xL, int yL, // position of TU in local CU
2298	2901	int log2TrafoSize,
2299	2902	int cIdx)
2300	2903	{

2306	2909	const seq_parameter_set* sps = &img->sps;
2307	2910	const pic_parameter_set* pps = &img->pps;
2308	2911
	2912	enum PredMode PredMode = img->get_pred_mode(x0,y0);
2309	2913
2310	2914	if (cIdx==0) {
2311	2915	img->set_nonzero_coefficient(x0,y0,log2TrafoSize);

2314	2918
2315	2919	if (pps->transform_skip_enabled_flag &&
2316	2920	!tctx->cu_transquant_bypass_flag &&
2317		(log2TrafoSize==2))
	2921	(log2TrafoSize <= pps->Log2MaxTransformSkipSize))
2318	2922	{
2319	2923	tctx->transform_skip_flag[cIdx] = decode_transform_skip_flag(tctx,cIdx);
2320	2924	}

2322	2926	{
2323	2927	tctx->transform_skip_flag[cIdx] = 0;
2324	2928	}
	2929
	2930
	2931	tctx->explicit_rdpcm_flag = false;
	2932
	2933	if (PredMode == MODE_INTER && sps->range_extension.explicit_rdpcm_enabled_flag &&
	2934	( tctx->transform_skip_flag[cIdx] \|\| tctx->cu_transquant_bypass_flag))
	2935	{
	2936	tctx->explicit_rdpcm_flag = decode_explicit_rdpcm_flag(tctx,cIdx);
	2937	if (tctx->explicit_rdpcm_flag) {
	2938	tctx->explicit_rdpcm_dir = decode_explicit_rdpcm_dir(tctx,cIdx);
	2939	}
	2940
	2941	//printf("EXPLICIT RDPCM %d;%d\n",x0,y0);
	2942	}
	2943	else
	2944	{
	2945	tctx->explicit_rdpcm_flag = false;
	2946	}
	2947
	2948
	2949
	2950	// sbType for persistent_rice_adaptation_enabled_flag
	2951
	2952	int sbType = (cIdx==0) ? 2 : 0;
	2953	if (tctx->transform_skip_flag[cIdx] \|\| tctx->cu_transquant_bypass_flag) {
	2954	sbType++;
	2955	}
2325	2956
2326	2957
2327	2958	// --- decode position of last coded coefficient ---

2367	2998
2368	2999	int scanIdx;
2369	3000
2370		enum PredMode PredMode = img->get_pred_mode(x0,y0);
2371
2372
2373	3001	if (PredMode == MODE_INTRA) {
2374	3002	if (cIdx==0) {
2375		if (log2TrafoSize==2 \|\| log2TrafoSize==3) {
2376		enum IntraPredMode predMode = img->get_IntraPredMode(x0,y0);
2377		logtrace(LogSlice,"IntraPredMode[%d,%d] = %d\n",x0,y0,predMode);
2378
2379		if (predMode >= 6 && predMode <= 14) scanIdx=2;
2380		else if (predMode >= 22 && predMode <= 30) scanIdx=1;
2381		else scanIdx=0;
2382		}
2383		else { scanIdx=0; }
	3003	scanIdx = get_intra_scan_idx(log2TrafoSize, img->get_IntraPredMode(x0,y0), cIdx, sps);
2384	3004	}
2385	3005	else {
2386		if (log2TrafoSize==1 \|\| log2TrafoSize==2) {
2387		enum IntraPredMode predMode = tctx->IntraPredModeC;
2388
2389		if (predMode >= 6 && predMode <= 14) scanIdx=2;
2390		else if (predMode >= 22 && predMode <= 30) scanIdx=1;
2391		else scanIdx=0;
2392		}
2393		else { scanIdx=0; }
2394		}
2395
2396		logtrace(LogSlice,"pred: %d -> scan: %d\n",PredMode,scanIdx);
	3006	scanIdx = get_intra_scan_idx(log2TrafoSize, img->get_IntraPredModeC(x0,y0), cIdx, sps);
	3007	}
2397	3008	}
2398	3009	else {
2399	3010	scanIdx=0;
2400	3011	}
2401	3012
2402
2403		// HM 9 only ?
2404	3013	if (scanIdx==2) {
2405		int t = LastSignificantCoeffX;
2406		LastSignificantCoeffX = LastSignificantCoeffY;
2407		LastSignificantCoeffY = t;
	3014	std::swap(LastSignificantCoeffX, LastSignificantCoeffY);
2408	3015	}
2409	3016
2410	3017	logtrace(LogSlice,"LastSignificantCoeff: x=%d;y=%d\n",LastSignificantCoeffX,LastSignificantCoeffY);

2532	3139
2533	3140	// for all AC coefficients in sub-block, a significant_coeff flag is coded
2534	3141
2535		int significant_coeff = decode_significant_coeff_flag_lookup(tctx,
2536		ctxIdxMap[xC+(yC<<log2TrafoSize)]);
2537		//ctxIdxMap[(i<<4)+n]);
	3142	int ctxInc;
	3143	if (sps->range_extension.transform_skip_context_enabled_flag &&
	3144	(tctx->cu_transquant_bypass_flag \|\| tctx->transform_skip_flag[cIdx])) {
	3145	ctxInc = ( cIdx == 0 ) ? 42 : (16+27);
	3146	}
	3147	else {
	3148	ctxInc = ctxIdxMap[xC+(yC<<log2TrafoSize)];
	3149	}
	3150
	3151	int significant_coeff = decode_significant_coeff_flag_lookup(tctx, ctxInc);
2538	3152
2539	3153	if (significant_coeff) {
2540	3154	coeff_value[nCoefficients] = 1;

2555	3169	{
2556	3170	if (inferSbDcSigCoeffFlag==0) {
2557	3171	// if we cannot infert the DC coefficient, it is coded
2558		int significant_coeff = decode_significant_coeff_flag_lookup(tctx,
2559		ctxIdxMap[x0+(y0<<log2TrafoSize)]);
2560		//ctxIdxMap[(i<<4)+0]);
	3172
	3173	int ctxInc;
	3174	if (sps->range_extension.transform_skip_context_enabled_flag &&
	3175	(tctx->cu_transquant_bypass_flag \|\| tctx->transform_skip_flag[cIdx])) {
	3176	ctxInc = ( cIdx == 0 ) ? 42 : (16+27);
	3177	}
	3178	else {
	3179	ctxInc = ctxIdxMap[x0+(y0<<log2TrafoSize)];
	3180	}
	3181
	3182	int significant_coeff = decode_significant_coeff_flag_lookup(tctx, ctxInc);
2561	3183
2562	3184
2563	3185	if (significant_coeff) {

2580	3202
2581	3203
2582	3204	/*
2583		logtrace(LogSlice,"significant_coeff_flags:\n");
2584		for (int y=0;y<4;y++) {
	3205	logtrace(LogSlice,"significant_coeff_flags:\n");
	3206	for (int y=0;y<4;y++) {
2585	3207	logtrace(LogSlice," ");
2586	3208	for (int x=0;x<4;x++) {
2587		logtrace(LogSlice,"*%d ",significant_coeff_flag[y][x]);
	3209	logtrace(LogSlice,"*%d ",significant_coeff_flag[y][x]);
2588	3210	}
2589	3211	logtrace(LogSlice,"*\n");
2590		}
	3212	}
2591	3213	*/
2592	3214
2593	3215

2648	3270
2649	3271	// --- decode coefficient signs ---
2650	3272
2651		int signHidden = (coeff_scan_pos[0]-coeff_scan_pos[nCoefficients-1] > 3 &&
2652		!tctx->cu_transquant_bypass_flag);
	3273	int signHidden;
	3274
	3275
	3276	IntraPredMode predModeIntra;
	3277	if (cIdx==0) predModeIntra = img->get_IntraPredMode(x0,y0);
	3278	else predModeIntra = img->get_IntraPredModeC(x0,y0);
	3279
	3280
	3281	if (tctx->cu_transquant_bypass_flag \|\|
	3282	(PredMode == MODE_INTRA &&
	3283	sps->range_extension.implicit_rdpcm_enabled_flag &&
	3284	tctx->transform_skip_flag[cIdx] &&
	3285	( predModeIntra == 10 \|\| predModeIntra == 26 )) \|\|
	3286	tctx->explicit_rdpcm_flag)
	3287	{
	3288	signHidden = 0;
	3289	}
	3290	else
	3291	{
	3292	signHidden = (coeff_scan_pos[0]-coeff_scan_pos[nCoefficients-1] > 3);
	3293	}
	3294
2653	3295
2654	3296	for (int n=0;n<nCoefficients-1;n++) {
2655	3297	coeff_sign[n] = decode_CABAC_bypass(&tctx->cabac_decoder);

2669	3311	// --- decode coefficient value ---
2670	3312
2671	3313	int sumAbsLevel=0;
2672		int uiGoRiceParam=0;
	3314	int uiGoRiceParam;
	3315
	3316	if (sps->range_extension.persistent_rice_adaptation_enabled_flag==0) {
	3317	uiGoRiceParam = 0;
	3318	}
	3319	else {
	3320	uiGoRiceParam = tctx->StatCoeff[sbType]/4;
	3321	}
	3322
	3323	// printf("initial uiGoRiceParam=%d\n",uiGoRiceParam);
	3324	bool firstCoeffWithAbsLevelRemaining = true;
2673	3325
2674	3326	for (int n=0;n<nCoefficients;n++) {
2675	3327	int baseLevel = coeff_value[n];
2676	3328
2677	3329	int coeff_abs_level_remaining;
	3330
	3331	// printf("coeff %d/%d, uiRiceParam: %d\n",n,nCoefficients,uiGoRiceParam);
2678	3332
2679	3333	if (coeff_has_max_base_level[n]) {
2680	3334	coeff_abs_level_remaining =
2681	3335	decode_coeff_abs_level_remaining(tctx, uiGoRiceParam);
2682	3336
2683		// (9-462)
2684		if (baseLevel + coeff_abs_level_remaining > 3*(1<<uiGoRiceParam)) {
2685		uiGoRiceParam++;
2686		if (uiGoRiceParam>4) uiGoRiceParam=4;
	3337	if (sps->range_extension.persistent_rice_adaptation_enabled_flag == 0) {
	3338	// (2014.10 / 9-20)
	3339	if (baseLevel + coeff_abs_level_remaining > 3*(1<<uiGoRiceParam)) {
	3340	uiGoRiceParam++;
	3341	if (uiGoRiceParam>4) uiGoRiceParam=4;
	3342	}
2687	3343	}
	3344	else {
	3345	if (baseLevel + coeff_abs_level_remaining > 3*(1<<uiGoRiceParam))
	3346	uiGoRiceParam++;
	3347	}
	3348
	3349	// persistent_rice_adaptation_enabled_flag
	3350	if (sps->range_extension.persistent_rice_adaptation_enabled_flag &&
	3351	firstCoeffWithAbsLevelRemaining) {
	3352	if (coeff_abs_level_remaining >= (3 << (tctx->StatCoeff[sbType]/4 ))) {
	3353	tctx->StatCoeff[sbType]++;
	3354	}
	3355	else if (2*coeff_abs_level_remaining < (1 << (tctx->StatCoeff[sbType]/4 )) &&
	3356	tctx->StatCoeff[sbType] > 0) {
	3357	tctx->StatCoeff[sbType]--;
	3358	}
	3359	}
	3360
	3361	firstCoeffWithAbsLevelRemaining=false;
2688	3362	}
2689	3363	else {
2690	3364	coeff_abs_level_remaining = 0;
2691	3365	}
	3366
	3367	logtrace(LogSlice, "coeff_abs_level_remaining=%d\n",coeff_abs_level_remaining);
2692	3368
2693	3369
2694	3370	int16_t currCoeff = baseLevel + coeff_abs_level_remaining;

2703	3379	currCoeff = -currCoeff;
2704	3380	}
2705	3381	}
	3382
	3383	logtrace(LogSlice, "quantized coefficient=%d\n",currCoeff);
2706	3384
2707	3385	#ifdef DE265_LOG_TRACE
2708	3386	//TransCoeffLevel[yC*CoeffStride + xC] = currCoeff;

2716	3394	tctx->coeffList[cIdx][ tctx->nCoeff[cIdx] ] = currCoeff;
2717	3395	tctx->coeffPos [cIdx][ tctx->nCoeff[cIdx] ] = xC + yC*CoeffStride;
2718	3396	tctx->nCoeff[cIdx]++;
	3397
	3398	//printf("%d ",currCoeff);
2719	3399	} // iterate through coefficients in sub-block
	3400
	3401	//printf(" (%d;%d)\n",x0,y0);
	3402
2720	3403	} // if nonZero
2721	3404	} // next sub-block
2722	3405
2723	3406	return DE265_OK;
	3407	}
	3408
	3409
	3410	static void decode_TU(thread_context* tctx,
	3411	int x0,int y0,
	3412	int xCUBase,int yCUBase,
	3413	int nT, int cIdx, enum PredMode cuPredMode, bool cbf)
	3414	{
	3415	de265_image* img = tctx->img;
	3416
	3417	int residualDpcm = 0;
	3418
	3419	if (cuPredMode == MODE_INTRA) // if intra mode
	3420	{
	3421	enum IntraPredMode intraPredMode;
	3422
	3423	if (cIdx==0) {
	3424	intraPredMode = img->get_IntraPredMode(x0,y0);
	3425	}
	3426	else {
	3427	const int SubWidthC = tctx->img->sps.SubWidthC;
	3428	const int SubHeightC = tctx->img->sps.SubHeightC;
	3429
	3430	intraPredMode = img->get_IntraPredModeC(x0SubWidthC,y0SubHeightC);
	3431	}
	3432
	3433	if (intraPredMode<0 \|\| intraPredMode>=35) {
	3434	// TODO: ERROR
	3435	intraPredMode = INTRA_DC;
	3436	}
	3437
	3438	decode_intra_prediction(img, x0,y0, intraPredMode, nT, cIdx);
	3439
	3440
	3441	residualDpcm = tctx->img->sps.range_extension.implicit_rdpcm_enabled_flag &&
	3442	(tctx->cu_transquant_bypass_flag \|\| tctx->transform_skip_flag[cIdx]) &&
	3443	(intraPredMode == 10 \|\| intraPredMode == 26);
	3444
	3445	if (residualDpcm && intraPredMode == 26)
	3446	residualDpcm = 2;
	3447	}
	3448	else // INTER
	3449	{
	3450	if (tctx->explicit_rdpcm_flag) {
	3451	residualDpcm = (tctx->explicit_rdpcm_dir ? 2 : 1);
	3452	}
	3453	}
	3454
	3455	if (cbf) {
	3456	scale_coefficients(tctx, x0,y0, xCUBase,yCUBase, nT, cIdx,
	3457	tctx->transform_skip_flag[cIdx], cuPredMode==MODE_INTRA, residualDpcm);
	3458	}
	3459	/*
	3460	else if (!cbf && cIdx==0) {
	3461	memset(tctx->residual_luma,0,3232sizeof(int32_t));
	3462	}
	3463	*/
	3464	else if (!cbf && cIdx!=0 && tctx->ResScaleVal) {
	3465	// --- cross-component-prediction when CBF==0 ---
	3466
	3467	tctx->nCoeff[cIdx] = 0;
	3468	residualDpcm=0;
	3469
	3470	scale_coefficients(tctx, x0,y0, xCUBase,yCUBase, nT, cIdx,
	3471	tctx->transform_skip_flag[cIdx], cuPredMode==MODE_INTRA, residualDpcm);
	3472	}
	3473	}
	3474
	3475
	3476	static int decode_log2_res_scale_abs_plus1(thread_context* tctx, int cIdxMinus1)
	3477	{
	3478	//const int context = (cIdx==0) ? 0 : 1;
	3479
	3480	logtrace(LogSlice,"# log2_res_scale_abs_plus1 (c=%d)\n",cIdxMinus1);
	3481
	3482	int value = 0;
	3483	int cMax = 4;
	3484	for (int binIdx=0;binIdx<cMax;binIdx++)
	3485	{
	3486	int ctxIdxInc = 4*cIdxMinus1 + binIdx;
	3487
	3488	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
	3489	&tctx->ctx_model[CONTEXT_MODEL_LOG2_RES_SCALE_ABS_PLUS1+ctxIdxInc]);
	3490	if (!bit) break;
	3491	value++;
	3492	}
	3493
	3494	logtrace(LogSymbols,"$1 log2_res_scale_abs_plus1=%d\n",value);
	3495
	3496	return value;
	3497	}
	3498
	3499
	3500	static int decode_res_scale_sign_flag(thread_context* tctx, int cIdxMinus1)
	3501	{
	3502	//const int context = (cIdx==0) ? 0 : 1;
	3503
	3504	logtrace(LogSlice,"# res_scale_sign_flag (c=%d)\n",cIdxMinus1);
	3505
	3506	int bit = decode_CABAC_bit(&tctx->cabac_decoder,
	3507	&tctx->ctx_model[CONTEXT_MODEL_RES_SCALE_SIGN_FLAG+cIdxMinus1]);
	3508
	3509	logtrace(LogSymbols,"$1 res_scale_sign_flag=%d\n",bit);
	3510
	3511	return bit;
	3512	}
	3513
	3514
	3515	static void read_cross_comp_pred(thread_context* tctx, int cIdxMinus1)
	3516	{
	3517	int log2_res_scale_abs_plus1 = decode_log2_res_scale_abs_plus1(tctx,cIdxMinus1);
	3518	int ResScaleVal;
	3519
	3520	if (log2_res_scale_abs_plus1 != 0) {
	3521	int res_scale_sign_flag = decode_res_scale_sign_flag(tctx,cIdxMinus1);
	3522
	3523	ResScaleVal = 1 << (log2_res_scale_abs_plus1 - 1);
	3524	ResScaleVal = 1 - 2 res_scale_sign_flag;
	3525	}
	3526	else {
	3527	ResScaleVal = 0;
	3528	}
	3529
	3530	tctx->ResScaleVal = ResScaleVal;
2724	3531	}
2725	3532
2726	3533

2740	3547	assert(cbf_cr != -1);
2741	3548	assert(cbf_luma != -1);
2742	3549
	3550	const int ChromaArrayType = tctx->img->sps.ChromaArrayType;
	3551
	3552	int log2TrafoSizeC = (ChromaArrayType==CHROMA_444 ? log2TrafoSize : log2TrafoSize-1);
	3553	log2TrafoSizeC = libde265_max(2, log2TrafoSizeC);
	3554
	3555	const int cbfLuma = cbf_luma;
	3556	const int cbfChroma = cbf_cb \| cbf_cr;
	3557
2743	3558	tctx->transform_skip_flag[0]=0;
2744	3559	tctx->transform_skip_flag[1]=0;
2745	3560	tctx->transform_skip_flag[2]=0;
2746	3561
2747
2748		if (cbf_luma \|\| cbf_cb \|\| cbf_cr)
	3562	tctx->explicit_rdpcm_flag = false;
	3563
	3564
	3565	enum PredMode cuPredMode = tctx->img->get_pred_mode(x0,y0);
	3566
	3567	if (cbfLuma \|\| cbfChroma)
2749	3568	{
	3569	bool doDecodeQuantParameters = false;
	3570
2750	3571	if (tctx->img->pps.cu_qp_delta_enabled_flag &&
2751	3572	!tctx->IsCuQpDeltaCoded) {
2752	3573

2765	3586	logtrace(LogSlice,"cu_qp_delta_sign = %d\n",cu_qp_delta_sign);
2766	3587	logtrace(LogSlice,"CuQpDelta = %d\n",tctx->CuQpDelta);
2767	3588
	3589	doDecodeQuantParameters = true;
	3590	//decode_quantization_parameters(tctx, x0,y0, xCUBase, yCUBase);
	3591	}
	3592
	3593	if (tctx->shdr->cu_chroma_qp_offset_enabled_flag && cbfChroma &&
	3594	!tctx->cu_transquant_bypass_flag && !tctx->IsCuChromaQpOffsetCoded ) {
	3595	logtrace(LogSlice,"# cu_chroma_qp_offset_flag\n");
	3596
	3597	int cu_chroma_qp_offset_flag = decode_CABAC_bit(&tctx->cabac_decoder,
	3598	&tctx->ctx_model[CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_FLAG]);
	3599
	3600
	3601	const pic_parameter_set* pps = &tctx->img->pps;
	3602
	3603	int cu_chroma_qp_offset_idx = 0;
	3604	if (cu_chroma_qp_offset_flag && pps->range_extension.chroma_qp_offset_list_len > 1) {
	3605	cu_chroma_qp_offset_idx = decode_CABAC_bit(&tctx->cabac_decoder,
	3606	&tctx->ctx_model[CONTEXT_MODEL_CU_CHROMA_QP_OFFSET_IDX]);
	3607	}
	3608
	3609	tctx->IsCuChromaQpOffsetCoded = 1;
	3610
	3611	if (cu_chroma_qp_offset_flag) {
	3612	tctx->CuQpOffsetCb = pps->range_extension.cb_qp_offset_list[ cu_chroma_qp_offset_idx ];
	3613	tctx->CuQpOffsetCr = pps->range_extension.cr_qp_offset_list[ cu_chroma_qp_offset_idx ];
	3614	}
	3615	else {
	3616	tctx->CuQpOffsetCb = 0;
	3617	tctx->CuQpOffsetCr = 0;
	3618	}
	3619
	3620	doDecodeQuantParameters = true;
	3621	//decode_quantization_parameters(tctx, x0,y0, xCUBase, yCUBase);
	3622	}
	3623
	3624
	3625	if (doDecodeQuantParameters) {
2768	3626	decode_quantization_parameters(tctx, x0,y0, xCUBase, yCUBase);
2769	3627	}
2770	3628	}
2771	3629
2772
2773		if (cbf_luma \|\| cbf_cb \|\| cbf_cr)
	3630	// position of TU in local CU
	3631	int xL = x0 - xCUBase;
	3632	int yL = y0 - yCUBase;
	3633	int nT = 1<<log2TrafoSize;
	3634	int nTC = 1<<log2TrafoSizeC;
	3635
	3636	const int SubWidthC = tctx->img->sps.SubWidthC;
	3637	const int SubHeightC = tctx->img->sps.SubHeightC;
	3638
	3639	// --- luma ---
	3640
	3641	tctx->ResScaleVal = 0;
	3642
	3643	int err;
	3644	if (cbf_luma) {
	3645	if ((err=residual_coding(tctx,x0,y0, log2TrafoSize,0)) != DE265_OK) return err;
	3646	}
	3647
	3648	decode_TU(tctx, x0,y0, xCUBase,yCUBase, nT, 0, cuPredMode, cbf_luma);
	3649
	3650
	3651	// --- chroma ---
	3652
	3653	const int yOffset422 = 1<<log2TrafoSizeC;
	3654
	3655	if (log2TrafoSize>2 \|\| ChromaArrayType == CHROMA_444) {
	3656	// TODO: cross-component prediction
	3657
	3658	const bool do_cross_component_prediction =
	3659	(tctx->img->pps.range_extension.cross_component_prediction_enabled_flag &&
	3660	cbf_luma &&
	3661	(cuPredMode == MODE_INTER \|\| tctx->img->is_IntraPredModeC_Mode4(x0,y0)));
	3662
	3663	if (do_cross_component_prediction) {
	3664	read_cross_comp_pred(tctx, 0);
	3665	}
	3666	else {
	3667	tctx->ResScaleVal = 0;
	3668	}
	3669
2774	3670	{
2775		// position of TU in local CU
2776		int xL = x0 - xCUBase;
2777		int yL = y0 - yCUBase;
2778
2779		int err;
2780		if (cbf_luma) {
2781		if ((err=residual_coding(tctx,x0,y0, xL,yL,log2TrafoSize,0)) != DE265_OK) return err;
2782		}
2783
2784		if (log2TrafoSize>2) {
2785		if (cbf_cb) {
2786		if ((err=residual_coding(tctx,x0,y0,xL,yL,log2TrafoSize-1,1)) != DE265_OK) return err;
2787		}
2788
2789		if (cbf_cr) {
2790		if ((err=residual_coding(tctx,x0,y0,xL,yL,log2TrafoSize-1,2)) != DE265_OK) return err;
2791		}
2792		}
2793		else if (blkIdx==3) {
2794		if (cbf_cb) {
2795		if ((err=residual_coding(tctx,xBase,yBase,xBase-xCUBase,yBase-yCUBase,
2796		log2TrafoSize,1)) != DE265_OK) return err;
2797		}
2798
2799		if (cbf_cr) {
2800		if ((err=residual_coding(tctx,xBase,yBase,xBase-xCUBase,yBase-yCUBase,
2801		log2TrafoSize,2)) != DE265_OK) return err;
2802		}
2803		}
2804		}
	3671	if (cbf_cb & 1) {
	3672	if ((err=residual_coding(tctx,x0,y0,log2TrafoSizeC,1)) != DE265_OK) return err;
	3673	}
	3674
	3675	if (tctx->img->sps.ChromaArrayType != CHROMA_MONO) {
	3676	decode_TU(tctx,
	3677	x0/SubWidthC,y0/SubHeightC,
	3678	xCUBase/SubWidthC,yCUBase/SubHeightC, nTC, 1, cuPredMode, cbf_cb & 1);
	3679	}
	3680	}
	3681
	3682	// 4:2:2
	3683	if (ChromaArrayType == CHROMA_422) {
	3684	const int yOffset = 1<<log2TrafoSizeC;
	3685
	3686	if (cbf_cb & 2) {
	3687	if ((err=residual_coding(tctx,
	3688	x0,y0+yOffset*SubHeightC,
	3689	log2TrafoSizeC,1)) != DE265_OK) return err;
	3690	}
	3691
	3692	decode_TU(tctx,
	3693	x0/SubWidthC,y0/SubHeightC + yOffset,
	3694	xCUBase/SubWidthC,yCUBase/SubHeightC +yOffset,
	3695	nTC, 1, cuPredMode, cbf_cb & 2);
	3696	}
	3697
	3698
	3699	if (do_cross_component_prediction) {
	3700	read_cross_comp_pred(tctx, 1);
	3701	}
	3702	else {
	3703	tctx->ResScaleVal = 0;
	3704	}
	3705
	3706	{
	3707	if (cbf_cr & 1) {
	3708	if ((err=residual_coding(tctx,x0,y0,log2TrafoSizeC,2)) != DE265_OK) return err;
	3709	}
	3710
	3711	if (tctx->img->sps.ChromaArrayType != CHROMA_MONO) {
	3712	decode_TU(tctx,
	3713	x0/SubWidthC,y0/SubHeightC,
	3714	xCUBase/SubWidthC,yCUBase/SubHeightC,
	3715	nTC, 2, cuPredMode, cbf_cr & 1);
	3716	}
	3717	}
	3718
	3719	// 4:2:2
	3720	if (ChromaArrayType == CHROMA_422) {
	3721	const int yOffset = 1<<log2TrafoSizeC;
	3722
	3723	if (cbf_cr & 2) {
	3724	if ((err=residual_coding(tctx,
	3725	x0,y0+yOffset*SubHeightC,
	3726	log2TrafoSizeC,2)) != DE265_OK) return err;
	3727	}
	3728
	3729	decode_TU(tctx,
	3730	x0/SubWidthC,y0/SubHeightC+yOffset,
	3731	xCUBase/SubWidthC,yCUBase/SubHeightC+yOffset,
	3732	nTC, 2, cuPredMode, cbf_cr & 2);
	3733	}
	3734	}
	3735	else if (blkIdx==3) {
	3736	if (cbf_cb & 1) {
	3737	if ((err=residual_coding(tctx,xBase,yBase,
	3738	log2TrafoSize,1)) != DE265_OK) return err;
	3739	}
	3740
	3741	if (tctx->img->sps.ChromaArrayType != CHROMA_MONO) {
	3742	decode_TU(tctx,
	3743	xBase/SubWidthC, yBase/SubHeightC,
	3744	xCUBase/SubWidthC,yCUBase/SubHeightC, nT, 1, cuPredMode, cbf_cb & 1);
	3745	}
	3746
	3747	// 4:2:2
	3748	if (cbf_cb & 2) {
	3749	if ((err=residual_coding(tctx,
	3750	xBase ,yBase +(1<<log2TrafoSize),
	3751	log2TrafoSize,1)) != DE265_OK) return err;
	3752	}
	3753
	3754	if (ChromaArrayType == CHROMA_422) {
	3755	decode_TU(tctx,
	3756	xBase/SubWidthC, yBase/SubHeightC + (1<<log2TrafoSize),
	3757	xCUBase/SubWidthC,yCUBase/SubHeightC, nT, 1, cuPredMode, cbf_cb & 2);
	3758	}
	3759
	3760	if (cbf_cr & 1) {
	3761	if ((err=residual_coding(tctx,xBase,yBase,
	3762	log2TrafoSize,2)) != DE265_OK) return err;
	3763	}
	3764
	3765	if (tctx->img->sps.ChromaArrayType != CHROMA_MONO) {
	3766	decode_TU(tctx,
	3767	xBase/SubWidthC, yBase/SubHeightC,
	3768	xCUBase/SubWidthC,yCUBase/SubHeightC, nT, 2, cuPredMode, cbf_cr & 1);
	3769	}
	3770
	3771	// 4:2:2
	3772	if (cbf_cr & 2) {
	3773	if ((err=residual_coding(tctx,
	3774	xBase ,yBase +(1<<log2TrafoSizeC),
	3775	log2TrafoSize,2)) != DE265_OK) return err;
	3776	}
	3777
	3778	if (ChromaArrayType == CHROMA_422) {
	3779	decode_TU(tctx,
	3780	xBase/SubWidthC, yBase/SubHeightC + (1<<log2TrafoSize),
	3781	xCUBase/SubWidthC,yCUBase/SubHeightC, nT, 2, cuPredMode, cbf_cr & 2);
	3782	}
	3783	}
	3784
2805	3785
2806	3786	return DE265_OK;
	3787	}
	3788
	3789
	3790	static void dump_cbsize(de265_image* img)
	3791	{
	3792	int w = img->get_width(0);
	3793	int h = img->get_height(0);
	3794
	3795	for (int y=0;y<h;y+=8) {
	3796	for (int x=0;x<w;x+=8) {
	3797	printf("%d",img->get_log2CbSize(x,y));
	3798	}
	3799	printf("\n");
	3800	}
2807	3801	}
2808	3802
2809	3803

2817	3811	int MaxTrafoDepth,
2818	3812	int IntraSplitFlag,
2819	3813	enum PredMode cuPredMode,
2820		bool parent_cbf_cb,bool parent_cbf_cr)
	3814	uint8_t parent_cbf_cb,uint8_t parent_cbf_cr)
2821	3815	{
2822	3816	logtrace(LogSlice,"- read_transform_tree (interleaved) x0:%d y0:%d xBase:%d yBase:%d "
2823		"log2TrafoSize:%d trafoDepth:%d MaxTrafoDepth:%d\n",
2824		x0,y0,xBase,yBase,log2TrafoSize,trafoDepth,MaxTrafoDepth);
	3817	"log2TrafoSize:%d trafoDepth:%d MaxTrafoDepth:%d parent-cbf-cb:%d parent-cbf-cr:%d\n",
	3818	x0,y0,xBase,yBase,log2TrafoSize,trafoDepth,MaxTrafoDepth,parent_cbf_cb,parent_cbf_cr);
2825	3819
2826	3820	de265_image* img = tctx->img;
2827	3821	const seq_parameter_set* sps = &img->sps;
2828	3822
	3823	int split_transform_flag;
	3824
2829	3825	enum PredMode PredMode = img->get_pred_mode(x0,y0);
2830		enum PartMode PartMode = img->get_PartMode(x0,y0);
2831
2832		int split_transform_flag;
2833
2834		int interSplitFlag= (sps->max_transform_hierarchy_depth_inter==0 &&
2835		PredMode == MODE_INTER &&
2836		PartMode != PART_2Nx2N &&
2837		trafoDepth == 0);
2838
	3826	assert(PredMode == cuPredMode);
2839	3827
2840	3828	/* If TrafoSize is larger than maximum size -> split automatically
2841	3829	If TrafoSize is at minimum size -> do not split

2852	3840	}
2853	3841	else
2854	3842	{
	3843	enum PartMode PartMode = img->get_PartMode(x0,y0);
	3844
	3845	int interSplitFlag= (sps->max_transform_hierarchy_depth_inter==0 &&
	3846	trafoDepth == 0 &&
	3847	PredMode == MODE_INTER &&
	3848	PartMode != PART_2Nx2N);
	3849
2855	3850	split_transform_flag = (log2TrafoSize > sps->Log2MaxTrafoSize \|\|
2856	3851	(IntraSplitFlag==1 && trafoDepth==0) \|\|
2857	3852	interSplitFlag==1) ? 1:0;
2858	3853	}
2859	3854
2860
2861	3855	if (split_transform_flag) {
2862	3856	logtrace(LogSlice,"set_split_transform_flag(%d,%d, %d)\n",x0,y0,trafoDepth);
2863	3857	img->set_split_transform_flag(x0,y0,trafoDepth);
2864	3858	}
2865	3859
2866
2867	3860	int cbf_cb=-1;
2868	3861	int cbf_cr=-1;
2869	3862
2870		if (log2TrafoSize>2) {
	3863	// CBF_CB/CR flags are encoded like this:
	3864	// 4:2:0 and 4:4:4 modes: binary flag in bit 0
	3865	// 4:2:2 mode: bit 0: top block, bit 1: bottom block
	3866
	3867	if ((log2TrafoSize>2 && sps->ChromaArrayType != CHROMA_MONO) \|\|
	3868	sps->ChromaArrayType == CHROMA_444) {
2871	3869	// we do not have to test for trafoDepth==0, because parent_cbf_cb is 1 at depth 0
2872	3870	if (/trafoDepth==0 \|\|/ parent_cbf_cb) {
2873	3871	cbf_cb = decode_cbf_chroma(tctx,trafoDepth);
	3872
	3873	if (sps->ChromaArrayType == CHROMA_422 && (!split_transform_flag \|\| log2TrafoSize==3)) {
	3874	cbf_cb \|= (decode_cbf_chroma(tctx,trafoDepth) << 1);
	3875	}
2874	3876	}
2875	3877
2876	3878	// we do not have to test for trafoDepth==0, because parent_cbf_cb is 1 at depth 0
2877	3879	if (/trafoDepth==0 \|\|/ parent_cbf_cr) {
2878	3880	cbf_cr = decode_cbf_chroma(tctx,trafoDepth);
2879		}
2880		}
2881
	3881
	3882	if (sps->ChromaArrayType == CHROMA_422 && (!split_transform_flag \|\| log2TrafoSize==3)) {
	3883	cbf_cr \|= (decode_cbf_chroma(tctx,trafoDepth) << 1);
	3884	}
	3885	}
	3886	}
	3887
	3888	//printf("CBF: cb:%d cr:%d\n",cbf_cb,cbf_cr);
2882	3889
2883	3890	// cbf_cr/cbf_cb not present in bitstream -> induce values
2884	3891
2885	3892	if (cbf_cb<0) {
	3893	assert(!(trafoDepth==0 && log2TrafoSize==2));
	3894
	3895	/* The standard specifies to check trafoDepth>0 AND log2TrafoSize==2.
	3896	However, I think that trafoDepth>0 is redundant as a CB is always
	3897	at least 8x8 and hence trafoDepth>0.
	3898	*/
	3899
2886	3900	if (trafoDepth>0 && log2TrafoSize==2) {
2887	3901	cbf_cb = parent_cbf_cb;
2888	3902	} else {

2914	3928	MaxTrafoDepth,IntraSplitFlag, cuPredMode, cbf_cb,cbf_cr);
2915	3929	}
2916	3930	else {
2917		int cbf_luma=1;
	3931	int cbf_luma;
2918	3932
2919	3933	if (PredMode==MODE_INTRA \|\| trafoDepth!=0 \|\| cbf_cb \|\| cbf_cr) {
2920	3934	cbf_luma = decode_cbf_luma(tctx,trafoDepth);
2921	3935	}
	3936	else {
	3937	/* There cannot be INTER blocks with no residual data.
	3938	That case is already handled with rqt_root_cbf.
	3939	*/
	3940
	3941	cbf_luma = 1;
	3942	}
2922	3943
2923	3944	logtrace(LogSlice,"call read_transform_unit %d/%d\n",x0,y0);
2924	3945
2925	3946	read_transform_unit(tctx, x0,y0,xBase,yBase, xCUBase,yCUBase, log2TrafoSize,trafoDepth, blkIdx,
2926	3947	cbf_luma, cbf_cb, cbf_cr);
2927
2928
2929		int nT = 1<<log2TrafoSize;
2930
2931
2932		if (cuPredMode == MODE_INTRA) // if intra mode
2933		{
2934		enum IntraPredMode intraPredMode = img->get_IntraPredMode(x0,y0);
2935
2936		decode_intra_prediction(img, x0,y0, intraPredMode, nT, 0);
2937
2938		enum IntraPredMode chromaPredMode = tctx->IntraPredModeC;
2939
2940		if (nT>=8) {
2941		decode_intra_prediction(img, x0/2,y0/2, chromaPredMode, nT/2, 1);
2942		decode_intra_prediction(img, x0/2,y0/2, chromaPredMode, nT/2, 2);
2943		}
2944		else if (blkIdx==3) {
2945		decode_intra_prediction(img, xBase/2,yBase/2, chromaPredMode, nT, 1);
2946		decode_intra_prediction(img, xBase/2,yBase/2, chromaPredMode, nT, 2);
2947		}
2948		}
2949
2950		// NOTE: disable MC-mode residuals:
2951		{ //if (cuPredMode == MODE_INTRA) {
2952		if (cbf_luma) {
2953		scale_coefficients(tctx, x0,y0, xCUBase,yCUBase, nT, 0,
2954		tctx->transform_skip_flag[0], PredMode==MODE_INTRA);
2955		}
2956
2957		if (nT>=8) {
2958		if (cbf_cb) {
2959		scale_coefficients(tctx, x0/2,y0/2, xCUBase/2,yCUBase/2, nT/2, 1,
2960		tctx->transform_skip_flag[1], PredMode==MODE_INTRA);
2961		}
2962		if (cbf_cr) {
2963		scale_coefficients(tctx, x0/2,y0/2, xCUBase/2,yCUBase/2, nT/2, 2,
2964		tctx->transform_skip_flag[2], PredMode==MODE_INTRA);
2965		}
2966		}
2967		else if (blkIdx==3) {
2968		if (cbf_cb) {
2969		scale_coefficients(tctx, xBase/2,yBase/2, xCUBase/2,yCUBase/2, nT, 1,
2970		tctx->transform_skip_flag[1], PredMode==MODE_INTRA);
2971		}
2972		if (cbf_cr) {
2973		scale_coefficients(tctx, xBase/2,yBase/2, xCUBase/2,yCUBase/2, nT, 2,
2974		tctx->transform_skip_flag[2], PredMode==MODE_INTRA);
2975		}
2976		}
2977		}
2978	3948	}
2979	3949	}
2980	3950

3049	4019	}
3050	4020
3051	4021	//set_mvd(tctx->decctx, x0,y0, refList, value[0],value[1]);
3052		tctx->mvd[refList][0] = value[0];
3053		tctx->mvd[refList][1] = value[1];
	4022	tctx->motion.mvd[refList][0] = value[0];
	4023	tctx->motion.mvd[refList][1] = value[1];
3054	4024
3055	4025	logtrace(LogSlice, "MVD[%d;%d\|%d] = %d;%d\n",x0,y0,refList, value[0],value[1]);
3056	4026	}

3060	4030	int x0, int y0,
3061	4031	int nPbW, int nPbH)
3062	4032	{
3063		slice_segment_header* shdr = tctx->shdr;
3064
3065		int merge_idx;
3066		if (shdr->MaxNumMergeCand>1) {
3067		merge_idx = decode_merge_idx(tctx);
3068		}
3069		else {
3070		merge_idx = 0;
3071		}
3072
3073		tctx->merge_idx = merge_idx;
3074		tctx->merge_flag = true;
	4033	int merge_idx = decode_merge_idx(tctx);
	4034
	4035	tctx->motion.merge_idx = merge_idx;
	4036	tctx->motion.merge_flag = true;
3075	4037
3076	4038	logtrace(LogSlice,"prediction skip 2Nx2N, merge_idx: %d\n",merge_idx);
3077	4039	}
3078	4040
3079	4041
	4042	/* xC/yC : CB position
	4043	xB/yB : position offset of the PB
	4044	nPbW/nPbH : size of PB
	4045	nCS : CB size
	4046	*/
3080	4047	void read_prediction_unit(thread_context* tctx,
3081	4048	int xC,int yC, int xB,int yB,
3082	4049	int nPbW, int nPbH,

3090	4057	slice_segment_header* shdr = tctx->shdr;
3091	4058
3092	4059	int merge_flag = decode_merge_flag(tctx);
3093		tctx->merge_flag = merge_flag;
	4060	tctx->motion.merge_flag = merge_flag;
3094	4061
3095	4062	if (merge_flag) {
3096		int merge_idx;
3097
3098		if (shdr->MaxNumMergeCand>1) {
3099		merge_idx = decode_merge_idx(tctx);
3100		}
3101		else {
3102		merge_idx = 0;
3103		}
	4063	int merge_idx = decode_merge_idx(tctx);
3104	4064
3105	4065	logtrace(LogSlice,"prediction unit %d,%d, merge mode, index: %d\n",x0,y0,merge_idx);
3106	4066
3107		tctx->merge_idx = merge_idx;
	4067	tctx->motion.merge_idx = merge_idx;
3108	4068	}
3109	4069	else { // no merge flag
3110	4070	enum InterPredIdc inter_pred_idc;

3116	4076	inter_pred_idc = PRED_L0;
3117	4077	}
3118	4078
3119		tctx->inter_pred_idc = inter_pred_idc; // set_inter_pred_idc(ctx,x0,y0, inter_pred_idc);
	4079	tctx->motion.inter_pred_idc = inter_pred_idc; // set_inter_pred_idc(ctx,x0,y0, inter_pred_idc);
3120	4080
3121	4081	if (inter_pred_idc != PRED_L1) {
3122	4082	int ref_idx_l0 = decode_ref_idx_lX(tctx, shdr->num_ref_idx_l0_active);
3123	4083
3124	4084	// NOTE: case for only one reference frame is handles in decode_ref_idx_lX()
3125		tctx->refIdx[0] = ref_idx_l0;
	4085	tctx->motion.refIdx[0] = ref_idx_l0;
3126	4086
3127	4087	read_mvd_coding(tctx,x0,y0, 0);
3128	4088
3129	4089	int mvp_l0_flag = decode_mvp_lx_flag(tctx); // l0
3130		tctx->mvp_lX_flag[0] = mvp_l0_flag;
	4090	tctx->motion.mvp_l0_flag = mvp_l0_flag;
3131	4091
3132	4092	logtrace(LogSlice,"prediction unit %d,%d, L0, refIdx=%d mvp_l0_flag:%d\n",
3133		x0,y0, tctx->refIdx[0], mvp_l0_flag);
	4093	x0,y0, tctx->motion.refIdx[0], mvp_l0_flag);
3134	4094	}
3135	4095
3136	4096	if (inter_pred_idc != PRED_L0) {
3137	4097	int ref_idx_l1 = decode_ref_idx_lX(tctx, shdr->num_ref_idx_l1_active);
3138	4098
3139	4099	// NOTE: case for only one reference frame is handles in decode_ref_idx_lX()
3140		tctx->refIdx[1] = ref_idx_l1;
	4100	tctx->motion.refIdx[1] = ref_idx_l1;
3141	4101
3142	4102	if (shdr->mvd_l1_zero_flag &&
3143	4103	inter_pred_idc == PRED_BI) {
3144		tctx->mvd[1][0] = 0;
3145		tctx->mvd[1][1] = 0;
	4104	tctx->motion.mvd[1][0] = 0;
	4105	tctx->motion.mvd[1][1] = 0;
3146	4106	}
3147	4107	else {
3148	4108	read_mvd_coding(tctx,x0,y0, 1);
3149	4109	}
3150	4110
3151	4111	int mvp_l1_flag = decode_mvp_lx_flag(tctx); // l1
3152		tctx->mvp_lX_flag[1] = mvp_l1_flag;
	4112	tctx->motion.mvp_l1_flag = mvp_l1_flag;
3153	4113
3154	4114	logtrace(LogSlice,"prediction unit %d,%d, L1, refIdx=%d mvp_l1_flag:%d\n",
3155		x0,y0, tctx->refIdx[1], mvp_l1_flag);
3156		}
3157		}
3158
3159
3160
3161		decode_prediction_unit(tctx, xC,yC,xB,yB, nCS, nPbW,nPbH, partIdx);
3162		}
3163
3164
3165
	4115	x0,y0, tctx->motion.refIdx[1], mvp_l1_flag);
	4116	}
	4117	}
	4118
	4119
	4120
	4121	decode_prediction_unit(tctx->decctx, tctx->shdr, tctx->img, tctx->motion,
	4122	xC,yC,xB,yB, nCS, nPbW,nPbH, partIdx);
	4123	}
	4124
	4125
	4126
	4127
	4128	template <class pixel_t>
	4129	void read_pcm_samples_internal(thread_context* tctx, int x0, int y0, int log2CbSize,
	4130	int cIdx, bitreader& br)
	4131	{
	4132	const seq_parameter_set* sps = &tctx->img->sps;
	4133
	4134	int nPcmBits;
	4135	int bitDepth;
	4136
	4137	int w = 1<<log2CbSize;
	4138	int h = 1<<log2CbSize;
	4139
	4140	if (cIdx>0) {
	4141	w /= tctx->img->sps.SubWidthC;
	4142	h /= tctx->img->sps.SubHeightC;
	4143
	4144	x0 /= tctx->img->sps.SubWidthC;
	4145	y0 /= tctx->img->sps.SubHeightC;
	4146
	4147	nPcmBits = sps->pcm_sample_bit_depth_chroma;
	4148	bitDepth = sps->BitDepth_C;
	4149	}
	4150	else {
	4151	nPcmBits = sps->pcm_sample_bit_depth_luma;
	4152	bitDepth = sps->BitDepth_Y;
	4153	}
	4154
	4155	pixel_t* ptr;
	4156	int stride;
	4157	ptr = tctx->img->get_image_plane_at_pos_NEW<pixel_t>(cIdx,x0,y0);
	4158	stride = tctx->img->get_image_stride(cIdx);
	4159
	4160	int shift = bitDepth - nPcmBits;
	4161
	4162	for (int y=0;y<h;y++)
	4163	for (int x=0;x<w;x++)
	4164	{
	4165	int value = get_bits(&br, nPcmBits);
	4166	ptr[y*stride+x] = value << shift;
	4167	}
	4168	}
3166	4169
3167	4170	static void read_pcm_samples(thread_context* tctx, int x0, int y0, int log2CbSize)
3168	4171	{

3172	4175	br.nextbits = 0;
3173	4176	br.nextbits_cnt = 0;
3174	4177
3175		const seq_parameter_set* sps = &tctx->img->sps;
3176		//fprintf(stderr,"PCM pos: %d %d (POC=%d)\n",x0,y0,tctx->decctx->img->PicOrderCntVal);
3177
3178		int nBitsY = sps->pcm_sample_bit_depth_luma;
3179		int nBitsC = sps->pcm_sample_bit_depth_chroma;
3180
3181		int wY = 1<<log2CbSize;
3182		int wC = 1<<(log2CbSize-1);
3183
3184		uint8_t* yPtr;
3185		uint8_t* cbPtr;
3186		uint8_t* crPtr;
3187		int stride;
3188		int chroma_stride;
3189		yPtr = tctx->img->get_image_plane(0);
3190		cbPtr = tctx->img->get_image_plane(1);
3191		crPtr = tctx->img->get_image_plane(2);
3192		stride = tctx->img->get_image_stride(0);
3193		chroma_stride = tctx->img->get_image_stride(1);
3194
3195		yPtr = &yPtr [y0*stride + x0];
3196		cbPtr = &cbPtr[y0/2*chroma_stride + x0/2];
3197		crPtr = &crPtr[y0/2*chroma_stride + x0/2];
3198
3199		int shiftY = sps->BitDepth_Y - nBitsY;
3200		int shiftC = sps->BitDepth_C - nBitsC;
3201
3202		for (int y=0;y<wY;y++)
3203		for (int x=0;x<wY;x++)
3204		{
3205		int value = get_bits(&br, nBitsY);
3206		yPtr[y*stride+x] = value << shiftY;
3207		}
3208
3209		for (int y=0;y<wC;y++)
3210		for (int x=0;x<wC;x++)
3211		{
3212		int value = get_bits(&br, nBitsC);
3213		cbPtr[y*chroma_stride+x] = value << shiftC;
3214		}
3215
3216		for (int y=0;y<wC;y++)
3217		for (int x=0;x<wC;x++)
3218		{
3219		int value = get_bits(&br, nBitsC);
3220		crPtr[y*chroma_stride+x] = value << shiftC;
3221		}
	4178
	4179	if (tctx->img->high_bit_depth(0)) {
	4180	read_pcm_samples_internal<uint16_t>(tctx,x0,y0,log2CbSize,0,br);
	4181	} else {
	4182	read_pcm_samples_internal<uint8_t>(tctx,x0,y0,log2CbSize,0,br);
	4183	}
	4184
	4185	if (tctx->img->sps.ChromaArrayType != CHROMA_MONO) {
	4186	if (tctx->img->high_bit_depth(1)) {
	4187	read_pcm_samples_internal<uint16_t>(tctx,x0,y0,log2CbSize,1,br);
	4188	read_pcm_samples_internal<uint16_t>(tctx,x0,y0,log2CbSize,2,br);
	4189	} else {
	4190	read_pcm_samples_internal<uint8_t>(tctx,x0,y0,log2CbSize,1,br);
	4191	read_pcm_samples_internal<uint8_t>(tctx,x0,y0,log2CbSize,2,br);
	4192	}
	4193	}
3222	4194
3223	4195	prepare_for_CABAC(&br);
3224	4196	tctx->cabac_decoder.bitstream_curr = br.data;
3225	4197	init_CABAC_decoder_2(&tctx->cabac_decoder);
3226	4198	}
3227	4199
	4200
	4201	int map_chroma_pred_mode(int intra_chroma_pred_mode, int IntraPredMode)
	4202	{
	4203	if (intra_chroma_pred_mode==4) {
	4204	return IntraPredMode;
	4205	}
	4206	else {
	4207	static const enum IntraPredMode IntraPredModeCCand[4] = {
	4208	INTRA_PLANAR,
	4209	INTRA_ANGULAR_26, // vertical
	4210	INTRA_ANGULAR_10, // horizontal
	4211	INTRA_DC
	4212	};
	4213
	4214	int IntraPredModeC = IntraPredModeCCand[intra_chroma_pred_mode];
	4215	if (IntraPredModeC == IntraPredMode) {
	4216	return INTRA_ANGULAR_34;
	4217	}
	4218	else {
	4219	return IntraPredModeC;
	4220	}
	4221	}
	4222	}
	4223
	4224	// h.265-V2 Table 8-3
	4225	static const uint8_t map_chroma_422[35] = {
	4226	0,1,2, 2, 2, 2, 3, 5, 7, 8,10,12,13,15,17,18,19,20,
	4227	21,22,23,23,24,24,25,25,26,27,27,28,28,29,29,30,31
	4228	};
3228	4229
3229	4230	void read_coding_unit(thread_context* tctx,
3230	4231	int x0, int y0, // position of coding unit in frame

3239	4240	logtrace(LogSlice,"- read_coding_unit %d;%d cbsize:%d\n",x0,y0,1<<log2CbSize);
3240	4241
3241	4242
3242		img->set_log2CbSize(x0,y0, log2CbSize);
	4243	//QQprintf("- read_coding_unit %d;%d cbsize:%d\n",x0,y0,1<<log2CbSize);
	4244
	4245	img->set_log2CbSize(x0,y0, log2CbSize, true);
	4246
	4247	/* This is only required on corrupted input streams.
	4248	It may happen that there are several slices in the image that overlap.
	4249	In this case, flags would accumulate from both slices.
	4250	*/
	4251	img->clear_split_transform_flags(x0,y0, log2CbSize);
3243	4252
3244	4253	int nCbS = 1<<log2CbSize; // number of coding block samples
3245	4254

3282	4291	// DECODE
3283	4292
3284	4293	int nCS_L = 1<<log2CbSize;
3285		decode_prediction_unit(tctx,x0,y0, 0,0, nCS_L, nCS_L,nCS_L, 0);
	4294	decode_prediction_unit(tctx->decctx,tctx->shdr,tctx->img,tctx->motion,
	4295	x0,y0, 0,0, nCS_L, nCS_L,nCS_L, 0);
3286	4296	}
3287	4297	else /* not skipped */ {
3288	4298	if (shdr->slice_type != SLICE_TYPE_I) {

3347	4357
3348	4358	int mpm_idx[4], rem_intra_luma_pred_mode[4];
3349	4359	idx=0;
	4360
	4361	int availableA0 = check_CTB_available(img, x0,y0, x0-1,y0);
	4362	int availableB0 = check_CTB_available(img, x0,y0, x0,y0-1);
3350	4363
3351	4364	for (int j=0;j<nCbS;j+=pbOffset)
3352	4365	for (int i=0;i<nCbS;i+=pbOffset)

3366	4379
3367	4380	int IntraPredMode;
3368	4381
3369		int availableA = check_CTB_available(img, shdr, x,y, x-1,y);
3370		int availableB = check_CTB_available(img, shdr, x,y, x,y-1);
	4382	int availableA = availableA0 \|\| (i>0); // left candidate always available for right blk
	4383	int availableB = availableB0 \|\| (j>0); // top candidate always available for bottom blk
	4384
	4385
3371	4386
3372	4387	int PUidx = (x>>sps->Log2MinPUSize) + (y>>sps->Log2MinPUSize)*sps->PicWidthInMinPUs;
3373	4388
3374		// block on left side
3375
3376		enum IntraPredMode candIntraPredModeA, candIntraPredModeB;
3377		if (availableA==false) {
3378		candIntraPredModeA=INTRA_DC;
3379		}
3380		else if (img->get_pred_mode(x-1,y) != MODE_INTRA \|\|
3381		img->get_pcm_flag (x-1,y)) {
3382		candIntraPredModeA=INTRA_DC;
3383		}
3384		else {
3385		candIntraPredModeA = img->get_IntraPredMode_atIndex(PUidx-1);
3386		}
3387
3388		// block above
3389
3390		if (availableB==false) {
3391		candIntraPredModeB=INTRA_DC;
3392		}
3393		else if (img->get_pred_mode(x,y-1) != MODE_INTRA \|\|
3394		img->get_pcm_flag (x,y-1)) {
3395		candIntraPredModeB=INTRA_DC;
3396		}
3397		else if (y-1 < ((y >> sps->Log2CtbSizeY) << sps->Log2CtbSizeY)) {
3398		candIntraPredModeB=INTRA_DC;
3399		}
3400		else {
3401		candIntraPredModeB = img->get_IntraPredMode_atIndex(PUidx-sps->PicWidthInMinPUs);
3402		}
3403
3404		// build candidate list
3405
3406	4389	int candModeList[3];
3407	4390
3408		logtrace(LogSlice,"availableA:%d candA:%d & availableB:%d candB:%d\n",
3409		availableA, candIntraPredModeA,
3410		availableB, candIntraPredModeB);
3411
3412		if (candIntraPredModeA == candIntraPredModeB) {
3413		if (candIntraPredModeA < 2) {
3414		candModeList[0] = INTRA_PLANAR;
3415		candModeList[1] = INTRA_DC;
3416		candModeList[2] = INTRA_ANGULAR_26;
3417		}
3418		else {
3419		candModeList[0] = candIntraPredModeA;
3420		candModeList[1] = 2 + ((candIntraPredModeA-2 -1 +32) % 32);
3421		candModeList[2] = 2 + ((candIntraPredModeA-2 +1 ) % 32);
3422		}
3423		}
3424		else {
3425		candModeList[0] = candIntraPredModeA;
3426		candModeList[1] = candIntraPredModeB;
3427
3428		if (candIntraPredModeA != INTRA_PLANAR &&
3429		candIntraPredModeB != INTRA_PLANAR) {
3430		candModeList[2] = INTRA_PLANAR;
3431		}
3432		else if (candIntraPredModeA != INTRA_DC &&
3433		candIntraPredModeB != INTRA_DC) {
3434		candModeList[2] = INTRA_DC;
3435		}
3436		else {
3437		candModeList[2] = INTRA_ANGULAR_26;
3438		}
3439		}
	4391	fillIntraPredModeCandidates(candModeList,x,y,PUidx,
	4392	availableA, availableB, img);
3440	4393
3441	4394	for (int i=0;i<3;i++)
3442	4395	logtrace(LogSlice,"candModeList[%d] = %d\n", i, candModeList[i]);

3448	4401	// sort candModeList
3449	4402
3450	4403	if (candModeList[0] > candModeList[1]) {
3451		int t = candModeList[0]; candModeList[0]=candModeList[1]; candModeList[1]=t;
	4404	std::swap(candModeList[0],candModeList[1]);
3452	4405	}
3453	4406	if (candModeList[0] > candModeList[2]) {
3454		int t = candModeList[0]; candModeList[0]=candModeList[2]; candModeList[2]=t;
	4407	std::swap(candModeList[0],candModeList[2]);
3455	4408	}
3456	4409	if (candModeList[1] > candModeList[2]) {
3457		int t = candModeList[1]; candModeList[1]=candModeList[2]; candModeList[2]=t;
	4410	std::swap(candModeList[1],candModeList[2]);
3458	4411	}
3459	4412
3460	4413	// skip modes in the list

3476	4429
3477	4430	// set chroma intra prediction mode
3478	4431
3479		int intra_chroma_pred_mode = decode_intra_chroma_pred_mode(tctx);
3480
3481		int IntraPredMode = img->get_IntraPredMode(x0,y0);
3482		logtrace(LogSlice,"IntraPredMode: %d\n",IntraPredMode);
3483
3484		int IntraPredModeC;
3485		if (intra_chroma_pred_mode==4) {
3486		IntraPredModeC = IntraPredMode;
3487		}
3488		else {
3489		static enum IntraPredMode IntraPredModeCCand[4] = {
3490		INTRA_PLANAR,
3491		INTRA_ANGULAR_26, // vertical
3492		INTRA_ANGULAR_10, // horizontal
3493		INTRA_DC
3494		};
3495
3496		IntraPredModeC = IntraPredModeCCand[intra_chroma_pred_mode];
3497		if (IntraPredModeC == IntraPredMode) {
3498		IntraPredModeC = INTRA_ANGULAR_34;
	4432	if (sps->ChromaArrayType == CHROMA_444) {
	4433	// chroma 4:4:4
	4434
	4435	idx = 0;
	4436	for (int j=0;j<nCbS;j+=pbOffset)
	4437	for (int i=0;i<nCbS;i+=pbOffset) {
	4438	int x = x0+i;
	4439	int y = y0+j;
	4440
	4441	int intra_chroma_pred_mode = decode_intra_chroma_pred_mode(tctx);
	4442	int IntraPredMode = img->get_IntraPredMode(x,y);
	4443
	4444	int IntraPredModeC = map_chroma_pred_mode(intra_chroma_pred_mode, IntraPredMode);
	4445
	4446	logtrace(LogSlice,"IntraPredModeC[%d][%d]: %d (blksize:%d)\n",x,y,IntraPredModeC,
	4447	1<<log2IntraPredSize);
	4448
	4449	img->set_IntraPredModeC(x,y, log2IntraPredSize,
	4450	(enum IntraPredMode)IntraPredModeC,
	4451	intra_chroma_pred_mode == 4);
	4452	idx++;
	4453	}
	4454	}
	4455	else if (sps->ChromaArrayType != CHROMA_MONO) {
	4456	// chroma 4:2:0 and 4:2:2
	4457
	4458	int intra_chroma_pred_mode = decode_intra_chroma_pred_mode(tctx);
	4459	int IntraPredMode = img->get_IntraPredMode(x0,y0);
	4460	logtrace(LogSlice,"IntraPredMode: %d\n",IntraPredMode);
	4461	int IntraPredModeC = map_chroma_pred_mode(intra_chroma_pred_mode, IntraPredMode);
	4462
	4463	if (sps->ChromaArrayType == CHROMA_422) {
	4464	IntraPredModeC = map_chroma_422[ IntraPredModeC ];
3499	4465	}
3500		}
3501
3502		logtrace(LogSlice,"IntraPredModeC[%d][%d]: %d\n",x0,y0,IntraPredModeC);
3503
3504		tctx->IntraPredModeC = (enum IntraPredMode) IntraPredModeC;
	4466
	4467	img->set_IntraPredModeC(x0,y0, log2CbSize,
	4468	(enum IntraPredMode)IntraPredModeC,
	4469	intra_chroma_pred_mode == 4);
	4470	}
3505	4471	}
3506	4472	}
3507	4473	else { // INTER

3551	4517	if (!pcm_flag) { // !pcm
3552	4518	bool rqt_root_cbf;
3553	4519
3554		uint8_t merge_flag = tctx->merge_flag; // !!get_merge_flag(ctx,x0,y0);
	4520	uint8_t merge_flag = tctx->motion.merge_flag; // !!get_merge_flag(ctx,x0,y0);
3555	4521
3556	4522	if (cuPredMode != MODE_INTRA &&
3557	4523	!(PartMode == PART_2Nx2N && merge_flag)) {

3559	4525	rqt_root_cbf = !!decode_rqt_root_cbf(tctx);
3560	4526	}
3561	4527	else {
	4528	/* rqt_root_cbf=1 is inferred for Inter blocks with 2Nx2N, merge mode.
	4529	These must be some residual data, because otherwise, the CB could
	4530	also be coded in SKIP mode.
	4531	*/
	4532
3562	4533	rqt_root_cbf = true;
3563	4534	}
3564	4535

3576	4547
3577	4548	logtrace(LogSlice,"MaxTrafoDepth: %d\n",MaxTrafoDepth);
3578	4549
	4550	uint8_t initial_chroma_cbf = 1;
	4551	if (sps->ChromaArrayType == CHROMA_MONO) {
	4552	initial_chroma_cbf = 0;
	4553	}
	4554
3579	4555	read_transform_tree(tctx, x0,y0, x0,y0, x0,y0, log2CbSize, 0,0,
3580		MaxTrafoDepth, IntraSplitFlag, cuPredMode, 1,1);
	4556	MaxTrafoDepth, IntraSplitFlag, cuPredMode,
	4557	initial_chroma_cbf, initial_chroma_cbf);
3581	4558	}
3582	4559	} // !pcm
3583	4560	}

3624	4601	// shdr->CuQpDelta = 0; // TODO check: is this the right place to set to default value ?
3625	4602	}
3626	4603
	4604
	4605	if (tctx->shdr->cu_chroma_qp_offset_enabled_flag &&
	4606	log2CbSize >= img->pps.Log2MinCuChromaQpOffsetSize) {
	4607	tctx->IsCuChromaQpOffsetCoded = 0;
	4608	}
	4609
3627	4610	if (split_flag) {
3628	4611	int x1 = x0 + (1<<(log2CbSize-1));
3629	4612	int y1 = y0 + (1<<(log2CbSize-1));

3674	4657
3675	4658	const int startCtbY = tctx->CtbY;
3676	4659
	4660	//printf("start decoding substream at %d;%d\n",tctx->CtbX,tctx->CtbY);
	4661
3677	4662	// in WPP mode: initialize CABAC model with stored model from row above
3678	4663
3679	4664	if ((!first_independent_substream \|\| tctx->CtbY != startCtbY) &&

3681	4666	tctx->CtbY>=1 && tctx->CtbX==0)
3682	4667	{
3683	4668	if (sps->PicWidthInCtbsY>1) {
	4669	if ((tctx->CtbY-1) >= tctx->imgunit->ctx_models.size()) {
	4670	return Decode_Error;
	4671	}
	4672
	4673	//printf("CTX wait on %d/%d\n",1,tctx->CtbY-1);
	4674
3684	4675	// we have to wait until the context model data is there
3685	4676	tctx->img->wait_for_progress(tctx->task, 1,tctx->CtbY-1,CTB_PROGRESS_PREFILTER);
3686	4677
3687	4678	// copy CABAC model from previous CTB row
3688		memcpy(tctx->ctx_model,
3689		&tctx->imgunit->ctx_models[(tctx->CtbY-1) * CONTEXT_MODEL_TABLE_LENGTH],
3690		CONTEXT_MODEL_TABLE_LENGTH * sizeof(context_model));
	4679	tctx->ctx_model = tctx->imgunit->ctx_models[(tctx->CtbY-1)];
	4680	tctx->imgunit->ctx_models[(tctx->CtbY-1)].release(); // not used anymore
3691	4681	}
3692	4682	else {
3693	4683	tctx->img->wait_for_progress(tctx->task, 0,tctx->CtbY-1,CTB_PROGRESS_PREFILTER);
3694		initialize_CABAC(tctx);
	4684	initialize_CABAC_models(tctx);
3695	4685	}
3696	4686	}
3697	4687

3700	4690	const int ctbx = tctx->CtbX;
3701	4691	const int ctby = tctx->CtbY;
3702	4692
	4693	if (ctbx+ctby*ctbW >= tctx->img->pps.CtbAddrRStoTS.size()) {
	4694	return Decode_Error;
	4695	}
	4696
	4697	if (ctbx >= tctx->img->sps.PicWidthInCtbsY \|\|
	4698	ctby >= tctx->img->sps.PicHeightInCtbsY) {
	4699	return Decode_Error;
	4700	}
	4701
3703	4702	if (block_wpp && ctby>0 && ctbx < ctbW-1) {
3704		//printf("wait on %d/%d\n",ctbx+1,ctby-1);
	4703
	4704	// TODO: if we are in tiles mode and at the right border, do not wait for x+1,y-1
	4705
	4706	//printf("wait on %d/%d (%d)\n",ctbx+1,ctby-1, ctbx+1+(ctby-1)*sps->PicWidthInCtbsY);
3705	4707
3706	4708	tctx->img->wait_for_progress(tctx->task, ctbx+1,ctby-1, CTB_PROGRESS_PREFILTER);
3707	4709	}
3708	4710
3709		//printf("%p: decode %d;%d\n", tctx, tctx->CtbY,tctx->CtbX);
	4711	//printf("%p: decode %d;%d\n", tctx, tctx->CtbX,tctx->CtbY);
3710	4712
3711	4713
3712	4714	// read and decode CTB
	4715
	4716	if (tctx->ctx_model.empty() == false) {
	4717	return Decode_Error;
	4718	}
3713	4719
3714	4720	read_coding_tree_unit(tctx);
3715	4721

3720	4726	ctbx == 1 &&
3721	4727	ctby < sps->PicHeightInCtbsY-1)
3722	4728	{
3723		context_model* ctx_store = &tctx->imgunit->ctx_models[ctby * CONTEXT_MODEL_TABLE_LENGTH];
3724
3725		memcpy(ctx_store,
3726		&tctx->ctx_model,
3727		CONTEXT_MODEL_TABLE_LENGTH * sizeof(context_model));
	4729	// no storage for context table has been allocated
	4730	if (tctx->imgunit->ctx_models.size() <= ctby) {
	4731	return Decode_Error;
	4732	}
	4733
	4734	tctx->imgunit->ctx_models[ctby] = tctx->ctx_model;
	4735	tctx->imgunit->ctx_models[ctby].decouple(); // store an independent copy
3728	4736	}
3729	4737
3730	4738
3731	4739	// end of slice segment ?
3732	4740
3733	4741	int end_of_slice_segment_flag = decode_CABAC_term_bit(&tctx->cabac_decoder);
	4742	//printf("end-of-slice flag: %d\n", end_of_slice_segment_flag);
3734	4743
3735	4744	if (end_of_slice_segment_flag) {
3736	4745	// at the end of the slice segment, we store the CABAC model if we need it
3737	4746	// because a dependent slice may follow
3738	4747
3739	4748	if (pps->dependent_slice_segments_enabled_flag) {
3740		memcpy(tctx->shdr->ctx_model_storage,
3741		tctx->ctx_model,
3742		CONTEXT_MODEL_TABLE_LENGTH * sizeof(context_model));
	4749	tctx->shdr->ctx_model_storage = tctx->ctx_model;
	4750	tctx->shdr->ctx_model_storage.decouple(); // store an independent copy
	4751
	4752	tctx->shdr->ctx_model_storage_defined = true;
3743	4753	}
3744	4754	}
3745	4755

3765	4775
3766	4776
3767	4777	if (end_of_slice_segment_flag) {
	4778	/* corrupted inputs may send the end_of_slice_segment_flag even if not all
	4779	CTBs in a row have been coded. Hence, we mark all of them as finished.
	4780	*/
	4781
	4782	/*
	4783	for (int x = ctbx+1 ; x<sps->PicWidthInCtbsY; x++) {
	4784	printf("mark skipped %d;%d\n",ctbx,ctby);
	4785	tctx->img->ctb_progress[ctbx+ctby*ctbW].set_progress(CTB_PROGRESS_PREFILTER);
	4786	}
	4787	*/
	4788
3768	4789	return Decode_EndOfSliceSegment;
3769	4790	}
3770	4791

3794	4815
3795	4816
3796	4817
3797		void initialize_CABAC_at_slice_segment_start(thread_context* tctx)
	4818	bool initialize_CABAC_at_slice_segment_start(thread_context* tctx)
3798	4819	{
3799	4820	de265_image* img = tctx->img;
3800	4821	const pic_parameter_set* pps = &img->pps;

3804	4825	if (shdr->dependent_slice_segment_flag) {
3805	4826	int prevCtb = pps->CtbAddrTStoRS[ pps->CtbAddrRStoTS[shdr->slice_segment_address] -1 ];
3806	4827
3807		slice_segment_header* prevCtbHdr = img->slices[ img->get_SliceHeaderIndex_atIndex(prevCtb) ];
	4828	int sliceIdx = img->get_SliceHeaderIndex_atIndex(prevCtb);
	4829	if (sliceIdx >= img->slices.size()) {
	4830	return false;
	4831	}
	4832	slice_segment_header* prevCtbHdr = img->slices[ sliceIdx ];
3808	4833
3809	4834	if (pps->is_tile_start_CTB(shdr->slice_segment_address % sps->PicWidthInCtbsY,
3810	4835	shdr->slice_segment_address / sps->PicWidthInCtbsY
3811	4836	)) {
3812		initialize_CABAC(tctx);
	4837	initialize_CABAC_models(tctx);
3813	4838	}
3814	4839	else {
	4840	// wait for previous slice to finish decoding
	4841
	4842	//printf("wait for previous slice to finish decoding\n");
	4843
	4844
	4845	slice_unit* prevSliceSegment = tctx->imgunit->get_prev_slice_segment(tctx->sliceunit);
	4846	//assert(prevSliceSegment);
	4847	if (prevSliceSegment==NULL) {
	4848	return false;
	4849	}
	4850
	4851	prevSliceSegment->finished_threads.wait_for_progress(prevSliceSegment->nThreads);
	4852
	4853
	4854	/*
	4855	printf("wait for %d,%d (init)\n",
	4856	prevCtb / sps->PicWidthInCtbsY,
	4857	prevCtb % sps->PicWidthInCtbsY);
3815	4858	tctx->img->wait_for_progress(tctx->task, prevCtb, CTB_PROGRESS_PREFILTER);
3816
3817		memcpy(tctx->ctx_model,
3818		prevCtbHdr->ctx_model_storage,
3819		CONTEXT_MODEL_TABLE_LENGTH * sizeof(context_model));
	4859	*/
	4860
	4861	if (!prevCtbHdr->ctx_model_storage_defined) {
	4862	return false;
	4863	}
	4864
	4865	tctx->ctx_model = prevCtbHdr->ctx_model_storage;
	4866	prevCtbHdr->ctx_model_storage.release();
3820	4867	}
3821	4868	}
3822	4869	else {
3823		initialize_CABAC(tctx);
3824		}
	4870	initialize_CABAC_models(tctx);
	4871	}
	4872
	4873	return true;
	4874	}
	4875
	4876
	4877	std::string thread_task_ctb_row::name() const {
	4878	char buf[100];
	4879	sprintf(buf,"ctb-row-%d",debug_startCtbRow);
	4880	return buf;
	4881	}
	4882
	4883
	4884	std::string thread_task_slice_segment::name() const {
	4885	char buf[100];
	4886	sprintf(buf,"slice-segment-%d;%d",debug_startCtbX,debug_startCtbY);
	4887	return buf;
3825	4888	}
3826	4889
3827	4890
3828	4891	void thread_task_slice_segment::work()
3829	4892	{
3830		struct thread_task_slice_segment* data = this;
	4893	thread_task_slice_segment* data = this;
3831	4894	thread_context* tctx = data->tctx;
3832	4895	de265_image* img = tctx->img;
3833	4896
3834	4897	state = Running;
3835		img->thread_run();
	4898	img->thread_run(this);
3836	4899
3837	4900	setCtbAddrFromTS(tctx);
3838	4901
3839	4902	//printf("%p: A start decoding at %d/%d\n", tctx, tctx->CtbX,tctx->CtbY);
3840	4903
3841	4904	if (data->firstSliceSubstream) {
3842		initialize_CABAC_at_slice_segment_start(tctx);
	4905	bool success = initialize_CABAC_at_slice_segment_start(tctx);
	4906	if (!success) {
	4907	state = Finished;
	4908	tctx->sliceunit->finished_threads.increase_progress(1);
	4909	img->thread_finishes(this);
	4910	return;
	4911	}
3843	4912	}
3844	4913	else {
3845		initialize_CABAC(tctx);
	4914	initialize_CABAC_models(tctx);
3846	4915	}
3847	4916
3848	4917	init_CABAC_decoder_2(&tctx->cabac_decoder);

3850	4919	/enum DecodeResult result =/ decode_substream(tctx, false, data->firstSliceSubstream);
3851	4920
3852	4921	state = Finished;
3853		img->thread_finishes();
	4922	tctx->sliceunit->finished_threads.increase_progress(1);
	4923	img->thread_finishes(this);
3854	4924
3855	4925	return; // DE265_OK;
3856	4926	}

3858	4928
3859	4929	void thread_task_ctb_row::work()
3860	4930	{
3861		struct thread_task_ctb_row* data = this;
	4931	thread_task_ctb_row* data = this;
3862	4932	thread_context* tctx = data->tctx;
3863	4933	de265_image* img = tctx->img;
3864	4934

3866	4936	int ctbW = sps->PicWidthInCtbsY;
3867	4937
3868	4938	state = Running;
3869		img->thread_run();
	4939	img->thread_run(this);
3870	4940
3871	4941	setCtbAddrFromTS(tctx);
3872	4942
3873	4943	int ctby = tctx->CtbAddrInRS / ctbW;
3874	4944	int myCtbRow = ctby;
3875	4945
3876		// printf("start decoding at %d/%d\n", ctbx,ctby);
	4946	//printf("start CTB-row decoding at row %d\n", ctby);
3877	4947
3878	4948	if (data->firstSliceSubstream) {
3879		initialize_CABAC_at_slice_segment_start(tctx);
	4949	bool success = initialize_CABAC_at_slice_segment_start(tctx);
	4950	if (!success) {
	4951	// could not decode this row, mark whole row as finished
	4952	for (int x=0;x<ctbW;x++) {
	4953	img->ctb_progress[myCtbRow*ctbW + x].set_progress(CTB_PROGRESS_PREFILTER);
	4954	}
	4955
	4956	state = Finished;
	4957	tctx->sliceunit->finished_threads.increase_progress(1);
	4958	img->thread_finishes(this);
	4959	return;
	4960	}
3880	4961	//initialize_CABAC(tctx);
3881	4962	}
3882	4963

3892	4973
3893	4974	// TODO: what about slices that end properly in the middle of a CTB row?
3894	4975
3895		#if 1
3896	4976	if (tctx->CtbY == myCtbRow) {
3897	4977	int lastCtbX = sps->PicWidthInCtbsY; // assume no tiles when WPP is on
3898	4978	for (int x = tctx->CtbX; x<lastCtbX ; x++) {
3899		img->ctb_progress[myCtbRow*ctbW + x].set_progress(CTB_PROGRESS_PREFILTER);
3900		}
3901		}
3902		#endif
	4979
	4980	if (x < img->sps.PicWidthInCtbsY &&
	4981	myCtbRow < img->sps.PicHeightInCtbsY) {
	4982	img->ctb_progress[myCtbRow*ctbW + x].set_progress(CTB_PROGRESS_PREFILTER);
	4983	}
	4984	}
	4985	}
3903	4986
3904	4987	state = Finished;
3905		img->thread_finishes();
	4988	tctx->sliceunit->finished_threads.increase_progress(1);
	4989	img->thread_finishes(this);
3906	4990	}
3907	4991
3908	4992

3915	4999	const seq_parameter_set* sps = &img->sps;
3916	5000	slice_segment_header* shdr = tctx->shdr;
3917	5001
3918		initialize_CABAC_at_slice_segment_start(tctx);
	5002	bool success = initialize_CABAC_at_slice_segment_start(tctx);
	5003	if (!success) {
	5004	return DE265_ERROR_UNSPECIFIED_DECODING_ERROR;
	5005	}
3919	5006
3920	5007	init_CABAC_decoder_2(&tctx->cabac_decoder);
3921	5008
3922		// printf("-----\n");
	5009	//printf("-----\n");
3923	5010
3924	5011	bool first_slice_substream = !shdr->dependent_slice_segment_flag;
3925	5012

3941	5028	}
3942	5029
3943	5030	substream++;
3944
	5031
3945	5032
3946	5033	result = decode_substream(tctx, false, first_slice_substream);
3947	5034

3954	5041	first_slice_substream = false;
3955	5042
3956	5043	if (pps->tiles_enabled_flag) {
3957		initialize_CABAC(tctx);
	5044	initialize_CABAC_models(tctx);
3958	5045	}
3959	5046	} while (true);
3960	5047

+81

-62

libde265/slice.h less more

28	28	#include "libde265/util.h"
29	29	#include "libde265/refpic.h"
30	30	#include "libde265/threads.h"
	31	#include "contextmodel.h"
31	32
32	33	#include <vector>
	34	#include <string.h>
33	35
34	36	#define MAX_NUM_REF_PICS 16
35	37
36		#define SLICE_TYPE_B 0
37		#define SLICE_TYPE_P 1
38		#define SLICE_TYPE_I 2
39
	38	class decoder_context;
	39	class thread_context;
	40	class error_queue;
	41	class seq_parameter_set;
	42	class pic_parameter_set;
	43
	44	enum SliceType
	45	{
	46	SLICE_TYPE_B = 0,
	47	SLICE_TYPE_P = 1,
	48	SLICE_TYPE_I = 2
	49	};
40	50
41	51	/*
42		2Nx2N 2NxN Nx2N NxN
	52	2Nx2N 2NxN Nx2N NxN
43	53	+-------+ +-------+ +---+---+ +---+---+
44	54	\| \| \| \| \| \| \| \| \| \|
45	55	\| \| \|_______\| \| \| \| \|___\|___\|

47	57	\| \| \| \| \| \| \| \| \| \|
48	58	+-------+ +-------+ +---+---+ +---+---+
49	59
50		2NxnU 2NxnD nLx2N nRx2N
	60	2NxnU 2NxnD nLx2N nRx2N
51	61	+-------+ +-------+ +-+-----+ +-----+-+
52	62	\|_______\| \| \| \| \| \| \| \| \|
53	63	\| \| \| \| \| \| \| \| \| \|

88	98	INTRA_ANGULAR_22 = 22, INTRA_ANGULAR_23 = 23, INTRA_ANGULAR_24 = 24, INTRA_ANGULAR_25 = 25,
89	99	INTRA_ANGULAR_26 = 26, INTRA_ANGULAR_27 = 27, INTRA_ANGULAR_28 = 28, INTRA_ANGULAR_29 = 29,
90	100	INTRA_ANGULAR_30 = 30, INTRA_ANGULAR_31 = 31, INTRA_ANGULAR_32 = 32, INTRA_ANGULAR_33 = 33,
91		INTRA_ANGULAR_34 = 34,
92		INTRA_CHROMA_EQ_LUMA = 100 // chroma := luma
93		};
	101	INTRA_ANGULAR_34 = 34
	102	};
	103
	104
	105	enum IntraChromaPredMode
	106	{
	107	INTRA_CHROMA_PLANAR_OR_34 = 0,
	108	INTRA_CHROMA_ANGULAR_26_OR_34 = 1,
	109	INTRA_CHROMA_ANGULAR_10_OR_34 = 2,
	110	INTRA_CHROMA_DC_OR_34 = 3,
	111	INTRA_CHROMA_LIKE_LUMA = 4
	112	};
	113
94	114
95	115	enum InterPredIdc
96	116	{
97		PRED_L0=0,
98		PRED_L1=1,
99		PRED_BI=2
100		};
101
102		enum context_model_indices {
103		CONTEXT_MODEL_SAO_MERGE_FLAG = 0,
104		CONTEXT_MODEL_SAO_TYPE_IDX = CONTEXT_MODEL_SAO_MERGE_FLAG +1,
105		CONTEXT_MODEL_SPLIT_CU_FLAG = CONTEXT_MODEL_SAO_TYPE_IDX + 1,
106		CONTEXT_MODEL_CU_SKIP_FLAG = CONTEXT_MODEL_SPLIT_CU_FLAG + 3,
107		CONTEXT_MODEL_PART_MODE = CONTEXT_MODEL_CU_SKIP_FLAG + 3,
108		CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG = CONTEXT_MODEL_PART_MODE + 4,
109		CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE = CONTEXT_MODEL_PREV_INTRA_LUMA_PRED_FLAG + 1,
110		CONTEXT_MODEL_CBF_LUMA = CONTEXT_MODEL_INTRA_CHROMA_PRED_MODE + 1,
111		CONTEXT_MODEL_CBF_CHROMA = CONTEXT_MODEL_CBF_LUMA + 2,
112		CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG = CONTEXT_MODEL_CBF_CHROMA + 4,
113		CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX = CONTEXT_MODEL_SPLIT_TRANSFORM_FLAG + 3,
114		CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX = CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_X_PREFIX + 18,
115		CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG = CONTEXT_MODEL_LAST_SIGNIFICANT_COEFFICIENT_Y_PREFIX + 18,
116		CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG = CONTEXT_MODEL_CODED_SUB_BLOCK_FLAG + 4,
117		CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG = CONTEXT_MODEL_SIGNIFICANT_COEFF_FLAG + 42,
118		CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG = CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER1_FLAG + 24,
119		CONTEXT_MODEL_CU_QP_DELTA_ABS = CONTEXT_MODEL_COEFF_ABS_LEVEL_GREATER2_FLAG + 6,
120		CONTEXT_MODEL_TRANSFORM_SKIP_FLAG = CONTEXT_MODEL_CU_QP_DELTA_ABS + 2,
121		CONTEXT_MODEL_MERGE_FLAG = CONTEXT_MODEL_TRANSFORM_SKIP_FLAG + 2,
122		CONTEXT_MODEL_MERGE_IDX = CONTEXT_MODEL_MERGE_FLAG + 1,
123		CONTEXT_MODEL_PRED_MODE_FLAG = CONTEXT_MODEL_MERGE_IDX + 1,
124		CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG = CONTEXT_MODEL_PRED_MODE_FLAG + 1,
125		CONTEXT_MODEL_MVP_LX_FLAG = CONTEXT_MODEL_ABS_MVD_GREATER01_FLAG + 2,
126		CONTEXT_MODEL_RQT_ROOT_CBF = CONTEXT_MODEL_MVP_LX_FLAG + 1,
127		CONTEXT_MODEL_REF_IDX_LX = CONTEXT_MODEL_RQT_ROOT_CBF + 1,
128		CONTEXT_MODEL_INTER_PRED_IDC = CONTEXT_MODEL_REF_IDX_LX + 2,
129		CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG = CONTEXT_MODEL_INTER_PRED_IDC + 5,
130		CONTEXT_MODEL_TABLE_LENGTH = CONTEXT_MODEL_CU_TRANSQUANT_BYPASS_FLAG + 1
131		};
132
133
134		typedef struct slice_segment_header {
135		slice_segment_header() { }
136
137		de265_error read(bitreader* br, struct decoder_context, bool continueDecoding);
	117	// note: values have to match the decoding function decode_inter_pred_idc()
	118	PRED_L0=1,
	119	PRED_L1=2,
	120	PRED_BI=3
	121	};
	122
	123
	124
	125	class slice_segment_header {
	126	public:
	127	slice_segment_header() {
	128	reset();
	129	}
	130
	131	de265_error read(bitreader* br, decoder_context, bool continueDecoding);
	132	de265_error write(error_queue*, CABAC_encoder&,
	133	const seq_parameter_set* sps,
	134	const pic_parameter_set* pps,
	135	uint8_t nal_unit_type);
	136
138	137	void dump_slice_segment_header(const decoder_context*, int fd) const;
139	138
140
141		int slice_index; // index through all slices in a picture
	139	void set_defaults();
	140	void reset();
	141
	142
	143	int slice_index; // index through all slices in a picture (internal only)
142	144
143	145	char first_slice_segment_in_pic_flag;
144	146	char no_output_of_prior_pics_flag;

202	204	int slice_cb_qp_offset;
203	205	int slice_cr_qp_offset;
204	206
	207	char cu_chroma_qp_offset_enabled_flag;
	208
205	209	char deblocking_filter_override_flag;
206	210	char slice_deblocking_filter_disabled_flag;
207	211	int slice_beta_offset; // = pps->beta_offset if undefined

218	222
219	223	// --- derived data ---
220	224
221		int SliceAddrRS; // start of last independent slice
222	225	int SliceQPY;
223
224	226	int initType;
225	227
226		int MaxNumMergeCand;
	228	void compute_derived_values(const pic_parameter_set* pps);
	229
	230
	231	// --- data for external modules ---
	232
	233	int SliceAddrRS; // slice_segment_address of last independent slice
	234
	235	int MaxNumMergeCand; // directly derived from 'five_minus_max_num_merge_cand'
227	236	int CurrRpsIdx;
228	237	ref_pic_set CurrRps; // the active reference-picture set
229	238	int NumPocTotalCurr;
230	239
231		int RefPicList[2][MAX_NUM_REF_PICS]; // contains indices into DPB
	240	// number of entries: num_ref_idx_l0_active / num_ref_idx_l1_active
	241	int RefPicList[2][MAX_NUM_REF_PICS]; // contains buffer IDs (D:indices into DPB/E:frame number)
232	242	int RefPicList_POC[2][MAX_NUM_REF_PICS];
233	243	int RefPicList_PicState[2][MAX_NUM_REF_PICS]; /* We have to save the PicState because the decoding
234	244	of an image may be delayed and the PicState can

239	249	is a long-term picture. */
240	250
241	251	// context storage for dependent slices (stores CABAC model at end of slice segment)
242		context_model ctx_model_storage[CONTEXT_MODEL_TABLE_LENGTH];
	252	context_model_table ctx_model_storage;
	253	bool ctx_model_storage_defined; // whether there is valid data in ctx_model_storage
243	254
244	255	std::vector<int> RemoveReferencesList; // images that can be removed from the DPB before decoding this slice
245	256
246		} slice_segment_header;
	257	};
247	258
248	259
249	260

254	265	unsigned char SaoEoClass; // use with (SaoTypeIdx>>(2*cIdx)) & 0x3
255	266
256	267	uint8_t sao_band_position[3];
257		int8_t saoOffsetVal[3][4]; // index with [][idx-1] as saoOffsetVal[][0]==0 always
	268	int8_t saoOffsetVal[3][4]; // index with [][idx-1] as saoOffsetVal[][0]==0 always
258	269	} sao_info;
259	270
260	271
261	272
262	273
263		de265_error read_slice_segment_data(struct thread_context* tctx);
	274	de265_error read_slice_segment_data(thread_context* tctx);
264	275
265	276	bool alloc_and_init_significant_coeff_ctxIdx_lookupTable();
266	277	void free_significant_coeff_ctxIdx_lookupTable();

270	281	{
271	282	public:
272	283	bool firstSliceSubstream;
273		struct thread_context* tctx;
	284	int debug_startCtbRow;
	285	thread_context* tctx;
274	286
275	287	virtual void work();
	288	virtual std::string name() const;
276	289	};
277	290
278	291	class thread_task_slice_segment : public thread_task
279	292	{
280	293	public:
281	294	bool firstSliceSubstream;
282		struct thread_context* tctx;
	295	int debug_startCtbX, debug_startCtbY;
	296	thread_context* tctx;
283	297
284	298	virtual void work();
	299	virtual std::string name() const;
285	300	};
286	301
	302
	303	int check_CTB_available(const de265_image* img,
	304	int xC,int yC, int xN,int yN);
	305
287	306	#endif

+544

-68

libde265/sps.cc less more

28	28
29	29	#define READ_VLC_OFFSET(variable, vlctype, offset) \
30	30	if ((vlc = get_ ## vlctype(br)) == UVLC_ERROR) { \
31		ctx->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false); \
	31	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false); \
32	32	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; \
33	33	} \
34	34	variable = vlc + offset;

45	45
46	46
47	47	// TODO: should be in some header-file of refpic.c
48		extern bool read_short_term_ref_pic_set(decoder_context* ctx,
	48	extern bool read_short_term_ref_pic_set(error_queue* errqueue,
49	49	const seq_parameter_set* sps,
50	50	bitreader* br,
51	51	ref_pic_set* out_set,

53	53	const std::vector<ref_pic_set>& sets,
54	54	bool sliceRefPicSet);
55	55
	56	extern bool write_short_term_ref_pic_set(error_queue* errqueue,
	57	const seq_parameter_set* sps,
	58	CABAC_encoder& out,
	59	const ref_pic_set* in_set, // which set to write
	60	int idxRps, // index of the set to be read
	61	const std::vector<ref_pic_set>& sets, // previously read sets
	62	bool sliceRefPicSet); // is this in the slice header?
	63
	64
	65	sps_range_extension::sps_range_extension()
	66	{
	67	transform_skip_rotation_enabled_flag = 0;
	68	transform_skip_context_enabled_flag = 0;
	69	implicit_rdpcm_enabled_flag = 0;
	70	explicit_rdpcm_enabled_flag = 0;
	71	extended_precision_processing_flag = 0;
	72	intra_smoothing_disabled_flag = 0;
	73	high_precision_offsets_enabled_flag = 0;
	74	persistent_rice_adaptation_enabled_flag = 0;
	75	cabac_bypass_alignment_enabled_flag = 0;
	76	}
	77
	78
56	79	seq_parameter_set::seq_parameter_set()
57	80	{
58	81	// TODO: this is dangerous

69	92	}
70	93
71	94
72		de265_error seq_parameter_set::read(decoder_context* ctx, bitreader* br)
	95	void seq_parameter_set::set_defaults(enum PresetSet)
	96	{
	97	video_parameter_set_id = 0;
	98	sps_max_sub_layers = 1;
	99	sps_temporal_id_nesting_flag = 1;
	100
	101	profile_tier_level_.general.set_defaults(Profile_Main, 6,2); // TODO
	102
	103	seq_parameter_set_id = 0;
	104	chroma_format_idc = 1;
	105	ChromaArrayType = chroma_format_idc;
	106
	107	separate_colour_plane_flag = 0;
	108	pic_width_in_luma_samples = 0;
	109	pic_height_in_luma_samples = 0;
	110	conformance_window_flag = 0;
	111
	112	conf_win_left_offset = 0;
	113	conf_win_right_offset = 0;
	114	conf_win_top_offset = 0;
	115	conf_win_bottom_offset = 0;
	116
	117	bit_depth_luma =8;
	118	bit_depth_chroma=8;
	119
	120	log2_max_pic_order_cnt_lsb = 8;
	121	sps_sub_layer_ordering_info_present_flag = 0;
	122
	123	sps_max_dec_pic_buffering[0] = 1;
	124	sps_max_num_reorder_pics[0] = 0;
	125	sps_max_latency_increase_plus1[0] = 0;
	126
	127	set_CB_log2size_range(4,4);
	128	set_TB_log2size_range(3,4);
	129	max_transform_hierarchy_depth_inter = 1;
	130	max_transform_hierarchy_depth_intra = 1;
	131
	132	scaling_list_enable_flag = 0;
	133	sps_scaling_list_data_present_flag = 0;
	134
	135	// TODO struct scaling_list_data scaling_list;
	136
	137	amp_enabled_flag = 0;
	138	sample_adaptive_offset_enabled_flag = 0;
	139	pcm_enabled_flag = 0;
	140
	141	pcm_sample_bit_depth_luma = 8;
	142	pcm_sample_bit_depth_chroma = 8;
	143	// TODO log2_min_pcm_luma_coding_block_size;
	144	// TODO log2_diff_max_min_pcm_luma_coding_block_size;
	145	pcm_loop_filter_disable_flag = 1;
	146
	147	// num_short_term_ref_pic_sets = 0;
	148	// std::vector<ref_pic_set> ref_pic_sets; // [0 ; num_short_term_ref_pic_set (<=MAX_REF_PIC_SETS) )
	149	ref_pic_sets.clear();
	150
	151	long_term_ref_pics_present_flag = 0;
	152
	153	num_long_term_ref_pics_sps = 0;
	154
	155	/* TODO
	156	int lt_ref_pic_poc_lsb_sps[MAX_NUM_LT_REF_PICS_SPS];
	157	char used_by_curr_pic_lt_sps_flag[MAX_NUM_LT_REF_PICS_SPS];
	158	*/
	159
	160	sps_temporal_mvp_enabled_flag = 0;
	161	strong_intra_smoothing_enable_flag = 0;
	162	vui_parameters_present_flag = 0;
	163
	164	/*
	165	if( vui_parameters_present_flag )
	166	vui_parameters()
	167	*/
	168
	169	sps_extension_present_flag = 0;
	170	sps_range_extension_flag = 0;
	171	sps_multilayer_extension_flag = 0;
	172	sps_extension_6bits = 0;
	173	}
	174
	175
	176	void seq_parameter_set::set_CB_log2size_range(int mini,int maxi)
	177	{
	178	log2_min_luma_coding_block_size = mini;
	179	log2_diff_max_min_luma_coding_block_size = maxi-mini;
	180	}
	181
	182
	183	void seq_parameter_set::set_TB_log2size_range(int mini,int maxi)
	184	{
	185	log2_min_transform_block_size = mini;
	186	log2_diff_max_min_transform_block_size = maxi-mini;
	187	}
	188
	189
	190	void seq_parameter_set::set_resolution(int w,int h)
	191	{
	192	pic_width_in_luma_samples = w;
	193	pic_height_in_luma_samples = h;
	194	}
	195
	196
	197	de265_error seq_parameter_set::read(error_queue* errqueue, bitreader* br)
73	198	{
74	199	int vlc;
75	200

81	206
82	207	sps_temporal_id_nesting_flag = get_bits(br,1);
83	208
84		read_profile_tier_level(br,&profile_tier_level, sps_max_sub_layers);
	209	profile_tier_level_.read(br, sps_max_sub_layers);
85	210
86	211	READ_VLC(seq_parameter_set_id, uvlc);
	212	if (seq_parameter_set_id >= DE265_MAX_SPS_SETS) {
	213	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	214	}
87	215
88	216
89	217	// --- decode chroma type ---

106	234
107	235	if (chroma_format_idc<0 \|\|
108	236	chroma_format_idc>3) {
109		ctx->add_warning(DE265_WARNING_INVALID_CHROMA_FORMAT, false);
110		return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
111		}
112
113		SubWidthC = SubWidthC_tab [chroma_format_idc];
114		SubHeightC = SubHeightC_tab[chroma_format_idc];
	237	errqueue->add_warning(DE265_WARNING_INVALID_CHROMA_FORMAT, false);
	238	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	239	}
115	240
116	241
117	242	// --- picture size ---
118	243
119	244	READ_VLC(pic_width_in_luma_samples, uvlc);
120	245	READ_VLC(pic_height_in_luma_samples, uvlc);
	246
	247	if (pic_width_in_luma_samples == 0 \|\|
	248	pic_height_in_luma_samples == 0) {
	249	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	250	}
	251
	252	if (pic_width_in_luma_samples > MAX_PICTURE_WIDTH \|\|
	253	pic_height_in_luma_samples> MAX_PICTURE_HEIGHT) {
	254	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	255	}
121	256
122	257	conformance_window_flag = get_bits(br,1);
123	258

134	269	conf_win_bottom_offset= 0;
135	270	}
136	271
137		if (ChromaArrayType==0) {
138		WinUnitX = 1;
139		WinUnitY = 1;
140		}
141		else {
142		WinUnitX = SubWidthC_tab [chroma_format_idc];
143		WinUnitY = SubHeightC_tab[chroma_format_idc];
144		}
145
146
147	272	READ_VLC_OFFSET(bit_depth_luma, uvlc, 8);
148	273	READ_VLC_OFFSET(bit_depth_chroma,uvlc, 8);
149	274

165	290	vlc=get_uvlc(br);
166	291	if (vlc == UVLC_ERROR \|\|
167	292	vlc+1 > MAX_NUM_REF_PICS) {
168		ctx->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	293	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
169	294	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
170	295	}
171	296

204	329	READ_VLC(log2_diff_max_min_transform_block_size, uvlc);
205	330	READ_VLC(max_transform_hierarchy_depth_inter, uvlc);
206	331	READ_VLC(max_transform_hierarchy_depth_intra, uvlc);
	332
	333	if (log2_min_luma_coding_block_size > 6) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	334	if (log2_min_luma_coding_block_size + log2_diff_max_min_luma_coding_block_size > 6) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	335	if (log2_min_transform_block_size > 5) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	336	if (log2_min_transform_block_size + log2_diff_max_min_transform_block_size > 5) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	337
207	338	scaling_list_enable_flag = get_bits(br,1);
208	339
209	340	if (scaling_list_enable_flag) {

239	370	pcm_loop_filter_disable_flag = 0;
240	371	}
241	372
	373	int num_short_term_ref_pic_sets;
242	374	READ_VLC(num_short_term_ref_pic_sets, uvlc);
243	375	if (num_short_term_ref_pic_sets < 0 \|\|
244	376	num_short_term_ref_pic_sets > 64) {
245		ctx->add_warning(DE265_WARNING_NUMBER_OF_SHORT_TERM_REF_PIC_SETS_OUT_OF_RANGE, false);
	377	errqueue->add_warning(DE265_WARNING_NUMBER_OF_SHORT_TERM_REF_PIC_SETS_OUT_OF_RANGE, false);
246	378	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
247	379	}
248	380

254	386
255	387	for (int i = 0; i < num_short_term_ref_pic_sets; i++) {
256	388
257		bool success = read_short_term_ref_pic_set(ctx,this,br,
	389	bool success = read_short_term_ref_pic_set(errqueue,this,br,
258	390	&ref_pic_sets[i], i,
259	391	ref_pic_sets,
260	392	false);

286	418
287	419	sps_temporal_mvp_enabled_flag = get_bits(br,1);
288	420	strong_intra_smoothing_enable_flag = get_bits(br,1);
	421
289	422	vui_parameters_present_flag = get_bits(br,1);
290
291		#if 0
292	423	if (vui_parameters_present_flag) {
293		assert(false);
294		/*
295		vui_parameters()
296
297		sps_extension_flag
298		u(1)
299		if( sps_extension_flag )
300
301		while( more_rbsp_data() )
302
303		sps_extension_data_flag
304		u(1)
305		rbsp_trailing_bits()
306		*/
307		}
308
	424	vui.read(errqueue, br, this);
	425	}
	426
	427
	428	sps_extension_present_flag = get_bits(br,1);
	429	if (sps_extension_present_flag) {
	430	sps_range_extension_flag = get_bits(br,1);
	431	sps_multilayer_extension_flag = get_bits(br,1);
	432	sps_extension_6bits = get_bits(br,6);
	433	}
	434	else {
	435	sps_range_extension_flag = 0;
	436	}
	437
	438	if (sps_range_extension_flag) {
	439	de265_error err = range_extension.read(errqueue, br);
	440	if (err != DE265_OK) { return err; }
	441	}
	442
	443	/*
309	444	sps_extension_flag = get_bits(br,1);
310	445	if (sps_extension_flag) {
311	446	assert(false);
312	447	}
313
314		check_rbsp_trailing_bits(br);
315		#endif
316
	448	*/
	449
	450
	451	de265_error err = compute_derived_values();
	452	if (err != DE265_OK) { return err; }
	453
	454	sps_read = true;
	455
	456	return DE265_OK;
	457	}
	458
	459
	460	de265_error seq_parameter_set::compute_derived_values()
	461	{
317	462	// --- compute derived values ---
	463
	464	SubWidthC = SubWidthC_tab [chroma_format_idc];
	465	SubHeightC = SubHeightC_tab[chroma_format_idc];
	466
	467	if (ChromaArrayType==0) {
	468	WinUnitX = 1;
	469	WinUnitY = 1;
	470	}
	471	else {
	472	WinUnitX = SubWidthC_tab [chroma_format_idc];
	473	WinUnitY = SubHeightC_tab[chroma_format_idc];
	474	}
	475
	476
318	477
319	478	BitDepth_Y = bit_depth_luma;
320	479	QpBdOffset_Y = 6*(bit_depth_luma-8);

325	484	Log2CtbSizeY = Log2MinCbSizeY + log2_diff_max_min_luma_coding_block_size;
326	485	MinCbSizeY = 1 << Log2MinCbSizeY;
327	486	CtbSizeY = 1 << Log2CtbSizeY;
328		PicWidthInMinCbsY = pic_width_in_luma_samples / MinCbSizeY;
	487
	488	PicWidthInMinCbsY = ceil_div(pic_width_in_luma_samples, MinCbSizeY);
329	489	PicWidthInCtbsY = ceil_div(pic_width_in_luma_samples, CtbSizeY);
330		PicHeightInMinCbsY = pic_height_in_luma_samples / MinCbSizeY;
	490	PicHeightInMinCbsY = ceil_div(pic_height_in_luma_samples, MinCbSizeY);
331	491	PicHeightInCtbsY = ceil_div(pic_height_in_luma_samples,CtbSizeY);
332	492	PicSizeInMinCbsY = PicWidthInMinCbsY * PicHeightInMinCbsY;
333	493	PicSizeInCtbsY = PicWidthInCtbsY * PicHeightInCtbsY;

345	505	Log2MinTrafoSize = log2_min_transform_block_size;
346	506	Log2MaxTrafoSize = log2_min_transform_block_size + log2_diff_max_min_transform_block_size;
347	507
	508	if (max_transform_hierarchy_depth_inter > Log2CtbSizeY - Log2MinTrafoSize) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	509	if (max_transform_hierarchy_depth_intra > Log2CtbSizeY - Log2MinTrafoSize) { return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; }
	510
348	511	Log2MinPUSize = Log2MinCbSizeY-1;
349	512	PicWidthInMinPUs = PicWidthInCtbsY << (Log2CtbSizeY - Log2MinPUSize);
350	513	PicHeightInMinPUs = PicHeightInCtbsY << (Log2CtbSizeY - Log2MinPUSize);

358	521	PicHeightInTbsY = PicHeightInCtbsY << (Log2CtbSizeY - Log2MinTrafoSize);
359	522	PicSizeInTbsY = PicWidthInTbsY * PicHeightInTbsY;
360	523
	524
	525	if (range_extension.high_precision_offsets_enabled_flag) {
	526	WpOffsetBdShiftY = 0;
	527	WpOffsetBdShiftC = 0;
	528	WpOffsetHalfRangeY = 1 << (BitDepth_Y - 1);
	529	WpOffsetHalfRangeC = 1 << (BitDepth_C - 1);
	530	}
	531	else {
	532	WpOffsetBdShiftY = ( BitDepth_Y - 8 );
	533	WpOffsetBdShiftC = ( BitDepth_C - 8 );
	534	WpOffsetHalfRangeY = 1 << 7;
	535	WpOffsetHalfRangeC = 1 << 7;
	536	}
	537
	538
	539	// --- check SPS sanity ---
	540
	541	if (pic_width_in_luma_samples % MinCbSizeY != 0 \|\|
	542	pic_height_in_luma_samples % MinCbSizeY != 0) {
	543	// TODO: warn that image size is coded wrong in bitstream (must be multiple of MinCbSizeY)
	544	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	545	}
	546
	547	if (Log2MinTrafoSize > Log2MinCbSizeY) {
	548	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	549	}
	550
	551	if (Log2MaxTrafoSize > libde265_min(Log2CtbSizeY,5)) {
	552	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	553	}
	554
	555
	556	if (BitDepth_Y < 8 \|\| BitDepth_Y > 16) {
	557	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	558	}
	559
	560	if (BitDepth_C < 8 \|\| BitDepth_C > 16) {
	561	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	562	}
	563
	564
361	565	sps_read = true;
362	566
363	567	return DE265_OK;

365	569
366	570
367	571
368		void seq_parameter_set::dump_sps(int fd) const
	572	void seq_parameter_set::dump(int fd) const
369	573	{
370	574	//#if (_MSC_VER >= 1500)
371	575	//#define LOG0(t) loginfo(LogHeaders, t)

389	593	LOG1("sps_max_sub_layers : %d\n", sps_max_sub_layers);
390	594	LOG1("sps_temporal_id_nesting_flag : %d\n", sps_temporal_id_nesting_flag);
391	595
392		dump_profile_tier_level(&profile_tier_level, sps_max_sub_layers, fh);
	596	profile_tier_level_.dump(sps_max_sub_layers, fh);
393	597
394	598	LOG1("seq_parameter_set_id : %d\n", seq_parameter_set_id);
395	599	LOG2("chroma_format_idc : %d (%s)\n", chroma_format_idc,
	600	chroma_format_idc == 0 ? "monochrome" :
396	601	chroma_format_idc == 1 ? "4:2:0" :
397	602	chroma_format_idc == 2 ? "4:2:2" :
398	603	chroma_format_idc == 3 ? "4:4:4" : "unknown");

459	664	LOG1("pcm_loop_filter_disable_flag : %d\n", pcm_loop_filter_disable_flag);
460	665	}
461	666
462		LOG1("num_short_term_ref_pic_sets : %d\n", num_short_term_ref_pic_sets);
463
464		for (int i = 0; i < num_short_term_ref_pic_sets; i++) {
	667	LOG1("num_short_term_ref_pic_sets : %d\n", ref_pic_sets.size());
	668
	669	for (int i = 0; i < ref_pic_sets.size(); i++) {
465	670	LOG1("ref_pic_set[ %2d ]: ",i);
466	671	dump_compact_short_term_ref_pic_set(&ref_pic_sets[i], 16, fh);
467	672	}

481	686	LOG1("sps_temporal_mvp_enabled_flag : %d\n", sps_temporal_mvp_enabled_flag);
482	687	LOG1("strong_intra_smoothing_enable_flag : %d\n", strong_intra_smoothing_enable_flag);
483	688	LOG1("vui_parameters_present_flag : %d\n", vui_parameters_present_flag);
	689
	690	LOG1("sps_extension_present_flag : %d\n", sps_extension_present_flag);
	691	LOG1("sps_range_extension_flag : %d\n", sps_range_extension_flag);
	692	LOG1("sps_multilayer_extension_flag : %d\n", sps_multilayer_extension_flag);
	693	LOG1("sps_extension_6bits : %d\n", sps_extension_6bits);
484	694
485	695	LOG1("CtbSizeY : %d\n", CtbSizeY);
486	696	LOG1("MinCbSizeY : %d\n", MinCbSizeY);

488	698	LOG1("MinTBSizeY : %d\n", 1<<log2_min_transform_block_size);
489	699	LOG1("MaxTBSizeY : %d\n", 1<<(log2_min_transform_block_size + log2_diff_max_min_transform_block_size));
490	700
	701	LOG1("PicWidthInCtbsY : %d\n", PicWidthInCtbsY);
	702	LOG1("PicHeightInCtbsY : %d\n", PicHeightInCtbsY);
491	703	LOG1("SubWidthC : %d\n", SubWidthC);
492	704	LOG1("SubHeightC : %d\n", SubHeightC);
493	705
494		return;
	706	if (sps_range_extension_flag) {
	707	range_extension.dump(fd);
	708	}
495	709
496	710	if (vui_parameters_present_flag) {
497		assert(false);
498		/*
499		vui_parameters()
500
501		sps_extension_flag
502		u(1)
503		if( sps_extension_flag )
504
505		while( more_rbsp_data() )
506
507		sps_extension_data_flag
508		u(1)
509		rbsp_trailing_bits()
510		*/
	711	vui.dump(fd);
511	712	}
512	713	#undef LOG0
513	714	#undef LOG1

645	846	char scaling_list_pred_mode_flag = get_bits(br,1);
646	847	if (!scaling_list_pred_mode_flag) {
647	848	int scaling_list_pred_matrix_id_delta = get_uvlc(br);
648		if (scaling_list_pred_matrix_id_delta < 0 \|\|
	849	if (scaling_list_pred_matrix_id_delta == UVLC_ERROR \|\|
649	850	scaling_list_pred_matrix_id_delta > matrixId) {
650	851	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
651	852	}

742	943	}
743	944
744	945
	946	de265_error write_scaling_list(CABAC_encoder& out, const seq_parameter_set* sps,
	947	scaling_list_data* sclist, bool inPPS)
	948	{
	949	assert(false);
	950	// TODO
	951
	952	return DE265_OK;
	953	}
	954
	955
745	956	void set_default_scaling_lists(scaling_list_data* sclist)
746	957	{
747	958	// 4x4

777	988	default_ScalingList_8x8_inter, 3);
778	989	}
779	990
	991
	992	de265_error seq_parameter_set::write(error_queue* errqueue, CABAC_encoder& out)
	993	{
	994	out.write_bits(video_parameter_set_id, 4);
	995	if (sps_max_sub_layers>7) {
	996	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	997	}
	998	out.write_bits(sps_max_sub_layers-1, 3);
	999
	1000	out.write_bit(sps_temporal_id_nesting_flag);
	1001
	1002	profile_tier_level_.write(out, sps_max_sub_layers);
	1003
	1004	out.write_uvlc(seq_parameter_set_id);
	1005
	1006
	1007	// --- encode chroma type ---
	1008
	1009	out.write_uvlc(chroma_format_idc);
	1010
	1011	if (chroma_format_idc<0 \|\|
	1012	chroma_format_idc>3) {
	1013	errqueue->add_warning(DE265_WARNING_INVALID_CHROMA_FORMAT, false);
	1014	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	1015	}
	1016
	1017	if (chroma_format_idc == 3) {
	1018	out.write_bit(separate_colour_plane_flag);
	1019	}
	1020
	1021
	1022	// --- picture size ---
	1023
	1024	out.write_uvlc(pic_width_in_luma_samples);
	1025	out.write_uvlc(pic_height_in_luma_samples);
	1026
	1027	out.write_bit(conformance_window_flag);
	1028
	1029	if (conformance_window_flag) {
	1030	out.write_uvlc(conf_win_left_offset);
	1031	out.write_uvlc(conf_win_right_offset);
	1032	out.write_uvlc(conf_win_top_offset);
	1033	out.write_uvlc(conf_win_bottom_offset);
	1034	}
	1035
	1036
	1037	out.write_uvlc(bit_depth_luma-8);
	1038	out.write_uvlc(bit_depth_chroma-8);
	1039
	1040	out.write_uvlc(log2_max_pic_order_cnt_lsb-4);
	1041
	1042
	1043	// --- sub_layer_ordering_info ---
	1044
	1045	out.write_bit(sps_sub_layer_ordering_info_present_flag);
	1046
	1047	int firstLayer = (sps_sub_layer_ordering_info_present_flag ?
	1048	0 : sps_max_sub_layers-1 );
	1049
	1050	for (int i=firstLayer ; i <= sps_max_sub_layers-1; i++ ) {
	1051
	1052	// sps_max_dec_pic_buffering[i]
	1053
	1054	if (sps_max_dec_pic_buffering[i] > MAX_NUM_REF_PICS) {
	1055	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	1056	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	1057	}
	1058
	1059	out.write_uvlc(sps_max_dec_pic_buffering[i]-1);
	1060
	1061	// sps_max_num_reorder_pics[i]
	1062
	1063	out.write_uvlc(sps_max_num_reorder_pics[i]);
	1064
	1065
	1066	// sps_max_latency_increase[i]
	1067
	1068	out.write_uvlc(sps_max_latency_increase_plus1[i]);
	1069	}
	1070
	1071
	1072	out.write_uvlc(log2_min_luma_coding_block_size-3);
	1073	out.write_uvlc(log2_diff_max_min_luma_coding_block_size);
	1074	out.write_uvlc(log2_min_transform_block_size-2);
	1075	out.write_uvlc(log2_diff_max_min_transform_block_size);
	1076	out.write_uvlc(max_transform_hierarchy_depth_inter);
	1077	out.write_uvlc(max_transform_hierarchy_depth_intra);
	1078	out.write_bit(scaling_list_enable_flag);
	1079
	1080	if (scaling_list_enable_flag) {
	1081
	1082	out.write_bit(sps_scaling_list_data_present_flag);
	1083	if (sps_scaling_list_data_present_flag) {
	1084
	1085	de265_error err;
	1086	if ((err=write_scaling_list(out,this, &scaling_list, false)) != DE265_OK) {
	1087	return err;
	1088	}
	1089	}
	1090	}
	1091
	1092	out.write_bit(amp_enabled_flag);
	1093	out.write_bit(sample_adaptive_offset_enabled_flag);
	1094	out.write_bit(pcm_enabled_flag);
	1095	if (pcm_enabled_flag) {
	1096	out.write_bits(pcm_sample_bit_depth_luma -1,4);
	1097	out.write_bits(pcm_sample_bit_depth_chroma-1,4);
	1098	out.write_uvlc(log2_min_pcm_luma_coding_block_size-3);
	1099	out.write_uvlc(log2_diff_max_min_pcm_luma_coding_block_size);
	1100	out.write_bit(pcm_loop_filter_disable_flag);
	1101	}
	1102
	1103	int num_short_term_ref_pic_sets = ref_pic_sets.size();
	1104	if (num_short_term_ref_pic_sets < 0 \|\|
	1105	num_short_term_ref_pic_sets > 64) {
	1106	errqueue->add_warning(DE265_WARNING_NUMBER_OF_SHORT_TERM_REF_PIC_SETS_OUT_OF_RANGE, false);
	1107	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	1108	}
	1109	out.write_uvlc(num_short_term_ref_pic_sets);
	1110
	1111	// --- allocate reference pic set ---
	1112
	1113	// we do not allocate the ref-pic-set for the slice header here, but in the slice header itself
	1114
	1115	for (int i = 0; i < num_short_term_ref_pic_sets; i++) {
	1116
	1117	bool success = write_short_term_ref_pic_set(errqueue,this,out,
	1118	&ref_pic_sets[i], i,
	1119	ref_pic_sets,
	1120	false);
	1121
	1122	if (!success) {
	1123	return DE265_WARNING_SPS_HEADER_INVALID;
	1124	}
	1125
	1126	// dump_short_term_ref_pic_set(&(*ref_pic_sets)[i], fh);
	1127	}
	1128
	1129	out.write_bit(long_term_ref_pics_present_flag);
	1130
	1131	if (long_term_ref_pics_present_flag) {
	1132
	1133	if (num_long_term_ref_pics_sps > MAX_NUM_LT_REF_PICS_SPS) {
	1134	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	1135	}
	1136	out.write_uvlc(num_long_term_ref_pics_sps);
	1137
	1138	for (int i = 0; i < num_long_term_ref_pics_sps; i++ ) {
	1139	out.write_bits(lt_ref_pic_poc_lsb_sps[i], log2_max_pic_order_cnt_lsb);
	1140	out.write_bit (used_by_curr_pic_lt_sps_flag[i]);
	1141	}
	1142	}
	1143
	1144	out.write_bit(sps_temporal_mvp_enabled_flag);
	1145	out.write_bit(strong_intra_smoothing_enable_flag);
	1146	out.write_bit(vui_parameters_present_flag);
	1147
	1148	#if 0
	1149	if (vui_parameters_present_flag) {
	1150	assert(false);
	1151	/*
	1152	vui_parameters()
	1153	sps_extension_flag
	1154	u(1)
	1155	if( sps_extension_flag )
	1156	while( more_rbsp_data() )
	1157	sps_extension_data_flag
	1158	u(1)
	1159	rbsp_trailing_bits()
	1160	*/
	1161	}
	1162	#endif
	1163
	1164	out.write_bit(sps_extension_present_flag);
	1165
	1166	#if 0
	1167	if (sps_extension_flag) {
	1168	assert(false);
	1169	}
	1170	check_rbsp_trailing_bits(br);
	1171	#endif
	1172
	1173	// --- compute derived values ---
	1174
	1175	#if 0
	1176	BitDepth_Y = bit_depth_luma;
	1177	QpBdOffset_Y = 6*(bit_depth_luma-8);
	1178	BitDepth_C = bit_depth_chroma;
	1179	QpBdOffset_C = 6*(bit_depth_chroma-8);
	1180	Log2MinCbSizeY = log2_min_luma_coding_block_size;
	1181	Log2CtbSizeY = Log2MinCbSizeY + log2_diff_max_min_luma_coding_block_size;
	1182	MinCbSizeY = 1 << Log2MinCbSizeY;
	1183	CtbSizeY = 1 << Log2CtbSizeY;
	1184	PicWidthInMinCbsY = pic_width_in_luma_samples / MinCbSizeY;
	1185	PicWidthInCtbsY = ceil_div(pic_width_in_luma_samples, CtbSizeY);
	1186	PicHeightInMinCbsY = pic_height_in_luma_samples / MinCbSizeY;
	1187	PicHeightInCtbsY = ceil_div(pic_height_in_luma_samples,CtbSizeY);
	1188	PicSizeInMinCbsY = PicWidthInMinCbsY * PicHeightInMinCbsY;
	1189	PicSizeInCtbsY = PicWidthInCtbsY * PicHeightInCtbsY;
	1190	PicSizeInSamplesY = pic_width_in_luma_samples * pic_height_in_luma_samples;
	1191	if (chroma_format_idc==0 \|\| separate_colour_plane_flag) {
	1192	CtbWidthC = 0;
	1193	CtbHeightC = 0;
	1194	}
	1195	else {
	1196	CtbWidthC = CtbSizeY / SubWidthC;
	1197	CtbHeightC = CtbSizeY / SubHeightC;
	1198	}
	1199	Log2MinTrafoSize = log2_min_transform_block_size;
	1200	Log2MaxTrafoSize = log2_min_transform_block_size + log2_diff_max_min_transform_block_size;
	1201	Log2MinPUSize = Log2MinCbSizeY-1;
	1202	PicWidthInMinPUs = PicWidthInCtbsY << (Log2CtbSizeY - Log2MinPUSize);
	1203	PicHeightInMinPUs = PicHeightInCtbsY << (Log2CtbSizeY - Log2MinPUSize);
	1204	Log2MinIpcmCbSizeY = log2_min_pcm_luma_coding_block_size;
	1205	Log2MaxIpcmCbSizeY = (log2_min_pcm_luma_coding_block_size +
	1206	log2_diff_max_min_pcm_luma_coding_block_size);
	1207	// the following are not in the standard
	1208	PicWidthInTbsY = PicWidthInCtbsY << (Log2CtbSizeY - Log2MinTrafoSize);
	1209	PicHeightInTbsY = PicHeightInCtbsY << (Log2CtbSizeY - Log2MinTrafoSize);
	1210	PicSizeInTbsY = PicWidthInTbsY * PicHeightInTbsY;
	1211	sps_read = true;
	1212	#endif
	1213
	1214	return DE265_OK;
	1215	}
	1216
	1217
	1218	de265_error sps_range_extension::read(error_queue* errqueue, bitreader* br)
	1219	{
	1220	transform_skip_rotation_enabled_flag = get_bits(br,1);
	1221	transform_skip_context_enabled_flag = get_bits(br,1);
	1222	implicit_rdpcm_enabled_flag = get_bits(br,1);
	1223	explicit_rdpcm_enabled_flag = get_bits(br,1);
	1224	extended_precision_processing_flag = get_bits(br,1);
	1225	intra_smoothing_disabled_flag = get_bits(br,1);
	1226	high_precision_offsets_enabled_flag = get_bits(br,1);
	1227	persistent_rice_adaptation_enabled_flag = get_bits(br,1);
	1228	cabac_bypass_alignment_enabled_flag = get_bits(br,1);
	1229
	1230	return DE265_OK;
	1231	}
	1232
	1233
	1234	#define LOG0(t) log2fh(fh, t)
	1235	#define LOG1(t,d) log2fh(fh, t,d)
	1236	void sps_range_extension::dump(int fd) const
	1237	{
	1238	FILE* fh;
	1239	if (fd==1) fh=stdout;
	1240	else if (fd==2) fh=stderr;
	1241	else { return; }
	1242
	1243	LOG0("----------------- SPS-range-extension -----------------\n");
	1244	LOG1("transform_skip_rotation_enabled_flag : %d\n", transform_skip_rotation_enabled_flag);
	1245	LOG1("transform_skip_context_enabled_flag : %d\n", transform_skip_context_enabled_flag);
	1246	LOG1("implicit_rdpcm_enabled_flag : %d\n", implicit_rdpcm_enabled_flag);
	1247	LOG1("explicit_rdpcm_enabled_flag : %d\n", explicit_rdpcm_enabled_flag);
	1248	LOG1("extended_precision_processing_flag : %d\n", extended_precision_processing_flag);
	1249	LOG1("intra_smoothing_disabled_flag : %d\n", intra_smoothing_disabled_flag);
	1250	LOG1("high_precision_offsets_enabled_flag : %d\n", high_precision_offsets_enabled_flag);
	1251	LOG1("persistent_rice_adaptation_enabled_flag : %d\n", persistent_rice_adaptation_enabled_flag);
	1252	LOG1("cabac_bypass_alignment_enabled_flag : %d\n", cabac_bypass_alignment_enabled_flag);
	1253	}
	1254	#undef LOG1
	1255	#undef LOG0

+86

-25

libde265/sps.h less more

21	21	#define DE265_SPS_H
22	22
23	23	#include "libde265/vps.h"
	24	#include "libde265/vui.h"
24	25	#include "libde265/bitstream.h"
25	26	#include "libde265/refpic.h"
26	27	#include "libde265/de265.h"
	28	#include "libde265/cabac.h"
27	29
28	30	#include <vector>
29	31
30		#define MAX_REF_PIC_SETS 64 // maximum according to standard
	32	class error_queue;
	33
	34	// #define MAX_REF_PIC_SETS 64 // maximum according to standard
31	35	#define MAX_NUM_LT_REF_PICS_SPS 32
32	36
	37	// this is just a safety range
	38	#define MAX_PICTURE_WIDTH 70000
	39	#define MAX_PICTURE_HEIGHT 70000
33	40
34	41	enum {
35	42	CHROMA_MONO = 0,

50	57	} scaling_list_data;
51	58
52	59
53		struct seq_parameter_set {
	60	enum PresetSet {
	61	Preset_Default
	62	};
	63
	64
	65	class sps_range_extension
	66	{
	67	public:
	68	sps_range_extension();
	69
	70	uint8_t transform_skip_rotation_enabled_flag;
	71	uint8_t transform_skip_context_enabled_flag;
	72	uint8_t implicit_rdpcm_enabled_flag;
	73	uint8_t explicit_rdpcm_enabled_flag;
	74	uint8_t extended_precision_processing_flag;
	75	uint8_t intra_smoothing_disabled_flag;
	76	uint8_t high_precision_offsets_enabled_flag;
	77	uint8_t persistent_rice_adaptation_enabled_flag;
	78	uint8_t cabac_bypass_alignment_enabled_flag;
	79
	80	de265_error read(error_queue, bitreader);
	81	void dump(int fd) const;
	82	};
	83
	84
	85	class seq_parameter_set {
	86	public:
54	87	seq_parameter_set();
55	88	~seq_parameter_set();
56	89
57		de265_error read(struct decoder_context, bitreader);
58
59		void dump_sps(int fd) const;
	90	de265_error read(error_queue, bitreader);
	91	de265_error write(error_queue*, CABAC_encoder&);
	92
	93	void dump(int fd) const;
	94
	95	void set_defaults(enum PresetSet = Preset_Default);
	96	void set_CB_log2size_range(int mini,int maxi);
	97	void set_TB_log2size_range(int mini,int maxi);
	98	void set_resolution(int w,int h);
60	99
61	100	bool sps_read; // whether the sps has been read from the bitstream
62	101
63	102
64	103	char video_parameter_set_id;
65		char sps_max_sub_layers;
	104	char sps_max_sub_layers; // [1;7]
66	105	char sps_temporal_id_nesting_flag;
67	106
68		struct profile_tier_level profile_tier_level;
	107	profile_tier_level profile_tier_level_;
69	108
70	109	int seq_parameter_set_id;
71	110	int chroma_format_idc;

90	129	int sps_max_num_reorder_pics[7];
91	130	int sps_max_latency_increase_plus1[7];
92	131
93		int log2_min_luma_coding_block_size;
94		int log2_diff_max_min_luma_coding_block_size;
95		int log2_min_transform_block_size;
96		int log2_diff_max_min_transform_block_size;
	132	int log2_min_luma_coding_block_size; // smallest CB size [3;6]
	133	int log2_diff_max_min_luma_coding_block_size; // largest CB size
	134	int log2_min_transform_block_size; // smallest TB size [2;5]
	135	int log2_diff_max_min_transform_block_size; // largest TB size
97	136	int max_transform_hierarchy_depth_inter;
98	137	int max_transform_hierarchy_depth_intra;
99	138

113	152	int log2_diff_max_min_pcm_luma_coding_block_size;
114	153	char pcm_loop_filter_disable_flag;
115	154
116		int num_short_term_ref_pic_sets;
	155	int num_short_term_ref_pic_sets() const { return ref_pic_sets.size(); }
117	156	std::vector<ref_pic_set> ref_pic_sets; // [0 ; num_short_term_ref_pic_set (<=MAX_REF_PIC_SETS) )
118
119		/*
120		for( i = 0; i < num_short_term_ref_pic_sets; i++)
121		short_term_ref_pic_set(i)
122		*/
123	157
124	158	char long_term_ref_pics_present_flag;
125	159

130	164
131	165	char sps_temporal_mvp_enabled_flag;
132	166	char strong_intra_smoothing_enable_flag;
	167
133	168	char vui_parameters_present_flag;
134
135		/*
136		if( vui_parameters_present_flag )
137		vui_parameters()
138		*/
139
140		char sps_extension_flag;
	169	video_usability_information vui;
	170
	171	char sps_extension_present_flag;
	172	char sps_range_extension_flag;
	173	char sps_multilayer_extension_flag;
	174	char sps_extension_6bits;
	175
	176	sps_range_extension range_extension;
141	177
142	178	/*
143	179	if( sps_extension_flag )

147	183	rbsp_trailing_bits()
148	184	*/
149	185
	186
	187	// --- derived values ---
	188
	189	de265_error compute_derived_values();
	190
150	191	int BitDepth_Y;
151	192	int QpBdOffset_Y;
152	193	int BitDepth_C;

157	198	int WinUnitX, WinUnitY;
158	199
159	200	int MaxPicOrderCntLsb;
160
	201
161	202	int Log2MinCbSizeY;
162	203	int Log2CtbSizeY;
163	204	int MinCbSizeY;

187	228	int Log2MaxIpcmCbSizeY;
188	229
189	230	int SpsMaxLatencyPictures[7]; // [temporal layer]
	231
	232	uint8_t WpOffsetBdShiftY;
	233	uint8_t WpOffsetBdShiftC;
	234	int32_t WpOffsetHalfRangeY;
	235	int32_t WpOffsetHalfRangeC;
	236
	237
	238	int getPUIndexRS(int pixelX,int pixelY) const {
	239	return (pixelX>>Log2MinPUSize) + (pixelY>>Log2MinPUSize)*PicWidthInMinPUs;
	240	}
	241
	242	int get_bit_depth(int cIdx) const {
	243	if (cIdx==0) return BitDepth_Y;
	244	else return BitDepth_C;
	245	}
	246
	247	int get_chroma_shift_W(int cIdx) const { return cIdx ? SubWidthC -1 : 0; }
	248	int get_chroma_shift_H(int cIdx) const { return cIdx ? SubHeightC-1 : 0; }
190	249	};
191	250
192	251	de265_error read_scaling_list(bitreader, const seq_parameter_set, scaling_list_data*, bool inPPS);
	252	de265_error write_scaling_list(CABAC_encoder& out, const seq_parameter_set* sps,
	253	scaling_list_data* sclist, bool inPPS);
193	254	void set_default_scaling_lists(scaling_list_data*);
194	255
195	256	#endif

+12

-0

libde265/threads.cc less more

34	34	#define THREAD_RESULT void*
35	35	#define THREAD_PARAM void*
36	36
	37	#include <stdio.h>
	38
37	39	int de265_thread_create(de265_thread* t, void (start_routine) (void ), void arg) { return pthread_create(t,NULL,start_routine,arg); }
38	40	void de265_thread_join(de265_thread t) { pthread_join(t,NULL); }
39	41	void de265_thread_destroy(de265_thread* t) { }

120	122	de265_mutex_unlock(&mutex);
121	123	}
122	124
	125	void de265_progress_lock::increase_progress(int progress)
	126	{
	127	de265_mutex_lock(&mutex);
	128
	129	mProgress += progress;
	130	de265_cond_broadcast(&cond, &mutex);
	131
	132	de265_mutex_unlock(&mutex);
	133	}
	134
123	135	int de265_progress_lock::get_progress() const
124	136	{
125	137	return mProgress;

+13

-0

libde265/threads.h less more

31	31	#endif
32	32
33	33	#include <deque>
	34	#include <string>
34	35
35	36	#ifndef _WIN32
36	37	#include <pthread.h>

42	43	#else // _WIN32
43	44	#include <windows.h>
44	45	#include "../extra/win32cond.h"
	46	#if _MSC_VER > 1310
45	47	#include <intrin.h>
	48	#else
	49	extern "C"
	50	{
	51	LONG __cdecl _InterlockedExchangeAdd(long volatile *Addend, LONG Value);
	52	}
	53	#pragma intrinsic (_InterlockedExchangeAdd)
	54	#define InterlockedExchangeAdd _InterlockedExchangeAdd
	55	#endif
46	56
47	57	typedef HANDLE de265_thread;
48	58	typedef HANDLE de265_mutex;

99	109
100	110	void wait_for_progress(int progress);
101	111	void set_progress(int progress);
	112	void increase_progress(int progress);
102	113	int get_progress() const;
103	114	void reset(int value=0) { mProgress=value; }
104	115

122	133	enum { Queued, Running, Blocked, Finished } state;
123	134
124	135	virtual void work() = 0;
	136
	137	virtual std::string name() const { return "noname"; }
125	138	};
126	139
127	140

+397

-75

libde265/transform.cc less more

67	67	int qPY_PRED;
68	68
69	69	// first QG in CTB row ?
70
	70
71	71	int ctbLSBMask = ((1<<sps->Log2CtbSizeY)-1);
72	72	bool firstInCTBRow = (xQG == 0 && ((yQG & ctbLSBMask)==0));
73	73

147	147
148	148	tctx->qPYPrime = QPY + sps->QpBdOffset_Y;
149	149
150		int qPiCb = Clip3(-sps->QpBdOffset_C,57, QPY+pps->pic_cb_qp_offset + shdr->slice_cb_qp_offset);
151		int qPiCr = Clip3(-sps->QpBdOffset_C,57, QPY+pps->pic_cr_qp_offset + shdr->slice_cr_qp_offset);
	150	int qPiCb = Clip3(-sps->QpBdOffset_C,57, QPY+pps->pic_cb_qp_offset + shdr->slice_cb_qp_offset + tctx->CuQpOffsetCb);
	151	int qPiCr = Clip3(-sps->QpBdOffset_C,57, QPY+pps->pic_cr_qp_offset + shdr->slice_cr_qp_offset + tctx->CuQpOffsetCr);
152	152
153	153	logtrace(LogTransform,"qPiCb:%d (%d %d), qPiCr:%d (%d %d)\n",
154	154	qPiCb, pps->pic_cb_qp_offset, shdr->slice_cb_qp_offset,
155	155	qPiCr, pps->pic_cr_qp_offset, shdr->slice_cr_qp_offset);
156	156
157		int qPCb = table8_22(qPiCb);
158		int qPCr = table8_22(qPiCr);
	157	int qPCb,qPCr;
	158
	159	if (sps->ChromaArrayType == CHROMA_420) {
	160	qPCb = table8_22(qPiCb);
	161	qPCr = table8_22(qPiCr);
	162	}
	163	else {
	164	qPCb = qPiCb;
	165	qPCr = qPiCr;
	166	}
	167
	168	//printf("q: %d %d\n",qPiCb, qPCb);
159	169
160	170	tctx->qPCbPrime = qPCb + sps->QpBdOffset_C;
161	171	tctx->qPCrPrime = qPCr + sps->QpBdOffset_C;
162	172
	173	/*
	174	printf("Q: %d (%d %d %d / %d %d) %d %d %d\n",QPY,
	175	sps->QpBdOffset_Y,
	176	pps->pic_cb_qp_offset + shdr->slice_cb_qp_offset,
	177	pps->pic_cr_qp_offset + shdr->slice_cr_qp_offset,
	178	sps->QpBdOffset_C, sps->QpBdOffset_C,
	179	tctx->qPYPrime, tctx->qPCbPrime, tctx->qPCrPrime);
	180	*/
	181
163	182	int log2CbSize = tctx->img->get_log2CbSize(xCUBase, yCUBase);
	183
	184	// TODO: On broken input, log2CbSize may be zero (multithreaded only). Not sure yet why.
	185	// Maybe another decoding thread is overwriting the value set in slice.cc:read_coding_unit.
	186	// id:000163,sig:06,src:002041,op:havoc,rep:16.bin
	187	if (log2CbSize<3) { log2CbSize=3; }
	188
164	189	tctx->img->set_QPY(xCUBase, yCUBase, log2CbSize, QPY);
165	190	tctx->currentQPY = QPY;
166	191

175	200
176	201
177	202
178		void transform_coefficients(decoder_context* ctx,
179		int16_t* coeff, int coeffStride, int nT, int trType, int postShift,
180		uint8_t* dst, int dstStride)
	203	template <class pixel_t>
	204	void transform_coefficients(acceleration_functions* acceleration,
	205	int16_t* coeff, int coeffStride, int nT, int trType,
	206	pixel_t* dst, int dstStride, int bit_depth)
181	207	{
182	208	logtrace(LogTransform,"transform --- trType: %d nT: %d\n",trType,nT);
183	209
	210
184	211	if (trType==1) {
185	212
186		ctx->acceleration.transform_4x4_luma_add_8(dst, coeff, dstStride);
	213	acceleration->transform_4x4_dst_add<pixel_t>(dst, coeff, dstStride, bit_depth);
187	214
188	215	} else {
189	216
190		/**/ if (nT==4) { ctx->acceleration.transform_4x4_add_8(dst,coeff,dstStride); }
191		else if (nT==8) { ctx->acceleration.transform_8x8_add_8(dst,coeff,dstStride); }
192		else if (nT==16) { ctx->acceleration.transform_16x16_add_8(dst,coeff,dstStride); }
193		else { ctx->acceleration.transform_32x32_add_8(dst,coeff,dstStride); }
194		}
195		}
	217	/**/ if (nT==4) { acceleration->transform_add<pixel_t>(0,dst,coeff,dstStride, bit_depth); }
	218	else if (nT==8) { acceleration->transform_add<pixel_t>(1,dst,coeff,dstStride, bit_depth); }
	219	else if (nT==16) { acceleration->transform_add<pixel_t>(2,dst,coeff,dstStride, bit_depth); }
	220	else { acceleration->transform_add<pixel_t>(3,dst,coeff,dstStride, bit_depth); }
	221	}
	222
	223	#if 0
	224	printf("decoded pixels:\n");
	225	for (int y=0;y<nT;y++,printf("\n"))
	226	for (int x=0;x<nT;x++) {
	227	printf("%02x ",dst[y*dstStride+x]);
	228	}
	229	#endif
	230	}
	231
	232
	233	// TODO: make this an accelerated function
	234	void cross_comp_pred(const thread_context* tctx, int32_t* residual, int nT)
	235	{
	236	const int BitDepthC = tctx->img->sps.BitDepth_C;
	237	const int BitDepthY = tctx->img->sps.BitDepth_Y;
	238
	239	for (int y=0;y<nT;y++)
	240	for (int x=0;x<nT;x++) {
	241	/* TODO: the most usual case is definitely BitDepthY == BitDepthC, in which case
	242	we could just omit two shifts. The second most common case is probably
	243	BitDepthY>BitDepthC, for which we could also eliminate one shift. The remaining
	244	case is also one shift only.
	245	*/
	246
	247	residual[ynT+x] += (tctx->ResScaleVal
	248	((tctx->residual_luma[y*nT+x] << BitDepthC ) >> BitDepthY ) ) >> 3;
	249	}
	250	}
	251
	252
	253	template <class pixel_t>
	254	void transform_coefficients_explicit(thread_context* tctx,
	255	int16_t* coeff, int coeffStride, int nT, int trType,
	256	pixel_t* dst, int dstStride, int bit_depth, int cIdx)
	257	{
	258	logtrace(LogTransform,"transform --- trType: %d nT: %d\n",trType,nT);
	259
	260	const acceleration_functions* acceleration = &tctx->decctx->acceleration;
	261
	262	int32_t residual_buffer[32*32];
	263	int32_t* residual;
	264	if (cIdx==0) {
	265	residual = tctx->residual_luma;
	266	}
	267	else {
	268	residual = residual_buffer;
	269	}
	270
	271
	272	// TODO
	273	int bdShift = 20 - bit_depth;
	274	int max_coeff_bits = 15;
	275
	276	if (trType==1) {
	277
	278	acceleration->transform_idst_4x4(residual, coeff, bdShift, max_coeff_bits);
	279
	280	} else {
	281
	282	/**/ if (nT==4) { acceleration->transform_idct_4x4(residual,coeff,bdShift,max_coeff_bits); }
	283	else if (nT==8) { acceleration->transform_idct_8x8(residual,coeff,bdShift,max_coeff_bits); }
	284	else if (nT==16) { acceleration->transform_idct_16x16(residual,coeff,bdShift,max_coeff_bits); }
	285	else { acceleration->transform_idct_32x32(residual,coeff,bdShift,max_coeff_bits); }
	286	}
	287
	288
	289	//printBlk("prediction",(uint8_t*)dst,nT,dstStride);
	290	//printBlk("residual",residual,nT,nT);
	291
	292	if (cIdx != 0) {
	293	if (tctx->ResScaleVal != 0) {
	294	cross_comp_pred(tctx, residual, nT);
	295	}
	296
	297	//printBlk("cross-comp-pred modified residual",residual,nT,nT);
	298	}
	299
	300	acceleration->add_residual(dst,dstStride, residual,nT, bit_depth);
	301	}
	302
	303
	304	void inv_transform(acceleration_functions* acceleration,
	305	uint8_t* dst, int dstStride, int16_t* coeff,
	306	int log2TbSize, int trType)
	307	{
	308	if (trType==1) {
	309	assert(log2TbSize==2);
	310
	311	acceleration->transform_4x4_dst_add_8(dst, coeff, dstStride);
	312
	313	} else {
	314	acceleration->transform_add_8[log2TbSize-2](dst,coeff,dstStride);
	315	}
	316
	317
	318	#if 0
	319	int nT = 1<<log2TbSize;
	320	printf("decoded pixels:\n");
	321	for (int y=0;y<nT;y++,printf("\n"))
	322	for (int x=0;x<nT;x++) {
	323	printf("%02x ",dst[y*dstStride+x]);
	324	}
	325	#endif
	326	}
	327
	328
	329	void fwd_transform(acceleration_functions* acceleration,
	330	int16_t* coeff, int coeffStride, int log2TbSize, int trType,
	331	const int16_t* src, int srcStride)
	332	{
	333	logtrace(LogTransform,"transform --- trType: %d nT: %d\n",trType,1<<log2TbSize);
	334
	335	if (trType==1) {
	336	// DST 4x4
	337
	338	acceleration->fwd_transform_4x4_dst_8(coeff, src, srcStride);
	339	} else {
	340	// DCT 4x4, 8x8, 16x16, 32x32
	341
	342	acceleration->fwd_transform_8[log2TbSize-2](coeff,src,srcStride);
	343	}
	344	}
	345
196	346
197	347
198	348	static const int levelScale[] = { 40,45,51,57,64,72 };
199	349
200	350	// (8.6.2) and (8.6.3)
201		void scale_coefficients(thread_context* tctx,
202		int xT,int yT, // position of TU in frame (chroma adapted)
203		int x0,int y0, // position of CU in frame (chroma adapted)
204		int nT, int cIdx,
205		bool transform_skip_flag, bool intra)
	351	template <class pixel_t>
	352	void scale_coefficients_internal(thread_context* tctx,
	353	int xT,int yT, // position of TU in frame (chroma adapted)
	354	int x0,int y0, // position of CU in frame (chroma adapted)
	355	int nT, int cIdx,
	356	bool transform_skip_flag, bool intra, int rdpcmMode)
206	357	{
207	358	seq_parameter_set* sps = &tctx->img->sps;
208	359	pic_parameter_set* pps = &tctx->img->pps;

217	368
218	369	logtrace(LogTransform,"qP: %d\n",qP);
219	370
220		//printf("residual %d;%d cIdx=%d qp=%d\n",xT * (cIdx?2:1),yT * (cIdx?2:1),cIdx,qP);
221
222	371
223	372	int16_t* coeff;
224	373	int coeffStride;

230	379
231	380
232	381
233		uint8_t* pred;
	382	pixel_t* pred;
234	383	int stride;
235		pred = tctx->img->get_image_plane_at_pos(cIdx, xT,yT);
	384	pred = tctx->img->get_image_plane_at_pos_NEW<pixel_t>(cIdx, xT,yT);
236	385	stride = tctx->img->get_image_stride(cIdx);
237	386
238		//fprintf(stderr,"POC=%d pred: %p (%d;%d stride=%d)\n",ctx->img->PicOrderCntVal,pred,xT,yT,stride);
239
240		/*
241		int x,y;
242		for (y=0;y<nT;y++)
243		{
244		printf("P: ");
245
246		for (x=0;x<nT;x++)
247		{
248		printf("%02x ",pred[x+y*stride]);
249		}
250
251		printf("\n");
252		}
253		*/
	387	// We explicitly include the case for sizeof(pixel_t)==1 so that the compiler
	388	// can optimize away a lot of code for 8-bit pixels.
	389	const int bit_depth = ((sizeof(pixel_t)==1) ? 8 : sps->get_bit_depth(cIdx));
	390
	391	//assert(intra == (tctx->img->get_pred_mode(xT,yT)==MODE_INTRA));
	392	int cuPredModeIntra = (tctx->img->get_pred_mode(xT,yT)==MODE_INTRA);
	393
	394	bool rotateCoeffs = (sps->range_extension.transform_skip_rotation_enabled_flag &&
	395	nT == 4 &&
	396	cuPredModeIntra);
254	397
255	398	if (tctx->cu_transquant_bypass_flag) {
256		//assert(false); // TODO
257
	399
	400	int32_t residual_buffer[32*32];
	401
	402	int32_t* residual;
	403	if (cIdx==0) residual = tctx->residual_luma;
	404	else residual = residual_buffer;
	405
	406
	407	// TODO: we could fold the coefficient rotation into the coefficient expansion here:
258	408	for (int i=0;i<tctx->nCoeff[cIdx];i++) {
259		int32_t currCoeff = tctx->coeffList[cIdx][i];
	409	int32_t currCoeff = tctx->coeffList[cIdx][i];
260	410	tctx->coeffBuf[ tctx->coeffPos[cIdx][i] ] = currCoeff;
261	411	}
262	412
263		tctx->decctx->acceleration.transform_bypass_8(pred, coeff, nT, stride);
	413	if (rotateCoeffs) {
	414	tctx->decctx->acceleration.rotate_coefficients(coeff, nT);
	415	}
	416
	417	if (rdpcmMode) {
	418	if (rdpcmMode==2)
	419	tctx->decctx->acceleration.transform_bypass_rdpcm_v(residual, coeff, nT);
	420	else
	421	tctx->decctx->acceleration.transform_bypass_rdpcm_h(residual, coeff, nT);
	422	}
	423	else {
	424	tctx->decctx->acceleration.transform_bypass(residual, coeff, nT);
	425	}
	426
	427	if (cIdx != 0) {
	428	if (tctx->ResScaleVal != 0) {
	429	cross_comp_pred(tctx, residual, nT);
	430	}
	431	}
	432
	433	tctx->decctx->acceleration.add_residual(pred,stride, residual,nT, bit_depth);
	434
	435	if (rotateCoeffs) {
	436	memset(coeff, 0, nTnTsizeof(int16_t)); // delete all, because we moved the coeffs around
	437	}
264	438	}
265	439	else {
266	440	// (8.6.3)

270	444	logtrace(LogTransform,"bdShift=%d\n",bdShift);
271	445
272	446	logtrace(LogTransform,"dequant %d;%d cIdx=%d qp=%d\n",xT(cIdx?2:1),yT(cIdx?2:1),cIdx,qP);
	447
	448
	449	// --- inverse quantization ---
273	450
274	451	if (sps->scaling_list_enable_flag==0) {
275	452

285	462	// usually, this needs to be 64bit, but because we modify the shift above, we can use 16 bit
286	463	int32_t currCoeff = tctx->coeffList[cIdx][i];
287	464
	465	//logtrace(LogTransform,"coefficient[%d] = %d\n",tctx->coeffPos[cIdx][i],
	466	//tctx->coeffList[cIdx][i]);
	467
288	468	currCoeff = Clip3(-32768,32767,
289	469	( (currCoeff * fact + offset ) >> bdShift));
	470
	471	//logtrace(LogTransform," -> %d\n",currCoeff);
290	472
291	473	tctx->coeffBuf[ tctx->coeffPos[cIdx][i] ] = currCoeff;
292	474	}

326	508	}
327	509	}
328	510
	511
	512	// --- do transform or skip ---
	513
329	514	logtrace(LogTransform,"coefficients OUT:\n");
330	515	for (int y=0;y<nT;y++) {
331	516	logtrace(LogTransform," ");

344	529
345	530	if (transform_skip_flag) {
346	531
347		tctx->decctx->acceleration.transform_skip_8(pred, coeff, stride);
	532	int extended_precision_processing_flag = 0;
	533	int Log2nTbS = Log2(nT);
	534	int bdShift = libde265_max( 20 - bit_depth, extended_precision_processing_flag ? 11 : 0 );
	535	int tsShift = (extended_precision_processing_flag ? libde265_min( 5, bdShift - 2 ) : 5 )
	536	+ Log2nTbS;
	537
	538	if (rotateCoeffs) {
	539	tctx->decctx->acceleration.rotate_coefficients(coeff, nT);
	540	}
	541
	542	int32_t residual_buffer[32*32];
	543
	544	int32_t* residual;
	545	if (cIdx==0) residual = tctx->residual_luma;
	546	else residual = residual_buffer;
	547
	548	if (rdpcmMode) {
	549	/*
	550	if (rdpcmMode==2)
	551	tctx->decctx->acceleration.transform_skip_rdpcm_v(pred,coeff, Log2(nT), stride, bit_depth);
	552	else
	553	tctx->decctx->acceleration.transform_skip_rdpcm_h(pred,coeff, Log2(nT), stride, bit_depth);
	554	*/
	555
	556	if (rdpcmMode==2)
	557	tctx->decctx->acceleration.rdpcm_v(residual, coeff,nT, tsShift,bdShift);
	558	else
	559	tctx->decctx->acceleration.rdpcm_h(residual, coeff,nT, tsShift,bdShift);
	560	}
	561	else {
	562	//tctx->decctx->acceleration.transform_skip(pred, coeff, stride, bit_depth);
	563
	564	tctx->decctx->acceleration.transform_skip_residual(residual, coeff, nT, tsShift, bdShift);
	565	}
	566
	567	if (cIdx != 0) {
	568	if (tctx->ResScaleVal != 0) {
	569	cross_comp_pred(tctx, residual, nT);
	570	}
	571	}
	572
	573	tctx->decctx->acceleration.add_residual(pred,stride, residual,nT, bit_depth);
	574
	575	if (rotateCoeffs) {
	576	memset(coeff, 0, nTnTsizeof(int16_t)); // delete all, because we moved the coeffs around
	577	}
348	578	}
349	579	else {
350	580	int trType;
351	581
352		if (nT==4 && cIdx==0 && tctx->img->get_pred_mode(xT,yT)==MODE_INTRA) {
	582	//if (nT==4 && cIdx==0 && tctx->img->get_pred_mode(xT,yT)==MODE_INTRA) {
	583	if (nT==4 && cIdx==0 && cuPredModeIntra) {
353	584	trType=1;
354	585	}
355	586	else {
356	587	trType=0;
357	588	}
358	589
359		transform_coefficients(tctx->decctx, coeff, coeffStride, nT, trType, bdShift2,
360		pred, stride);
	590	assert(rdpcmMode==0);
	591
	592
	593	if (tctx->img->pps.range_extension.cross_component_prediction_enabled_flag) {
	594	// cross-component-prediction: transform to residual buffer and add in a separate step
	595
	596	transform_coefficients_explicit(tctx, coeff, coeffStride, nT, trType,
	597	pred, stride, bit_depth, cIdx);
	598	}
	599	else {
	600	transform_coefficients(&tctx->decctx->acceleration, coeff, coeffStride, nT, trType,
	601	pred, stride, bit_depth);
	602	}
361	603	}
362	604	}
363	605

365	607	logtrace(LogTransform,"pixels (cIdx:%d), position %d %d:\n",cIdx, xT,yT);
366	608
367	609	for (int y=0;y<nT;y++) {
368		logtrace(LogTransform,"RECO-%d-%d-%d ",xT,yT+y,cIdx);
	610	logtrace(LogTransform,"RECO-%3d-%3d-%d ",xT,yT+y,cIdx);
369	611
370	612	for (int x=0;x<nT;x++) {
371		logtrace(LogTransform,"%02x ", pred[x+ystride]);
	613	logtrace(LogTransform,"%03x ", pred[x+ystride]);
372	614	}
373	615
374	616	logtrace(LogTransform,"*\n");
375		}
376
377		/*
378		for (y=0;y<nT;y++)
379		{
380		printf("C: ");
381
382		for (x=0;x<nT;x++)
383		{
384		printf("%4d ",coeff[x+y*nT]);
385		}
386
387		printf("\n");
388		}
389
390		for (y=0;y<nT;y++)
391		{
392		for (x=0;x<nT;x++)
393		{
394		printf("%02x ",pred[x+y*stride]);
395		}
396
397		printf("\n");
398		}
399		*/
	617	}
400	618
401	619	// zero out scrap coefficient buffer again
402	620

404	622	tctx->coeffBuf[ tctx->coeffPos[cIdx][i] ] = 0;
405	623	}
406	624	}
	625
	626
	627	void scale_coefficients(thread_context* tctx,
	628	int xT,int yT, // position of TU in frame (chroma adapted)
	629	int x0,int y0, // position of CU in frame (chroma adapted)
	630	int nT, int cIdx,
	631	bool transform_skip_flag, bool intra,
	632	int rdpcmMode // 0 - off, 1 - Horizontal, 2 - Vertical
	633	)
	634	{
	635	if (tctx->img->high_bit_depth(cIdx)) {
	636	scale_coefficients_internal<uint16_t>(tctx, xT,yT, x0,y0, nT,cIdx, transform_skip_flag, intra,
	637	rdpcmMode);
	638	} else {
	639	scale_coefficients_internal<uint8_t> (tctx, xT,yT, x0,y0, nT,cIdx, transform_skip_flag, intra,
	640	rdpcmMode);
	641	}
	642	}
	643
	644
	645	//#define QUANT_IQUANT_SHIFT 20 // Q(QP%6) * IQ(QP%6) = 2^20
	646	#define QUANT_SHIFT 14 // Q(4) = 2^14
	647	//#define SCALE_BITS 15 // Inherited from TMuC, pressumably for fractional bit estimates in RDOQ
	648	#define MAX_TR_DYNAMIC_RANGE 15 // Maximum transform dynamic range (excluding sign bit)
	649
	650
	651	const static uint16_t g_quantScales[6] = {
	652	26214,23302,20560,18396,16384,14564
	653	};
	654
	655	void quant_coefficients(//encoder_context* ectx,
	656	int16_t* out_coeff,
	657	const int16_t* in_coeff,
	658	int log2TrSize, int qp,
	659	bool intra)
	660	{
	661	const int qpDiv6 = qp / 6;
	662	const int qpMod6 = qp % 6;
	663
	664	//int uiLog2TrSize = xLog2( iWidth - 1);
	665
	666	int uiQ = g_quantScales[qpMod6];
	667	int bitDepth = 8;
	668	int transformShift = MAX_TR_DYNAMIC_RANGE - bitDepth - log2TrSize; // Represents scaling through forward transform
	669	int qBits = QUANT_SHIFT + qpDiv6 + transformShift;
	670
	671	/* TODO: originally, this was checking for intra slices, why not for intra mode ?
	672	*/
	673	int rnd = (intra ? 171 : 85) << (qBits-9);
	674
	675	int x, y;
	676	int uiAcSum = 0;
	677
	678	int nStride = (1<<log2TrSize);
	679
	680	for (y=0; y < (1<<log2TrSize) ; y++) {
	681	for (x=0; x < (1<<log2TrSize) ; x++) {
	682	int level;
	683	int sign;
	684	int blockPos = y * nStride + x;
	685	level = in_coeff[blockPos];
	686	//logtrace(LogTransform,"(%d,%d) %d -> ", x,y,level);
	687	sign = (level < 0 ? -1: 1);
	688
	689	level = (abs_value(level) * uiQ + rnd ) >> qBits;
	690	uiAcSum += level;
	691	level *= sign;
	692	out_coeff[blockPos] = Clip3(-32768, 32767, level);
	693	//logtrace(LogTransform,"%d\n", out_coeff[blockPos]);
	694	}
	695	}
	696	}
	697
	698
	699	void dequant_coefficients(int16_t* out_coeff,
	700	const int16_t* in_coeff,
	701	int log2TrSize, int qP)
	702	{
	703	const int m_x_y = 1;
	704	int bitDepth = 8;
	705	int bdShift = bitDepth + log2TrSize - 5;
	706	bdShift -= 4; // this is equivalent to having a m_x_y of 16 and we can use 32bit integers
	707
	708	const int offset = (1<<(bdShift-1));
	709	const int fact = m_x_y * levelScale[qP%6] << (qP/6);
	710
	711	int blkSize = (1<<log2TrSize);
	712	int nCoeff = (1<<(log2TrSize<<1));
	713
	714	for (int i=0;i<nCoeff;i++) {
	715
	716	// usually, this needs to be 64bit, but because we modify the shift above, we can use 16 bit
	717	int32_t currCoeff = in_coeff[i];
	718
	719	//logtrace(LogTransform,"coefficient[%d] = %d\n",i,currCoeff);
	720
	721	currCoeff = Clip3(-32768,32767,
	722	( (currCoeff * fact + offset ) >> bdShift));
	723
	724	//logtrace(LogTransform," -> %d\n",currCoeff);
	725
	726	out_coeff[i] = currCoeff;
	727	}
	728	}

+19

-1

libde265/transform.h less more

41	41	int xT,int yT, // position of TU in frame (chroma adapted)
42	42	int x0,int y0, // position of CU in frame (chroma adapted)
43	43	int nT, int cIdx,
44		bool transform_skip_flag, bool intra);
	44	bool transform_skip_flag, bool intra, int rdpcmMode);
	45
	46
	47	void inv_transform(acceleration_functions* acceleration,
	48	uint8_t* dst, int dstStride, int16_t* coeff,
	49	int log2TbSize, int trType);
	50
	51	void fwd_transform(acceleration_functions* acceleration,
	52	int16_t* coeff, int coeffStride, int log2TbSize, int trType,
	53	const int16_t* src, int srcStride);
	54
	55	void quant_coefficients(int16_t* out_coeff,
	56	const int16_t* in_coeff,
	57	int log2TrSize, int qp,
	58	bool intra);
	59
	60	void dequant_coefficients(int16_t* out_coeff,
	61	const int16_t* in_coeff,
	62	int log2TrSize, int qP);
45	63
46	64	#endif

+117

-18

libde265/util.cc less more

22	22
23	23	#include <stdarg.h>
24	24	#include <stdio.h>
	25	#include <string.h>
	26
	27
	28	void copy_subimage(uint8_t* dst,int dststride,
	29	const uint8_t* src,int srcstride,
	30	int w, int h)
	31	{
	32	for (int y=0;y<h;y++) {
	33	memcpy(dst, src, w);
	34	dst += dststride;
	35	src += srcstride;
	36	}
	37	}
	38
25	39
26	40
27	41	#ifdef DE265_LOGGING
28	42	static int current_poc=0;
29	43	static int log_poc_start=-9999; // frame-numbers can be negative
30		static int enable_log = 1;
	44	static bool disable_log[NUMBER_OF_LogModules];
31	45	void log_set_current_POC(int poc) { current_poc=poc; }
32	46	#endif
33	47
34	48
35		static int disable_logging=0;
	49	static int disable_logging_OLD=0;
36	50	static int verbosity = 0;
37	51
	52
38	53	LIBDE265_API void de265_disable_logging() // DEPRECATED
39	54	{
40		disable_logging=1;
41		}
	55	disable_logging_OLD=1;
	56	}
	57
42	58
43	59	LIBDE265_API void de265_set_verbosity(int level)
44	60	{

46	62	}
47	63
48	64	#if defined(DE265_LOG_ERROR) \|\| defined(DE265_LOG_INFO) \|\| defined(DE265_LOG_DEBUG) \|\| defined(DE265_LOG_INFO)
49		void enablelog() { enable_log=1; }
50		#endif
	65	void enable_logging(enum LogModule module)
	66	{
	67	disable_log[module]=false;
	68	}
	69	void disable_logging(enum LogModule module)
	70	{
	71	disable_log[module]=true;
	72	}
	73	#endif
	74
	75	static long logcnt[10];
51	76
52	77	#ifdef DE265_LOG_ERROR
53	78	void logerror(enum LogModule module, const char* string, ...)
54	79	{
55		if (disable_logging) return;
56		if (current_poc < log_poc_start) { return; }
57		if (!enable_log) return;
	80	if (current_poc < log_poc_start) { return; }
	81	if (disable_log[module]) return;
58	82
59	83	va_list va;
60	84

71	95	void loginfo (enum LogModule module, const char* string, ...)
72	96	{
73	97	if (verbosity<1) return;
74		if (disable_logging) return;
75		if (current_poc < log_poc_start) { return; }
76		if (!enable_log) return;
	98	if (current_poc < log_poc_start) { return; }
	99	if (disable_log[module]) return;
77	100
78	101	va_list va;
79	102

90	113	void logdebug(enum LogModule module, const char* string, ...)
91	114	{
92	115	if (verbosity<2) return;
93		if (disable_logging) return;
94		if (current_poc < log_poc_start) { return; }
95		if (!enable_log) return;
	116	if (current_poc < log_poc_start) { return; }
	117	if (disable_log[module]) return;
96	118
97	119	va_list va;
98	120

109	131	void logtrace(enum LogModule module, const char* string, ...)
110	132	{
111	133	if (verbosity<3) return;
112		if (disable_logging) return;
113		if (current_poc < log_poc_start) { return; }
114		if (!enable_log) return;
	134	if (current_poc < log_poc_start) { return; }
	135	if (disable_log[module]) return;
	136
	137	//if (module != LogSymbols /&& module != LogCABAC/) { return; }
	138	//if (logcnt<319500) return;
115	139
116	140	//if (module != LogCABAC) return;
117	141
118	142	va_list va;
119	143
	144	if (string[0]=='$') {
	145	int id = string[1]-'0';
	146	logcnt[id]++;
	147	fprintf(stdout, "[%ld] ",logcnt[id]);
	148
	149	string += 3;
	150	}
	151
120	152	int noPrefix = (string[0]=='*');
121	153	if (!noPrefix) { } // fprintf(stdout, "ERR: ");
122	154	va_start(va, string);

137	169	va_end(va);
138	170	fflush(stdout);
139	171	}
	172
	173
	174
	175	void printBlk(const char* title, const int16_t* data, int blksize, int stride)
	176	{
	177	printf("%s:\n",title);
	178
	179	for (int y=0;y<blksize;y++) {
	180	//logtrace(LogTransform," ");
	181	printf(" ");
	182	for (int x=0;x<blksize;x++) {
	183	//logtrace(LogTransform,"%3d ", data[x+ystride]);
	184	printf("%4d ", data[x+y*stride]);
	185	}
	186	//logtrace(LogTransform,"*\n");
	187	printf("\n");
	188	}
	189	}
	190
	191
	192	void printBlk(const char* title, const int32_t* data, int blksize, int stride)
	193	{
	194	printf("%s:\n",title);
	195
	196	for (int y=0;y<blksize;y++) {
	197	//logtrace(LogTransform," ");
	198	printf(" ");
	199	for (int x=0;x<blksize;x++) {
	200	//logtrace(LogTransform,"%3d ", data[x+ystride]);
	201	printf("%4d ", data[x+y*stride]);
	202	}
	203	//logtrace(LogTransform,"*\n");
	204	printf("\n");
	205	}
	206	}
	207
	208
	209	void printBlk(const char* title, const uint8_t* data, int blksize, int stride)
	210	{
	211	printf("%s:\n",title);
	212
	213	for (int y=0;y<blksize;y++) {
	214	//logtrace(LogTransform," ");
	215	printf(" ");
	216	for (int x=0;x<blksize;x++) {
	217	//logtrace(LogTransform,"%3d ", data[x+ystride]);
	218	printf("%02x ", data[x+y*stride]);
	219	}
	220	//logtrace(LogTransform,"*\n");
	221	printf("\n");
	222	}
	223	}
	224
	225
	226	static void (debug_image_output_func)(const struct de265_image, int slot) = NULL;
	227
	228	void debug_set_image_output(void (func)(const struct de265_image, int slot))
	229	{
	230	debug_image_output_func = func;
	231	}
	232
	233	void debug_show_image(const struct de265_image* img, int slot)
	234	{
	235	if (debug_image_output_func) {
	236	debug_image_output_func(img,slot);
	237	}
	238	}

+69

-6

libde265/util.h less more

28	28
29	29	#include "libde265/de265.h"
30	30
	31	#ifdef __GNUC__
	32	#define GCC_VERSION (__GNUC__ * 10000 + __GNUC_MINOR__ * 100 + __GNUC_PATCHLEVEL__)
	33	#endif
31	34
32	35	#ifdef _MSC_VER
33	36	#define LIBDE265_DECLARE_ALIGNED( var, n ) __declspec(align(n)) var

39	42	#define unlikely(x) __builtin_expect(!!(x), 0)
40	43	#endif
41	44
	45	#if defined(__GNUC__) && (__GNUC__ >= 4)
	46	#define LIBDE265_CHECK_RESULT __attribute__ ((warn_unused_result))
	47	#elif defined(_MSC_VER) && (_MSC_VER >= 1700)
	48	#define LIBDE265_CHECK_RESULT _Check_return_
	49	#else
	50	#define LIBDE265_CHECK_RESULT
	51	#endif
	52
42	53	#define ALIGNED_32( var ) LIBDE265_DECLARE_ALIGNED( var, 32 )
43	54	#define ALIGNED_16( var ) LIBDE265_DECLARE_ALIGNED( var, 16 )
44	55	#define ALIGNED_8( var ) LIBDE265_DECLARE_ALIGNED( var, 8 )
45	56	#define ALIGNED_4( var ) LIBDE265_DECLARE_ALIGNED( var, 4 )
46	57
47
	58	// C++11 specific features
	59	#if defined(_MSC_VER) \|\| (!__clang__ && __GNUC__ && GCC_VERSION < 40600)
	60	#define FOR_LOOP(type, var, list) for each (type var in list)
	61	#undef FOR_LOOP_AUTO_SUPPORT
	62	#else
	63	#define FOR_LOOP(type, var, list) for (type var : list)
	64	#define FOR_LOOP_AUTO_SUPPORT 1
	65	#endif
	66
	67	#ifdef USE_STD_TR1_NAMESPACE
	68	#include <tr1/memory>
	69	namespace std { using namespace std::tr1; }
	70	#endif
	71
	72	#if __GNUC__ && GCC_VERSION < 40600
	73	// nullptr was introduced in gcc 4.6, a simple alias should be fine for our use case
	74	#define nullptr NULL
	75	#endif
	76
	77	#ifdef _MSC_VER
	78	#ifdef _CPPRTTI
	79	#define RTTI_ENABLED
	80	#endif
	81	#else
	82	#ifdef __GXX_RTTI
	83	#define RTTI_ENABLED
	84	#endif
	85	#endif
48	86
49	87	//inline uint8_t Clip1_8bit(int16_t value) { if (value<=0) return 0; else if (value>=255) return 255; else return value; }
50	88	#define Clip1_8bit(value) ((value)<0 ? 0 : (value)>255 ? 255 : (value))
	89	#define Clip_BitDepth(value, bit_depth) ((value)<0 ? 0 : (value)>((1<<bit_depth)-1) ? ((1<<bit_depth)-1) : (value))
51	90	#define Clip3(low,high,value) ((value)<(low) ? (low) : (value)>(high) ? (high) : (value))
52	91	#define Sign(value) (((value)<0) ? -1 : ((value)>0) ? 1 : 0)
53	92	#define abs_value(a) (((a)<0) ? -(a) : (a))

59	98	num += denom-1;
60	99	return num/denom;
61	100	}
	101
62	102	LIBDE265_INLINE static int ceil_log2(int val)
63	103	{
64	104	int n=0;

68	108
69	109	return n;
70	110	}
	111
71	112	LIBDE265_INLINE static int Log2(int v)
72	113	{
73	114	int n=0;

78	119
79	120	return n;
80	121	}
	122
	123	LIBDE265_INLINE static int Log2SizeToArea(int v)
	124	{
	125	return (1<<(v<<1));
	126	}
	127
	128	void copy_subimage(uint8_t* dst,int dststride,
	129	const uint8_t* src,int srcstride,
	130	int w, int h);
81	131
82	132
83	133	// === logging ===

94	144	LogSEI,
95	145	LogIntraPred,
96	146	LogPixels,
97		LogCABAC
	147	LogSymbols,
	148	LogCABAC,
	149	LogEncoder,
	150	NUMBER_OF_LogModules
98	151	};
99	152
100	153
101	154	#if defined(DE265_LOG_ERROR) \|\| defined(DE265_LOG_INFO) \|\| defined(DE265_LOG_DEBUG) \|\| defined(DE265_LOG_TRACE)
102	155	# define DE265_LOGGING 1
103		void enablelog();
104		#else
105		#define enablelog() { }
	156	void enable_logging(enum LogModule);
	157	void disable_logging(enum LogModule);
	158	#else
	159	#define enable_logging(x) { }
	160	#define disable_logging(x) { }
106	161	#endif
107	162
108	163	#ifdef DE265_LOGGING

137	192
138	193	void log2fh(FILE* fh, const char* string, ...);
139	194
140		#endif
	195
	196	void printBlk(const char* title,const int32_t* data, int blksize, int stride);
	197	void printBlk(const char* title,const int16_t* data, int blksize, int stride);
	198	void printBlk(const char* title,const uint8_t* data, int blksize, int stride);
	199
	200	void debug_set_image_output(void ()(const struct de265_image, int slot));
	201	void debug_show_image(const struct de265_image*, int slot);
	202
	203	#endif

-2

libde265/visualize.cc less more

291	291	tint_rect(img,stride, x0,y0,w,h, cols[predMode], pixelSize);
292	292	}
293	293	else if (what == PBMotionVectors) {
294		const PredVectorInfo* mvi = srcimg->get_mv_info(x0,y0);
	294	const MotionVectorSpec* mvi = srcimg->get_mv_info(x0,y0);
295	295	int x = x0+w/2;
296	296	int y = y0+h/2;
297	297	if (mvi->predFlag[0]) {

553	553	}
554	554	}
555	555	}
556

+415

-186

libde265/vps.cc less more

24	24	#include <assert.h>
25	25
26	26
27		de265_error read_vps(decoder_context* ctx, bitreader* reader, video_parameter_set* vps)
	27	void profile_data::set_defaults(enum profile_idc profile, int level_major, int level_minor)
	28	{
	29	profile_present_flag = 1;
	30
	31	profile_space = 0;
	32	tier_flag = 0;
	33	profile_idc = profile;
	34
	35	for (int i=0;i<32;i++) {
	36	profile_compatibility_flag[i]=0;
	37	}
	38
	39	switch (profile) {
	40	case Profile_Main:
	41	profile_compatibility_flag[Profile_Main]=1;
	42	profile_compatibility_flag[Profile_Main10]=1;
	43	break;
	44	case Profile_Main10:
	45	profile_compatibility_flag[Profile_Main10]=1;
	46	break;
	47	default:
	48	assert(0);
	49	}
	50
	51	progressive_source_flag = 0;
	52	interlaced_source_flag = 0;
	53	non_packed_constraint_flag = 0;
	54	frame_only_constraint_flag = 0;
	55
	56
	57	// --- level ---
	58
	59	level_present_flag = 1;
	60	level_idc = level_major30 + level_minor3;
	61	}
	62
	63
	64	void video_parameter_set::set_defaults(enum profile_idc profile, int level_major, int level_minor)
	65	{
	66	video_parameter_set_id = 0;
	67	vps_max_layers = 1; // always =1 in current version of standard
	68	vps_max_sub_layers = 1; // temporal sub-layers
	69	vps_temporal_id_nesting_flag = 1;
	70
	71	profile_tier_level_.general.set_defaults(profile,level_major,level_minor);
	72
	73	vps_sub_layer_ordering_info_present_flag = 0;
	74	layer[0].vps_max_dec_pic_buffering = 1;
	75	layer[0].vps_max_num_reorder_pics = 0;
	76	layer[0].vps_max_latency_increase = 0;
	77
	78	vps_max_layer_id = 0;
	79	vps_num_layer_sets = 1;
	80
	81	layer_id_included_flag.resize(vps_num_layer_sets);
	82
	83
	84	// --- timing info ---
	85
	86	vps_timing_info_present_flag = 0;
	87	vps_num_units_in_tick = 0;
	88	vps_time_scale = 0;
	89	vps_poc_proportional_to_timing_flag = 0;
	90
	91	vps_num_ticks_poc_diff_one = 0;
	92	vps_num_hrd_parameters = 0;
	93
	94
	95	// --- vps extension ---
	96
	97	vps_extension_flag = 0;
	98	}
	99
	100
	101	de265_error video_parameter_set::read(error_queue* errqueue, bitreader* reader)
28	102	{
29	103	int vlc;
30	104
31		vps->video_parameter_set_id = vlc = get_bits(reader, 4);
	105	video_parameter_set_id = vlc = get_bits(reader, 4);
32	106	if (vlc >= DE265_MAX_VPS_SETS) return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
33	107
34	108	skip_bits(reader, 2);
35		vps->vps_max_layers = vlc = get_bits(reader,6) +1;
36		if (vlc != 1) return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; // TODO: out of specification
37
38		vps->vps_max_sub_layers = vlc = get_bits(reader,3) +1;
	109	vps_max_layers = vlc = get_bits(reader,6) +1;
	110	if (vlc > 63) return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; // vps_max_layers_minus1 (range 0...63)
	111
	112	vps_max_sub_layers = vlc = get_bits(reader,3) +1;
39	113	if (vlc >= MAX_TEMPORAL_SUBLAYERS) return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
40	114
41		vps->vps_temporal_id_nesting_flag = get_bits(reader,1);
	115	vps_temporal_id_nesting_flag = get_bits(reader,1);
42	116	skip_bits(reader, 16);
43	117
44		read_profile_tier_level(reader, &vps->profile_tier_level,
45		vps->vps_max_sub_layers);
	118	profile_tier_level_.read(reader, vps_max_sub_layers);
46	119
47	120	/*
48		read_bit_rate_pic_rate_info(reader, &vps->bit_rate_pic_rate_info,
49		0, vps->vps_max_sub_layers-1);
	121	read_bit_rate_pic_rate_info(reader, &bit_rate_pic_rate_info,
	122	0, vps_max_sub_layers-1);
50	123	*/
51	124
52		vps->vps_sub_layer_ordering_info_present_flag = get_bits(reader,1);
53		//assert(vps->vps_max_sub_layers-1 < MAX_TEMPORAL_SUBLAYERS);
54
55		int firstLayerRead = vps->vps_sub_layer_ordering_info_present_flag ? 0 : (vps->vps_max_sub_layers-1);
56
57		for (int i=firstLayerRead;i<vps->vps_max_sub_layers;i++) {
58		vps->layer[i].vps_max_dec_pic_buffering = get_uvlc(reader);
59		vps->layer[i].vps_max_num_reorder_pics = get_uvlc(reader);
60		vps->layer[i].vps_max_latency_increase = get_uvlc(reader);
61		}
62
63		if (!vps->vps_sub_layer_ordering_info_present_flag) {
	125	vps_sub_layer_ordering_info_present_flag = get_bits(reader,1);
	126	//assert(vps_max_sub_layers-1 < MAX_TEMPORAL_SUBLAYERS);
	127
	128	int firstLayerRead = vps_sub_layer_ordering_info_present_flag ? 0 : (vps_max_sub_layers-1);
	129
	130	for (int i=firstLayerRead;i<vps_max_sub_layers;i++) {
	131	layer[i].vps_max_dec_pic_buffering = get_uvlc(reader);
	132	layer[i].vps_max_num_reorder_pics = get_uvlc(reader);
	133	layer[i].vps_max_latency_increase = get_uvlc(reader);
	134
	135	if (layer[i].vps_max_dec_pic_buffering == UVLC_ERROR \|\|
	136	layer[i].vps_max_num_reorder_pics == UVLC_ERROR \|\|
	137	layer[i].vps_max_latency_increase == UVLC_ERROR) {
	138	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	139	}
	140	}
	141
	142	if (!vps_sub_layer_ordering_info_present_flag) {
64	143	assert(firstLayerRead < MAX_TEMPORAL_SUBLAYERS);
65	144
66	145	for (int i=0;i<firstLayerRead;i++) {
67		vps->layer[i].vps_max_dec_pic_buffering = vps->layer[firstLayerRead].vps_max_dec_pic_buffering;
68		vps->layer[i].vps_max_num_reorder_pics = vps->layer[firstLayerRead].vps_max_num_reorder_pics;
69		vps->layer[i].vps_max_latency_increase = vps->layer[firstLayerRead].vps_max_latency_increase;
70		}
71		}
72
73
74		vps->vps_max_layer_id = get_bits(reader,6);
75		vps->vps_num_layer_sets = get_uvlc(reader)+1;
76
77		if (vps->vps_num_layer_sets<0 \|\|
78		vps->vps_num_layer_sets>=1024) {
79		ctx->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	146	layer[i].vps_max_dec_pic_buffering = layer[firstLayerRead].vps_max_dec_pic_buffering;
	147	layer[i].vps_max_num_reorder_pics = layer[firstLayerRead].vps_max_num_reorder_pics;
	148	layer[i].vps_max_latency_increase = layer[firstLayerRead].vps_max_latency_increase;
	149	}
	150	}
	151
	152
	153	vps_max_layer_id = get_bits(reader,6);
	154	vps_num_layer_sets = get_uvlc(reader);
	155
	156	if (vps_num_layer_sets+1<0 \|\|
	157	vps_num_layer_sets+1>=1024 \|\|
	158	vps_num_layer_sets == UVLC_ERROR) {
	159	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
80	160	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
81	161	}
82
83		for (int i=1; i <= vps->vps_num_layer_sets-1; i++)
84		for (int j=0; j <= vps->vps_max_layer_id; j++)
85		{
86		vps->layer_id_included_flag[i][j] = get_bits(reader,1);
87		}
88
89		vps->vps_timing_info_present_flag = get_bits(reader,1);
90
91		if (vps->vps_timing_info_present_flag) {
92		vps->vps_num_units_in_tick = get_bits(reader,32);
93		vps->vps_time_scale = get_bits(reader,32);
94		vps->vps_poc_proportional_to_timing_flag = get_bits(reader,1);
95
96		if (vps->vps_poc_proportional_to_timing_flag) {
97		vps->vps_num_ticks_poc_diff_one = get_uvlc(reader)+1;
98		vps->vps_num_hrd_parameters = get_uvlc(reader);
99
100		if (vps->vps_num_hrd_parameters >= 1024) {
	162	vps_num_layer_sets += 1;
	163
	164	layer_id_included_flag.resize(vps_num_layer_sets);
	165
	166	for (int i=1; i <= vps_num_layer_sets-1; i++)
	167	{
	168	layer_id_included_flag[i].resize(vps_max_layer_id+1);
	169
	170	for (int j=0; j <= vps_max_layer_id; j++)
	171	{
	172	layer_id_included_flag[i][j] = get_bits(reader,1);
	173	}
	174	}
	175
	176	vps_timing_info_present_flag = get_bits(reader,1);
	177
	178	if (vps_timing_info_present_flag) {
	179	vps_num_units_in_tick = get_bits(reader,32);
	180	vps_time_scale = get_bits(reader,32);
	181	vps_poc_proportional_to_timing_flag = get_bits(reader,1);
	182
	183	if (vps_poc_proportional_to_timing_flag) {
	184	vps_num_ticks_poc_diff_one = get_uvlc(reader)+1;
	185	vps_num_hrd_parameters = get_uvlc(reader);
	186
	187	if (vps_num_hrd_parameters >= 1024) {
101	188	assert(false); // TODO: return bitstream error
102	189	}
103	190
104		for (int i=0; i<vps->vps_num_hrd_parameters; i++) {
105		vps->hrd_layer_set_idx[i] = get_uvlc(reader);
	191	hrd_layer_set_idx .resize(vps_num_hrd_parameters);
	192	cprms_present_flag.resize(vps_num_hrd_parameters);
	193
	194	for (int i=0; i<vps_num_hrd_parameters; i++) {
	195	hrd_layer_set_idx[i] = get_uvlc(reader);
106	196
107	197	if (i > 0) {
108		vps->cprms_present_flag[i] = get_bits(reader,1);
	198	cprms_present_flag[i] = get_bits(reader,1);
109	199	}
110	200
111	201	//hrd_parameters(cprms_present_flag[i], vps_max_sub_layers_minus1)

114	204	}
115	205	}
116	206	}
117
118		vps->vps_extension_flag = get_bits(reader,1);
119
120		if (vps->vps_extension_flag) {
	207
	208	vps_extension_flag = get_bits(reader,1);
	209
	210	if (vps_extension_flag) {
	211	/*
	212	while( more_rbsp_data() )
	213	vps_extension_data_flag u(1)
	214	rbsp_trailing_bits()
	215	*/
	216	}
	217
	218	return DE265_OK;
	219	}
	220
	221
	222	de265_error video_parameter_set::write(error_queue* errqueue, CABAC_encoder& out) const
	223	{
	224	if (video_parameter_set_id >= DE265_MAX_VPS_SETS) return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	225	out.write_bits(video_parameter_set_id,4);
	226
	227	out.write_bits(0x3,2);
	228	out.write_bits(vps_max_layers-1,6);
	229
	230	if (vps_max_sub_layers >= MAX_TEMPORAL_SUBLAYERS) return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	231	out.write_bits(vps_max_sub_layers-1,3);
	232
	233	out.write_bit(vps_temporal_id_nesting_flag);
	234	out.write_bits(0xFFFF, 16);
	235
	236	profile_tier_level_.write(out, vps_max_sub_layers);
	237
	238	/*
	239	read_bit_rate_pic_rate_info(reader, &bit_rate_pic_rate_info,
	240	0, vps_max_sub_layers-1);
	241	*/
	242
	243	out.write_bit(vps_sub_layer_ordering_info_present_flag);
	244	//assert(vps_max_sub_layers-1 < MAX_TEMPORAL_SUBLAYERS);
	245
	246	int firstLayerRead = vps_sub_layer_ordering_info_present_flag ? 0 : (vps_max_sub_layers-1);
	247
	248	for (int i=firstLayerRead;i<vps_max_sub_layers;i++) {
	249	out.write_uvlc(layer[i].vps_max_dec_pic_buffering);
	250	out.write_uvlc(layer[i].vps_max_num_reorder_pics);
	251	out.write_uvlc(layer[i].vps_max_latency_increase);
	252	}
	253
	254	if (vps_num_layer_sets<0 \|\|
	255	vps_num_layer_sets>=1024) {
	256	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	257	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE;
	258	}
	259
	260	out.write_bits(vps_max_layer_id,6);
	261	out.write_uvlc(vps_num_layer_sets-1);
	262
	263	for (int i=1; i <= vps_num_layer_sets-1; i++)
	264	for (int j=0; j <= vps_max_layer_id; j++)
	265	{
	266	out.write_bit(layer_id_included_flag[i][j]);
	267	}
	268
	269	out.write_bit(vps_timing_info_present_flag);
	270
	271	if (vps_timing_info_present_flag) {
	272	out.write_bits(vps_num_units_in_tick,32);
	273	out.write_bits(vps_time_scale ,32);
	274	out.write_bit (vps_poc_proportional_to_timing_flag);
	275
	276	if (vps_poc_proportional_to_timing_flag) {
	277	out.write_uvlc(vps_num_ticks_poc_diff_one-1);
	278	out.write_uvlc(vps_num_hrd_parameters);
	279
	280	for (int i=0; i<vps_num_hrd_parameters; i++) {
	281	out.write_uvlc(hrd_layer_set_idx[i]);
	282
	283	if (i > 0) {
	284	out.write_bit(cprms_present_flag[i]);
	285	}
	286
	287	//hrd_parameters(cprms_present_flag[i], vps_max_sub_layers_minus1)
	288
	289	return DE265_OK; // TODO: decode hrd_parameters()
	290	}
	291	}
	292	}
	293
	294	out.write_bit(vps_extension_flag);
	295
	296	if (vps_extension_flag) {
121	297	/*
122	298	while( more_rbsp_data() )
123	299	vps_extension_data_flag u(1)

129	305	}
130	306
131	307
132		void read_profile_tier_level(bitreader* reader,
133		struct profile_tier_level* hdr,
134		int max_sub_layers)
135		{
136		hdr->general_profile_space = get_bits(reader,2);
137		hdr->general_tier_flag = get_bits(reader,1);
138		hdr->general_profile_idc = get_bits(reader,5);
139
140		for (int i=0; i<32; i++) {
141		hdr->general_profile_compatibility_flag[i] = get_bits(reader,1);
142		}
143
144		hdr->general_progressive_source_flag = get_bits(reader,1);
145		hdr->general_interlaced_source_flag = get_bits(reader,1);
146		hdr->general_non_packed_constraint_flag = get_bits(reader,1);
147		hdr->general_frame_only_constraint_flag = get_bits(reader,1);
148		skip_bits(reader,44);
149
150		hdr->general_level_idc = get_bits(reader,8);
151
	308	void profile_data::read(bitreader* reader)
	309	{
	310	if (profile_present_flag) {
	311	profile_space = get_bits(reader,2);
	312	tier_flag = get_bits(reader,1);
	313	profile_idc = (enum profile_idc)get_bits(reader,5);
	314
	315	for (int i=0; i<32; i++) {
	316	profile_compatibility_flag[i] = get_bits(reader,1);
	317	}
	318
	319	progressive_source_flag = get_bits(reader,1);
	320	interlaced_source_flag = get_bits(reader,1);
	321	non_packed_constraint_flag = get_bits(reader,1);
	322	frame_only_constraint_flag = get_bits(reader,1);
	323	skip_bits(reader,44);
	324	}
	325
	326	if (level_present_flag) {
	327	level_idc = get_bits(reader,8);
	328	}
	329	}
	330
	331
	332	void profile_tier_level::read(bitreader* reader,
	333	int max_sub_layers)
	334	{
	335	// --- read the general profile ---
	336
	337	general.profile_present_flag = true;
	338	general.level_present_flag = true;
	339	general.read(reader);
	340
	341
	342	// --- read the profile/levels of the sub-layers ---
152	343
153	344	for (int i=0; i<max_sub_layers-1; i++)
154	345	{
155		hdr->profile[i].sub_layer_profile_present_flag = get_bits(reader,1);
156		hdr->profile[i].sub_layer_level_present_flag = get_bits(reader,1);
	346	sub_layer[i].profile_present_flag = get_bits(reader,1);
	347	sub_layer[i].level_present_flag = get_bits(reader,1);
157	348	}
158	349
159	350	if (max_sub_layers > 1)

166	357
167	358	for (int i=0; i<max_sub_layers-1; i++)
168	359	{
169		if (hdr->profile[i].sub_layer_profile_present_flag)
	360	sub_layer[i].read(reader);
	361	}
	362	}
	363
	364
	365	void profile_data::write(CABAC_encoder& out) const
	366	{
	367	if (profile_present_flag)
	368	{
	369	out.write_bits(profile_space,2);
	370	out.write_bit (tier_flag);
	371	out.write_bits(profile_idc,5);
	372
	373	for (int j=0; j<32; j++)
170	374	{
171		hdr->profile[i].sub_layer_profile_space = get_bits(reader,2);
172		hdr->profile[i].sub_layer_tier_flag = get_bits(reader,1);
173		hdr->profile[i].sub_layer_profile_idc = get_bits(reader,5);
174
175		for (int j=0; j<32; j++)
176		{
177		hdr->profile[i].sub_layer_profile_compatibility_flag[j] = get_bits(reader,1);
178		}
179
180		hdr->profile[i].sub_layer_progressive_source_flag = get_bits(reader,1);
181		hdr->profile[i].sub_layer_interlaced_source_flag = get_bits(reader,1);
182		hdr->profile[i].sub_layer_non_packed_constraint_flag = get_bits(reader,1);
183		hdr->profile[i].sub_layer_frame_only_constraint_flag = get_bits(reader,1);
184		skip_bits(reader,44);
	375	out.write_bit(profile_compatibility_flag[j]);
185	376	}
186	377
187		if (hdr->profile[i].sub_layer_level_present_flag)
	378	out.write_bit(progressive_source_flag);
	379	out.write_bit(interlaced_source_flag);
	380	out.write_bit(non_packed_constraint_flag);
	381	out.write_bit(frame_only_constraint_flag);
	382	out.skip_bits(44);
	383	}
	384
	385	if (level_present_flag)
	386	{
	387	out.write_bits(level_idc,8);
	388	}
	389	}
	390
	391	void profile_tier_level::write(CABAC_encoder& out, int max_sub_layers) const
	392	{
	393	assert(general.profile_present_flag==true);
	394	assert(general.level_present_flag==true);
	395
	396	general.write(out);
	397
	398	for (int i=0; i<max_sub_layers-1; i++)
	399	{
	400	out.write_bit(sub_layer[i].profile_present_flag);
	401	out.write_bit(sub_layer[i].level_present_flag);
	402	}
	403
	404	if (max_sub_layers > 1)
	405	{
	406	for (int i=max_sub_layers-1; i<8; i++)
188	407	{
189		hdr->profile[i].sub_layer_level_idc = get_bits(reader,8);
	408	out.skip_bits(2);
190	409	}
	410	}
	411
	412	for (int i=0; i<max_sub_layers-1; i++)
	413	{
	414	sub_layer[i].write(out);
191	415	}
192	416	}
193	417

223	447	#define LOG2(t,d1,d2) log2fh(fh, t,d1,d2)
224	448	#define LOG3(t,d1,d2,d3) log2fh(fh, t,d1,d2,d3)
225	449
226		void dump_vps(video_parameter_set* vps, int fd)
	450	void video_parameter_set::dump(int fd) const
227	451	{
228	452	FILE* fh;
229	453	if (fd==1) fh=stdout;

231	455	else { return; }
232	456
233	457	LOG0("----------------- VPS -----------------\n");
234		LOG1("video_parameter_set_id : %d\n", vps->video_parameter_set_id);
235		LOG1("vps_max_layers : %d\n", vps->vps_max_layers);
236		LOG1("vps_max_sub_layers : %d\n", vps->vps_max_sub_layers);
237		LOG1("vps_temporal_id_nesting_flag : %d\n", vps->vps_temporal_id_nesting_flag);
238
239		dump_profile_tier_level(&vps->profile_tier_level, vps->vps_max_sub_layers, fh);
240		//dump_bit_rate_pic_rate_info(&vps->bit_rate_pic_rate_info, 0, vps->vps_max_sub_layers-1);
	458	LOG1("video_parameter_set_id : %d\n", video_parameter_set_id);
	459	LOG1("vps_max_layers : %d\n", vps_max_layers);
	460	LOG1("vps_max_sub_layers : %d\n", vps_max_sub_layers);
	461	LOG1("vps_temporal_id_nesting_flag : %d\n", vps_temporal_id_nesting_flag);
	462
	463	profile_tier_level_.dump(vps_max_sub_layers, fh);
	464	//dump_bit_rate_pic_rate_info(&bit_rate_pic_rate_info, 0, vps_max_sub_layers-1);
241	465
242	466	LOG1("vps_sub_layer_ordering_info_present_flag : %d\n",
243		vps->vps_sub_layer_ordering_info_present_flag);
244
245		if (vps->vps_sub_layer_ordering_info_present_flag) {
246		for (int i=0;i<vps->vps_max_sub_layers;i++) {
247		LOG2("layer %d: vps_max_dec_pic_buffering = %d\n",i,vps->layer[i].vps_max_dec_pic_buffering);
248		LOG1(" vps_max_num_reorder_pics = %d\n",vps->layer[i].vps_max_num_reorder_pics);
249		LOG1(" vps_max_latency_increase = %d\n",vps->layer[i].vps_max_latency_increase);
	467	vps_sub_layer_ordering_info_present_flag);
	468
	469	if (vps_sub_layer_ordering_info_present_flag) {
	470	for (int i=0;i<vps_max_sub_layers;i++) {
	471	LOG2("layer %d: vps_max_dec_pic_buffering = %d\n",i,layer[i].vps_max_dec_pic_buffering);
	472	LOG1(" vps_max_num_reorder_pics = %d\n",layer[i].vps_max_num_reorder_pics);
	473	LOG1(" vps_max_latency_increase = %d\n",layer[i].vps_max_latency_increase);
250	474	}
251	475	}
252	476	else {
253		LOG1("layer (all): vps_max_dec_pic_buffering = %d\n",vps->layer[0].vps_max_dec_pic_buffering);
254		LOG1(" vps_max_num_reorder_pics = %d\n",vps->layer[0].vps_max_num_reorder_pics);
255		LOG1(" vps_max_latency_increase = %d\n",vps->layer[0].vps_max_latency_increase);
256		}
257
258
259		LOG1("vps_max_layer_id = %d\n", vps->vps_max_layer_id);
260		LOG1("vps_num_layer_sets = %d\n", vps->vps_num_layer_sets);
261
262		for (int i=1; i <= vps->vps_num_layer_sets-1; i++)
263		for (int j=0; j <= vps->vps_max_layer_id; j++)
	477	LOG1("layer (all): vps_max_dec_pic_buffering = %d\n",layer[0].vps_max_dec_pic_buffering);
	478	LOG1(" vps_max_num_reorder_pics = %d\n",layer[0].vps_max_num_reorder_pics);
	479	LOG1(" vps_max_latency_increase = %d\n",layer[0].vps_max_latency_increase);
	480	}
	481
	482
	483	LOG1("vps_max_layer_id = %d\n", vps_max_layer_id);
	484	LOG1("vps_num_layer_sets = %d\n", vps_num_layer_sets);
	485
	486	for (int i=1; i <= vps_num_layer_sets-1; i++)
	487	for (int j=0; j <= vps_max_layer_id; j++)
264	488	{
265	489	LOG3("layer_id_included_flag[%d][%d] = %d\n",i,j,
266		vps->layer_id_included_flag[i][j]);
	490	layer_id_included_flag[i][j]);
267	491	}
268	492
269	493	LOG1("vps_timing_info_present_flag = %d\n",
270		vps->vps_timing_info_present_flag);
271
272		if (vps->vps_timing_info_present_flag) {
273		LOG1("vps_num_units_in_tick = %d\n", vps->vps_num_units_in_tick);
274		LOG1("vps_time_scale = %d\n", vps->vps_time_scale);
275		LOG1("vps_poc_proportional_to_timing_flag = %d\n", vps->vps_poc_proportional_to_timing_flag);
276
277		if (vps->vps_poc_proportional_to_timing_flag) {
278		LOG1("vps_num_ticks_poc_diff_one = %d\n", vps->vps_num_ticks_poc_diff_one);
279		LOG1("vps_num_hrd_parameters = %d\n", vps->vps_num_hrd_parameters);
280
281		for (int i=0; i<vps->vps_num_hrd_parameters; i++) {
282		LOG2("hrd_layer_set_idx[%d] = %d\n", i, vps->hrd_layer_set_idx[i]);
	494	vps_timing_info_present_flag);
	495
	496	if (vps_timing_info_present_flag) {
	497	LOG1("vps_num_units_in_tick = %d\n", vps_num_units_in_tick);
	498	LOG1("vps_time_scale = %d\n", vps_time_scale);
	499	LOG1("vps_poc_proportional_to_timing_flag = %d\n", vps_poc_proportional_to_timing_flag);
	500
	501	if (vps_poc_proportional_to_timing_flag) {
	502	LOG1("vps_num_ticks_poc_diff_one = %d\n", vps_num_ticks_poc_diff_one);
	503	LOG1("vps_num_hrd_parameters = %d\n", vps_num_hrd_parameters);
	504
	505	for (int i=0; i<vps_num_hrd_parameters; i++) {
	506	LOG2("hrd_layer_set_idx[%d] = %d\n", i, hrd_layer_set_idx[i]);
283	507
284	508	if (i > 0) {
285		LOG2("cprms_present_flag[%d] = %d\n", i, vps->cprms_present_flag[i]);
	509	LOG2("cprms_present_flag[%d] = %d\n", i, cprms_present_flag[i]);
286	510	}
287	511
288	512	//hrd_parameters(cprms_present_flag[i], vps_max_sub_layers_minus1)

291	515	}
292	516	}
293	517	}
294
295		LOG1("vps_extension_flag = %d\n", vps->vps_extension_flag);
296		}
297
298
299		void dump_profile_tier_level(const struct profile_tier_level* hdr,
300		int max_sub_layers, FILE* fh)
301		{
302		LOG1(" general_profile_space : %d\n", hdr->general_profile_space);
303		LOG1(" general_tier_flag : %d\n", hdr->general_tier_flag);
304		LOG1(" general_profile_idc : %d\n", hdr->general_profile_idc);
305
306		LOG0(" general_profile_compatibility_flags: ");
307		for (int i=0; i<32; i++) {
308		if (i) LOG0("*,");
309		LOG1("*%d",hdr->general_profile_compatibility_flag[i]);
310		}
311		LOG0("*\n");
312
313		LOG1(" general_level_idc : %d\n", hdr->general_level_idc);
	518
	519	LOG1("vps_extension_flag = %d\n", vps_extension_flag);
	520	}
	521
	522
	523	static const char* profile_name(profile_idc p)
	524	{
	525	switch (p) {
	526	case Profile_Main: return "Main";
	527	case Profile_Main10: return "Main10";
	528	case Profile_MainStillPicture: return "MainStillPicture";
	529	case Profile_FormatRangeExtensions: return "FormatRangeExtensions";
	530	default:
	531	return "(unknown)";
	532	}
	533	}
	534
	535
	536	void profile_data::dump(bool general, FILE* fh) const
	537	{
	538	const char* prefix = (general ? "general" : "sub_layer");
	539
	540	if (profile_present_flag) {
	541	LOG2(" %s_profile_space : %d\n", prefix,profile_space);
	542	LOG2(" %s_tier_flag : %d\n", prefix,tier_flag);
	543	LOG2(" %s_profile_idc : %s\n", prefix, profile_name(profile_idc));
	544
	545	LOG1(" %s_profile_compatibility_flags: ", prefix);
	546	for (int i=0; i<32; i++) {
	547	if (i) LOG0("*,");
	548	LOG1("*%d",profile_compatibility_flag[i]);
	549	}
	550	LOG0("*\n");
	551	LOG2(" %s_progressive_source_flag : %d\n",prefix,progressive_source_flag);
	552	LOG2(" %s_interlaced_source_flag : %d\n",prefix,interlaced_source_flag);
	553	LOG2(" %s_non_packed_constraint_flag : %d\n",prefix,non_packed_constraint_flag);
	554	LOG2(" %s_frame_only_constraint_flag : %d\n",prefix,frame_only_constraint_flag);
	555	}
	556
	557	if (level_present_flag) {
	558	LOG3(" %s_level_idc : %d (%4.2f)\n", prefix,level_idc, level_idc/30.0f);
	559	}
	560	}
	561
	562
	563	void profile_tier_level::dump(int max_sub_layers, FILE* fh) const
	564	{
	565	general.dump(true, fh);
314	566
315	567	for (int i=0; i<max_sub_layers-1; i++)
316	568	{
317	569	LOG1(" Profile/Tier/Level [Layer %d]\n",i);
318
319		if (hdr->profile[i].sub_layer_profile_present_flag) {
320
321		LOG1(" sub_layer_profile_space : %d\n",hdr->profile[i].sub_layer_profile_space);
322		LOG1(" sub_layer_tier_flag : %d\n",hdr->profile[i].sub_layer_tier_flag);
323		LOG1(" sub_layer_profile_idc : %d\n",hdr->profile[i].sub_layer_profile_idc);
324
325		LOG0(" sub_layer_profile_compatibility_flags: ");
326		for (int j=0; j<32; j++) {
327		if (j) LOG0(",");
328		LOG1("%d",hdr->profile[i].sub_layer_profile_compatibility_flag[j]);
329		}
330		LOG0("\n");
331
332		LOG1(" sub_layer_progressive_source_flag : %d\n",hdr->profile[i].sub_layer_progressive_source_flag);
333		LOG1(" sub_layer_interlaced_source_flag : %d\n",hdr->profile[i].sub_layer_interlaced_source_flag);
334		LOG1(" sub_layer_non_packed_constraint_flag : %d\n",hdr->profile[i].sub_layer_non_packed_constraint_flag);
335		LOG1(" sub_layer_frame_only_constraint_flag : %d\n",hdr->profile[i].sub_layer_frame_only_constraint_flag);
336		}
337
338
339		if (hdr->profile[i].sub_layer_level_present_flag) {
340		LOG1(" sub_layer_level_idc : %d\n", hdr->profile[i].sub_layer_level_idc);
341		}
	570	sub_layer[i].dump(false, fh);
342	571	}
343	572	}
344	573

+73

-55

libde265/vps.h less more

30	30
31	31	#include "libde265/bitstream.h"
32	32	#include "libde265/de265.h"
	33	#include "libde265/cabac.h"
	34
	35	#include <vector>
	36
	37	class error_queue;
33	38
34	39	#define MAX_TEMPORAL_SUBLAYERS 8
35	40
36	41
37		struct profile_data {
	42	enum profile_idc {
	43	Profile_Main = 1,
	44	Profile_Main10 = 2,
	45	Profile_MainStillPicture = 3,
	46	Profile_FormatRangeExtensions = 4
	47	};
	48
	49
	50	class profile_data {
	51	public:
	52	void read(bitreader* reader);
	53	void write(CABAC_encoder& writer) const;
	54	void dump(bool general, FILE* fh) const;
	55
	56	void set_defaults(enum profile_idc, int level_major, int level_minor);
	57
38	58	// --- profile ---
39	59
40		char sub_layer_profile_present_flag;
	60	char profile_present_flag; // always true for general profile
41	61
42		char sub_layer_profile_space;
43		char sub_layer_tier_flag;
44		char sub_layer_profile_idc;
	62	char profile_space; // currently always 0
	63	char tier_flag; // main tier or low tier (see Table A-66/A-67)
	64	enum profile_idc profile_idc; // profile
45	65
46		char sub_layer_profile_compatibility_flag[32];
	66	char profile_compatibility_flag[32]; // to which profile we are compatible
47	67
48		char sub_layer_progressive_source_flag;
49		char sub_layer_interlaced_source_flag;
50		char sub_layer_non_packed_constraint_flag;
51		char sub_layer_frame_only_constraint_flag;
	68	char progressive_source_flag;
	69	char interlaced_source_flag;
	70	char non_packed_constraint_flag;
	71	char frame_only_constraint_flag;
52	72
53	73
54	74	// --- level ---
55	75
56		char sub_layer_level_present_flag;
57		int sub_layer_level_idc;
	76	char level_present_flag; // always true for general level
	77	int level_idc; // level * 30
58	78	};
59	79
60	80
61		struct profile_tier_level {
62		int general_profile_space;
63		int general_tier_flag;
64		int general_profile_idc;
	81	class profile_tier_level
	82	{
	83	public:
	84	void read(bitreader* reader, int max_sub_layers);
	85	void write(CABAC_encoder& writer, int max_sub_layers) const;
	86	void dump(int max_sub_layers, FILE* fh) const;
65	87
66		char general_profile_compatibility_flag[32];
	88	profile_data general;
67	89
68		char general_progressive_source_flag;
69		char general_interlaced_source_flag;
70		char general_non_packed_constraint_flag;
71		char general_frame_only_constraint_flag;
	90	//bool sub_layer_profile_present[MAX_TEMPORAL_SUBLAYERS];
	91	//bool sub_layer_level_present[MAX_TEMPORAL_SUBLAYERS];
72	92
73		int general_level_idc;
74
75		struct profile_data profile[MAX_TEMPORAL_SUBLAYERS];
	93	profile_data sub_layer[MAX_TEMPORAL_SUBLAYERS];
76	94	};
77
78
79		void read_profile_tier_level(bitreader* reader,
80		struct profile_tier_level* hdr,
81		int max_sub_layers);
82
83		void dump_profile_tier_level(const struct profile_tier_level* hdr,
84		int max_sub_layers, FILE* fh);
85	95
86	96
87	97	/*

109	119
110	120
111	121	typedef struct {
112		int vps_max_dec_pic_buffering;
113		int vps_max_num_reorder_pics;
114		int vps_max_latency_increase;
	122	int vps_max_dec_pic_buffering; // [1 ; ]
	123	int vps_max_num_reorder_pics; // [0 ; ]
	124	int vps_max_latency_increase; // 0 -> no limit, otherwise value is (x-1)
115	125	} layer_data;
116	126
117		typedef struct {
	127
	128	class video_parameter_set
	129	{
	130	public:
	131	de265_error read(error_queue* errqueue, bitreader* reader);
	132	de265_error write(error_queue* errqueue, CABAC_encoder& out) const;
	133	void dump(int fd) const;
	134
	135	void set_defaults(enum profile_idc profile, int level_major, int level_minor);
	136
118	137	int video_parameter_set_id;
119		int vps_max_layers;
120		int vps_max_sub_layers;
121		int vps_temporal_id_nesting_flag;
122		struct profile_tier_level profile_tier_level;
123		//struct bit_rate_pic_rate_info bit_rate_pic_rate_info;
	138	int vps_max_layers; // [1;?] currently always 1
	139	int vps_max_sub_layers; // [1;7] number of temporal sub-layers
	140	int vps_temporal_id_nesting_flag; // indicate temporal up-switching always possible
	141	profile_tier_level profile_tier_level_;
	142
124	143	int vps_sub_layer_ordering_info_present_flag;
125
126	144	layer_data layer[MAX_TEMPORAL_SUBLAYERS];
127	145
128		uint8_t vps_max_layer_id;
129		int vps_num_layer_sets;
	146	uint8_t vps_max_layer_id; // max value for nuh_layer_id in NALs
	147	int vps_num_layer_sets; // [1;1024], currently always 1
130	148
131		char layer_id_included_flag[1024][64];
	149	std::vector<std::vector<bool> > layer_id_included_flag; // max size = [1024][64]
	150
	151
	152	// --- timing info ---
132	153
133	154	char vps_timing_info_present_flag;
134	155	uint32_t vps_num_units_in_tick;
135	156	uint32_t vps_time_scale;
136	157	char vps_poc_proportional_to_timing_flag;
	158	uint32_t vps_num_ticks_poc_diff_one;
137	159
138		int vps_num_ticks_poc_diff_one;
139		int vps_num_hrd_parameters;
	160	int vps_num_hrd_parameters; // currently [0;1]
140	161
141		uint16_t hrd_layer_set_idx[1024];
142		char cprms_present_flag[1024];
	162	std::vector<uint16_t> hrd_layer_set_idx; // max size = 1024
	163	std::vector<char> cprms_present_flag; // max size = 1024
143	164
144		// hrd_parameters(cprms_present_flag[i], vps_max_sub_layers-1)
	165
	166	// --- vps extension ---
145	167
146	168	char vps_extension_flag;
147
148		} video_parameter_set;
	169	};
149	170
150	171
151		de265_error read_vps(struct decoder_context* ctx, bitreader* reader, video_parameter_set* vps);
152		void dump_vps(video_parameter_set*, int fd);
153
154	172	#endif

+425

-0

libde265/vui.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include "vui.h"
	21	#include "decctx.h"
	22
	23	#include <assert.h>
	24	#include <stdlib.h>
	25	#include <string.h>
	26
	27	#define READ_VLC_OFFSET(variable, vlctype, offset) \
	28	if ((vlc = get_ ## vlctype(br)) == UVLC_ERROR) { \
	29	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false); \
	30	return DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE; \
	31	} \
	32	variable = vlc + offset;
	33
	34	#define READ_VLC(variable, vlctype) READ_VLC_OFFSET(variable,vlctype,0)
	35
	36
	37	#define NUM_SAR_PRESETS 17
	38
	39	static uint16_t sar_presets[NUM_SAR_PRESETS+1][2] = {
	40	{ 0,0 },
	41	{ 1,1 },
	42	{ 12,11 },
	43	{ 10,11 },
	44	{ 16,11 },
	45	{ 40,33 },
	46	{ 24,11 },
	47	{ 20,11 },
	48	{ 32,11 },
	49	{ 80,33 },
	50	{ 18,11 },
	51	{ 15,11 },
	52	{ 64,33 },
	53	{ 160,99 },
	54	{ 4,3 },
	55	{ 3,2 },
	56	{ 2,1 }
	57	};
	58
	59	#define EXTENDED_SAR 255
	60
	61
	62	const char* get_video_format_name(enum VideoFormat format)
	63	{
	64	switch (format) {
	65	case VideoFormat_Component: return "component";
	66	case VideoFormat_PAL: return "PAL";
	67	case VideoFormat_NTSC: return "NTSC";
	68	case VideoFormat_SECAM: return "SECAM";
	69	case VideoFormat_MAC: return "MAC";
	70	default: return "unspecified";
	71	}
	72	}
	73
	74
	75	video_usability_information::video_usability_information()
	76	{
	77	aspect_ratio_info_present_flag = false;
	78	sar_width = 0;
	79	sar_height = 0;
	80
	81
	82	// --- overscan ---
	83
	84	overscan_info_present_flag = false;
	85	overscan_appropriate_flag = false;
	86
	87
	88	// --- video signal type ---
	89
	90	video_signal_type_present_flag = false;
	91	video_format = VideoFormat_Unspecified;
	92	video_full_range_flag = false;
	93	colour_description_present_flag = false;
	94	colour_primaries = 2;
	95	transfer_characteristics = 2;
	96	matrix_coeffs = 2;
	97
	98	// --- chroma / interlaced ---
	99
	100	chroma_loc_info_present_flag = false;
	101	chroma_sample_loc_type_top_field = 0;
	102	chroma_sample_loc_type_bottom_field = 0;
	103
	104	neutral_chroma_indication_flag = false;
	105	field_seq_flag = false;
	106	frame_field_info_present_flag = false;
	107
	108	// --- default display window ---
	109
	110	default_display_window_flag = false;
	111	def_disp_win_left_offset = 0;
	112	def_disp_win_right_offset = 0;
	113	def_disp_win_top_offset = 0;
	114	def_disp_win_bottom_offset = 0;
	115
	116
	117	// --- timing ---
	118
	119	vui_timing_info_present_flag = false;
	120	vui_num_units_in_tick = 0;
	121	vui_time_scale = 0;
	122
	123	vui_poc_proportional_to_timing_flag = false;
	124	vui_num_ticks_poc_diff_one = 1;
	125
	126
	127	// --- hrd parameters ---
	128
	129	vui_hrd_parameters_present_flag = false;
	130	//hrd_parameters vui_hrd_parameters;
	131
	132
	133	// --- bitstream restriction ---
	134
	135	bitstream_restriction_flag = false;
	136	tiles_fixed_structure_flag = false;
	137	motion_vectors_over_pic_boundaries_flag = true;
	138	restricted_ref_pic_lists_flag = false;
	139	min_spatial_segmentation_idc = 0;
	140	max_bytes_per_pic_denom = 2;
	141	max_bits_per_min_cu_denom = 1;
	142	log2_max_mv_length_horizontal = 15;
	143	log2_max_mv_length_vertical = 15;
	144	}
	145
	146
	147	de265_error video_usability_information::read(error_queue* errqueue, bitreader* br,
	148	const seq_parameter_set* sps)
	149	{
	150	int vlc;
	151
	152
	153	// --- sample aspect ratio (SAR) ---
	154
	155	aspect_ratio_info_present_flag = get_bits(br,1);
	156	if (aspect_ratio_info_present_flag) {
	157	int aspect_ratio_idc = get_bits(br,8);
	158	if (aspect_ratio_idc <= NUM_SAR_PRESETS) {
	159	sar_width = sar_presets[aspect_ratio_idc][0];
	160	sar_height = sar_presets[aspect_ratio_idc][1];
	161	}
	162	else if (aspect_ratio_idc == EXTENDED_SAR) {
	163	sar_width = get_bits(br,16);
	164	sar_height = get_bits(br,16);
	165	}
	166	else {
	167	sar_width = 0;
	168	sar_height = 0;
	169	}
	170	}
	171	else {
	172	sar_width = 0;
	173	sar_height = 0;
	174	}
	175
	176
	177	// --- overscan ---
	178
	179	overscan_info_present_flag = get_bits(br,1);
	180	if (overscan_info_present_flag) {
	181	overscan_appropriate_flag = get_bits(br,1);
	182	}
	183
	184
	185	// --- video signal type ---
	186
	187	{ // defaults
	188	video_format = VideoFormat_Unspecified;
	189	video_full_range_flag = false;
	190	colour_primaries = 2;
	191	transfer_characteristics = 2;
	192	matrix_coeffs = 2;
	193	}
	194
	195	video_signal_type_present_flag = get_bits(br,1);
	196	if (video_signal_type_present_flag) {
	197	int video_format_idc = get_bits(br,3);
	198	if (video_format_idc > 5) {
	199	video_format_idc = VideoFormat_Unspecified;
	200	}
	201	video_format = (VideoFormat)video_format_idc;
	202
	203	video_full_range_flag = get_bits(br,1);
	204
	205	colour_description_present_flag = get_bits(br,1);
	206	if (colour_description_present_flag) {
	207	colour_primaries = get_bits(br,8);
	208	if (colour_primaries == 0 \|\|
	209	colour_primaries == 3 \|\|
	210	colour_primaries >= 11) {
	211	colour_primaries = 2;
	212	}
	213
	214	transfer_characteristics = get_bits(br,8);
	215	if (transfer_characteristics == 0 \|\|
	216	transfer_characteristics == 3 \|\|
	217	transfer_characteristics >= 18) {
	218	transfer_characteristics = 2;
	219	}
	220
	221	matrix_coeffs = get_bits(br,8);
	222	if (matrix_coeffs == 0 \|\|
	223	matrix_coeffs >= 11) {
	224	matrix_coeffs = 2;
	225	}
	226	}
	227	}
	228
	229
	230	// --- chroma / interlaced ---
	231
	232	chroma_loc_info_present_flag = get_bits(br,1);
	233	if (chroma_loc_info_present_flag) {
	234	READ_VLC(chroma_sample_loc_type_top_field, uvlc);
	235	READ_VLC(chroma_sample_loc_type_bottom_field, uvlc);
	236	}
	237	else {
	238	chroma_sample_loc_type_top_field = 0;
	239	chroma_sample_loc_type_bottom_field = 0;
	240	}
	241
	242	neutral_chroma_indication_flag = get_bits(br,1);
	243	field_seq_flag = get_bits(br,1);
	244	frame_field_info_present_flag = get_bits(br,1);
	245
	246
	247	// --- default display window ---
	248
	249	default_display_window_flag = get_bits(br,1);
	250	if (default_display_window_flag) {
	251	READ_VLC(def_disp_win_left_offset ,uvlc);
	252	READ_VLC(def_disp_win_right_offset ,uvlc);
	253	READ_VLC(def_disp_win_top_offset ,uvlc);
	254	READ_VLC(def_disp_win_bottom_offset,uvlc);
	255	}
	256	else {
	257	def_disp_win_left_offset =0;
	258	def_disp_win_right_offset =0;
	259	def_disp_win_top_offset =0;
	260	def_disp_win_bottom_offset=0;
	261	}
	262
	263
	264	// --- timing ---
	265
	266	vui_timing_info_present_flag = get_bits(br,1);
	267	if (vui_timing_info_present_flag) {
	268	vui_num_units_in_tick = get_bits(br,32);
	269	vui_time_scale = get_bits(br,32);
	270	}
	271
	272	vui_poc_proportional_to_timing_flag = get_bits(br,1);
	273	READ_VLC_OFFSET(vui_num_ticks_poc_diff_one, uvlc, 1);
	274
	275
	276	// --- hrd parameters ---
	277
	278	vui_hrd_parameters_present_flag = get_bits(br,1);
	279	if (vui_hrd_parameters_present_flag) {
	280	return DE265_ERROR_NOT_IMPLEMENTED_YET;
	281	//hrd_parameters vui_hrd_parameters;
	282	}
	283
	284
	285	// --- bitstream restriction ---
	286
	287	bitstream_restriction_flag = get_bits(br,1);
	288	if (bitstream_restriction_flag) {
	289	tiles_fixed_structure_flag = get_bits(br,1);
	290	motion_vectors_over_pic_boundaries_flag = get_bits(br,1);
	291	restricted_ref_pic_lists_flag = get_bits(br,1);
	292
	293	READ_VLC(min_spatial_segmentation_idc, uvlc);
	294	if (min_spatial_segmentation_idc > 4095) {
	295	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	296	min_spatial_segmentation_idc = 0;
	297	}
	298
	299	READ_VLC(max_bytes_per_pic_denom, uvlc);
	300	if (max_bytes_per_pic_denom > 16) {
	301	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	302	max_bytes_per_pic_denom = 2;
	303	}
	304
	305	READ_VLC(max_bits_per_min_cu_denom, uvlc);
	306	if (max_bits_per_min_cu_denom > 16) {
	307	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	308	max_bits_per_min_cu_denom = 1;
	309	}
	310
	311	READ_VLC(log2_max_mv_length_horizontal, uvlc);
	312	if (log2_max_mv_length_horizontal > 15) {
	313	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	314	log2_max_mv_length_horizontal = 15;
	315	}
	316
	317	READ_VLC(log2_max_mv_length_vertical, uvlc);
	318	if (log2_max_mv_length_vertical > 15) {
	319	errqueue->add_warning(DE265_ERROR_CODED_PARAMETER_OUT_OF_RANGE, false);
	320	log2_max_mv_length_vertical = 15;
	321	}
	322	}
	323	else {
	324	tiles_fixed_structure_flag = false;
	325	motion_vectors_over_pic_boundaries_flag = true;
	326	restricted_ref_pic_lists_flag = false; // NOTE: default not specified in standard 2014/10
	327
	328	min_spatial_segmentation_idc = 0;
	329	max_bytes_per_pic_denom = 2;
	330	max_bits_per_min_cu_denom = 1;
	331	log2_max_mv_length_horizontal = 15;
	332	log2_max_mv_length_vertical = 15;
	333	}
	334
	335	//vui_read = true;
	336
	337	return DE265_OK;
	338	}
	339
	340
	341	void video_usability_information::dump(int fd) const
	342	{
	343	//#if (_MSC_VER >= 1500)
	344	//#define LOG0(t) loginfo(LogHeaders, t)
	345	//#define LOG1(t,d) loginfo(LogHeaders, t,d)
	346	//#define LOG2(t,d1,d2) loginfo(LogHeaders, t,d1,d2)
	347	//#define LOG3(t,d1,d2,d3) loginfo(LogHeaders, t,d1,d2,d3)
	348
	349	FILE* fh;
	350	if (fd==1) fh=stdout;
	351	else if (fd==2) fh=stderr;
	352	else { return; }
	353
	354	#define LOG0(t) log2fh(fh, t)
	355	#define LOG1(t,d) log2fh(fh, t,d)
	356	#define LOG2(t,d1,d2) log2fh(fh, t,d1,d2)
	357	#define LOG3(t,d1,d2,d3) log2fh(fh, t,d1,d2,d3)
	358
	359	LOG0("----------------- VUI -----------------\n");
	360	LOG2("sample aspect ratio : %d:%d\n", sar_width,sar_height);
	361	LOG1("overscan_info_present_flag : %d\n", overscan_info_present_flag);
	362	LOG1("overscan_appropriate_flag : %d\n", overscan_appropriate_flag);
	363
	364	LOG1("video_signal_type_present_flag: %d\n", video_signal_type_present_flag);
	365	if (video_signal_type_present_flag) {
	366	LOG1(" video_format : %s\n", get_video_format_name(video_format));
	367	LOG1(" video_full_range_flag : %d\n", video_full_range_flag);
	368	LOG1(" colour_description_present_flag : %d\n", colour_description_present_flag);
	369	LOG1(" colour_primaries : %d\n", colour_primaries);
	370	LOG1(" transfer_characteristics : %d\n", transfer_characteristics);
	371	LOG1(" matrix_coeffs : %d\n", matrix_coeffs);
	372	}
	373
	374	LOG1("chroma_loc_info_present_flag: %d\n", chroma_loc_info_present_flag);
	375	if (chroma_loc_info_present_flag) {
	376	LOG1(" chroma_sample_loc_type_top_field : %d\n", chroma_sample_loc_type_top_field);
	377	LOG1(" chroma_sample_loc_type_bottom_field: %d\n", chroma_sample_loc_type_bottom_field);
	378	}
	379
	380	LOG1("neutral_chroma_indication_flag: %d\n", neutral_chroma_indication_flag);
	381	LOG1("field_seq_flag : %d\n", field_seq_flag);
	382	LOG1("frame_field_info_present_flag : %d\n", frame_field_info_present_flag);
	383
	384	LOG1("default_display_window_flag : %d\n", default_display_window_flag);
	385	LOG1(" def_disp_win_left_offset : %d\n", def_disp_win_left_offset);
	386	LOG1(" def_disp_win_right_offset : %d\n", def_disp_win_right_offset);
	387	LOG1(" def_disp_win_top_offset : %d\n", def_disp_win_top_offset);
	388	LOG1(" def_disp_win_bottom_offset : %d\n", def_disp_win_bottom_offset);
	389
	390	LOG1("vui_timing_info_present_flag : %d\n", vui_timing_info_present_flag);
	391	if (vui_timing_info_present_flag) {
	392	LOG1(" vui_num_units_in_tick : %d\n", vui_num_units_in_tick);
	393	LOG1(" vui_time_scale : %d\n", vui_time_scale);
	394	}
	395
	396	LOG1("vui_poc_proportional_to_timing_flag : %d\n", vui_poc_proportional_to_timing_flag);
	397	LOG1("vui_num_ticks_poc_diff_one : %d\n", vui_num_ticks_poc_diff_one);
	398
	399	LOG1("vui_hrd_parameters_present_flag : %d\n", vui_hrd_parameters_present_flag);
	400	if (vui_hrd_parameters_present_flag) {
	401	//hrd_parameters vui_hrd_parameters;
	402	}
	403
	404
	405	// --- bitstream restriction ---
	406
	407	LOG1("bitstream_restriction_flag : %d\n", bitstream_restriction_flag);
	408	if (bitstream_restriction_flag) {
	409	LOG1(" tiles_fixed_structure_flag : %d\n", tiles_fixed_structure_flag);
	410	LOG1(" motion_vectors_over_pic_boundaries_flag : %d\n", motion_vectors_over_pic_boundaries_flag);
	411	LOG1(" restricted_ref_pic_lists_flag : %d\n", restricted_ref_pic_lists_flag);
	412	LOG1(" min_spatial_segmentation_idc : %d\n", min_spatial_segmentation_idc);
	413	LOG1(" max_bytes_per_pic_denom : %d\n", max_bytes_per_pic_denom);
	414	LOG1(" max_bits_per_min_cu_denom : %d\n", max_bits_per_min_cu_denom);
	415	LOG1(" log2_max_mv_length_horizontal : %d\n", log2_max_mv_length_horizontal);
	416	LOG1(" log2_max_mv_length_vertical : %d\n", log2_max_mv_length_vertical);
	417	}
	418
	419	#undef LOG0
	420	#undef LOG1
	421	#undef LOG2
	422	#undef LOG3
	423	//#endif
	424	}

+126

-0

libde265/vui.h less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU Lesser General Public License as
	8	* published by the Free Software Foundation, either version 3 of
	9	* the License, or (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU Lesser General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU Lesser General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifndef DE265_VUI_H
	21	#define DE265_VUI_H
	22
	23	#include "libde265/de265.h"
	24	#include "libde265/bitstream.h"
	25
	26	#include <vector>
	27
	28	class error_queue;
	29	class seq_parameter_set;
	30
	31
	32	enum VideoFormat {
	33	VideoFormat_Component = 0,
	34	VideoFormat_PAL = 1,
	35	VideoFormat_NTSC = 2,
	36	VideoFormat_SECAM = 3,
	37	VideoFormat_MAC = 4,
	38	VideoFormat_Unspecified = 5
	39	};
	40
	41	const char* get_video_format_name(enum VideoFormat);
	42
	43
	44	class video_usability_information
	45	{
	46	public:
	47	video_usability_information();
	48
	49	de265_error read(error_queue, bitreader, const seq_parameter_set*);
	50	void dump(int fd) const;
	51
	52
	53	// --- sample aspect ratio (SAR) ---
	54
	55	bool aspect_ratio_info_present_flag;
	56	uint16_t sar_width; // sar_width and sar_height are zero if unspecified
	57	uint16_t sar_height;
	58
	59
	60	// --- overscan ---
	61
	62	bool overscan_info_present_flag;
	63	bool overscan_appropriate_flag;
	64
	65
	66	// --- video signal type ---
	67
	68	bool video_signal_type_present_flag;
	69	enum VideoFormat video_format;
	70	bool video_full_range_flag;
	71	bool colour_description_present_flag;
	72	uint8_t colour_primaries;
	73	uint8_t transfer_characteristics;
	74	uint8_t matrix_coeffs;
	75
	76	// --- chroma / interlaced ---
	77
	78	bool chroma_loc_info_present_flag;
	79	uint8_t chroma_sample_loc_type_top_field;
	80	uint8_t chroma_sample_loc_type_bottom_field;
	81
	82	bool neutral_chroma_indication_flag;
	83	bool field_seq_flag;
	84	bool frame_field_info_present_flag;
	85
	86	// --- default display window ---
	87
	88	bool default_display_window_flag;
	89	uint32_t def_disp_win_left_offset;
	90	uint32_t def_disp_win_right_offset;
	91	uint32_t def_disp_win_top_offset;
	92	uint32_t def_disp_win_bottom_offset;
	93
	94
	95	// --- timing ---
	96
	97	bool vui_timing_info_present_flag;
	98	uint32_t vui_num_units_in_tick;
	99	uint32_t vui_time_scale;
	100
	101	bool vui_poc_proportional_to_timing_flag;
	102	uint32_t vui_num_ticks_poc_diff_one;
	103
	104
	105	// --- hrd parameters ---
	106
	107	bool vui_hrd_parameters_present_flag;
	108	//hrd_parameters vui_hrd_parameters;
	109
	110
	111	// --- bitstream restriction ---
	112
	113	bool bitstream_restriction_flag;
	114	bool tiles_fixed_structure_flag;
	115	bool motion_vectors_over_pic_boundaries_flag;
	116	bool restricted_ref_pic_lists_flag;
	117	uint16_t min_spatial_segmentation_idc;
	118	uint8_t max_bytes_per_pic_denom;
	119	uint8_t max_bits_per_min_cu_denom;
	120	uint8_t log2_max_mv_length_horizontal;
	121	uint8_t log2_max_mv_length_vertical;
	122	};
	123
	124
	125	#endif

+24

-0

libde265/x86/CMakeLists.txt less more

	0	set (x86_sources
	1	sse.cc sse.h
	2	)
	3
	4	set (x86_sse_sources
	5	sse-motion.cc sse-motion.h sse-dct.h sse-dct.cc
	6	)
	7
	8	add_library(x86 STATIC ${x86_sources})
	9
	10	add_library(x86_sse STATIC ${x86_sse_sources})
	11
	12	set(sse_flags "")
	13
	14	if(NOT MSVC)
	15	set(sse_flags "${sse_flags} -msse4.1")
	16	endif()
	17
	18	target_link_libraries(x86 x86_sse)
	19
	20	if(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
	21	SET_TARGET_PROPERTIES(x86 PROPERTIES COMPILE_FLAGS "-fPIC")
	22	SET_TARGET_PROPERTIES(x86_sse PROPERTIES COMPILE_FLAGS "-fPIC ${sse_flags}")
	23	endif(CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")

-0

libde265/x86/Makefile.am less more

17	17	libde265_x86_sse_la_CXXFLAGS += -DHAVE_VISIBILITY
18	18	endif
19	19
	20	EXTRA_DIST = \
	21	CMakeLists.txt

+10

-3

libde265/x86/Makefile.in less more

85	85	$(top_srcdir)/depcomp
86	86	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
87	87	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	88	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
88	89	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
89	90	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
90	91	$(top_srcdir)/m4/lt~obsolete.m4 \
91	92	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
92		$(top_srcdir)/m4/visibility.m4 $(top_srcdir)/configure.ac
	93	$(top_srcdir)/configure.ac
93	94	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
94	95	$(ACLOCAL_M4)
95	96	mkinstalldirs = $(install_sh) -d

197	198	CTAGS = ctags
198	199	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
199	200	ACLOCAL = @ACLOCAL@
	201	ALLOCA = @ALLOCA@
200	202	AMTAR = @AMTAR@
201	203	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
202	204	AR = @AR@

205	207	AUTOMAKE = @AUTOMAKE@
206	208	AWK = @AWK@
207	209	CC = @CC@
	210	CCAS = @CCAS@
	211	CCASDEPMODE = @CCASDEPMODE@
	212	CCASFLAGS = @CCASFLAGS@
208	213	CCDEPMODE = @CCDEPMODE@
209	214	CFLAGS = @CFLAGS@
210		CFLAG_VISIBILITY = @CFLAG_VISIBILITY@
211	215	CPP = @CPP@
212	216	CPPFLAGS = @CPPFLAGS@
213	217	CXX = @CXX@

227	231	EXEEXT = @EXEEXT@
228	232	FGREP = @FGREP@
229	233	GREP = @GREP@
230		HAVE_VISIBILITY = @HAVE_VISIBILITY@
	234	HAVE_CXX11 = @HAVE_CXX11@
231	235	INSTALL = @INSTALL@
232	236	INSTALL_DATA = @INSTALL_DATA@
233	237	INSTALL_PROGRAM = @INSTALL_PROGRAM@

347	351	libde265_x86_sse_la_CXXFLAGS = -msse4.1 -I.. $(CFLAG_VISIBILITY) \
348	352	$(am__append_2)
349	353	libde265_x86_sse_la_SOURCES = sse-motion.cc sse-motion.h sse-dct.h sse-dct.cc
	354	EXTRA_DIST = \
	355	CMakeLists.txt
	356
350	357	all: all-am
351	358
352	359	.SUFFIXES:

+16

-16

libde265/x86/sse-dct.cc less more

281	281	#define add_1st (1 << (shift_1st - 1))
282	282
283	283
284		void ff_hevc_transform_skip_8_sse(uint8_t _dst, int16_t coeffs, ptrdiff_t _stride)
	284	void ff_hevc_transform_skip_8_sse(uint8_t _dst, const int16_t coeffs, ptrdiff_t _stride)
285	285	{
286	286	uint8_t dst = (uint8_t)_dst;
287	287	ptrdiff_t stride = _stride;

341	341
342	342
343	343	#if HAVE_SSE4_1
344		void ff_hevc_transform_4x4_luma_add_8_sse4(uint8_t _dst, int16_t coeffs,
	344	void ff_hevc_transform_4x4_luma_add_8_sse4(uint8_t _dst, const int16_t coeffs,
345	345	ptrdiff_t _stride) {
346	346
347	347	uint8_t shift_2nd = 12; // 20 - Bit depth

349	349
350	350	uint8_t dst = (uint8_t) _dst;
351	351	ptrdiff_t stride = _stride;
352		int16_t *src = coeffs;
	352	const int16_t *src = coeffs;
353	353	__m128i m128iAdd, S0, S8, m128iTmp1, m128iTmp2, m128iAC, m128iBD, m128iA,
354	354	m128iD;
355	355	m128iAdd = _mm_set1_epi32(64);

495	495	#endif // SSE4.1
496	496
497	497	#if 0
498		void ff_hevc_transform_4x4_luma_add_10_sse4(uint8_t _dst, int16_t coeffs,
	498	void ff_hevc_transform_4x4_luma_add_10_sse4(uint8_t _dst, const int16_t coeffs,
499	499	ptrdiff_t _stride) {
500	500	int i,j;
501	501	uint8_t shift_2nd = 10; // 20 - Bit depth

622	622
623	623
624	624	#if HAVE_SSE4_1
625		void ff_hevc_transform_4x4_add_8_sse4(uint8_t _dst, int16_t coeffs,
	625	void ff_hevc_transform_4x4_add_8_sse4(uint8_t _dst, const int16_t coeffs,
626	626	ptrdiff_t _stride) {
627	627	uint8_t shift_2nd = 12; // 20 - Bit depth
628	628	uint16_t add_2nd = 1 << 11; //(1 << (shift_2nd - 1))
629	629
630	630	uint8_t dst = (uint8_t) _dst;
631	631	ptrdiff_t stride = _stride;
632		int16_t *src = coeffs;
	632	const int16_t *src = coeffs;
633	633
634	634	__m128i S0, S8, m128iAdd, m128Tmp, E1, E2, O1, O2, m128iA, m128iD, m128iTmp1,m128iTmp2;
635	635	S0 = _mm_load_si128((__m128i *) (src));

743	743	#endif
744	744
745	745	#if 0
746		void ff_hevc_transform_4x4_add_10_sse4(uint8_t _dst, int16_t coeffs,
	746	void ff_hevc_transform_4x4_add_10_sse4(uint8_t _dst, const int16_t coeffs,
747	747	ptrdiff_t _stride) {
748	748	int i;
749	749	uint8_t shift_2nd = 10; // 20 - Bit depth

838	838	#endif
839	839
840	840	#if HAVE_SSE4_1
841		void ff_hevc_transform_8x8_add_8_sse4(uint8_t _dst, int16_t coeffs,
	841	void ff_hevc_transform_8x8_add_8_sse4(uint8_t _dst, const int16_t coeffs,
842	842	ptrdiff_t _stride) {
843	843	uint8_t shift_2nd = 12; // 20 - Bit depth
844	844	uint16_t add_2nd = 1 << 11; //(1 << (shift_2nd - 1))
845	845
846	846	uint8_t dst = (uint8_t) _dst;
847	847	ptrdiff_t stride = _stride / sizeof(uint8_t);
848		int16_t *src = coeffs;
	848	const int16_t *src = coeffs;
849	849	__m128i m128iS0, m128iS1, m128iS2, m128iS3, m128iS4, m128iS5, m128iS6,
850	850	m128iS7, m128iAdd, m128Tmp0, m128Tmp1, m128Tmp2, m128Tmp3, E0h, E1h,
851	851	E2h, E3h, E0l, E1l, E2l, E3l, O0h, O1h, O2h, O3h, O0l, O1l, O2l,

1173	1173	#endif
1174	1174
1175	1175	#if 0
1176		void ff_hevc_transform_8x8_add_10_sse4(uint8_t _dst, int16_t coeffs,
	1176	void ff_hevc_transform_8x8_add_10_sse4(uint8_t _dst, const int16_t coeffs,
1177	1177	ptrdiff_t _stride) {
1178	1178	int i;
1179	1179	uint16_t dst = (uint16_t) _dst;

1488	1488
1489	1489
1490	1490	#if HAVE_SSE4_1
1491		void ff_hevc_transform_16x16_add_8_sse4(uint8_t _dst, int16_t coeffs,
	1491	void ff_hevc_transform_16x16_add_8_sse4(uint8_t _dst, const int16_t coeffs,
1492	1492	ptrdiff_t _stride) {
1493	1493	uint8_t shift_2nd = 12; // 20 - Bit depth
1494	1494	uint16_t add_2nd = 1 << 11; //(1 << (shift_2nd - 1))
1495	1495	int i;
1496	1496	uint8_t dst = (uint8_t) _dst;
1497	1497	ptrdiff_t stride = _stride / sizeof(uint8_t);
1498		int16_t *src = coeffs;
	1498	const int16_t *src = coeffs;
1499	1499	int32_t shift;
1500	1500	__m128i m128iS0, m128iS1, m128iS2, m128iS3, m128iS4, m128iS5, m128iS6,
1501	1501	m128iS7, m128iS8, m128iS9, m128iS10, m128iS11, m128iS12, m128iS13,

2245	2245
2246	2246
2247	2247	#if 0
2248		void ff_hevc_transform_16x16_add_10_sse4(uint8_t _dst, int16_t coeffs,
	2248	void ff_hevc_transform_16x16_add_10_sse4(uint8_t _dst, const int16_t coeffs,
2249	2249	ptrdiff_t _stride) {
2250	2250	int i;
2251	2251	uint16_t dst = (uint16_t) _dst;

2908	2908
2909	2909
2910	2910	#if HAVE_SSE4_1
2911		void ff_hevc_transform_32x32_add_8_sse4(uint8_t _dst, int16_t coeffs,
	2911	void ff_hevc_transform_32x32_add_8_sse4(uint8_t _dst, const int16_t coeffs,
2912	2912	ptrdiff_t _stride) {
2913	2913	uint8_t shift_2nd = 12; // 20 - Bit depth
2914	2914	uint16_t add_2nd = 1 << 11; //(1 << (shift_2nd - 1))

2916	2916	uint8_t dst = (uint8_t) _dst;
2917	2917	ptrdiff_t stride = _stride / sizeof(uint8_t);
2918	2918	int shift;
2919		int16_t *src = coeffs;
	2919	const int16_t *src = coeffs;
2920	2920
2921	2921	__m128i m128iS0, m128iS1, m128iS2, m128iS3, m128iS4, m128iS5, m128iS6,
2922	2922	m128iS7, m128iS8, m128iS9, m128iS10, m128iS11, m128iS12, m128iS13,

5200	5200
5201	5201
5202	5202	#if 0
5203		void ff_hevc_transform_32x32_add_10_sse4(uint8_t _dst, int16_t coeffs,
	5203	void ff_hevc_transform_32x32_add_10_sse4(uint8_t _dst, const int16_t coeffs,
5204	5204	ptrdiff_t _stride) {
5205	5205	int i, j;
5206	5206	uint16_t dst = (uint16_t) _dst;

-6

libde265/x86/sse-dct.h less more

24	24	#include <stddef.h>
25	25	#include <stdint.h>
26	26
27		void ff_hevc_transform_skip_8_sse(uint8_t _dst, int16_t coeffs, ptrdiff_t _stride);
28		void ff_hevc_transform_4x4_luma_add_8_sse4(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
29		void ff_hevc_transform_4x4_add_8_sse4(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
30		void ff_hevc_transform_8x8_add_8_sse4(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
31		void ff_hevc_transform_16x16_add_8_sse4(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
32		void ff_hevc_transform_32x32_add_8_sse4(uint8_t dst, int16_t coeffs, ptrdiff_t stride);
	27	void ff_hevc_transform_skip_8_sse(uint8_t _dst, const int16_t coeffs, ptrdiff_t _stride);
	28	void ff_hevc_transform_4x4_luma_add_8_sse4(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	29	void ff_hevc_transform_4x4_add_8_sse4(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	30	void ff_hevc_transform_8x8_add_8_sse4(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	31	void ff_hevc_transform_16x16_add_8_sse4(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
	32	void ff_hevc_transform_32x32_add_8_sse4(uint8_t dst, const int16_t coeffs, ptrdiff_t stride);
33	33
34	34	#endif

+67

-56

libde265/x86/sse-motion.cc less more

85	85	#define BIT_DEPTH 8
86	86
87	87	void ff_hevc_put_unweighted_pred_8_sse(uint8_t *_dst, ptrdiff_t dststride,
88		int16_t *src, ptrdiff_t srcstride, int width, int height) {
	88	const int16_t *src, ptrdiff_t srcstride,
	89	int width, int height) {
89	90	int x, y;
90	91	uint8_t dst = (uint8_t) _dst;
91	92	__m128i r0, r1, f0;

168	169	}
169	170
170	171	void ff_hevc_put_unweighted_pred_sse(uint8_t *_dst, ptrdiff_t _dststride,
171		int16_t *src, ptrdiff_t srcstride, int width, int height) {
	172	const int16_t *src, ptrdiff_t srcstride,
	173	int width, int height) {
172	174	int x, y;
173	175	uint8_t dst = (uint8_t) _dst;
174	176	ptrdiff_t dststride = _dststride / sizeof(uint8_t);

202	204	}
203	205
204	206	void ff_hevc_put_weighted_pred_avg_8_sse(uint8_t *_dst, ptrdiff_t dststride,
205		int16_t src1, int16_t src2, ptrdiff_t srcstride, int width,
206		int height) {
	207	const int16_t src1, const int16_t src2,
	208	ptrdiff_t srcstride, int width,
	209	int height) {
207	210	int x, y;
208	211	uint8_t dst = (uint8_t) _dst;
209	212	__m128i r0, r1, f0, r2, r3;

306	309	}
307	310
308	311	void ff_hevc_put_weighted_pred_avg_sse(uint8_t *_dst, ptrdiff_t _dststride,
309		int16_t src1, int16_t src2, ptrdiff_t srcstride, int width,
310		int height) {
	312	const int16_t src1, const int16_t src2,
	313	ptrdiff_t srcstride, int width,
	314	int height) {
311	315	int x, y;
312	316	uint8_t dst = (uint8_t) _dst;
313	317	ptrdiff_t dststride = _dststride / sizeof(uint8_t);

345	349
346	350	#if 0
347	351	void ff_hevc_weighted_pred_8_sse4(uint8_t denom, int16_t wlxFlag, int16_t olxFlag,
348		uint8_t _dst, ptrdiff_t _dststride, int16_t src, ptrdiff_t srcstride,
349		int width, int height) {
	352	uint8_t *_dst, ptrdiff_t _dststride,
	353	const int16_t *src, ptrdiff_t srcstride,
	354	int width, int height) {
350	355
351	356	int log2Wd;
352	357	int x, y;

586	591
587	592	#if 0
588	593	void ff_hevc_weighted_pred_sse(uint8_t denom, int16_t wlxFlag, int16_t olxFlag,
589		uint8_t _dst, ptrdiff_t _dststride, int16_t src, ptrdiff_t srcstride,
590		int width, int height) {
	594	uint8_t *_dst, ptrdiff_t _dststride,
	595	const int16_t *src, ptrdiff_t srcstride,
	596	int width, int height) {
591	597
592	598	int log2Wd;
593	599	int x, y;

671	677
672	678	#if HAVE_SSE4_1
673	679	void ff_hevc_weighted_pred_avg_8_sse4(uint8_t denom, int16_t wl0Flag,
674		int16_t wl1Flag, int16_t ol0Flag, int16_t ol1Flag, uint8_t *_dst,
675		ptrdiff_t _dststride, int16_t src1, int16_t src2, ptrdiff_t srcstride,
676		int width, int height) {
	680	int16_t wl1Flag, int16_t ol0Flag, int16_t ol1Flag,
	681	uint8_t *_dst, ptrdiff_t _dststride,
	682	const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
	683	int width, int height) {
677	684	int shift, shift2;
678	685	int log2Wd;
679	686	int o0;

868	875	#if 0
869	876	void ff_hevc_weighted_pred_avg_sse(uint8_t denom, int16_t wl0Flag,
870	877	int16_t wl1Flag, int16_t ol0Flag, int16_t ol1Flag, uint8_t *_dst,
871		ptrdiff_t _dststride, int16_t src1, int16_t src2, ptrdiff_t srcstride,
	878	ptrdiff_t _dststride, const int16_t src1, const int16_t src2, ptrdiff_t srcstride,
872	879	int width, int height) {
873	880	int shift, shift2;
874	881	int log2Wd;

942	949
943	950
944	951	void ff_hevc_put_hevc_epel_pixels_8_sse(int16_t *dst, ptrdiff_t dststride,
945		uint8_t *_src, ptrdiff_t srcstride, int width, int height, int mx,
946		int my, int16_t* mcbuffer) {
	952	const uint8_t *_src, ptrdiff_t srcstride,
	953	int width, int height, int mx,
	954	int my, int16_t* mcbuffer) {
947	955	int x, y;
948	956	__m128i x1, x2,x3;
949	957	uint8_t src = (uint8_t) _src;

1019	1027
1020	1028	#ifndef __native_client__
1021	1029	void ff_hevc_put_hevc_epel_pixels_10_sse(int16_t *dst, ptrdiff_t dststride,
1022		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1023		int my, int16_t* mcbuffer) {
	1030	const uint8_t *_src, ptrdiff_t _srcstride,
	1031	int width, int height, int mx,
	1032	int my, int16_t* mcbuffer) {
1024	1033	int x, y;
1025	1034	__m128i x2;
1026	1035	uint16_t src = (uint16_t) _src;

1070	1079	#endif
1071	1080
1072	1081	void ff_hevc_put_hevc_epel_h_8_sse(int16_t *dst, ptrdiff_t dststride,
1073		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1074		int my, int16_t* mcbuffer) {
	1082	const uint8_t *_src, ptrdiff_t _srcstride,
	1083	int width, int height, int mx,
	1084	int my, int16_t* mcbuffer, int bit_depth) {
1075	1085	int x, y;
1076		uint8_t src = (uint8_t) _src;
	1086	const uint8_t src = (const uint8_t) _src;
1077	1087	ptrdiff_t srcstride = _srcstride;
1078	1088	const int8_t *filter = epel_filters[mx - 1];
1079	1089	__m128i r0, bshuffle1, bshuffle2, x1, x2, x3;

1167	1177
1168	1178	#ifndef __native_client__
1169	1179	void ff_hevc_put_hevc_epel_h_10_sse(int16_t *dst, ptrdiff_t dststride,
1170		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1171		int my, int16_t* mcbuffer) {
	1180	const uint8_t *_src, ptrdiff_t _srcstride,
	1181	int width, int height, int mx,
	1182	int my, int16_t* mcbuffer) {
1172	1183	int x, y;
1173	1184	uint16_t src = (uint16_t) _src;
1174	1185	ptrdiff_t srcstride = _srcstride>>1;

1229	1240
1230	1241
1231	1242	void ff_hevc_put_hevc_epel_v_8_sse(int16_t *dst, ptrdiff_t dststride,
1232		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1233		int my, int16_t* mcbuffer) {
	1243	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
	1244	int my, int16_t* mcbuffer, int bit_depth) {
1234	1245	int x, y;
1235	1246	__m128i x0, x1, x2, x3, t0, t1, t2, t3, r0, f0, f1, f2, f3, r1;
1236	1247	uint8_t src = (uint8_t) _src;

1368	1379
1369	1380	#ifndef __native_client__
1370	1381	void ff_hevc_put_hevc_epel_v_10_sse(int16_t *dst, ptrdiff_t dststride,
1371		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
	1382	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1372	1383	int my, int16_t* mcbuffer) {
1373	1384	int x, y;
1374	1385	__m128i x0, x1, x2, x3, t0, t1, t2, t3, r0, f0, f1, f2, f3, r1, r2, r3;

1535	1546	#endif
1536	1547
1537	1548	void ff_hevc_put_hevc_epel_hv_8_sse(int16_t *dst, ptrdiff_t dststride,
1538		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1539		int my, int16_t* mcbuffer) {
	1549	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
	1550	int my, int16_t* mcbuffer, int bit_depth) {
1540	1551	int x, y;
1541	1552	uint8_t src = (uint8_t) _src;
1542	1553	ptrdiff_t srcstride = _srcstride;

1762	1773
1763	1774	#ifndef __native_client__
1764	1775	void ff_hevc_put_hevc_epel_hv_10_sse(int16_t *dst, ptrdiff_t dststride,
1765		uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
	1776	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height, int mx,
1766	1777	int my, int16_t* mcbuffer) {
1767	1778	int x, y;
1768	1779	uint16_t src = (uint16_t) _src;

1924	1935	#endif
1925	1936
1926	1937	void ff_hevc_put_hevc_qpel_pixels_8_sse(int16_t *dst, ptrdiff_t dststride,
1927		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	1938	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
1928	1939	int16_t* mcbuffer) {
1929	1940	int x, y;
1930	1941	__m128i x1, x2, x3, x0;

1998	2009
1999	2010	#ifndef __native_client__
2000	2011	void ff_hevc_put_hevc_qpel_pixels_10_sse(int16_t *dst, ptrdiff_t dststride,
2001		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2012	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2002	2013	int16_t* mcbuffer) {
2003	2014	int x, y;
2004	2015	__m128i x1, x2, x4;

2045	2056
2046	2057
2047	2058	void ff_hevc_put_hevc_qpel_h_1_8_sse(int16_t *dst, ptrdiff_t dststride,
2048		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2059	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2049	2060	int16_t* mcbuffer) {
2050	2061	int x, y;
2051		uint8_t *src = _src;
	2062	const uint8_t *src = _src;
2052	2063	ptrdiff_t srcstride = _srcstride / sizeof(uint8_t);
2053	2064	__m128i x1, r0, x2, x3, x4, x5;
2054	2065

2144	2155	* @TODO : Valgrind to see if it's useful to use SSE or wait for AVX2 implementation
2145	2156	*/
2146	2157	void ff_hevc_put_hevc_qpel_h_1_10_sse(int16_t *dst, ptrdiff_t dststride,
2147		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2158	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2148	2159	int16_t* mcbuffer) {
2149	2160	int x, y;
2150	2161	uint16_t src = (uint16_t)_src;

2178	2189
2179	2190
2180	2191	void ff_hevc_put_hevc_qpel_h_2_8_sse(int16_t *dst, ptrdiff_t dststride,
2181		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2192	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2182	2193	int16_t* mcbuffer) {
2183	2194	int x, y;
2184		uint8_t *src = _src;
	2195	const uint8_t *src = _src;
2185	2196	ptrdiff_t srcstride = _srcstride / sizeof(uint8_t);
2186	2197	__m128i x1, r0, x2, x3, x4, x5;
2187	2198

2248	2259
2249	2260	#if 0
2250	2261	static void ff_hevc_put_hevc_qpel_h_2_sse(int16_t *dst, ptrdiff_t dststride,
2251		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2262	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2252	2263	int16_t* mcbuffer) {
2253	2264	int x, y;
2254	2265	uint8_t *src = _src;

2316	2327
2317	2328	}
2318	2329	static void ff_hevc_put_hevc_qpel_h_3_sse(int16_t *dst, ptrdiff_t dststride,
2319		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2330	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2320	2331	int16_t* mcbuffer) {
2321	2332	int x, y;
2322	2333	uint8_t *src = _src;

2383	2394	#endif
2384	2395
2385	2396	void ff_hevc_put_hevc_qpel_h_3_8_sse(int16_t *dst, ptrdiff_t dststride,
2386		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2397	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2387	2398	int16_t* mcbuffer) {
2388	2399	int x, y;
2389		uint8_t *src = _src;
	2400	const uint8_t *src = _src;
2390	2401	ptrdiff_t srcstride = _srcstride / sizeof(uint8_t);
2391	2402	__m128i x1, r0, x2, x3, x4, x5;
2392	2403

2451	2462
2452	2463	*/
2453	2464	void ff_hevc_put_hevc_qpel_v_1_8_sse(int16_t *dst, ptrdiff_t dststride,
2454		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2465	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2455	2466	int16_t* mcbuffer) {
2456	2467	int x, y;
2457	2468	uint8_t src = (uint8_t) _src;

2610	2621
2611	2622	#if 0
2612	2623	void ff_hevc_put_hevc_qpel_v_1_10_sse4(int16_t *dst, ptrdiff_t dststride,
2613		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2624	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2614	2625	int16_t* mcbuffer) {
2615	2626	int x, y;
2616	2627	uint16_t src = (uint16_t) _src;

2687	2698
2688	2699
2689	2700	void ff_hevc_put_hevc_qpel_v_2_8_sse(int16_t *dst, ptrdiff_t dststride,
2690		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2701	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2691	2702	int16_t* mcbuffer) {
2692	2703	int x, y;
2693	2704	uint8_t src = (uint8_t) _src;

2862	2873
2863	2874	#if 0
2864	2875	void ff_hevc_put_hevc_qpel_v_2_10_sse(int16_t *dst, ptrdiff_t dststride,
2865		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2876	cosnt uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2866	2877	int16_t* mcbuffer) {
2867	2878	int x, y;
2868	2879	uint16_t src = (uint16_t) _src;

2953	2964
2954	2965	#if 0
2955	2966	static void ff_hevc_put_hevc_qpel_v_3_sse(int16_t *dst, ptrdiff_t dststride,
2956		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	2967	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
2957	2968	int16_t* mcbuffer) {
2958	2969	int x, y;
2959	2970	uint8_t src = (uint8_t) _src;

3120	3131	#endif
3121	3132
3122	3133	void ff_hevc_put_hevc_qpel_v_3_8_sse(int16_t *dst, ptrdiff_t dststride,
3123		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	3134	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
3124	3135	int16_t* mcbuffer) {
3125	3136	int x, y;
3126	3137	uint8_t src = (uint8_t) _src;

3275	3286
3276	3287	#if 0
3277	3288	void ff_hevc_put_hevc_qpel_v_3_10_sse(int16_t *dst, ptrdiff_t dststride,
3278		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	3289	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
3279	3290	int16_t* mcbuffer) {
3280	3291	int x, y;
3281	3292	uint16_t src = (uint16_t) _src;

3351	3362
3352	3363
3353	3364	void ff_hevc_put_hevc_qpel_h_1_v_1_sse(int16_t *dst, ptrdiff_t dststride,
3354		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	3365	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
3355	3366	int16_t* mcbuffer) {
3356	3367	int x, y;
3357	3368	uint8_t* src = (uint8_t*) _src;

3523	3534	}
3524	3535	}
3525	3536	void ff_hevc_put_hevc_qpel_h_1_v_2_sse(int16_t *dst, ptrdiff_t dststride,
3526		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	3537	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
3527	3538	int16_t* mcbuffer) {
3528	3539	int x, y;
3529	3540	uint8_t src = (uint8_t) _src;

3706	3717	}
3707	3718	}
3708	3719	void ff_hevc_put_hevc_qpel_h_1_v_3_sse(int16_t *dst, ptrdiff_t dststride,
3709		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	3720	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
3710	3721	int16_t* mcbuffer) {
3711	3722	int x, y;
3712	3723	uint8_t src = (uint8_t) _src;

3883	3894	}
3884	3895	}
3885	3896	void ff_hevc_put_hevc_qpel_h_2_v_1_sse(int16_t *dst, ptrdiff_t dststride,
3886		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	3897	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
3887	3898	int16_t* mcbuffer) {
3888	3899	int x, y;
3889	3900	uint8_t src = (uint8_t) _src;

4054	4065	}
4055	4066	}
4056	4067	void ff_hevc_put_hevc_qpel_h_2_v_2_sse(int16_t *dst, ptrdiff_t dststride,
4057		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	4068	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
4058	4069	int16_t* mcbuffer) {
4059	4070	int x, y;
4060	4071	uint8_t src = (uint8_t) _src;

4237	4248	}
4238	4249	}
4239	4250	void ff_hevc_put_hevc_qpel_h_2_v_3_sse(int16_t *dst, ptrdiff_t dststride,
4240		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	4251	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
4241	4252	int16_t* mcbuffer) {
4242	4253	int x, y;
4243	4254	uint8_t src = (uint8_t) _src;

4415	4426	}
4416	4427	}
4417	4428	void ff_hevc_put_hevc_qpel_h_3_v_1_sse(int16_t *dst, ptrdiff_t dststride,
4418		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	4429	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
4419	4430	int16_t* mcbuffer) {
4420	4431	int x, y;
4421	4432	uint8_t src = (uint8_t) _src;

4588	4599	}
4589	4600	}
4590	4601	void ff_hevc_put_hevc_qpel_h_3_v_2_sse(int16_t *dst, ptrdiff_t dststride,
4591		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	4602	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
4592	4603	int16_t* mcbuffer) {
4593	4604	int x, y;
4594	4605	uint8_t src = (uint8_t) _src;

4774	4785	}
4775	4786	}
4776	4787	void ff_hevc_put_hevc_qpel_h_3_v_3_sse(int16_t *dst, ptrdiff_t dststride,
4777		uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
	4788	const uint8_t *_src, ptrdiff_t _srcstride, int width, int height,
4778	4789	int16_t* mcbuffer) {
4779	4790	int x, y;
4780	4791	uint8_t src = (uint8_t) _src;

+41

-41

libde265/x86/sse-motion.h less more

26	26
27	27
28	28	void ff_hevc_put_unweighted_pred_8_sse(uint8_t *_dst, ptrdiff_t dststride,
29		int16_t *src, ptrdiff_t srcstride,
30		int width, int height);
	29	const int16_t *src, ptrdiff_t srcstride,
	30	int width, int height);
31	31
32	32	void ff_hevc_put_weighted_pred_avg_8_sse(uint8_t *_dst, ptrdiff_t dststride,
33		int16_t src1, int16_t src2,
34		ptrdiff_t srcstride, int width,
35		int height);
	33	const int16_t src1, const int16_t src2,
	34	ptrdiff_t srcstride, int width,
	35	int height);
36	36
37	37	void ff_hevc_put_hevc_epel_pixels_8_sse(int16_t *dst, ptrdiff_t dststride,
38		uint8_t *_src, ptrdiff_t srcstride,
	38	const uint8_t *_src, ptrdiff_t srcstride,
39	39	int width, int height,
40	40	int mx, int my, int16_t* mcbuffer);
41	41	void ff_hevc_put_hevc_epel_h_8_sse(int16_t *dst, ptrdiff_t dststride,
42		uint8_t *_src, ptrdiff_t srcstride,
	42	const uint8_t *_src, ptrdiff_t srcstride,
43	43	int width, int height,
44		int mx, int my, int16_t* mcbuffer);
	44	int mx, int my, int16_t* mcbuffer, int bit_depth);
45	45	void ff_hevc_put_hevc_epel_v_8_sse(int16_t *dst, ptrdiff_t dststride,
46		uint8_t *_src, ptrdiff_t srcstride,
	46	const uint8_t *_src, ptrdiff_t srcstride,
47	47	int width, int height,
48		int mx, int my, int16_t* mcbuffer);
	48	int mx, int my, int16_t* mcbuffer, int bit_depth);
49	49	void ff_hevc_put_hevc_epel_hv_8_sse(int16_t *dst, ptrdiff_t dststride,
50		uint8_t *_src, ptrdiff_t srcstride,
	50	const uint8_t *_src, ptrdiff_t srcstride,
51	51	int width, int height,
52		int mx, int my, int16_t* mcbuffer);
	52	int mx, int my, int16_t* mcbuffer, int bit_depth);
53	53
54	54	void ff_hevc_put_hevc_qpel_pixels_8_sse(int16_t *dst, ptrdiff_t dststride,
55		uint8_t *src, ptrdiff_t srcstride,
56		int width, int height, int16_t* mcbuffer);
	55	const uint8_t *src, ptrdiff_t srcstride,
	56	int width, int height, int16_t* mcbuffer);
57	57	void ff_hevc_put_hevc_qpel_v_1_8_sse(int16_t *dst, ptrdiff_t dststride,
58		uint8_t *src, ptrdiff_t srcstride,
59		int width, int height, int16_t* mcbuffer);
	58	const uint8_t *src, ptrdiff_t srcstride,
	59	int width, int height, int16_t* mcbuffer);
60	60	void ff_hevc_put_hevc_qpel_v_2_8_sse(int16_t *dst, ptrdiff_t dststride,
61		uint8_t *src, ptrdiff_t srcstride,
62		int width, int height, int16_t* mcbuffer);
	61	const uint8_t *src, ptrdiff_t srcstride,
	62	int width, int height, int16_t* mcbuffer);
63	63	void ff_hevc_put_hevc_qpel_v_3_8_sse(int16_t *dst, ptrdiff_t dststride,
64		uint8_t *src, ptrdiff_t srcstride,
65		int width, int height, int16_t* mcbuffer);
	64	const uint8_t *src, ptrdiff_t srcstride,
	65	int width, int height, int16_t* mcbuffer);
66	66	void ff_hevc_put_hevc_qpel_h_1_8_sse(int16_t *dst, ptrdiff_t dststride,
67		uint8_t *src, ptrdiff_t srcstride,
	67	const uint8_t *src, ptrdiff_t srcstride,
68	68	int width, int height, int16_t* mcbuffer);
69	69	void ff_hevc_put_hevc_qpel_h_1_v_1_sse(int16_t *dst, ptrdiff_t dststride,
70		uint8_t *src, ptrdiff_t srcstride,
71		int width, int height, int16_t* mcbuffer);
	70	const uint8_t *src, ptrdiff_t srcstride,
	71	int width, int height, int16_t* mcbuffer);
72	72	void ff_hevc_put_hevc_qpel_h_1_v_2_sse(int16_t *dst, ptrdiff_t dststride,
73		uint8_t *src, ptrdiff_t srcstride,
74		int width, int height, int16_t* mcbuffer);
	73	const uint8_t *src, ptrdiff_t srcstride,
	74	int width, int height, int16_t* mcbuffer);
75	75	void ff_hevc_put_hevc_qpel_h_1_v_3_sse(int16_t *dst, ptrdiff_t dststride,
76		uint8_t *src, ptrdiff_t srcstride,
77		int width, int height, int16_t* mcbuffer);
	76	const uint8_t *src, ptrdiff_t srcstride,
	77	int width, int height, int16_t* mcbuffer);
78	78	void ff_hevc_put_hevc_qpel_h_2_8_sse(int16_t *dst, ptrdiff_t dststride,
79		uint8_t *src, ptrdiff_t srcstride,
	79	const uint8_t *src, ptrdiff_t srcstride,
80	80	int width, int height, int16_t* mcbuffer);
81	81	void ff_hevc_put_hevc_qpel_h_2_v_1_sse(int16_t *dst, ptrdiff_t dststride,
82		uint8_t *src, ptrdiff_t srcstride,
83		int width, int height, int16_t* mcbuffer);
	82	const uint8_t *src, ptrdiff_t srcstride,
	83	int width, int height, int16_t* mcbuffer);
84	84	void ff_hevc_put_hevc_qpel_h_2_v_2_sse(int16_t *dst, ptrdiff_t dststride,
85		uint8_t *src, ptrdiff_t srcstride,
86		int width, int height, int16_t* mcbuffer);
	85	const uint8_t *src, ptrdiff_t srcstride,
	86	int width, int height, int16_t* mcbuffer);
87	87	void ff_hevc_put_hevc_qpel_h_2_v_3_sse(int16_t *dst, ptrdiff_t dststride,
88		uint8_t *src, ptrdiff_t srcstride,
89		int width, int height, int16_t* mcbuffer);
	88	const uint8_t *src, ptrdiff_t srcstride,
	89	int width, int height, int16_t* mcbuffer);
90	90	void ff_hevc_put_hevc_qpel_h_3_8_sse(int16_t *dst, ptrdiff_t dststride,
91		uint8_t *src, ptrdiff_t srcstride,
	91	const uint8_t *src, ptrdiff_t srcstride,
92	92	int width, int height, int16_t* mcbuffer);
93	93	void ff_hevc_put_hevc_qpel_h_3_v_1_sse(int16_t *dst, ptrdiff_t dststride,
94		uint8_t *src, ptrdiff_t srcstride,
95		int width, int height, int16_t* mcbuffer);
	94	const uint8_t *src, ptrdiff_t srcstride,
	95	int width, int height, int16_t* mcbuffer);
96	96	void ff_hevc_put_hevc_qpel_h_3_v_2_sse(int16_t *dst, ptrdiff_t dststride,
97		uint8_t *src, ptrdiff_t srcstride,
98		int width, int height, int16_t* mcbuffer);
	97	const uint8_t *src, ptrdiff_t srcstride,
	98	int width, int height, int16_t* mcbuffer);
99	99	void ff_hevc_put_hevc_qpel_h_3_v_3_sse(int16_t *dst, ptrdiff_t dststride,
100		uint8_t *src, ptrdiff_t srcstride,
101		int width, int height, int16_t* mcbuffer);
	100	const uint8_t *src, ptrdiff_t srcstride,
	101	int width, int height, int16_t* mcbuffer);
102	102
103	103	#endif

-3

libde265/x86/sse.cc less more

94	94	//accel->transform_4x4_luma_add_8 = ff_hevc_transform_4x4_luma_add_8_sse4; // SSE-4 only TODO
95	95	//accel->transform_4x4_add_8 = ff_hevc_transform_4x4_add_8_sse4;
96	96
97		accel->transform_8x8_add_8 = ff_hevc_transform_8x8_add_8_sse4;
98		accel->transform_16x16_add_8 = ff_hevc_transform_16x16_add_8_sse4;
99		accel->transform_32x32_add_8 = ff_hevc_transform_32x32_add_8_sse4;
	97	accel->transform_add_8[1] = ff_hevc_transform_8x8_add_8_sse4;
	98	accel->transform_add_8[2] = ff_hevc_transform_16x16_add_8_sse4;
	99	accel->transform_add_8[3] = ff_hevc_transform_32x32_add_8_sse4;
100	100	}
101	101	#endif
102	102	}

+142

-0

m4/ax_cxx_compile_stdcxx_11.m4 less more

	0	# ============================================================================
	1	# http://www.gnu.org/software/autoconf-archive/ax_cxx_compile_stdcxx_11.html
	2	# ============================================================================
	3	#
	4	# SYNOPSIS
	5	#
	6	# AX_CXX_COMPILE_STDCXX_11([ext\|noext],[mandatory\|optional])
	7	#
	8	# DESCRIPTION
	9	#
	10	# Check for baseline language coverage in the compiler for the C++11
	11	# standard; if necessary, add switches to CXXFLAGS to enable support.
	12	#
	13	# The first argument, if specified, indicates whether you insist on an
	14	# extended mode (e.g. -std=gnu++11) or a strict conformance mode (e.g.
	15	# -std=c++11). If neither is specified, you get whatever works, with
	16	# preference for an extended mode.
	17	#
	18	# The second argument, if specified 'mandatory' or if left unspecified,
	19	# indicates that baseline C++11 support is required and that the macro
	20	# should error out if no mode with that support is found. If specified
	21	# 'optional', then configuration proceeds regardless, after defining
	22	# HAVE_CXX11 if and only if a supporting mode is found.
	23	#
	24	# LICENSE
	25	#
	26	# Copyright (c) 2008 Benjamin Kosnik <bkoz@redhat.com>
	27	# Copyright (c) 2012 Zack Weinberg <zackw@panix.com>
	28	# Copyright (c) 2013 Roy Stogner <roystgnr@ices.utexas.edu>
	29	# Copyright (c) 2014 Alexey Sokolov <sokolov@google.com>
	30	#
	31	# Copying and distribution of this file, with or without modification, are
	32	# permitted in any medium without royalty provided the copyright notice
	33	# and this notice are preserved. This file is offered as-is, without any
	34	# warranty.
	35
	36	#serial 4
	37
	38	m4_define([_AX_CXX_COMPILE_STDCXX_11_testbody], [[
	39	template <typename T>
	40	struct check
	41	{
	42	static_assert(sizeof(int) <= sizeof(T), "not big enough");
	43	};
	44
	45	struct Base {
	46	virtual void f() {}
	47	};
	48	struct Child : public Base {
	49	virtual void f() {} // DiFa: override {} # override not supported in gcc 4.6
	50	};
	51
	52	typedef check<check<bool>> right_angle_brackets;
	53
	54	int a;
	55	decltype(a) b;
	56
	57	typedef check<int> check_type;
	58	check_type c;
	59	check_type&& cr = static_cast<check_type&&>(c);
	60
	61	auto d = a;
	62	auto l = [](){};
	63	]])
	64
	65	AC_DEFUN([AX_CXX_COMPILE_STDCXX_11], [dnl
	66	m4_if([$1], [], [],
	67	[$1], [ext], [],
	68	[$1], [noext], [],
	69	[m4_fatal([invalid argument `$1' to AX_CXX_COMPILE_STDCXX_11])])dnl
	70	m4_if([$2], [], [ax_cxx_compile_cxx11_required=true],
	71	[$2], [mandatory], [ax_cxx_compile_cxx11_required=true],
	72	[$2], [optional], [ax_cxx_compile_cxx11_required=false],
	73	[m4_fatal([invalid second argument `$2' to AX_CXX_COMPILE_STDCXX_11])])
	74	AC_LANG_PUSH([C++])dnl
	75	ac_success=no
	76	AC_CACHE_CHECK(whether $CXX supports C++11 features by default,
	77	ax_cv_cxx_compile_cxx11,
	78	[AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_11_testbody])],
	79	[ax_cv_cxx_compile_cxx11=yes],
	80	[ax_cv_cxx_compile_cxx11=no])])
	81	if test x$ax_cv_cxx_compile_cxx11 = xyes; then
	82	ac_success=yes
	83	fi
	84
	85	m4_if([$1], [noext], [], [dnl
	86	if test x$ac_success = xno; then
	87	for switch in -std=gnu++11 -std=gnu++0x; do
	88	cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx11_$switch])
	89	AC_CACHE_CHECK(whether $CXX supports C++11 features with $switch,
	90	$cachevar,
	91	[ac_save_CXXFLAGS="$CXXFLAGS"
	92	CXXFLAGS="$CXXFLAGS $switch"
	93	AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_11_testbody])],
	94	[eval $cachevar=yes],
	95	[eval $cachevar=no])
	96	CXXFLAGS="$ac_save_CXXFLAGS"])
	97	if eval test x\$$cachevar = xyes; then
	98	CXXFLAGS="$CXXFLAGS $switch"
	99	ac_success=yes
	100	break
	101	fi
	102	done
	103	fi])
	104
	105	m4_if([$1], [ext], [], [dnl
	106	if test x$ac_success = xno; then
	107	for switch in -std=c++11 -std=c++0x; do
	108	cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx11_$switch])
	109	AC_CACHE_CHECK(whether $CXX supports C++11 features with $switch,
	110	$cachevar,
	111	[ac_save_CXXFLAGS="$CXXFLAGS"
	112	CXXFLAGS="$CXXFLAGS $switch"
	113	AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_11_testbody])],
	114	[eval $cachevar=yes],
	115	[eval $cachevar=no])
	116	CXXFLAGS="$ac_save_CXXFLAGS"])
	117	if eval test x\$$cachevar = xyes; then
	118	CXXFLAGS="$CXXFLAGS $switch"
	119	ac_success=yes
	120	break
	121	fi
	122	done
	123	fi])
	124	AC_LANG_POP([C++])
	125	if test x$ax_cxx_compile_cxx11_required = xtrue; then
	126	if test x$ac_success = xno; then
	127	AC_MSG_ERROR([*** A compiler with support for C++11 language features is required.])
	128	fi
	129	else
	130	if test x$ac_success = xno; then
	131	HAVE_CXX11=0
	132	AC_MSG_NOTICE([No compiler with C++11 support was found])
	133	else
	134	HAVE_CXX11=1
	135	AC_DEFINE(HAVE_CXX11,1,
	136	[define if the compiler supports basic C++11 syntax])
	137	fi
	138
	139	AC_SUBST(HAVE_CXX11)
	140	fi
	141	])

-52

~~m4/visibility.m4~~ less more

0		# visibility.m4 serial 1 (gettext-0.15)
1		dnl Copyright (C) 2005 Free Software Foundation, Inc.
2		dnl This file is free software; the Free Software Foundation
3		dnl gives unlimited permission to copy and/or distribute it,
4		dnl with or without modifications, as long as this notice is preserved.
5
6		dnl From Bruno Haible.
7
8		dnl Tests whether the compiler supports the command-line option
9		dnl -fvisibility=hidden and the function and variable attributes
10		dnl __attribute__((__visibility__("hidden"))) and
11		dnl __attribute__((__visibility__("default"))).
12		dnl Does not test for __visibility__("protected") - which has tricky
13		dnl semantics (see the 'vismain' test in glibc) and does not exist e.g. on
14		dnl MacOS X.
15		dnl Does not test for __visibility__("internal") - which has processor
16		dnl dependent semantics.
17		dnl Does not test for #pragma GCC visibility push(hidden) - which is
18		dnl "really only recommended for legacy code".
19		dnl Set the variable CFLAG_VISIBILITY.
20		dnl Defines and sets the variable HAVE_VISIBILITY.
21
22		AC_DEFUN([gl_VISIBILITY],
23		[
24		AC_REQUIRE([AC_PROG_CC])
25		CFLAG_VISIBILITY=
26		HAVE_VISIBILITY=0
27		if test -n "$GCC"; then
28		AC_MSG_CHECKING([for simple visibility declarations])
29		AC_CACHE_VAL(gl_cv_cc_visibility, [
30		gl_save_CFLAGS="$CFLAGS"
31		CFLAGS="$CFLAGS -fvisibility=hidden"
32		AC_TRY_COMPILE(
33		[extern __attribute__((__visibility__("hidden"))) int hiddenvar;
34		extern __attribute__((__visibility__("default"))) int exportedvar;
35		extern __attribute__((__visibility__("hidden"))) int hiddenfunc (void);
36		extern __attribute__((__visibility__("default"))) int exportedfunc (void);],
37		[],
38		gl_cv_cc_visibility=yes,
39		gl_cv_cc_visibility=no)
40		CFLAGS="$gl_save_CFLAGS"])
41		AC_MSG_RESULT([$gl_cv_cc_visibility])
42		if test $gl_cv_cc_visibility = yes; then
43		CFLAG_VISIBILITY="-fvisibility=hidden"
44		HAVE_VISIBILITY=1
45		fi
46		fi
47		AC_SUBST([CFLAG_VISIBILITY])
48		AC_SUBST([HAVE_VISIBILITY])
49		AC_DEFINE_UNQUOTED([HAVE_VISIBILITY], [$HAVE_VISIBILITY],
50		[Define to 1 or 0, depending whether the compiler supports simple visibility declarations.])
51		])

-3

sherlock265/Makefile.in less more

88	88	$(top_srcdir)/depcomp COPYING README
89	89	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
90	90	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	91	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
91	92	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
92	93	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
93	94	$(top_srcdir)/m4/lt~obsolete.m4 \
94	95	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
95		$(top_srcdir)/m4/visibility.m4 $(top_srcdir)/configure.ac
	96	$(top_srcdir)/configure.ac
96	97	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
97	98	$(ACLOCAL_M4)
98	99	mkinstalldirs = $(install_sh) -d

197	198	CTAGS = ctags
198	199	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
199	200	ACLOCAL = @ACLOCAL@
	201	ALLOCA = @ALLOCA@
200	202	AMTAR = @AMTAR@
201	203	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
202	204	AR = @AR@

205	207	AUTOMAKE = @AUTOMAKE@
206	208	AWK = @AWK@
207	209	CC = @CC@
	210	CCAS = @CCAS@
	211	CCASDEPMODE = @CCASDEPMODE@
	212	CCASFLAGS = @CCASFLAGS@
208	213	CCDEPMODE = @CCDEPMODE@
209	214	CFLAGS = @CFLAGS@
210		CFLAG_VISIBILITY = @CFLAG_VISIBILITY@
211	215	CPP = @CPP@
212	216	CPPFLAGS = @CPPFLAGS@
213	217	CXX = @CXX@

227	231	EXEEXT = @EXEEXT@
228	232	FGREP = @FGREP@
229	233	GREP = @GREP@
230		HAVE_VISIBILITY = @HAVE_VISIBILITY@
	234	HAVE_CXX11 = @HAVE_CXX11@
231	235	INSTALL = @INSTALL@
232	236	INSTALL_DATA = @INSTALL_DATA@
233	237	INSTALL_PROGRAM = @INSTALL_PROGRAM@

-5

sherlock265/VideoDecoder.cc less more

32	32
33	33
34	34	VideoDecoder::VideoDecoder()
35		: ctx(NULL),
	35	: mFH(NULL),
	36	ctx(NULL),
36	37	img(NULL),
37	38	mNextBuffer(0),
38	39	mFrameCount(0),

40	41	mVideoEnded(false),
41	42	mSingleStep(false),
42	43	mShowDecodedImage(true),
	44	mShowQuantPY(false),
43	45	mCBShowPartitioning(false),
44	46	mTBShowPartitioning(false),
45	47	mPBShowPartitioning(false),
	48	mShowIntraPredMode(false),
46	49	mShowPBPredMode(false),
47		mShowIntraPredMode(false),
48		mShowQuantPY(false),
49	50	mShowMotionVec(false),
50		mShowSlices(false),
51	51	mShowTiles(false),
52		mFH(NULL)
	52	mShowSlices(false)
53	53	#ifdef HAVE_SWSCALE
54	54	, sws(NULL)
55	55	, width(0)

+46

-0

tools/Makefile.am less more

	0
	1	bin_PROGRAMS = gen-enc-table yuv-distortion rd-curves block-rate-estim tests bjoentegaard
	2
	3	AM_CPPFLAGS = -I../libde265
	4
	5	gen_enc_table_DEPENDENCIES = ../libde265/libde265.la
	6	gen_enc_table_CXXFLAGS =
	7	gen_enc_table_LDFLAGS =
	8	gen_enc_table_LDADD = ../libde265/libde265.la -lstdc++
	9	gen_enc_table_SOURCES = gen-entropy-table.cc
	10
	11	yuv_distortion_DEPENDENCIES = ../libde265/libde265.la
	12	yuv_distortion_CXXFLAGS =
	13	yuv_distortion_LDFLAGS =
	14	yuv_distortion_LDADD = ../libde265/libde265.la -lstdc++
	15	yuv_distortion_SOURCES = yuv-distortion.cc
	16
	17	if HAVE_VIDEOGFX
	18	yuv_distortion_CXXFLAGS += $(VIDEOGFX_CFLAGS)
	19	yuv_distortion_LDFLAGS += $(VIDEOGFX_LIBS)
	20	endif
	21
	22	rd_curves_DEPENDENCIES = ../libde265/libde265.la
	23	rd_curves_CXXFLAGS =
	24	rd_curves_LDFLAGS =
	25	rd_curves_LDADD = ../libde265/libde265.la -lstdc++
	26	rd_curves_SOURCES = rd-curves.cc
	27
	28	block_rate_estim_DEPENDENCIES = ../libde265/libde265.la
	29	block_rate_estim_CXXFLAGS =
	30	block_rate_estim_LDFLAGS =
	31	block_rate_estim_LDADD = ../libde265/libde265.la -lstdc++
	32	block_rate_estim_SOURCES = block-rate-estim.cc
	33
	34	tests_DEPENDENCIES = ../libde265/libde265.la
	35	tests_CXXFLAGS =
	36	tests_LDFLAGS =
	37	tests_LDADD = ../libde265/libde265.la -lstdc++
	38	tests_SOURCES = tests.cc
	39
	40	bjoentegaard_DEPENDENCIES = ../libde265/libde265.la
	41	bjoentegaard_CXXFLAGS =
	42	bjoentegaard_LDFLAGS =
	43	bjoentegaard_LDADD = ../libde265/libde265.la -lstdc++
	44	bjoentegaard_SOURCES = bjoentegaard.cc
	45

+825

-0

tools/Makefile.in less more

	0	# Makefile.in generated by automake 1.14.1 from Makefile.am.
	1	# @configure_input@
	2
	3	# Copyright (C) 1994-2013 Free Software Foundation, Inc.
	4
	5	# This Makefile.in is free software; the Free Software Foundation
	6	# gives unlimited permission to copy and/or distribute it,
	7	# with or without modifications, as long as this notice is preserved.
	8
	9	# This program is distributed in the hope that it will be useful,
	10	# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
	11	# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
	12	# PARTICULAR PURPOSE.
	13
	14	@SET_MAKE@
	15
	16	VPATH = @srcdir@
	17	am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
	18	am__make_running_with_option = \
	19	case $${target_option-} in \
	20	?) ;; \
	21	*) echo "am__make_running_with_option: internal error: invalid" \
	22	"target option '$${target_option-}' specified" >&2; \
	23	exit 1;; \
	24	esac; \
	25	has_opt=no; \
	26	sane_makeflags=$$MAKEFLAGS; \
	27	if $(am__is_gnu_make); then \
	28	sane_makeflags=$$MFLAGS; \
	29	else \
	30	case $$MAKEFLAGS in \
	31	\\[\ \ ]) \
	32	bs=\\; \
	33	sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
	34	\| sed "s/$$bs$$bs[$$bs $$bs ]*//g"`;; \
	35	esac; \
	36	fi; \
	37	skip_next=no; \
	38	strip_trailopt () \
	39	{ \
	40	flg=`printf '%s\n' "$$flg" \| sed "s/$$1.*$$//"`; \
	41	}; \
	42	for flg in $$sane_makeflags; do \
	43	test $$skip_next = yes && { skip_next=no; continue; }; \
	44	case $$flg in \
	45	=\|--*) continue;; \
	46	-*I) strip_trailopt 'I'; skip_next=yes;; \
	47	-I?) strip_trailopt 'I';; \
	48	-*O) strip_trailopt 'O'; skip_next=yes;; \
	49	-O?) strip_trailopt 'O';; \
	50	-*l) strip_trailopt 'l'; skip_next=yes;; \
	51	-l?) strip_trailopt 'l';; \
	52	-[dEDm]) skip_next=yes;; \
	53	-[JT]) skip_next=yes;; \
	54	esac; \
	55	case $$flg in \
	56	$$target_option) has_opt=yes; break;; \
	57	esac; \
	58	done; \
	59	test $$has_opt = yes
	60	am__make_dryrun = (target_option=n; $(am__make_running_with_option))
	61	am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
	62	pkgdatadir = $(datadir)/@PACKAGE@
	63	pkgincludedir = $(includedir)/@PACKAGE@
	64	pkglibdir = $(libdir)/@PACKAGE@
	65	pkglibexecdir = $(libexecdir)/@PACKAGE@
	66	am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
	67	install_sh_DATA = $(install_sh) -c -m 644
	68	install_sh_PROGRAM = $(install_sh) -c
	69	install_sh_SCRIPT = $(install_sh) -c
	70	INSTALL_HEADER = $(INSTALL_DATA)
	71	transform = $(program_transform_name)
	72	NORMAL_INSTALL = :
	73	PRE_INSTALL = :
	74	POST_INSTALL = :
	75	NORMAL_UNINSTALL = :
	76	PRE_UNINSTALL = :
	77	POST_UNINSTALL = :
	78	build_triplet = @build@
	79	host_triplet = @host@
	80	target_triplet = @target@
	81	bin_PROGRAMS = gen-enc-table$(EXEEXT) yuv-distortion$(EXEEXT) \
	82	rd-curves$(EXEEXT) block-rate-estim$(EXEEXT) tests$(EXEEXT) \
	83	bjoentegaard$(EXEEXT)
	84	@HAVE_VIDEOGFX_TRUE@am__append_1 = $(VIDEOGFX_CFLAGS)
	85	@HAVE_VIDEOGFX_TRUE@am__append_2 = $(VIDEOGFX_LIBS)
	86	subdir = tools
	87	DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/Makefile.am \
	88	$(top_srcdir)/depcomp
	89	ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
	90	am__aclocal_m4_deps = $(top_srcdir)/m4/ax_compare_version.m4 \
	91	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
	92	$(top_srcdir)/m4/libtool.m4 $(top_srcdir)/m4/ltoptions.m4 \
	93	$(top_srcdir)/m4/ltsugar.m4 $(top_srcdir)/m4/ltversion.m4 \
	94	$(top_srcdir)/m4/lt~obsolete.m4 \
	95	$(top_srcdir)/m4/m4_ax_check_compile_flag.m4 \
	96	$(top_srcdir)/configure.ac
	97	am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
	98	$(ACLOCAL_M4)
	99	mkinstalldirs = $(install_sh) -d
	100	CONFIG_HEADER = $(top_builddir)/config.h
	101	CONFIG_CLEAN_FILES =
	102	CONFIG_CLEAN_VPATH_FILES =
	103	am__installdirs = "$(DESTDIR)$(bindir)"
	104	PROGRAMS = $(bin_PROGRAMS)
	105	am_bjoentegaard_OBJECTS = bjoentegaard-bjoentegaard.$(OBJEXT)
	106	bjoentegaard_OBJECTS = $(am_bjoentegaard_OBJECTS)
	107	AM_V_lt = $(am__v_lt_@AM_V@)
	108	am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
	109	am__v_lt_0 = --silent
	110	am__v_lt_1 =
	111	bjoentegaard_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	112	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(bjoentegaard_CXXFLAGS) \
	113	$(CXXFLAGS) $(bjoentegaard_LDFLAGS) $(LDFLAGS) -o $@
	114	am_block_rate_estim_OBJECTS = \
	115	block_rate_estim-block-rate-estim.$(OBJEXT)
	116	block_rate_estim_OBJECTS = $(am_block_rate_estim_OBJECTS)
	117	block_rate_estim_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
	118	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
	119	$(block_rate_estim_CXXFLAGS) $(CXXFLAGS) \
	120	$(block_rate_estim_LDFLAGS) $(LDFLAGS) -o $@
	121	am_gen_enc_table_OBJECTS = gen_enc_table-gen-entropy-table.$(OBJEXT)
	122	gen_enc_table_OBJECTS = $(am_gen_enc_table_OBJECTS)
	123	gen_enc_table_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
	124	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
	125	$(gen_enc_table_CXXFLAGS) $(CXXFLAGS) $(gen_enc_table_LDFLAGS) \
	126	$(LDFLAGS) -o $@
	127	am_rd_curves_OBJECTS = rd_curves-rd-curves.$(OBJEXT)
	128	rd_curves_OBJECTS = $(am_rd_curves_OBJECTS)
	129	rd_curves_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	130	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(rd_curves_CXXFLAGS) \
	131	$(CXXFLAGS) $(rd_curves_LDFLAGS) $(LDFLAGS) -o $@
	132	am_tests_OBJECTS = tests-tests.$(OBJEXT)
	133	tests_OBJECTS = $(am_tests_OBJECTS)
	134	tests_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	135	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(tests_CXXFLAGS) \
	136	$(CXXFLAGS) $(tests_LDFLAGS) $(LDFLAGS) -o $@
	137	am_yuv_distortion_OBJECTS = yuv_distortion-yuv-distortion.$(OBJEXT)
	138	yuv_distortion_OBJECTS = $(am_yuv_distortion_OBJECTS)
	139	yuv_distortion_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
	140	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
	141	$(yuv_distortion_CXXFLAGS) $(CXXFLAGS) \
	142	$(yuv_distortion_LDFLAGS) $(LDFLAGS) -o $@
	143	AM_V_P = $(am__v_P_@AM_V@)
	144	am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
	145	am__v_P_0 = false
	146	am__v_P_1 = :
	147	AM_V_GEN = $(am__v_GEN_@AM_V@)
	148	am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
	149	am__v_GEN_0 = @echo " GEN " $@;
	150	am__v_GEN_1 =
	151	AM_V_at = $(am__v_at_@AM_V@)
	152	am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
	153	am__v_at_0 = @
	154	am__v_at_1 =
	155	DEFAULT_INCLUDES = -I.@am__isrc@ -I$(top_builddir)
	156	depcomp = $(SHELL) $(top_srcdir)/depcomp
	157	am__depfiles_maybe = depfiles
	158	am__mv = mv -f
	159	CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
	160	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
	161	LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	162	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
	163	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
	164	$(AM_CXXFLAGS) $(CXXFLAGS)
	165	AM_V_CXX = $(am__v_CXX_@AM_V@)
	166	am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
	167	am__v_CXX_0 = @echo " CXX " $@;
	168	am__v_CXX_1 =
	169	CXXLD = $(CXX)
	170	CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
	171	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
	172	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
	173	AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
	174	am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
	175	am__v_CXXLD_0 = @echo " CXXLD " $@;
	176	am__v_CXXLD_1 =
	177	SOURCES = $(bjoentegaard_SOURCES) $(block_rate_estim_SOURCES) \
	178	$(gen_enc_table_SOURCES) $(rd_curves_SOURCES) $(tests_SOURCES) \
	179	$(yuv_distortion_SOURCES)
	180	DIST_SOURCES = $(bjoentegaard_SOURCES) $(block_rate_estim_SOURCES) \
	181	$(gen_enc_table_SOURCES) $(rd_curves_SOURCES) $(tests_SOURCES) \
	182	$(yuv_distortion_SOURCES)
	183	am__can_run_installinfo = \
	184	case $$AM_UPDATE_INFO_DIR in \
	185	n\|no\|NO) false;; \
	186	*) (install-info --version) >/dev/null 2>&1;; \
	187	esac
	188	am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
	189	# Read a list of newline-separated strings from the standard input,
	190	# and print each of them once, without duplicates. Input order is
	191	# not preserved.
	192	am__uniquify_input = $(AWK) '\
	193	BEGIN { nonempty = 0; } \
	194	{ items[$$0] = 1; nonempty = 1; } \
	195	END { if (nonempty) { for (i in items) print i; }; } \
	196	'
	197	# Make sure the list of sources is unique. This is necessary because,
	198	# e.g., the same source file might be shared among _SOURCES variables
	199	# for different programs/libraries.
	200	am__define_uniq_tagged_files = \
	201	list='$(am__tagged_files)'; \
	202	unique=`for i in $$list; do \
	203	if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
	204	done \| $(am__uniquify_input)`
	205	ETAGS = etags
	206	CTAGS = ctags
	207	DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
	208	ACLOCAL = @ACLOCAL@
	209	ALLOCA = @ALLOCA@
	210	AMTAR = @AMTAR@
	211	AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
	212	AR = @AR@
	213	AUTOCONF = @AUTOCONF@
	214	AUTOHEADER = @AUTOHEADER@
	215	AUTOMAKE = @AUTOMAKE@
	216	AWK = @AWK@
	217	CC = @CC@
	218	CCAS = @CCAS@
	219	CCASDEPMODE = @CCASDEPMODE@
	220	CCASFLAGS = @CCASFLAGS@
	221	CCDEPMODE = @CCDEPMODE@
	222	CFLAGS = @CFLAGS@
	223	CPP = @CPP@
	224	CPPFLAGS = @CPPFLAGS@
	225	CXX = @CXX@
	226	CXXCPP = @CXXCPP@
	227	CXXDEPMODE = @CXXDEPMODE@
	228	CXXFLAGS = @CXXFLAGS@
	229	CYGPATH_W = @CYGPATH_W@
	230	DEFS = @DEFS@
	231	DEPDIR = @DEPDIR@
	232	DLLTOOL = @DLLTOOL@
	233	DSYMUTIL = @DSYMUTIL@
	234	DUMPBIN = @DUMPBIN@
	235	ECHO_C = @ECHO_C@
	236	ECHO_N = @ECHO_N@
	237	ECHO_T = @ECHO_T@
	238	EGREP = @EGREP@
	239	EXEEXT = @EXEEXT@
	240	FGREP = @FGREP@
	241	GREP = @GREP@
	242	HAVE_CXX11 = @HAVE_CXX11@
	243	INSTALL = @INSTALL@
	244	INSTALL_DATA = @INSTALL_DATA@
	245	INSTALL_PROGRAM = @INSTALL_PROGRAM@
	246	INSTALL_SCRIPT = @INSTALL_SCRIPT@
	247	INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
	248	LD = @LD@
	249	LDFLAGS = @LDFLAGS@
	250	LIBDE265_AGE = @LIBDE265_AGE@
	251	LIBDE265_CURRENT = @LIBDE265_CURRENT@
	252	LIBDE265_REVISION = @LIBDE265_REVISION@
	253	LIBOBJS = @LIBOBJS@
	254	LIBS = @LIBS@
	255	LIBTOOL = @LIBTOOL@
	256	LIPO = @LIPO@
	257	LN_S = @LN_S@
	258	LTLIBOBJS = @LTLIBOBJS@
	259	MAKEINFO = @MAKEINFO@
	260	MANIFEST_TOOL = @MANIFEST_TOOL@
	261	MKDIR_P = @MKDIR_P@
	262	NM = @NM@
	263	NMEDIT = @NMEDIT@
	264	NUMERIC_VERSION = @NUMERIC_VERSION@
	265	OBJDUMP = @OBJDUMP@
	266	OBJEXT = @OBJEXT@
	267	OTOOL = @OTOOL@
	268	OTOOL64 = @OTOOL64@
	269	PACKAGE = @PACKAGE@
	270	PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
	271	PACKAGE_NAME = @PACKAGE_NAME@
	272	PACKAGE_STRING = @PACKAGE_STRING@
	273	PACKAGE_TARNAME = @PACKAGE_TARNAME@
	274	PACKAGE_URL = @PACKAGE_URL@
	275	PACKAGE_VERSION = @PACKAGE_VERSION@
	276	PATH_SEPARATOR = @PATH_SEPARATOR@
	277	PKG_CONFIG = @PKG_CONFIG@
	278	PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
	279	PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
	280	QTCHOOSER = @QTCHOOSER@
	281	QTMOC = @QTMOC@
	282	QT_CFLAGS = @QT_CFLAGS@
	283	QT_LIBS = @QT_LIBS@
	284	RANLIB = @RANLIB@
	285	SDL_CFLAGS = @SDL_CFLAGS@
	286	SDL_LIBS = @SDL_LIBS@
	287	SED = @SED@
	288	SET_MAKE = @SET_MAKE@
	289	SHELL = @SHELL@
	290	STRIP = @STRIP@
	291	SWSCALE_CFLAGS = @SWSCALE_CFLAGS@
	292	SWSCALE_LIBS = @SWSCALE_LIBS@
	293	VERSION = @VERSION@
	294	VIDEOGFX_CFLAGS = @VIDEOGFX_CFLAGS@
	295	VIDEOGFX_LIBS = @VIDEOGFX_LIBS@
	296	abs_builddir = @abs_builddir@
	297	abs_srcdir = @abs_srcdir@
	298	abs_top_builddir = @abs_top_builddir@
	299	abs_top_srcdir = @abs_top_srcdir@
	300	ac_ct_AR = @ac_ct_AR@
	301	ac_ct_CC = @ac_ct_CC@
	302	ac_ct_CXX = @ac_ct_CXX@
	303	ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
	304	am__include = @am__include@
	305	am__leading_dot = @am__leading_dot@
	306	am__quote = @am__quote@
	307	am__tar = @am__tar@
	308	am__untar = @am__untar@
	309	bindir = @bindir@
	310	build = @build@
	311	build_alias = @build_alias@
	312	build_cpu = @build_cpu@
	313	build_os = @build_os@
	314	build_vendor = @build_vendor@
	315	builddir = @builddir@
	316	datadir = @datadir@
	317	datarootdir = @datarootdir@
	318	docdir = @docdir@
	319	dvidir = @dvidir@
	320	exec_prefix = @exec_prefix@
	321	host = @host@
	322	host_alias = @host_alias@
	323	host_cpu = @host_cpu@
	324	host_os = @host_os@
	325	host_vendor = @host_vendor@
	326	htmldir = @htmldir@
	327	includedir = @includedir@
	328	infodir = @infodir@
	329	install_sh = @install_sh@
	330	libdir = @libdir@
	331	libexecdir = @libexecdir@
	332	localedir = @localedir@
	333	localstatedir = @localstatedir@
	334	mandir = @mandir@
	335	mkdir_p = @mkdir_p@
	336	oldincludedir = @oldincludedir@
	337	pdfdir = @pdfdir@
	338	prefix = @prefix@
	339	program_transform_name = @program_transform_name@
	340	psdir = @psdir@
	341	sbindir = @sbindir@
	342	sharedstatedir = @sharedstatedir@
	343	srcdir = @srcdir@
	344	sysconfdir = @sysconfdir@
	345	target = @target@
	346	target_alias = @target_alias@
	347	target_cpu = @target_cpu@
	348	target_os = @target_os@
	349	target_vendor = @target_vendor@
	350	top_build_prefix = @top_build_prefix@
	351	top_builddir = @top_builddir@
	352	top_srcdir = @top_srcdir@
	353	AM_CPPFLAGS = -I../libde265
	354	gen_enc_table_DEPENDENCIES = ../libde265/libde265.la
	355	gen_enc_table_CXXFLAGS =
	356	gen_enc_table_LDFLAGS =
	357	gen_enc_table_LDADD = ../libde265/libde265.la -lstdc++
	358	gen_enc_table_SOURCES = gen-entropy-table.cc
	359	yuv_distortion_DEPENDENCIES = ../libde265/libde265.la
	360	yuv_distortion_CXXFLAGS = $(am__append_1)
	361	yuv_distortion_LDFLAGS = $(am__append_2)
	362	yuv_distortion_LDADD = ../libde265/libde265.la -lstdc++
	363	yuv_distortion_SOURCES = yuv-distortion.cc
	364	rd_curves_DEPENDENCIES = ../libde265/libde265.la
	365	rd_curves_CXXFLAGS =
	366	rd_curves_LDFLAGS =
	367	rd_curves_LDADD = ../libde265/libde265.la -lstdc++
	368	rd_curves_SOURCES = rd-curves.cc
	369	block_rate_estim_DEPENDENCIES = ../libde265/libde265.la
	370	block_rate_estim_CXXFLAGS =
	371	block_rate_estim_LDFLAGS =
	372	block_rate_estim_LDADD = ../libde265/libde265.la -lstdc++
	373	block_rate_estim_SOURCES = block-rate-estim.cc
	374	tests_DEPENDENCIES = ../libde265/libde265.la
	375	tests_CXXFLAGS =
	376	tests_LDFLAGS =
	377	tests_LDADD = ../libde265/libde265.la -lstdc++
	378	tests_SOURCES = tests.cc
	379	bjoentegaard_DEPENDENCIES = ../libde265/libde265.la
	380	bjoentegaard_CXXFLAGS =
	381	bjoentegaard_LDFLAGS =
	382	bjoentegaard_LDADD = ../libde265/libde265.la -lstdc++
	383	bjoentegaard_SOURCES = bjoentegaard.cc
	384	all: all-am
	385
	386	.SUFFIXES:
	387	.SUFFIXES: .cc .lo .o .obj
	388	$(srcdir)/Makefile.in: $(srcdir)/Makefile.am $(am__configure_deps)
	389	@for dep in $?; do \
	390	case '$(am__configure_deps)' in \
	391	$$dep) \
	392	( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
	393	&& { if test -f $@; then exit 0; else break; fi; }; \
	394	exit 1;; \
	395	esac; \
	396	done; \
	397	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu tools/Makefile'; \
	398	$(am__cd) $(top_srcdir) && \
	399	$(AUTOMAKE) --gnu tools/Makefile
	400	.PRECIOUS: Makefile
	401	Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
	402	@case '$?' in \
	403	config.status) \
	404	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
	405	*) \
	406	echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
	407	cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
	408	esac;
	409
	410	$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
	411	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	412
	413	$(top_srcdir)/configure: $(am__configure_deps)
	414	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	415	$(ACLOCAL_M4): $(am__aclocal_m4_deps)
	416	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
	417	$(am__aclocal_m4_deps):
	418	install-binPROGRAMS: $(bin_PROGRAMS)
	419	@$(NORMAL_INSTALL)
	420	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" \|\| list=; \
	421	if test -n "$$list"; then \
	422	echo " $(MKDIR_P) '$(DESTDIR)$(bindir)'"; \
	423	$(MKDIR_P) "$(DESTDIR)$(bindir)" \|\| exit 1; \
	424	fi; \
	425	for p in $$list; do echo "$$p $$p"; done \| \
	426	sed 's/$(EXEEXT)$$//' \| \
	427	while read p p1; do if test -f $$p \
	428	\|\| test -f $$p1 \
	429	; then echo "$$p"; echo "$$p"; else :; fi; \
	430	done \| \
	431	sed -e 'p;s,.*/,,;n;h' \
	432	-e 's\|.*\|.\|' \
	433	-e 'p;x;s,.*/,,;s/$(EXEEXT)$$//;$(transform);s/$$/$(EXEEXT)/' \| \
	434	sed 'N;N;N;s,\n, ,g' \| \
	435	$(AWK) 'BEGIN { files["."] = ""; dirs["."] = 1 } \
	436	{ d=$$3; if (dirs[d] != 1) { print "d", d; dirs[d] = 1 } \
	437	if ($$2 == $$4) files[d] = files[d] " " $$1; \
	438	else { print "f", $$3 "/" $$4, $$1; } } \
	439	END { for (d in files) print "f", d, files[d] }' \| \
	440	while read type dir files; do \
	441	if test "$$dir" = .; then dir=; else dir=/$$dir; fi; \
	442	test -z "$$files" \|\| { \
	443	echo " $(INSTALL_PROGRAM_ENV) $(LIBTOOL) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=install $(INSTALL_PROGRAM) $$files '$(DESTDIR)$(bindir)$$dir'"; \
	444	$(INSTALL_PROGRAM_ENV) $(LIBTOOL) $(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=install $(INSTALL_PROGRAM) $$files "$(DESTDIR)$(bindir)$$dir" \|\| exit $$?; \
	445	} \
	446	; done
	447
	448	uninstall-binPROGRAMS:
	449	@$(NORMAL_UNINSTALL)
	450	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" \|\| list=; \
	451	files=`for p in $$list; do echo "$$p"; done \| \
	452	sed -e 'h;s,^.*/,,;s/$(EXEEXT)$$//;$(transform)' \
	453	-e 's/$$/$(EXEEXT)/' \
	454	`; \
	455	test -n "$$list" \|\| exit 0; \
	456	echo " ( cd '$(DESTDIR)$(bindir)' && rm -f" $$files ")"; \
	457	cd "$(DESTDIR)$(bindir)" && rm -f $$files
	458
	459	clean-binPROGRAMS:
	460	@list='$(bin_PROGRAMS)'; test -n "$$list" \|\| exit 0; \
	461	echo " rm -f" $$list; \
	462	rm -f $$list \|\| exit $$?; \
	463	test -n "$(EXEEXT)" \|\| exit 0; \
	464	list=`for p in $$list; do echo "$$p"; done \| sed 's/$(EXEEXT)$$//'`; \
	465	echo " rm -f" $$list; \
	466	rm -f $$list
	467
	468	bjoentegaard$(EXEEXT): $(bjoentegaard_OBJECTS) $(bjoentegaard_DEPENDENCIES) $(EXTRA_bjoentegaard_DEPENDENCIES)
	469	@rm -f bjoentegaard$(EXEEXT)
	470	$(AM_V_CXXLD)$(bjoentegaard_LINK) $(bjoentegaard_OBJECTS) $(bjoentegaard_LDADD) $(LIBS)
	471
	472	block-rate-estim$(EXEEXT): $(block_rate_estim_OBJECTS) $(block_rate_estim_DEPENDENCIES) $(EXTRA_block_rate_estim_DEPENDENCIES)
	473	@rm -f block-rate-estim$(EXEEXT)
	474	$(AM_V_CXXLD)$(block_rate_estim_LINK) $(block_rate_estim_OBJECTS) $(block_rate_estim_LDADD) $(LIBS)
	475
	476	gen-enc-table$(EXEEXT): $(gen_enc_table_OBJECTS) $(gen_enc_table_DEPENDENCIES) $(EXTRA_gen_enc_table_DEPENDENCIES)
	477	@rm -f gen-enc-table$(EXEEXT)
	478	$(AM_V_CXXLD)$(gen_enc_table_LINK) $(gen_enc_table_OBJECTS) $(gen_enc_table_LDADD) $(LIBS)
	479
	480	rd-curves$(EXEEXT): $(rd_curves_OBJECTS) $(rd_curves_DEPENDENCIES) $(EXTRA_rd_curves_DEPENDENCIES)
	481	@rm -f rd-curves$(EXEEXT)
	482	$(AM_V_CXXLD)$(rd_curves_LINK) $(rd_curves_OBJECTS) $(rd_curves_LDADD) $(LIBS)
	483
	484	tests$(EXEEXT): $(tests_OBJECTS) $(tests_DEPENDENCIES) $(EXTRA_tests_DEPENDENCIES)
	485	@rm -f tests$(EXEEXT)
	486	$(AM_V_CXXLD)$(tests_LINK) $(tests_OBJECTS) $(tests_LDADD) $(LIBS)
	487
	488	yuv-distortion$(EXEEXT): $(yuv_distortion_OBJECTS) $(yuv_distortion_DEPENDENCIES) $(EXTRA_yuv_distortion_DEPENDENCIES)
	489	@rm -f yuv-distortion$(EXEEXT)
	490	$(AM_V_CXXLD)$(yuv_distortion_LINK) $(yuv_distortion_OBJECTS) $(yuv_distortion_LDADD) $(LIBS)
	491
	492	mostlyclean-compile:
	493	-rm -f *.$(OBJEXT)
	494
	495	distclean-compile:
	496	-rm -f *.tab.c
	497
	498	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/bjoentegaard-bjoentegaard.Po@am__quote@
	499	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/block_rate_estim-block-rate-estim.Po@am__quote@
	500	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/gen_enc_table-gen-entropy-table.Po@am__quote@
	501	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/rd_curves-rd-curves.Po@am__quote@
	502	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tests-tests.Po@am__quote@
	503	@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/yuv_distortion-yuv-distortion.Po@am__quote@
	504
	505	.cc.o:
	506	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	507	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	508	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	509	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	510	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
	511
	512	.cc.obj:
	513	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
	514	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Po
	515	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
	516	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	517	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
	518
	519	.cc.lo:
	520	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
	521	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/$.Tpo $(DEPDIR)/$.Plo
	522	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
	523	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	524	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
	525
	526	bjoentegaard-bjoentegaard.o: bjoentegaard.cc
	527	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(bjoentegaard_CXXFLAGS) $(CXXFLAGS) -MT bjoentegaard-bjoentegaard.o -MD -MP -MF $(DEPDIR)/bjoentegaard-bjoentegaard.Tpo -c -o bjoentegaard-bjoentegaard.o `test -f 'bjoentegaard.cc' \|\| echo '$(srcdir)/'`bjoentegaard.cc
	528	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/bjoentegaard-bjoentegaard.Tpo $(DEPDIR)/bjoentegaard-bjoentegaard.Po
	529	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='bjoentegaard.cc' object='bjoentegaard-bjoentegaard.o' libtool=no @AMDEPBACKSLASH@
	530	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	531	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(bjoentegaard_CXXFLAGS) $(CXXFLAGS) -c -o bjoentegaard-bjoentegaard.o `test -f 'bjoentegaard.cc' \|\| echo '$(srcdir)/'`bjoentegaard.cc
	532
	533	bjoentegaard-bjoentegaard.obj: bjoentegaard.cc
	534	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(bjoentegaard_CXXFLAGS) $(CXXFLAGS) -MT bjoentegaard-bjoentegaard.obj -MD -MP -MF $(DEPDIR)/bjoentegaard-bjoentegaard.Tpo -c -o bjoentegaard-bjoentegaard.obj `if test -f 'bjoentegaard.cc'; then $(CYGPATH_W) 'bjoentegaard.cc'; else $(CYGPATH_W) '$(srcdir)/bjoentegaard.cc'; fi`
	535	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/bjoentegaard-bjoentegaard.Tpo $(DEPDIR)/bjoentegaard-bjoentegaard.Po
	536	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='bjoentegaard.cc' object='bjoentegaard-bjoentegaard.obj' libtool=no @AMDEPBACKSLASH@
	537	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	538	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(bjoentegaard_CXXFLAGS) $(CXXFLAGS) -c -o bjoentegaard-bjoentegaard.obj `if test -f 'bjoentegaard.cc'; then $(CYGPATH_W) 'bjoentegaard.cc'; else $(CYGPATH_W) '$(srcdir)/bjoentegaard.cc'; fi`
	539
	540	block_rate_estim-block-rate-estim.o: block-rate-estim.cc
	541	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(block_rate_estim_CXXFLAGS) $(CXXFLAGS) -MT block_rate_estim-block-rate-estim.o -MD -MP -MF $(DEPDIR)/block_rate_estim-block-rate-estim.Tpo -c -o block_rate_estim-block-rate-estim.o `test -f 'block-rate-estim.cc' \|\| echo '$(srcdir)/'`block-rate-estim.cc
	542	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/block_rate_estim-block-rate-estim.Tpo $(DEPDIR)/block_rate_estim-block-rate-estim.Po
	543	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='block-rate-estim.cc' object='block_rate_estim-block-rate-estim.o' libtool=no @AMDEPBACKSLASH@
	544	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	545	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(block_rate_estim_CXXFLAGS) $(CXXFLAGS) -c -o block_rate_estim-block-rate-estim.o `test -f 'block-rate-estim.cc' \|\| echo '$(srcdir)/'`block-rate-estim.cc
	546
	547	block_rate_estim-block-rate-estim.obj: block-rate-estim.cc
	548	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(block_rate_estim_CXXFLAGS) $(CXXFLAGS) -MT block_rate_estim-block-rate-estim.obj -MD -MP -MF $(DEPDIR)/block_rate_estim-block-rate-estim.Tpo -c -o block_rate_estim-block-rate-estim.obj `if test -f 'block-rate-estim.cc'; then $(CYGPATH_W) 'block-rate-estim.cc'; else $(CYGPATH_W) '$(srcdir)/block-rate-estim.cc'; fi`
	549	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/block_rate_estim-block-rate-estim.Tpo $(DEPDIR)/block_rate_estim-block-rate-estim.Po
	550	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='block-rate-estim.cc' object='block_rate_estim-block-rate-estim.obj' libtool=no @AMDEPBACKSLASH@
	551	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	552	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(block_rate_estim_CXXFLAGS) $(CXXFLAGS) -c -o block_rate_estim-block-rate-estim.obj `if test -f 'block-rate-estim.cc'; then $(CYGPATH_W) 'block-rate-estim.cc'; else $(CYGPATH_W) '$(srcdir)/block-rate-estim.cc'; fi`
	553
	554	gen_enc_table-gen-entropy-table.o: gen-entropy-table.cc
	555	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(gen_enc_table_CXXFLAGS) $(CXXFLAGS) -MT gen_enc_table-gen-entropy-table.o -MD -MP -MF $(DEPDIR)/gen_enc_table-gen-entropy-table.Tpo -c -o gen_enc_table-gen-entropy-table.o `test -f 'gen-entropy-table.cc' \|\| echo '$(srcdir)/'`gen-entropy-table.cc
	556	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/gen_enc_table-gen-entropy-table.Tpo $(DEPDIR)/gen_enc_table-gen-entropy-table.Po
	557	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='gen-entropy-table.cc' object='gen_enc_table-gen-entropy-table.o' libtool=no @AMDEPBACKSLASH@
	558	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	559	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(gen_enc_table_CXXFLAGS) $(CXXFLAGS) -c -o gen_enc_table-gen-entropy-table.o `test -f 'gen-entropy-table.cc' \|\| echo '$(srcdir)/'`gen-entropy-table.cc
	560
	561	gen_enc_table-gen-entropy-table.obj: gen-entropy-table.cc
	562	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(gen_enc_table_CXXFLAGS) $(CXXFLAGS) -MT gen_enc_table-gen-entropy-table.obj -MD -MP -MF $(DEPDIR)/gen_enc_table-gen-entropy-table.Tpo -c -o gen_enc_table-gen-entropy-table.obj `if test -f 'gen-entropy-table.cc'; then $(CYGPATH_W) 'gen-entropy-table.cc'; else $(CYGPATH_W) '$(srcdir)/gen-entropy-table.cc'; fi`
	563	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/gen_enc_table-gen-entropy-table.Tpo $(DEPDIR)/gen_enc_table-gen-entropy-table.Po
	564	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='gen-entropy-table.cc' object='gen_enc_table-gen-entropy-table.obj' libtool=no @AMDEPBACKSLASH@
	565	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	566	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(gen_enc_table_CXXFLAGS) $(CXXFLAGS) -c -o gen_enc_table-gen-entropy-table.obj `if test -f 'gen-entropy-table.cc'; then $(CYGPATH_W) 'gen-entropy-table.cc'; else $(CYGPATH_W) '$(srcdir)/gen-entropy-table.cc'; fi`
	567
	568	rd_curves-rd-curves.o: rd-curves.cc
	569	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rd_curves_CXXFLAGS) $(CXXFLAGS) -MT rd_curves-rd-curves.o -MD -MP -MF $(DEPDIR)/rd_curves-rd-curves.Tpo -c -o rd_curves-rd-curves.o `test -f 'rd-curves.cc' \|\| echo '$(srcdir)/'`rd-curves.cc
	570	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/rd_curves-rd-curves.Tpo $(DEPDIR)/rd_curves-rd-curves.Po
	571	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='rd-curves.cc' object='rd_curves-rd-curves.o' libtool=no @AMDEPBACKSLASH@
	572	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	573	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rd_curves_CXXFLAGS) $(CXXFLAGS) -c -o rd_curves-rd-curves.o `test -f 'rd-curves.cc' \|\| echo '$(srcdir)/'`rd-curves.cc
	574
	575	rd_curves-rd-curves.obj: rd-curves.cc
	576	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rd_curves_CXXFLAGS) $(CXXFLAGS) -MT rd_curves-rd-curves.obj -MD -MP -MF $(DEPDIR)/rd_curves-rd-curves.Tpo -c -o rd_curves-rd-curves.obj `if test -f 'rd-curves.cc'; then $(CYGPATH_W) 'rd-curves.cc'; else $(CYGPATH_W) '$(srcdir)/rd-curves.cc'; fi`
	577	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/rd_curves-rd-curves.Tpo $(DEPDIR)/rd_curves-rd-curves.Po
	578	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='rd-curves.cc' object='rd_curves-rd-curves.obj' libtool=no @AMDEPBACKSLASH@
	579	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	580	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rd_curves_CXXFLAGS) $(CXXFLAGS) -c -o rd_curves-rd-curves.obj `if test -f 'rd-curves.cc'; then $(CYGPATH_W) 'rd-curves.cc'; else $(CYGPATH_W) '$(srcdir)/rd-curves.cc'; fi`
	581
	582	tests-tests.o: tests.cc
	583	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(tests_CXXFLAGS) $(CXXFLAGS) -MT tests-tests.o -MD -MP -MF $(DEPDIR)/tests-tests.Tpo -c -o tests-tests.o `test -f 'tests.cc' \|\| echo '$(srcdir)/'`tests.cc
	584	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/tests-tests.Tpo $(DEPDIR)/tests-tests.Po
	585	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='tests.cc' object='tests-tests.o' libtool=no @AMDEPBACKSLASH@
	586	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	587	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(tests_CXXFLAGS) $(CXXFLAGS) -c -o tests-tests.o `test -f 'tests.cc' \|\| echo '$(srcdir)/'`tests.cc
	588
	589	tests-tests.obj: tests.cc
	590	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(tests_CXXFLAGS) $(CXXFLAGS) -MT tests-tests.obj -MD -MP -MF $(DEPDIR)/tests-tests.Tpo -c -o tests-tests.obj `if test -f 'tests.cc'; then $(CYGPATH_W) 'tests.cc'; else $(CYGPATH_W) '$(srcdir)/tests.cc'; fi`
	591	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/tests-tests.Tpo $(DEPDIR)/tests-tests.Po
	592	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='tests.cc' object='tests-tests.obj' libtool=no @AMDEPBACKSLASH@
	593	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	594	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(tests_CXXFLAGS) $(CXXFLAGS) -c -o tests-tests.obj `if test -f 'tests.cc'; then $(CYGPATH_W) 'tests.cc'; else $(CYGPATH_W) '$(srcdir)/tests.cc'; fi`
	595
	596	yuv_distortion-yuv-distortion.o: yuv-distortion.cc
	597	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(yuv_distortion_CXXFLAGS) $(CXXFLAGS) -MT yuv_distortion-yuv-distortion.o -MD -MP -MF $(DEPDIR)/yuv_distortion-yuv-distortion.Tpo -c -o yuv_distortion-yuv-distortion.o `test -f 'yuv-distortion.cc' \|\| echo '$(srcdir)/'`yuv-distortion.cc
	598	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/yuv_distortion-yuv-distortion.Tpo $(DEPDIR)/yuv_distortion-yuv-distortion.Po
	599	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='yuv-distortion.cc' object='yuv_distortion-yuv-distortion.o' libtool=no @AMDEPBACKSLASH@
	600	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	601	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(yuv_distortion_CXXFLAGS) $(CXXFLAGS) -c -o yuv_distortion-yuv-distortion.o `test -f 'yuv-distortion.cc' \|\| echo '$(srcdir)/'`yuv-distortion.cc
	602
	603	yuv_distortion-yuv-distortion.obj: yuv-distortion.cc
	604	@am__fastdepCXX_TRUE@ $(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(yuv_distortion_CXXFLAGS) $(CXXFLAGS) -MT yuv_distortion-yuv-distortion.obj -MD -MP -MF $(DEPDIR)/yuv_distortion-yuv-distortion.Tpo -c -o yuv_distortion-yuv-distortion.obj `if test -f 'yuv-distortion.cc'; then $(CYGPATH_W) 'yuv-distortion.cc'; else $(CYGPATH_W) '$(srcdir)/yuv-distortion.cc'; fi`
	605	@am__fastdepCXX_TRUE@ $(AM_V_at)$(am__mv) $(DEPDIR)/yuv_distortion-yuv-distortion.Tpo $(DEPDIR)/yuv_distortion-yuv-distortion.Po
	606	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ $(AM_V_CXX)source='yuv-distortion.cc' object='yuv_distortion-yuv-distortion.obj' libtool=no @AMDEPBACKSLASH@
	607	@AMDEP_TRUE@@am__fastdepCXX_FALSE@ DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
	608	@am__fastdepCXX_FALSE@ $(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(yuv_distortion_CXXFLAGS) $(CXXFLAGS) -c -o yuv_distortion-yuv-distortion.obj `if test -f 'yuv-distortion.cc'; then $(CYGPATH_W) 'yuv-distortion.cc'; else $(CYGPATH_W) '$(srcdir)/yuv-distortion.cc'; fi`
	609
	610	mostlyclean-libtool:
	611	-rm -f *.lo
	612
	613	clean-libtool:
	614	-rm -rf .libs _libs
	615
	616	ID: $(am__tagged_files)
	617	$(am__define_uniq_tagged_files); mkid -fID $$unique
	618	tags: tags-am
	619	TAGS: tags
	620
	621	tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	622	set x; \
	623	here=`pwd`; \
	624	$(am__define_uniq_tagged_files); \
	625	shift; \
	626	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
	627	test -n "$$unique" \|\| unique=$$empty_fix; \
	628	if test $$# -gt 0; then \
	629	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	630	"$$@" $$unique; \
	631	else \
	632	$(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
	633	$$unique; \
	634	fi; \
	635	fi
	636	ctags: ctags-am
	637
	638	CTAGS: ctags
	639	ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
	640	$(am__define_uniq_tagged_files); \
	641	test -z "$(CTAGS_ARGS)$$unique" \
	642	\|\| $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
	643	$$unique
	644
	645	GTAGS:
	646	here=`$(am__cd) $(top_builddir) && pwd` \
	647	&& $(am__cd) $(top_srcdir) \
	648	&& gtags -i $(GTAGS_ARGS) "$$here"
	649	cscopelist: cscopelist-am
	650
	651	cscopelist-am: $(am__tagged_files)
	652	list='$(am__tagged_files)'; \
	653	case "$(srcdir)" in \
	654	[\\/]* \| ?:[\\/]*) sdir="$(srcdir)" ;; \
	655	*) sdir=$(subdir)/$(srcdir) ;; \
	656	esac; \
	657	for i in $$list; do \
	658	if test -f "$$i"; then \
	659	echo "$(subdir)/$$i"; \
	660	else \
	661	echo "$$sdir/$$i"; \
	662	fi; \
	663	done >> $(top_builddir)/cscope.files
	664
	665	distclean-tags:
	666	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
	667
	668	distdir: $(DISTFILES)
	669	@srcdirstrip=`echo "$(srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	670	topsrcdirstrip=`echo "$(top_srcdir)" \| sed 's/[].[^$$\\*]/\\\\&/g'`; \
	671	list='$(DISTFILES)'; \
	672	dist_files=`for file in $$list; do echo $$file; done \| \
	673	sed -e "s\|^$$srcdirstrip/\|\|;t" \
	674	-e "s\|^$$topsrcdirstrip/\|$(top_builddir)/\|;t"`; \
	675	case $$dist_files in \
	676	/) $(MKDIR_P) `echo "$$dist_files" \| \
	677	sed '/\//!d;s\|^\|$(distdir)/\|;s,/[^/]*$$,,' \| \
	678	sort -u` ;; \
	679	esac; \
	680	for file in $$dist_files; do \
	681	if test -f $$file \|\| test -d $$file; then d=.; else d=$(srcdir); fi; \
	682	if test -d $$d/$$file; then \
	683	dir=`echo "/$$file" \| sed -e 's,/[^/]*$$,,'`; \
	684	if test -d "$(distdir)/$$file"; then \
	685	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	686	fi; \
	687	if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
	688	cp -fpR $(srcdir)/$$file "$(distdir)$$dir" \|\| exit 1; \
	689	find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
	690	fi; \
	691	cp -fpR $$d/$$file "$(distdir)$$dir" \|\| exit 1; \
	692	else \
	693	test -f "$(distdir)/$$file" \
	694	\|\| cp -p $$d/$$file "$(distdir)/$$file" \
	695	\|\| exit 1; \
	696	fi; \
	697	done
	698	check-am: all-am
	699	check: check-am
	700	all-am: Makefile $(PROGRAMS)
	701	installdirs:
	702	for dir in "$(DESTDIR)$(bindir)"; do \
	703	test -z "$$dir" \|\| $(MKDIR_P) "$$dir"; \
	704	done
	705	install: install-am
	706	install-exec: install-exec-am
	707	install-data: install-data-am
	708	uninstall: uninstall-am
	709
	710	install-am: all-am
	711	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
	712
	713	installcheck: installcheck-am
	714	install-strip:
	715	if test -z '$(STRIP)'; then \
	716	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	717	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	718	install; \
	719	else \
	720	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
	721	install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
	722	"INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
	723	fi
	724	mostlyclean-generic:
	725
	726	clean-generic:
	727
	728	distclean-generic:
	729	-test -z "$(CONFIG_CLEAN_FILES)" \|\| rm -f $(CONFIG_CLEAN_FILES)
	730	-test . = "$(srcdir)" \|\| test -z "$(CONFIG_CLEAN_VPATH_FILES)" \|\| rm -f $(CONFIG_CLEAN_VPATH_FILES)
	731
	732	maintainer-clean-generic:
	733	@echo "This command is intended for maintainers to use"
	734	@echo "it deletes files that may require special tools to rebuild."
	735	clean: clean-am
	736
	737	clean-am: clean-binPROGRAMS clean-generic clean-libtool mostlyclean-am
	738
	739	distclean: distclean-am
	740	-rm -rf ./$(DEPDIR)
	741	-rm -f Makefile
	742	distclean-am: clean-am distclean-compile distclean-generic \
	743	distclean-tags
	744
	745	dvi: dvi-am
	746
	747	dvi-am:
	748
	749	html: html-am
	750
	751	html-am:
	752
	753	info: info-am
	754
	755	info-am:
	756
	757	install-data-am:
	758
	759	install-dvi: install-dvi-am
	760
	761	install-dvi-am:
	762
	763	install-exec-am: install-binPROGRAMS
	764
	765	install-html: install-html-am
	766
	767	install-html-am:
	768
	769	install-info: install-info-am
	770
	771	install-info-am:
	772
	773	install-man:
	774
	775	install-pdf: install-pdf-am
	776
	777	install-pdf-am:
	778
	779	install-ps: install-ps-am
	780
	781	install-ps-am:
	782
	783	installcheck-am:
	784
	785	maintainer-clean: maintainer-clean-am
	786	-rm -rf ./$(DEPDIR)
	787	-rm -f Makefile
	788	maintainer-clean-am: distclean-am maintainer-clean-generic
	789
	790	mostlyclean: mostlyclean-am
	791
	792	mostlyclean-am: mostlyclean-compile mostlyclean-generic \
	793	mostlyclean-libtool
	794
	795	pdf: pdf-am
	796
	797	pdf-am:
	798
	799	ps: ps-am
	800
	801	ps-am:
	802
	803	uninstall-am: uninstall-binPROGRAMS
	804
	805	.MAKE: install-am install-strip
	806
	807	.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean \
	808	clean-binPROGRAMS clean-generic clean-libtool cscopelist-am \
	809	ctags ctags-am distclean distclean-compile distclean-generic \
	810	distclean-libtool distclean-tags distdir dvi dvi-am html \
	811	html-am info info-am install install-am install-binPROGRAMS \
	812	install-data install-data-am install-dvi install-dvi-am \
	813	install-exec install-exec-am install-html install-html-am \
	814	install-info install-info-am install-man install-pdf \
	815	install-pdf-am install-ps install-ps-am install-strip \
	816	installcheck installcheck-am installdirs maintainer-clean \
	817	maintainer-clean-generic mostlyclean mostlyclean-compile \
	818	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
	819	tags tags-am uninstall uninstall-am uninstall-binPROGRAMS
	820
	821
	822	# Tell versions [3.59,3.63) of GNU make to not export all variables.
	823	# Otherwise a system limit (for SysV at least) may be exceeded.
	824	.NOEXPORT:

+344

-0

tools/bjoentegaard.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include <stdio.h>
	21	#include <stdlib.h>
	22	#include <fstream>
	23	#include <sstream>
	24	#include <string>
	25	#include <vector>
	26	#include <math.h>
	27	#include <unistd.h>
	28
	29
	30	const bool D = false;
	31
	32
	33	/* There are numerical stability problems in the matrix inverse.
	34	Switching to long double seems to provide enough accuracy.
	35	TODO: in the long term, use a better regression algorithm.
	36	*/
	37	typedef long double FP;
	38
	39
	40	struct datapoint
	41	{
	42	double rate;
	43	double distortion;
	44	};
	45
	46	struct BjoentegaardParams
	47	{
	48	// alog^3 R + blog^2 R + c*log R + d
	49	double a,b,c,d;
	50
	51	double minRate, maxRate;
	52	};
	53
	54	std::vector<datapoint> curveA,curveB;
	55	BjoentegaardParams paramsA,paramsB;
	56
	57	#define RATE_NORMALIZATION_FACTOR 1 //(1/1000.0)
	58
	59
	60
	61	FP invf(int i,int j,const FP* m)
	62	{
	63	int o = 2+(j-i);
	64
	65	i += 4+o;
	66	j += 4-o;
	67
	68	#define e(a,b) m[ ((j+b)%4)*4 + ((i+a)%4) ]
	69
	70	FP inv =
	71	+ e(+1,-1)e(+0,+0)e(-1,+1)
	72	+ e(+1,+1)e(+0,-1)e(-1,+0)
	73	+ e(-1,-1)e(+1,+0)e(+0,+1)
	74	- e(-1,-1)e(+0,+0)e(+1,+1)
	75	- e(-1,+1)e(+0,-1)e(+1,+0)
	76	- e(+1,-1)e(-1,+0)e(+0,+1);
	77
	78	return (o%2)?inv : -inv;
	79
	80	#undef e
	81
	82	}
	83
	84	bool inverseMatrix4x4(const FP m, FP out)
	85	{
	86	FP inv[16];
	87
	88	for(int i=0;i<4;i++)
	89	for(int j=0;j<4;j++)
	90	inv[j*4+i] = invf(i,j,m);
	91
	92	FP D = 0;
	93
	94	for(int k=0;k<4;k++) D += m[k] * inv[k*4];
	95
	96	if (D == 0) return false;
	97
	98	D = 1.0 / D;
	99
	100	for (int i = 0; i < 16; i++)
	101	out[i] = inv[i] * D;
	102
	103	return true;
	104
	105	}
	106
	107
	108
	109	BjoentegaardParams fitParams(const std::vector<datapoint>& curve)
	110	{
	111	// build regression matrix
	112
	113	int n = curve.size();
	114
	115	FP X[4*n]; // regression matrix
	116	FP XT[n*4]; // transpose of X
	117
	118	for (int i=0;i<n;i++) {
	119	FP x = log(curve[i].rate) * RATE_NORMALIZATION_FACTOR;
	120
	121	X[4*i + 0] = 1;
	122	X[4*i + 1] = x;
	123	X[4i + 2] = xx;
	124	X[4i + 3] = xx*x;
	125
	126	if (D) printf("%f %f %f %f ;\n",1.0,(double)x,(double)(xx),(double)(xx*x));
	127
	128	XT[i+0*n] = 1;
	129	XT[i+1*n] = x;
	130	XT[i+2n] = xx;
	131	XT[i+3n] = xx*x;
	132	}
	133
	134	if (D) {
	135	printf("rate: ");
	136	for (int i=0;i<n;i++) {
	137	printf("%f ; ",curve[i].rate);
	138	}
	139	printf("\n");
	140
	141	printf("distortion: ");
	142	for (int i=0;i<n;i++) {
	143	printf("%f ; ",curve[i].distortion);
	144	}
	145	printf("\n");
	146	}
	147
	148	// calc X^T * X
	149
	150	FP XTX[4*4];
	151	for (int y=0;y<4;y++)
	152	for (int x=0;x<4;x++) {
	153	FP sum=0;
	154
	155	for (int i=0;i<n;i++)
	156	{
	157	sum += XT[yn + i] X[x + i*4];
	158	}
	159
	160	XTX[y*4+x] = sum;
	161	}
	162
	163	FP XTXinv[4*4];
	164
	165	inverseMatrix4x4(XTX, XTXinv);
	166
	167	if (D) {
	168	for (int y=0;y<4;y++) {
	169	for (int x=0;x<4;x++) {
	170	printf("%f ",(double)XTXinv[y*4+x]);
	171	}
	172	printf("\n");
	173	}
	174	}
	175
	176	// calculate pseudo-inverse XP = (X^T * X)^-1 * X^T
	177
	178	FP XP[n*4];
	179
	180	for (int y=0;y<4;y++) {
	181	for (int x=0;x<n;x++) {
	182	FP sum=0;
	183
	184	for (int i=0;i<4;i++)
	185	{
	186	sum += XTXinv[y4 + i] XT[x + i*n];
	187	}
	188
	189	XP[y*n+x] = sum;
	190	}
	191	}
	192
	193	// calculate regression parameters
	194
	195	FP p[4];
	196
	197	for (int k=0;k<4;k++)
	198	{
	199	FP sum=0;
	200
	201	for (int i=0;i<n;i++) {
	202	sum += XP[kn + i] curve[i].distortion;
	203	}
	204
	205	p[k]=sum;
	206	}
	207
	208
	209	BjoentegaardParams param;
	210	param.d = p[0];
	211	param.c = p[1];
	212	param.b = p[2];
	213	param.a = p[3];
	214
	215
	216	// find min and max rate
	217
	218	param.minRate = curve[0].rate;
	219	param.maxRate = curve[0].rate;
	220
	221	for (int i=1;i<n;i++) {
	222	param.minRate = std::min(param.minRate, curve[i].rate);
	223	param.maxRate = std::max(param.maxRate, curve[i].rate);
	224	}
	225
	226	return param;
	227	}
	228
	229	FP evalIntegralAt(const BjoentegaardParams& p, double x)
	230	{
	231	FP sum = 0;
	232
	233	// integral of: d
	234
	235	sum += p.d * x;
	236
	237	// integral of: c*log(x)
	238
	239	sum += p.c * x* (log(x)-1);
	240
	241	// integral of: b*log(x)^2
	242
	243	sum += p.b * x * ((log(x)-2)*log(x)+2);
	244
	245	// integral of: a*log(x)^3
	246
	247	sum += p.a * x * (log(x)((log(x)-3)log(x)+6)-6);
	248
	249	return sum;
	250	}
	251
	252
	253	double calcBjoentegaard(const BjoentegaardParams& paramsA,
	254	const BjoentegaardParams& paramsB,
	255	double min_rate, double max_rate)
	256	{
	257	double mini = std::max(paramsA.minRate, paramsB.minRate);
	258	double maxi = std::min(paramsA.maxRate, paramsB.maxRate);
	259
	260	if (min_rate >= 0) mini = std::max(mini, min_rate);
	261	if (max_rate >= 0) maxi = std::min(maxi, max_rate);
	262
	263	if (D) printf("range: %f %f\n",mini,maxi);
	264
	265	FP intA = evalIntegralAt(paramsA, maxi) - evalIntegralAt(paramsA, mini);
	266	FP intB = evalIntegralAt(paramsB, maxi) - evalIntegralAt(paramsB, mini);
	267
	268	if (D) printf("int1:%f int2:%f\n",(double)intA,(double)intB);
	269
	270	return (intA-intB)/(maxi-mini);
	271	}
	272
	273
	274	std::vector<datapoint> readRDFile(const char* filename, float min_rate, float max_rate)
	275	{
	276	std::vector<datapoint> curve;
	277	std::ifstream istr(filename);
	278
	279	for (;;)
	280	{
	281	std::string line;
	282	getline(istr, line);
	283	if (istr.eof())
	284	break;
	285
	286	if (line[0]=='#') continue;
	287
	288	std::stringstream sstr(line);
	289	datapoint p;
	290	sstr >> p.rate >> p.distortion;
	291
	292	if (min_rate>=0 && p.rate < min_rate) continue;
	293	if (max_rate>=0 && p.rate > max_rate) continue;
	294
	295	curve.push_back(p);
	296	}
	297
	298	return curve;
	299	}
	300
	301
	302	int main(int argc, char** argv)
	303	{
	304	float min_rate = -1;
	305	float max_rate = -1;
	306
	307	int c;
	308	while ((c=getopt(argc,argv, "l:h:")) != -1) {
	309	switch (c) {
	310	case 'l': min_rate = atof(optarg); break;
	311	case 'h': max_rate = atof(optarg); break;
	312	}
	313	}
	314
	315	curveA = readRDFile(argv[optind], min_rate, max_rate);
	316	paramsA = fitParams(curveA);
	317
	318	printf("params A: %f %f %f %f\n",paramsA.a,paramsA.b,paramsA.c,paramsA.d);
	319
	320	printf("gnuplot: %flog(x)3+%flog(x)*2+%flog(x)+%f\n",paramsA.a,paramsA.b,paramsA.c,paramsA.d);
	321
	322	if (optind+1<argc) {
	323	curveB = readRDFile(argv[optind+1], min_rate, max_rate);
	324	paramsB = fitParams(curveB);
	325
	326	printf("params B: %f %f %f %f\n",paramsB.a,paramsB.b,paramsB.c,paramsB.d);
	327
	328	printf("gnuplot: %flog(x)3+%flog(x)*2+%flog(x)+%f\n",paramsB.a,paramsB.b,paramsB.c,paramsB.d);
	329
	330	double delta = calcBjoentegaard(paramsA,paramsB, min_rate,max_rate);
	331
	332	printf("Bjoentegaard delta: %f dB (A-B -> >0 -> first (A) is better)\n",delta);
	333
	334	if (delta>=0) {
	335	printf("-> first is better by %f dB\n",delta);
	336	}
	337	else {
	338	printf("-> second is better by %f dB\n",-delta);
	339	}
	340	}
	341
	342	return 0;
	343	}

+127

-0

tools/block-rate-estim.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include <vector>
	21	#include <string>
	22	#include <fstream>
	23	#include <math.h>
	24
	25
	26	struct datapoint {
	27	int log2blksize;
	28	float rate;
	29	float estim;
	30	};
	31
	32	std::vector<datapoint> pts;
	33
	34	#define NBINS 100
	35
	36	/*
	37	#define ESTIMDIV 100
	38	#define MAXESTIM 80000
	39	std::vector<float> bitestim2[MAXESTIM/ESTIMDIV];
	40	*/
	41
	42
	43
	44	void print_bitestim_results(int log2blksize)
	45	{
	46	float max_estim=0;
	47
	48	for (int i=0;i<pts.size();i++) {
	49	if (log2blksize==0 \|\| pts[i].log2blksize==log2blksize) {
	50	max_estim = std::max(max_estim, pts[i].estim);
	51	}
	52	}
	53
	54
	55
	56	float epsilon = 0.0001;
	57	float interval = (max_estim+epsilon) / NBINS;
	58
	59	for (int b=0;b<NBINS;b++) {
	60
	61	int cnt=0;
	62	double sum=0;
	63	float mini=999999;
	64	float maxi=0;
	65
	66	for (int i=0;i<pts.size();i++)
	67	if (log2blksize==0 \|\| pts[i].log2blksize==log2blksize) {
	68	int bin = pts[i].estim/interval;
	69	if (bin==b) {
	70	sum += pts[i].rate;
	71
	72	mini = std::min(mini,pts[i].rate);
	73	maxi = std::max(maxi,pts[i].rate);
	74	cnt++;
	75	}
	76	}
	77
	78	if (cnt>0) {
	79	double mean = sum/cnt;
	80
	81	double var = 0;
	82
	83	for (int i=0;i<pts.size();i++)
	84	if (log2blksize==0 \|\| pts[i].log2blksize==log2blksize) {
	85	int bin = pts[i].estim/interval;
	86	if (bin==b) {
	87	var += (pts[i].rate-mean)*(pts[i].rate-mean);
	88	}
	89	}
	90
	91	var /= cnt;
	92	double stddev = sqrt(var);
	93
	94	printf("%f %f %f %f %f %f %f %d\n",
	95	(b+0.5)*interval,mean,var,
	96	mean-stddev,mean+stddev, mini,maxi,
	97	cnt);
	98	}
	99	}
	100	}
	101
	102
	103	int main(int argc,char** argv)
	104	{
	105	std::string tag = argv[1];
	106
	107	std::ifstream istr(argv[2]);
	108	for (;;)
	109	{
	110	std::string t;
	111	int log2blksize;
	112	datapoint pt;
	113
	114	istr >> t >> pt.log2blksize >> pt.rate >> pt.estim;
	115
	116	if (istr.eof()) break;
	117
	118	if (t == tag) {
	119	pts.push_back(pt);
	120	}
	121	}
	122
	123	print_bitestim_results(0);
	124
	125	return 0;
	126	}

+477

-0

tools/gen-entropy-table.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20
	21	#include "libde265/cabac.h"
	22	#include <assert.h>
	23	#include <time.h>
	24	#include <stdlib.h>
	25	#include <stdio.h>
	26
	27
	28	void simple_getline(char** lineptr,size_t* linelen,FILE* fh)
	29	{
	30	const int LINESIZE=1000;
	31
	32	if (*lineptr==NULL) {
	33	*linelen = LINESIZE;
	34	lineptr = (char)malloc(LINESIZE);
	35	}
	36
	37	char* p = *lineptr;
	38
	39	for (;;) {
	40	assert(p - *lineptr < LINESIZE);
	41
	42	int c = fgetc(fh);
	43	if (c == EOF \|\| c == '\n') {
	44	*p = 0;
	45	break;
	46	}
	47
	48	*p++ = c;
	49	}
	50	}
	51
	52
	53	void generate_entropy_table()
	54	{
	55	#if 000
	56	const int nSymbols=1000100010;
	57	const int oversample = 10;
	58
	59	double tab[64][2];
	60
	61	for (int i=0;i<64;i++)
	62	for (int k=0;k<2;k++)
	63	tab[i][k]=0;
	64
	65	srand(time(0));
	66	//srand(123);
	67
	68	int cnt=1;
	69	for (;;cnt++) {
	70	printf("-------------------- %d --------------------\n",cnt);
	71
	72	for (int s=0;s<63;s++) {
	73	CABAC_encoder_bitstream cabac_mix0;
	74	CABAC_encoder_bitstream cabac_mix1;
	75	CABAC_encoder_bitstream cabac_ref;
	76
	77	for (int i=0;i<nSymbols*oversample;i++) {
	78	int r = rand();
	79	int n = (r>>2) % 63;
	80	int m = (r>>1) & 1;
	81	int b = r & 1;
	82
	83	context_model model;
	84	model.MPSbit = 1;
	85	model.state = n;
	86	cabac_ref.write_CABAC_bit(&model, b);
	87
	88	model.MPSbit = 1;
	89	model.state = n;
	90	cabac_mix0.write_CABAC_bit(&model, b);
	91
	92	model.MPSbit = 1;
	93	model.state = n;
	94	cabac_mix1.write_CABAC_bit(&model, b);
	95
	96	if (i%oversample == oversample/2) {
	97	model.MPSbit = 1;
	98	model.state = s;
	99	cabac_mix0.write_CABAC_bit(&model, 0);
	100
	101	model.MPSbit = 1;
	102	model.state = s;
	103	cabac_mix1.write_CABAC_bit(&model, 1);
	104
	105	//b = rand() & 1;
	106	//cabac_mix.write_CABAC_bypass(1);
	107	}
	108
	109	}
	110
	111	cabac_ref.flush_CABAC();
	112	cabac_mix0.flush_CABAC();
	113	cabac_mix1.flush_CABAC();
	114
	115	int bits_ref = cabac_ref.size()*8;
	116	int bits_mix0 = cabac_mix0.size()*8;
	117	int bits_mix1 = cabac_mix1.size()*8;
	118
	119	//printf("bits: %d %d\n",bits_ref,bits_mix);
	120	int bits_diff0 = bits_mix0-bits_ref;
	121	int bits_diff1 = bits_mix1-bits_ref;
	122	//printf("bits diff: %d\n",bits_diff);
	123
	124	double bits_per_symbol0 = bits_diff0 / double(nSymbols);
	125	double bits_per_symbol1 = bits_diff1 / double(nSymbols);
	126
	127	tab[s][0] += bits_per_symbol0;
	128	tab[s][1] += bits_per_symbol1;
	129
	130	double bps0 = tab[s][0]/cnt;
	131	double bps1 = tab[s][1]/cnt;
	132
	133	printf("/* state=%2d / 0x%05x / %f /, 0x%05x / %f */,\n", s,
	134	(int)(bps1*0x8000), bps1,
	135	(int)(bps0*0x8000), bps0);
	136	}
	137
	138	printf(" 0x0010c, 0x3bfbb /* dummy, should never be used */\n");
	139	}
	140	#endif
	141	}
	142
	143	int probTab[128+2] = {
	144	1537234,1602970,
	145	1608644,1815493,
	146	1822246,2245961,
	147	916773,1329391,
	148	1337504,1930659,
	149	1063692,1707588,
	150	868294,1532108,
	151	842934,1555538,
	152	689043,1396941,
	153	860184,1789964,
	154	534165,1258482,
	155	672508,1598821,
	156	578782,1476240,
	157	602247,1613140,
	158	409393,1206638,
	159	459294,1356779,
	160	430124,1359893,
	161	308326,1050647,
	162	313100,1099956,
	163	293887,1088978,
	164	220901,869582,
	165	214967,881695,
	166	197226,856990,
	167	166131,767761,
	168	152514,737406,
	169	128332,663998,
	170	117638,632653,
	171	106178,595539,
	172	90898,539506,
	173	83437,509231,
	174	76511,492801,
	175	64915,443096,
	176	57847,409809,
	177	52730,385395,
	178	45707,354059,
	179	42018,333028,
	180	37086,308073,
	181	33256,284497,
	182	36130,299172,
	183	28831,270716,
	184	25365,244840,
	185	22850,221896,
	186	19732,201462,
	187	17268,183729,
	188	15252,168106,
	189	13787,153979,
	190	12187,141455,
	191	10821,130337,
	192	9896,120165,
	193	8626,112273,
	194	8162,103886,
	195	7201,96441,
	196	6413,89805,
	197	5886,83733,
	198	5447,78084,
	199	4568,73356,
	200	4388,68831,
	201	3959,64688,
	202	3750,60804,
	203	3407,57271,
	204	3109,54024,
	205	2820,51099,
	206	48569,1451987,
	207	0, 0,
	208	0*22686225, 0
	209	};
	210
	211
	212	void generate_entropy_table_probTableWeighted()
	213	{
	214	#if 000
	215	int64_t probTabSum=0;
	216	for (int i=0;i<130;i++)
	217	probTabSum += probTab[i];
	218
	219
	220	const int nSymbols=1000100010;
	221	const int oversample = 10;
	222
	223	double tab[64][2];
	224
	225	for (int i=0;i<64;i++)
	226	for (int k=0;k<2;k++)
	227	tab[i][k]=0;
	228
	229	srand(time(0));
	230	//srand(123);
	231
	232	int cnt=1;
	233	for (;;cnt++) {
	234	printf("-------------------- %d --------------------\n",cnt);
	235
	236	for (int s=0;s<63;s++) {
	237	CABAC_encoder_bitstream cabac_mix0;
	238	CABAC_encoder_bitstream cabac_mix1;
	239	CABAC_encoder_bitstream cabac_ref;
	240
	241	for (int i=0;i<nSymbols*oversample;i++) {
	242	int r = rand();
	243
	244	r %= probTabSum;
	245	int idx=0;
	246	while (r>probTab[idx]) {
	247	r-=probTab[idx];
	248	idx++;
	249	}
	250
	251	assert(idx<=128);
	252
	253	int n = idx/2;
	254	int b = idx&1;
	255	bool bypass = (idx==128);
	256
	257	printf("%d %d %d\n",n,b,bypass);
	258
	259	context_model model;
	260	model.MPSbit = 1;
	261	model.state = n;
	262	if (bypass) cabac_ref.write_CABAC_bypass(1);
	263	else cabac_ref.write_CABAC_bit(&model, b);
	264
	265	model.MPSbit = 1;
	266	model.state = n;
	267	if (bypass) cabac_mix0.write_CABAC_bypass(1);
	268	else cabac_mix0.write_CABAC_bit(&model, b);
	269
	270	model.MPSbit = 1;
	271	model.state = n;
	272	if (bypass) cabac_mix1.write_CABAC_bypass(1);
	273	else cabac_mix1.write_CABAC_bit(&model, b);
	274
	275	if (i%oversample == oversample/2) {
	276	model.MPSbit = 1;
	277	model.state = s;
	278	cabac_mix0.write_CABAC_bit(&model, 0);
	279
	280	model.MPSbit = 1;
	281	model.state = s;
	282	cabac_mix1.write_CABAC_bit(&model, 1);
	283
	284	//b = rand() & 1;
	285	//cabac_mix.write_CABAC_bypass(1);
	286	}
	287
	288	}
	289
	290	cabac_ref.flush_CABAC();
	291	cabac_mix0.flush_CABAC();
	292	cabac_mix1.flush_CABAC();
	293
	294	int bits_ref = cabac_ref.size()*8;
	295	int bits_mix0 = cabac_mix0.size()*8;
	296	int bits_mix1 = cabac_mix1.size()*8;
	297
	298	//printf("bits: %d %d\n",bits_ref,bits_mix);
	299	int bits_diff0 = bits_mix0-bits_ref;
	300	int bits_diff1 = bits_mix1-bits_ref;
	301	//printf("bits diff: %d\n",bits_diff);
	302
	303	double bits_per_symbol0 = bits_diff0 / double(nSymbols);
	304	double bits_per_symbol1 = bits_diff1 / double(nSymbols);
	305
	306	tab[s][0] += bits_per_symbol0;
	307	tab[s][1] += bits_per_symbol1;
	308
	309	double bps0 = tab[s][0]/cnt;
	310	double bps1 = tab[s][1]/cnt;
	311
	312	printf("/* state=%2d / 0x%05x / %f /, 0x%05x / %f */,\n", s,
	313	(int)(bps1*0x8000), bps1,
	314	(int)(bps0*0x8000), bps0);
	315	}
	316
	317	printf(" 0x0010c, 0x3bfbb /* dummy, should never be used */\n");
	318	}
	319	#endif
	320	}
	321
	322
	323	void generate_entropy_table_replay()
	324	{
	325	#if 000
	326	const int oversample = 10;
	327
	328	char* lineptr = NULL;
	329	size_t linelen = 0;
	330
	331	for (int s=0;s<63;s++) {
	332	CABAC_encoder_bitstream cabac_mix0;
	333	CABAC_encoder_bitstream cabac_mix1;
	334	CABAC_encoder_bitstream cabac_ref;
	335
	336	int nSymbols = 0;
	337
	338	FILE* fh = fopen("streamdump-paris-intra","r");
	339
	340	for (int i=0;i<80000000;i++) {
	341	simple_getline(&lineptr,&linelen,fh);
	342	if (feof(fh))
	343	break;
	344
	345	int n,b;
	346	sscanf(lineptr,"%d %d",&n,&b);
	347
	348	bool bypass = (n==64);
	349
	350	if ((i%10000)==0)
	351	{ printf("%d %d %d \r",i,n,b);
	352	}
	353
	354	//printf("%d %d %d\n",n,b,bypass);
	355
	356	context_model model;
	357	model.MPSbit = 1;
	358	model.state = n;
	359	if (bypass) cabac_ref.write_CABAC_bypass(1);
	360	else cabac_ref.write_CABAC_bit(&model, b);
	361
	362	model.MPSbit = 1;
	363	model.state = n;
	364	if (bypass) cabac_mix0.write_CABAC_bypass(1);
	365	else cabac_mix0.write_CABAC_bit(&model, b);
	366
	367	model.MPSbit = 1;
	368	model.state = n;
	369	if (bypass) cabac_mix1.write_CABAC_bypass(1);
	370	else cabac_mix1.write_CABAC_bit(&model, b);
	371
	372	if (i%oversample == oversample/2) {
	373	model.MPSbit = 1;
	374	model.state = s;
	375	cabac_mix0.write_CABAC_bit(&model, 0);
	376
	377	model.MPSbit = 1;
	378	model.state = s;
	379	cabac_mix1.write_CABAC_bit(&model, 1);
	380
	381	nSymbols++;
	382
	383	//b = rand() & 1;
	384	//cabac_mix.write_CABAC_bypass(1);
	385	}
	386	}
	387
	388	fclose(fh);
	389
	390	cabac_ref.flush_CABAC();
	391	cabac_mix0.flush_CABAC();
	392	cabac_mix1.flush_CABAC();
	393
	394	int bits_ref = cabac_ref.size()*8;
	395	int bits_mix0 = cabac_mix0.size()*8;
	396	int bits_mix1 = cabac_mix1.size()*8;
	397
	398	//printf("bits: %d %d\n",bits_ref,bits_mix);
	399	int bits_diff0 = bits_mix0-bits_ref;
	400	int bits_diff1 = bits_mix1-bits_ref;
	401	//printf("bits diff: %d\n",bits_diff);
	402
	403	double bits_per_symbol0 = bits_diff0 / double(nSymbols);
	404	double bits_per_symbol1 = bits_diff1 / double(nSymbols);
	405
	406	double bps0 = bits_per_symbol0;
	407	double bps1 = bits_per_symbol1;
	408
	409	printf("/* state=%2d / 0x%05x / %f /, 0x%05x / %f */,\n", s,
	410	(int)(bps1*0x8000), bps1,
	411	(int)(bps0*0x8000), bps0);
	412	}
	413
	414	printf(" 0x0010c, 0x3bfbb /* dummy, should never be used */\n");
	415	#endif
	416	}
	417
	418
	419	void test_entropy_table_replay()
	420	{
	421	#if 000
	422	char* lineptr = NULL;
	423	size_t linelen = 0;
	424
	425
	426	CABAC_encoder_bitstream cabac_bs;
	427	CABAC_encoder_estim cabac_estim;
	428
	429	//FILE* fh = fopen("y","r");
	430	//FILE* fh = fopen("own-dump","r");
	431	//FILE* fh = fopen("rawstream-dump","r");
	432	//FILE* fh = fopen("johnny-stream-dump","r");
	433	FILE* fh = fopen("streamdump-paris-intra","r");
	434
	435	for (int i=0;i<80000000;i++) {
	436	simple_getline(&lineptr,&linelen,fh);
	437	if (feof(fh))
	438	break;
	439
	440	int n,b;
	441	sscanf(lineptr,"%d %d",&n,&b);
	442	b=!b;
	443	bool bypass = (n==64);
	444
	445	if ((i%10000)==0)
	446	{ printf("%d %d %d \n",i,n,b);
	447	}
	448
	449	context_model model;
	450	model.MPSbit = 1;
	451	model.state = n;
	452	if (bypass) cabac_bs.write_CABAC_bypass(1);
	453	else cabac_bs.write_CABAC_bit(&model, b);
	454
	455	model.MPSbit = 1;
	456	model.state = n;
	457	if (bypass) cabac_estim.write_CABAC_bypass(1);
	458	else cabac_estim.write_CABAC_bit(&model, b);
	459	}
	460
	461	fclose(fh);
	462
	463	printf("bs:%d estim:%d\n",cabac_bs.size(),cabac_estim.size());
	464	#endif
	465	}
	466
	467
	468	int main(int argc, char** argv)
	469	{
	470	//generate_entropy_table();
	471	//generate_entropy_table_replay();
	472
	473	test_entropy_table_replay();
	474
	475	return 0;
	476	}

+1075

-0

tools/rd-curves.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifdef HAVE_CONFIG_H
	21	#include "config.h"
	22	#endif
	23
	24	#include <stdio.h>
	25	#include <stdlib.h>
	26	#include <sys/types.h>
	27	#include <sys/stat.h>
	28	#include <unistd.h>
	29	#include <assert.h>
	30	#include <stdint.h>
	31	#include <string.h>
	32	#include <getopt.h>
	33
	34	#include <sys/time.h>
	35
	36	#ifndef WIN32
	37	#include <sys/times.h>
	38	#endif
	39
	40	#include <string>
	41	#include <sstream>
	42	#include <iostream>
	43	#include <iomanip>
	44	#include <fstream>
	45	#include <vector>
	46
	47
	48
	49	static struct {
	50	const char* name;
	51	const char* value;
	52	} variables[] = {
	53	{ "$HOME" , "/home/domain/farindk" },
	54	{ "$ROOT" , "/home/domain/farindk/prog/h265" },
	55	{ "$ENC265" , "$ROOT/libde265/enc265/enc265" },
	56	{ "$DEC265" , "$ROOT/libde265/dec265/dec265" },
	57	{ "$YUVDIST" , "$ROOT/libde265/tools/yuv-distortion" },
	58	{ "$YUVTMP" , "/mnt/temp/dirk/yuv/ftp.tnt.uni-hannover.de/testsequences" },
	59	{ "$YUV" , "/storage/users/farindk/yuv" },
	60	{ "$HMENC" , "HM13enc" },
	61	{ "$HM13CFG" , "$ROOT/HM/HM-13.0-dev/cfg" },
	62	{ "$HMSCCENC", "HM-SCC-enc" },
	63	{ "$HMSCCCFG", "$ROOT/HM/HM-SCC-extensions/cfg" },
	64	{ "$X265ENC" , "$ROOT/x265/build/linux/x265" },
	65	{ "$X264" , "x264" },
	66	{ "$FFMPEG" , "ffmpeg" },
	67	{ "$F265" , "$ROOT/f265/build/f265cli" },
	68	{ 0,0 }
	69	};
	70
	71
	72	bool keepStreams = false;
	73	int maxFrames = 0;
	74	std::string encoderParameters;
	75
	76
	77	std::string replace_variables(std::string str)
	78	{
	79	bool replaced = false;
	80	for (int i=0;variables[i].name;i++) {
	81	size_t pos = str.find(variables[i].name);
	82	if (pos != std::string::npos) {
	83	replaced = true;
	84	str = str.replace(pos, strlen(variables[i].name), variables[i].value);
	85	break;
	86	}
	87	}
	88
	89	if (!replaced) return str;
	90	else return replace_variables(str);
	91	}
	92
	93
	94	// ---------------------------------------------------------------------------
	95
	96	struct Preset
	97	{
	98	const int ID;
	99	const char* name;
	100	const char* descr;
	101
	102	const char* options_de265;
	103	const char* options_hm;
	104	const char* options_hm_scc;
	105	const char* options_x265;
	106	const char* options_f265;
	107	const char* options_x264;
	108	const char* options_x264_ffmpeg;
	109	const char* options_ffmpeg_mpeg2;
	110
	111	//int nFrames;
	112	};
	113
	114
	115	Preset preset[] = {
	116	{ 1, "pre01-intra-noLF", "intra, no LF, no SBH, CTB-size 32, min CB=8",
	117	/* de265 */ "--sop-structure intra",
	118	/* HM */ "-c $HM13CFG/encoder_intra_main.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	119	/* HM SCC */ "-c $HMSCCCFG/encoder_intra_main_scc.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	120	/* x265 */ "--no-lft -I 1 --no-signhide",
	121	/* f265 */ "key-frame-spacing=1",
	122	/* x264 */ "-I 1",
	123	/* ffmpeg */ "-g 1",
	124	/* mpeg-2 */ "-g 1"
	125	// 0 // all frames
	126	},
	127
	128	{ 2, "pre02-fastIntra", "intra, no LF, no SBH, CTB-size 32, min CB=8",
	129	/* de265 */ "--sop-structure intra --TB-IntraPredMode minSSD",
	130	/* HM */ "-c $HM13CFG/encoder_intra_main.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	131	/* HM SCC */ "-c $HMSCCCFG/encoder_intra_main_scc.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	132	/* x265 */ "--no-lft -I 1 --no-signhide",
	133	/* f265 */ "key-frame-spacing=1",
	134	/* x264 */ "-I 1",
	135	/* ffmpeg */ "-g 1",
	136	/* mpeg-2 */ "-g 1"
	137	// 0 // all frames
	138	},
	139
	140	{ 3, "pre03-fastIntra", "pre02, but fast-brute",
	141	/* de265 */ "--sop-structure intra --TB-IntraPredMode fast-brute",
	142	/* HM */ "-c $HM13CFG/encoder_intra_main.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	143	/* HM SCC */ "-c $HMSCCCFG/encoder_intra_main_scc.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	144	/* x265 */ "--no-lft -I 1 --no-signhide",
	145	/* f265 */ "key-frame-spacing=1",
	146	/* x264 */ "-I 1",
	147	/* ffmpeg */ "-g 1",
	148	/* mpeg-2 */ "-g 1"
	149	// 0 // all frames
	150	},
	151
	152	{ 50, "cb-auto16", "(development test)",
	153	/* de265 */ "--max-cb-size 16 --min-cb-size 8",
	154	/* HM */ "-c $HM13CFG/encoder_intra_main.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	155	/* HM SCC */ "-c $HMSCCCFG/encoder_intra_main_scc.cfg -SBH 0 --SAO=0 --LoopFilterDisable --DeblockingFilterControlPresent --MaxCUSize=32 --MaxPartitionDepth=2",
	156	/* x265 */ "--no-lft -I 1 --no-signhide",
	157	/* f265 */ "key-frame-spacing=1",
	158	/* x264 */ "-I 1",
	159	/* ffmpeg */ "-g 1",
	160	/* mpeg-2 */ "-g 1"
	161	// 0 // all frames
	162	},
	163
	164	{ 80, "lowdelay", "default (low-default) encoder parameters",
	165	"--MEMode search --max-cb-size 32 --min-cb-size 8 --min-tb-size 4 --CB-IntraPartMode-Fixed-partMode 2Nx2N --CB-IntraPartMode fixed --TB-IntraPredMode min-residual --PB-MV-TestMode zero",
	166	/* de265 */ //"--sop-structure low-delay --MEMode search --max-cb-size 32 --min-cb-size 8 --min-tb-size 4 --CB-IntraPartMode fixed --TB-IntraPredMode min-residual",
	167	/* HM */ "-c $HM13CFG/encoder_lowdelay_main.cfg -ip 248",
	168	/* HM SCC */ "-c $HMSCCCFG/encoder_lowdelay_main_scc.cfg -ip 248",
	169	/* x265 */ "-I 248 --no-wpp --bframes 0", // GOP size: 248
	170	/* f265 */ 0, //"key-frame-spacing=248",
	171	/* x264 */ "",
	172	/* ffmpeg */ "-g 248 -bf 0",
	173	/* mpeg-2 */ "" // GOP size 248 does not make sense here
	174	// 0 // all frames
	175	},
	176
	177	{ 98, "best", "default (random-access) encoder parameters",
	178	/* de265 */ "--max-cb-size 16 --min-cb-size 8",
	179	/* HM */ "-c $HM13CFG/encoder_randomaccess_main.cfg",
	180	/* HM SCC */ "-c $HMSCCCFG/encoder_randomaccess_main_scc.cfg",
	181	/* x265 */ "",
	182	/* f265 */ "",
	183	/* x264 */ "",
	184	/* ffmpeg */ "",
	185	/* mpeg-2 */ ""
	186	// 0 // all frames
	187	},
	188
	189	{ 99, "besteq", "default (random-access) encoder parameters, I-frame distance = 248",
	190	/* de265 */ "",
	191	/* HM */ "-c $HM13CFG/encoder_randomaccess_main.cfg -ip 248",
	192	/* HM SCC */ "-c $HMSCCCFG/encoder_randomaccess_main_scc.cfg -ip 248",
	193	/* x265 */ "-I 248 --no-wpp", // GOP size: 248
	194	/* f265 */ "key-frame-spacing=248",
	195	/* x264 */ "",
	196	/* ffmpeg */ "-g 248",
	197	/* mpeg-2 */ "" // GOP size 248 does not make sense here
	198	// 0 // all frames
	199	},
	200
	201	{ 0, NULL }
	202	};
	203
	204	// ---------------------------------------------------------------------------
	205
	206	class Input
	207	{
	208	public:
	209	Input() {
	210	width=height=0;
	211	maxFrames=0;
	212	}
	213
	214	void setInput(const char* yuvfilename,int w,int h, float fps) {
	215	mInputFilename = yuvfilename;
	216	width = w;
	217	height = h;
	218	mFPS = fps;
	219	}
	220
	221	void setMaxFrames(int n) { maxFrames=n; }
	222
	223	std::string options_de265() const {
	224	std::stringstream sstr;
	225	sstr << " -i " << mInputFilename << " --width " << width << " --height " << height;
	226	if (maxFrames) sstr << " --frames " << maxFrames;
	227
	228	return sstr.str();
	229	}
	230
	231	std::string options_HM() const {
	232	std::stringstream sstr;
	233	sstr << "-i " << mInputFilename << " -wdt " << width << " -hgt " << height
	234	<< " -fr " << mFPS;
	235	if (maxFrames) sstr << " -f " << maxFrames;
	236
	237	return sstr.str();
	238	}
	239
	240	std::string options_x265() const {
	241	std::stringstream sstr;
	242	sstr << mInputFilename << " --input-res " << width << "x" << height
	243	<< " --fps " << mFPS;
	244	if (maxFrames) sstr << " -f " << maxFrames;
	245
	246	return sstr.str();
	247	}
	248
	249	std::string options_x264() const {
	250	std::stringstream sstr;
	251	sstr << mInputFilename << " --input-res " << width << "x" << height;
	252	sstr << " --fps 25"; // TODO: check why crf/qp rate-control freaks out when fps is != 25
	253	if (maxFrames) sstr << " --frames " << maxFrames;
	254
	255	return sstr.str();
	256	}
	257
	258	std::string options_ffmpeg() const {
	259	std::stringstream sstr;
	260	sstr << "-f rawvideo -vcodec rawvideo -s " << width << "x" << height; // << " -r " << mFPS
	261	sstr << " -pix_fmt yuv420p -i " << mInputFilename;
	262	if (maxFrames) sstr << " -vframes " << maxFrames;
	263
	264	return sstr.str();
	265	}
	266
	267	std::string options_f265() const {
	268	std::stringstream sstr;
	269	sstr << mInputFilename << " -w " << width << ":" << height;
	270	if (maxFrames) sstr << " -c " << maxFrames;
	271
	272	return sstr.str();
	273	}
	274
	275	std::string getFilename() const { return mInputFilename; }
	276	float getFPS() const { return mFPS; }
	277	int getNFrames() const { return maxFrames; }
	278	int getWidth() const { return width; }
	279	int getHeight() const { return height; }
	280
	281	private:
	282	std::string mInputFilename;
	283	int width, height;
	284	int maxFrames;
	285	float mFPS;
	286	};
	287
	288	Input input;
	289
	290	struct InputSpec
	291	{
	292	const char* name;
	293	const char* filename;
	294	int width,height, nFrames;
	295	float fps;
	296	} inputSpec[] = {
	297	{ "paris", "$YUV/paris_cif.yuv",352,288,1065, 30.0 },
	298	{ "paris10", "$YUV/paris_cif.yuv",352,288, 10, 30.0 },
	299	{ "paris100", "$YUV/paris_cif.yuv",352,288, 100, 30.0 },
	300	{ "johnny", "$YUV/Johnny_1280x720_60.yuv",1280,720,600,60.0 },
	301	{ "johnny10", "$YUV/Johnny_1280x720_60.yuv",1280,720, 10,60.0 },
	302	{ "johnny100", "$YUV/Johnny_1280x720_60.yuv",1280,720,100,60.0 },
	303	{ "cactus", "$YUV/Cactus_1920x1080_50.yuv",1920,1080,500,50.0 },
	304	{ "cactus10", "$YUV/Cactus_1920x1080_50.yuv",1920,1080, 10,50.0 },
	305	{ "4people", "$YUVTMP/FourPeople_1280x720_60.yuv",1280,720,600,60.0 },
	306	{ "4people100", "$YUVTMP/FourPeople_1280x720_60.yuv",1280,720,100,60.0 },
	307	{ "slideedit", "$YUVTMP/SlideEditing_1280x720_30.yuv",1280,720,300,30.0 },
	308	{ "slideedit100","$YUVTMP/SlideEditing_1280x720_30.yuv",1280,720,100,30.0 },
	309	{ "slideshow", "$YUVTMP/SlideShow_1280x720_20.yuv",1280,720,500,20.0 },
	310	{ "slideshow100","$YUVTMP/SlideShow_1280x720_20.yuv",1280,720,100,20.0 },
	311	{ "screensharing","$HOME/test-screensharing-encoding/Screensharing.yuv",1360,768,4715,60.0 },
	312	{ NULL }
	313	};
	314
	315
	316	void setInput(const char* input_preset)
	317	{
	318	bool presetFound=false;
	319
	320	for (int i=0;inputSpec[i].name;i++) {
	321	if (strcmp(input_preset, inputSpec[i].name)==0) {
	322	input.setInput(inputSpec[i].filename,
	323	inputSpec[i].width,
	324	inputSpec[i].height,
	325	inputSpec[i].fps);
	326	input.setMaxFrames(inputSpec[i].nFrames);
	327	presetFound=true;
	328	break;
	329	}
	330	}
	331
	332	if (!presetFound) {
	333	fprintf(stderr,"no input preset '%s'\n",input_preset);
	334	exit(5);
	335	}
	336	}
	337
	338
	339	float bitrate(const char* filename)
	340	{
	341	struct stat s;
	342	stat(filename,&s);
	343
	344	long size = s.st_size;
	345
	346	int frames = input.getNFrames();
	347	assert(frames!=0);
	348
	349	float bitrate = size*8/(frames/input.getFPS());
	350	return bitrate;
	351	}
	352
	353
	354	// ---------------------------------------------------------------------------
	355
	356	class Quality
	357	{
	358	public:
	359	virtual ~Quality() { }
	360
	361	virtual void measure(const char* h265filename);
	362	virtual void measure_yuv(const char* yuvfilename);
	363
	364	float psnr, ssim;
	365	};
	366
	367
	368	void Quality::measure(const char* h265filename)
	369	{
	370	std::stringstream sstr;
	371	sstr << "$DEC265 " << h265filename << " -q -t6 -m " << input.getFilename() << " \| grep total "
	372	//"\| awk '{print $2}' "
	373	">/tmp/xtmp";
	374
	375	//std::cout << sstr.str() << "\n";
	376	int retval = system(replace_variables(sstr.str()).c_str());
	377
	378	std::ifstream istr;
	379	istr.open("/tmp/xtmp");
	380	std::string dummy;
	381	istr >> dummy >> psnr >> dummy >> dummy >> ssim;
	382
	383	unlink("/tmp/xtmp");
	384	}
	385
	386
	387	void Quality::measure_yuv(const char* yuvfilename)
	388	{
	389	std::stringstream sstr;
	390
	391	sstr << "$YUVDIST " << input.getFilename() << " " << yuvfilename
	392	<< " " << input.getWidth() << " " << input.getHeight()
	393	<< "\|grep total "
	394	//"\|awk '{print $2}' "
	395	">/tmp/ytmp";
	396
	397	//std::cout << sstr.str() << "\n";
	398	int retval = system(replace_variables(sstr.str()).c_str());
	399
	400	std::ifstream istr;
	401	istr.open("/tmp/ytmp");
	402	std::string dummy;
	403	istr >> dummy >> psnr >> ssim;
	404
	405	unlink("/tmp/ytmp");
	406	}
	407
	408	Quality quality;
	409
	410	// ---------------------------------------------------------------------------
	411
	412	long ticks_per_second;
	413
	414	void init_clock()
	415	{
	416	#ifndef WIN32
	417	ticks_per_second = sysconf(_SC_CLK_TCK);
	418	#endif
	419	}
	420
	421	double get_cpu_time()
	422	{
	423	#ifndef WIN32
	424	struct tms t;
	425	times(&t);
	426	return double(t.tms_cutime)/ticks_per_second;
	427	#else
	428	return 0; // not supported on windows (TODO)
	429	#endif
	430	}
	431
	432	double get_wall_time()
	433	{
	434	struct timeval tv;
	435	gettimeofday(&tv, NULL);
	436	double t = tv.tv_sec;
	437	double ut = tv.tv_usec/1000000.0f;
	438	t += ut;
	439	return t;
	440	}
	441
	442
	443	struct RDPoint
	444	{
	445	float rate;
	446	float psnr;
	447	float ssim;
	448	double cpu_time; // computation time in seconds
	449	double wall_time;
	450
	451
	452	RDPoint() { }
	453
	454	void compute_from_h265(std::string stream_name) {
	455	rate = bitrate(stream_name.c_str());
	456	quality.measure(stream_name.c_str());
	457	psnr = quality.psnr;
	458	ssim = quality.ssim;
	459	}
	460
	461	void compute_from_yuv(std::string stream_name, std::string yuv_name) {
	462	rate = bitrate(stream_name.c_str());
	463	quality.measure_yuv(yuv_name.c_str());
	464	psnr = quality.psnr;
	465	ssim = quality.ssim;
	466	}
	467
	468	void start_timer() {
	469	cpu_time = get_cpu_time();
	470	wall_time= get_wall_time();
	471	}
	472
	473	void end_timer() {
	474	cpu_time = get_cpu_time() - cpu_time;
	475	wall_time= get_wall_time()- wall_time;
	476	}
	477	};
	478
	479
	480	FILE* output_fh;
	481
	482	void write_rd_line(RDPoint p)
	483	{
	484	fprintf(output_fh,"%9.2f %6.4f %5.3f %5.4f %5.4f\n",
	485	p.rate/1024, p.psnr, p.ssim,
	486	p.cpu_time/60, p.wall_time/60);
	487	fflush(output_fh);
	488	}
	489
	490
	491
	492
	493	class Encoder
	494	{
	495	public:
	496	virtual ~Encoder() { }
	497
	498	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const = 0;
	499
	500	private:
	501	};
	502
	503
	504	class Encoder_de265 : public Encoder
	505	{
	506	public:
	507	Encoder_de265();
	508	void setQPRange(int low,int high,int step) { mQPLow=low; mQPHigh=high; mQPStep=step; }
	509
	510	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const;
	511
	512	private:
	513	RDPoint encode(const Preset& preset,int qp) const;
	514
	515	int mQPLow,mQPHigh,mQPStep;
	516	};
	517
	518
	519	Encoder_de265::Encoder_de265()
	520	{
	521	mQPLow = 14;
	522	mQPHigh= 40;
	523	mQPStep= 2;
	524	}
	525
	526
	527	std::vector<RDPoint> Encoder_de265::encode_curve(const Preset& preset) const
	528	{
	529	std::vector<RDPoint> curve;
	530
	531	for (int qp=mQPHigh ; qp>=mQPLow ; qp-=mQPStep) {
	532	curve.push_back(encode(preset, qp));
	533	}
	534
	535	return curve;
	536	}
	537
	538
	539	RDPoint Encoder_de265::encode(const Preset& preset,int qp) const
	540	{
	541	std::stringstream streamname;
	542	streamname << "de265-" << preset.name << "-" << qp << ".265";
	543
	544	std::stringstream cmd1;
	545	cmd1 << "$ENC265 " << input.options_de265()
	546	<< " " << preset.options_de265
	547	<< " -q " << qp << " -o " << streamname.str()
	548	<< " " << encoderParameters;
	549
	550	std::string cmd2 = replace_variables(cmd1.str());
	551
	552	printf("cmdline: %s\n",cmd2.c_str());
	553
	554	RDPoint rd;
	555	rd.start_timer();
	556	int retval = system(cmd2.c_str());
	557	rd.end_timer();
	558
	559	rd.compute_from_h265(streamname.str());
	560
	561	if (!keepStreams) { unlink(streamname.str().c_str()); }
	562
	563	write_rd_line(rd);
	564
	565	return rd;
	566	}
	567
	568
	569
	570
	571	class Encoder_HM : public Encoder
	572	{
	573	public:
	574	Encoder_HM();
	575
	576	void enableSCC(bool flag=true) { useSCC = flag; }
	577	void setQPRange(int low,int high,int step) { mQPLow=low; mQPHigh=high; mQPStep=step; }
	578
	579	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const;
	580
	581	private:
	582	RDPoint encode(const Preset& preset,int qp) const;
	583
	584	bool useSCC;
	585	int mQPLow,mQPHigh,mQPStep;
	586	};
	587
	588
	589	Encoder_HM::Encoder_HM()
	590	{
	591	mQPLow = 14;
	592	mQPHigh= 40;
	593	mQPStep= 2;
	594
	595	useSCC = false;
	596	}
	597
	598
	599	std::vector<RDPoint> Encoder_HM::encode_curve(const Preset& preset) const
	600	{
	601	std::vector<RDPoint> curve;
	602
	603	for (int qp=mQPHigh ; qp>=mQPLow ; qp-=mQPStep) {
	604	curve.push_back(encode(preset, qp));
	605	}
	606
	607	return curve;
	608	}
	609
	610
	611	RDPoint Encoder_HM::encode(const Preset& preset,int qp) const
	612	{
	613	std::stringstream streamname;
	614	streamname << (useSCC ? "hmscc-" : "hm-") << preset.name << "-" << qp << ".265";
	615
	616	char recoyuv_prefix[] = "/tmp/reco-XXXXXX";
	617	char *tempfile = mktemp(recoyuv_prefix);
	618	assert(tempfile != NULL && tempfile[0] != 0);
	619	std::string recoyuv = std::string(recoyuv_prefix) + ".yuv";
	620
	621	std::stringstream cmd1;
	622	cmd1 << (useSCC ? "$HMSCCENC " : "$HMENC ")
	623	<< input.options_HM()
	624	<< " " << (useSCC ? preset.options_hm_scc : preset.options_hm)
	625	<< " -q " << qp << " -o " << recoyuv << " -b " << streamname.str()
	626	<< " " << encoderParameters << " >&2";
	627
	628	std::string cmd2 = replace_variables(cmd1.str());
	629
	630	std::cout << "CMD: '" << cmd2 << "'\n";
	631	RDPoint rd;
	632	rd.start_timer();
	633	int retval = system(cmd2.c_str());
	634	rd.end_timer();
	635
	636	rd.compute_from_yuv(streamname.str(), recoyuv);
	637	if (!keepStreams) { unlink(streamname.str().c_str()); }
	638	unlink(recoyuv.c_str());
	639
	640	write_rd_line(rd);
	641
	642	return rd;
	643	}
	644
	645
	646
	647	class Encoder_x265 : public Encoder
	648	{
	649	public:
	650	Encoder_x265();
	651	void setQPRange(int low,int high,int step) { mQPLow=low; mQPHigh=high; mQPStep=step; }
	652
	653	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const;
	654
	655	private:
	656	RDPoint encode(const Preset& preset,int qp) const;
	657
	658	int mQPLow,mQPHigh,mQPStep;
	659	};
	660
	661
	662	Encoder_x265::Encoder_x265()
	663	{
	664	/* CRF
	665	mQPLow = 4;
	666	mQPHigh= 34;
	667	mQPStep= 2;
	668	*/
	669
	670	mQPLow = 14;
	671	mQPHigh= 40;
	672	mQPStep= 2;
	673	}
	674
	675
	676	std::vector<RDPoint> Encoder_x265::encode_curve(const Preset& preset) const
	677	{
	678	std::vector<RDPoint> curve;
	679
	680	for (int qp=mQPHigh ; qp>=mQPLow ; qp-=mQPStep) {
	681	curve.push_back(encode(preset, qp));
	682	}
	683
	684	return curve;
	685	}
	686
	687
	688	RDPoint Encoder_x265::encode(const Preset& preset,int qp) const
	689	{
	690	std::stringstream streamname;
	691	streamname << "x265-" << preset.name << "-" << qp << ".265";
	692
	693	std::stringstream cmd1;
	694	cmd1 << "$X265ENC " << input.options_x265()
	695	<< " " << preset.options_x265
	696	<< " --qp " << qp << " " << streamname.str()
	697	<< " " << encoderParameters
	698	<< " >&2";
	699
	700	std::string cmd2 = replace_variables(cmd1.str());
	701
	702	//std::cout << "CMD: '" << cmd2 << "'\n";
	703	RDPoint rd;
	704	rd.start_timer();
	705	int retval = system(cmd2.c_str());
	706	rd.end_timer();
	707
	708	rd.compute_from_h265(streamname.str());
	709	if (!keepStreams) { unlink(streamname.str().c_str()); }
	710
	711	write_rd_line(rd);
	712
	713	return rd;
	714	}
	715
	716
	717
	718
	719	class Encoder_f265 : public Encoder
	720	{
	721	public:
	722	Encoder_f265();
	723	void setQPRange(int low,int high,int step) { mQPLow=low; mQPHigh=high; mQPStep=step; }
	724
	725	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const;
	726
	727	private:
	728	RDPoint encode(const Preset& preset,int qp) const;
	729
	730	int mQPLow,mQPHigh,mQPStep;
	731	};
	732
	733
	734	Encoder_f265::Encoder_f265()
	735	{
	736	mQPLow = 14;
	737	mQPHigh= 40;
	738	mQPStep= 2;
	739	}
	740
	741
	742	std::vector<RDPoint> Encoder_f265::encode_curve(const Preset& preset) const
	743	{
	744	std::vector<RDPoint> curve;
	745
	746	for (int qp=mQPHigh ; qp>=mQPLow ; qp-=mQPStep) {
	747	curve.push_back(encode(preset, qp));
	748	}
	749
	750	return curve;
	751	}
	752
	753
	754	RDPoint Encoder_f265::encode(const Preset& preset,int qp) const
	755	{
	756	std::stringstream cmd1;
	757	cmd1 << "$F265 " << input.options_f265()
	758	<< " f265.out -v -p\"" << preset.options_f265 << " qp=" << qp
	759	<< " " << encoderParameters
	760	<< "\" >&2";
	761
	762	std::string cmd2 = replace_variables(cmd1.str());
	763
	764	std::cout << "CMD: '" << cmd2 << "'\n";
	765	RDPoint rd;
	766	rd.start_timer();
	767	int retval = system(cmd2.c_str());
	768	rd.end_timer();
	769
	770	rd.compute_from_h265("f265.out");
	771	if (!keepStreams) { unlink("f265.out"); }
	772
	773	write_rd_line(rd);
	774
	775	return rd;
	776	}
	777
	778
	779
	780	class Encoder_x264 : public Encoder
	781	{
	782	public:
	783	Encoder_x264();
	784	//void setCRFRange(int low,int high,int step) { mCRFLow=low; mCRFHigh=high; mCRFStep=step; }
	785
	786	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const;
	787
	788	private:
	789	RDPoint encode(const Preset& preset,int crf) const;
	790
	791	int mCRFLow,mCRFMid,mCRFHigh;
	792	int mCRFStepHigh, mCRFStepLow;
	793	};
	794
	795
	796	Encoder_x264::Encoder_x264()
	797	{
	798	// in the upper bit-rate range [mid;high], use larger CRF step-size 'StepHigh'
	799	// in the lower bit-rate range [low;mid], use smaller CRF step-size 'StepLow'
	800
	801	mCRFLow = 10;
	802	mCRFMid = 20;
	803	mCRFHigh= 36;
	804	mCRFStepHigh= 2;
	805	mCRFStepLow = 1;
	806	}
	807
	808
	809	std::vector<RDPoint> Encoder_x264::encode_curve(const Preset& preset) const
	810	{
	811	std::vector<RDPoint> curve;
	812
	813	for (int crf=mCRFLow ; crf<mCRFMid ; crf+=mCRFStepHigh) {
	814	curve.push_back(encode(preset, crf));
	815	}
	816
	817	for (int crf=mCRFMid ; crf<=mCRFHigh ; crf+=mCRFStepLow) {
	818	curve.push_back(encode(preset, crf));
	819	}
	820
	821	return curve;
	822	}
	823
	824
	825	RDPoint Encoder_x264::encode(const Preset& preset,int qp_crf) const
	826	{
	827	std::stringstream streamname;
	828	streamname << "x264-" << preset.name << "-" << qp_crf << ".264";
	829
	830	std::stringstream cmd1;
	831	#if 0
	832	cmd1 << "$X264 " << input.options_x264()
	833	<< " " << preset.options_x264
	834	<< " --crf " << qp_crf
	835	<< " -o " << streamname.str();
	836	#else
	837	cmd1 << "$FFMPEG " << input.options_ffmpeg()
	838	<< " " << preset.options_x264_ffmpeg
	839	<< " -crf " << qp_crf
	840	<< " -threads 6"
	841	<< " -f h264 " << streamname.str()
	842	<< " " << encoderParameters;
	843	#endif
	844
	845	std::string cmd2 = replace_variables(cmd1.str());
	846
	847	std::cerr << "-----------------------------\n";
	848
	849	std::cerr << "CMD: '" << cmd2 << "'\n";
	850
	851	RDPoint rd;
	852	rd.start_timer();
	853	int retval = system(cmd2.c_str());
	854	rd.end_timer();
	855
	856	char tmpyuv_prefix[] = "/tmp/rdout-XXXXXX";
	857	char *tempfile = mktemp(tmpyuv_prefix);
	858	assert(tempfile != NULL && tempfile[0] != 0);
	859	std::string tmpyuv = std::string(tmpyuv_prefix) + ".yuv";
	860
	861	std::string cmd3 = "ffmpeg -i " + streamname.str() + " -threads 6 " + tmpyuv;
	862
	863	retval = system(cmd3.c_str());
	864
	865	rd.compute_from_yuv(streamname.str(), tmpyuv);
	866
	867	unlink(tmpyuv.c_str());
	868	if (!keepStreams) { unlink(streamname.str().c_str()); }
	869
	870	write_rd_line(rd);
	871
	872	return rd;
	873	}
	874
	875
	876	class Encoder_mpeg2 : public Encoder
	877	{
	878	public:
	879	Encoder_mpeg2();
	880
	881	virtual std::vector<RDPoint> encode_curve(const Preset& preset) const;
	882
	883	private:
	884	RDPoint encode(const Preset& preset,int bitrate) const;
	885	};
	886
	887
	888	Encoder_mpeg2::Encoder_mpeg2()
	889	{
	890	}
	891
	892
	893	std::vector<RDPoint> Encoder_mpeg2::encode_curve(const Preset& preset) const
	894	{
	895	std::vector<RDPoint> curve;
	896
	897	int bitrates[] = { 250,500,750,1000,1250,1500,1750,2000,2500,3000,3500,4000,4500,5000,
	898	6000,7000,8000,9000,10000,12000,14000,16000,18000,20000,25000,30000,
	899	-1 };
	900
	901	for (int i=0; bitrates[i]>0; i++) {
	902	curve.push_back(encode(preset, bitrates[i]));
	903	}
	904
	905	return curve;
	906	}
	907
	908
	909	RDPoint Encoder_mpeg2::encode(const Preset& preset,int br) const
	910	{
	911	std::stringstream streamname;
	912	streamname << "mpeg2-" << preset.name << "-"
	913	<< std::setfill('0') << std::setw(5) << br << ".mp2";
	914
	915	std::stringstream cmd1;
	916	cmd1 << "$FFMPEG " << input.options_ffmpeg()
	917	<< " " << preset.options_x264_ffmpeg
	918	<< " -b " << br << "k "
	919	<< " -threads 6"
	920	<< " -f mpeg2video " << streamname.str()
	921	<< " " << encoderParameters;
	922
	923	std::string cmd2 = replace_variables(cmd1.str());
	924
	925	std::cerr << "-----------------------------\n";
	926
	927	std::cerr << "CMD: '" << cmd2 << "'\n";
	928
	929	RDPoint rd;
	930	rd.start_timer();
	931	int retval = system(cmd2.c_str());
	932	rd.end_timer();
	933
	934	char tmpyuv_prefix[] = "/tmp/rdout-XXXXXX";
	935	char *tempfile = mktemp(tmpyuv_prefix);
	936	assert(tempfile != NULL && tempfile[0] != 0);
	937	std::string tmpyuv = std::string(tmpyuv_prefix) + ".yuv";
	938
	939	std::string cmd3 = "ffmpeg -i " + streamname.str() + " -threads 6 " + tmpyuv;
	940
	941	retval = system(cmd3.c_str());
	942
	943	rd.compute_from_yuv(streamname.str(), tmpyuv);
	944
	945	unlink(tmpyuv.c_str());
	946	if (!keepStreams) { unlink(streamname.str().c_str()); }
	947
	948	write_rd_line(rd);
	949
	950	return rd;
	951	}
	952
	953
	954	Encoder_de265 enc_de265;
	955	Encoder_HM enc_hm;
	956	Encoder_x265 enc_x265;
	957	Encoder_f265 enc_f265;
	958	Encoder_x264 enc_x264;
	959	Encoder_mpeg2 enc_mpeg2;
	960
	961	// ---------------------------------------------------------------------------
	962
	963	static struct option long_options[] = {
	964	{"keep-streams", no_argument, 0, 'k' },
	965	//{"write-bytestream", required_argument,0, 'B' },
	966	{0, 0, 0, 0 }
	967	};
	968
	969
	970	void show_usage()
	971	{
	972	fprintf(stderr,
	973	"usage: rd-curves 'preset_id' 'input_preset' 'encoder'\n"
	974	"supported encoders: de265 / hm / hmscc / x265 / f265 / x264 / mpeg2\n");
	975	fprintf(stderr,
	976	"presets:\n");
	977
	978	for (int i=0;preset[i].name!=NULL;i++) {
	979	fprintf(stderr,
	980	" %2d %-20s %s\n",preset[i].ID,preset[i].name,preset[i].descr);
	981	}
	982
	983	fprintf(stderr,
	984	"\ninput presets:\n");
	985	for (int i=0;inputSpec[i].name;i++) {
	986	fprintf(stderr,
	987	" %-12s %-30s %4dx%4d, %4d frames, %5.2f fps\n",
	988	inputSpec[i].name,
	989	inputSpec[i].filename,
	990	inputSpec[i].width,
	991	inputSpec[i].height,
	992	inputSpec[i].nFrames,
	993	inputSpec[i].fps);
	994	}
	995	}
	996
	997	int main(int argc, char** argv)
	998	{
	999	init_clock();
	1000
	1001	while (1) {
	1002	int option_index = 0;
	1003
	1004	int c = getopt_long(argc, argv, "kf:p:",
	1005	long_options, &option_index);
	1006	if (c == -1)
	1007	break;
	1008
	1009	switch (c) {
	1010	case 'k': keepStreams=true; break;
	1011	case 'f': maxFrames=atoi(optarg); break;
	1012	case 'p': encoderParameters=optarg; break;
	1013	}
	1014	}
	1015
	1016	if (optind != argc-3) {
	1017	show_usage();
	1018	exit(5);
	1019	}
	1020
	1021	int presetID = atoi( argv[optind] );
	1022	const char* inputName = argv[optind+1];
	1023	const char* encoderName = argv[optind+2];
	1024
	1025	int presetIdx = -1;
	1026
	1027	for (int i=0;preset[i].name != NULL;i++) {
	1028	if (preset[i].ID == presetID) {
	1029	presetIdx = i;
	1030	break;
	1031	}
	1032	}
	1033
	1034	if (presetIdx == -1) {
	1035	fprintf(stderr,"preset ID %d does not exist\n",presetID);
	1036	exit(5);
	1037	}
	1038
	1039	setInput(inputName);
	1040	if (maxFrames) input.setMaxFrames(maxFrames);
	1041
	1042
	1043	Encoder* enc = NULL;
	1044	/**/ if (strcmp(encoderName,"de265")==0) { enc = &enc_de265; }
	1045	else if (strcmp(encoderName,"hm" )==0) { enc = &enc_hm; }
	1046	else if (strcmp(encoderName,"hmscc")==0) { enc = &enc_hm; enc_hm.enableSCC(); }
	1047	else if (strcmp(encoderName,"x265" )==0) { enc = &enc_x265; }
	1048	else if (strcmp(encoderName,"f265" )==0) { enc = &enc_f265; }
	1049	else if (strcmp(encoderName,"x264" )==0) { enc = &enc_x264; }
	1050	else if (strcmp(encoderName,"mpeg2")==0) { enc = &enc_mpeg2; }
	1051
	1052	if (enc==NULL) {
	1053	fprintf(stderr, "unknown encoder");
	1054	exit(5);
	1055	}
	1056
	1057
	1058	std::stringstream data_filename;
	1059	data_filename << encoderName << "-" << inputName << "-" << preset[presetIdx].name << ".rd";
	1060	output_fh = fopen(data_filename.str().c_str(), "wb");
	1061
	1062	fprintf(output_fh,"# %s\n", preset[presetIdx].descr);
	1063	fprintf(output_fh,"# 1:rate 2:psnr 3:ssim 4:cputime(min) 5:walltime(min)\n");
	1064
	1065	std::vector<RDPoint> curve = enc->encode_curve(preset[presetIdx]);
	1066
	1067	for (int i=0;i<curve.size();i++) {
	1068	//fprintf(out_fh,"%7.2f %6.4f\n", curve[i].rate/1024, curve[i].psnr);
	1069	}
	1070
	1071	fclose(output_fh);
	1072
	1073	return 0;
	1074	}

+100

-0

tools/tests.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include <stdio.h>
	21	#include <iostream>
	22	#include <string.h>
	23
	24
	25	class Test
	26	{
	27	public:
	28	Test() { next=s_firstTest; s_firstTest=this; }
	29	virtual ~Test() { }
	30
	31	virtual const char* getName() const { return "noname"; }
	32	virtual const char* getDescription() const { return "no description"; }
	33	virtual bool work(bool quiet=false) = 0;
	34
	35	static void runTest(const char* name) {
	36	Test* t = s_firstTest;
	37	while (t) {
	38	if (strcmp(t->getName(), name)==0) {
	39	t->work();
	40	break;
	41	}
	42	t=t->next;
	43	}
	44	}
	45
	46	static void runAllTests() {
	47	Test* t = s_firstTest;
	48	while (t) {
	49	printf("%s ... ",t->getName());
	50	fflush(stdout);
	51	if (t->work(true) == false) {
	52	printf("* FAILED *\n");
	53	}
	54	else {
	55	printf("passed\n");
	56	}
	57
	58	t=t->next;
	59	}
	60	}
	61
	62	public:
	63	Test* next;
	64	static Test* s_firstTest;
	65	};
	66
	67	Test* Test::s_firstTest = NULL;
	68
	69
	70	class ListTests : public Test
	71	{
	72	public:
	73	const char* getName() const { return "list"; }
	74	const char* getDescription() const { return "list all available tests"; }
	75	bool work(bool quiet) {
	76	if (!quiet) {
	77	Test* t = s_firstTest;
	78	while (t) {
	79	printf("- %s: %s\n",t->getName(), t->getDescription());
	80	t=t->next;
	81	}
	82	}
	83	return true;
	84	}
	85	} listtest;
	86
	87
	88
	89	int main(int argc,char** argv)
	90	{
	91	if (argc>=2) {
	92	Test::runTest(argv[1]);
	93	}
	94	else {
	95	Test::runAllTests();
	96	}
	97
	98	return 0;
	99	}

+113

-0

tools/yuv-distortion.cc less more

	0	/*
	1	* H.265 video codec.
	2	* Copyright (c) 2013-2014 struktur AG, Dirk Farin <farin@struktur.de>
	3	*
	4	* This file is part of libde265.
	5	*
	6	* libde265 is free software: you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation, either version 3 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* libde265 is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with libde265. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#ifdef HAVE_CONFIG_H
	21	#include "config.h"
	22	#endif
	23
	24	#include <stdio.h>
	25	#include <stdlib.h>
	26
	27	#include <libde265/quality.h>
	28
	29	#if HAVE_VIDEOGFX
	30	#include <libvideogfx.hh>
	31	using namespace videogfx;
	32	#endif
	33
	34
	35	float ssim(const uint8_t* img1,
	36	const uint8_t* img2,
	37	int width, int height)
	38	{
	39	#if HAVE_VIDEOGFX
	40	Bitmap<Pixel> ref, coded;
	41	ref .Create(width, height); // reference image
	42	coded.Create(width, height); // coded image
	43
	44	for (int y=0;y<height;y++) {
	45	memcpy(coded[y], img1 + y*width, width);
	46	memcpy(ref[y], img2 + y*width, width);
	47	}
	48
	49	SSIM ssimAlgo;
	50	return ssimAlgo.calcMSSIM(ref,coded);
	51	#else
	52	return 0;
	53	#endif
	54	}
	55
	56
	57	int main(int argc, char** argv)
	58	{
	59	if (argc != 5) {
	60	fprintf(stderr,"need two YUV files and image size as input: FILE1 FILE2 WIDTH HEIGHT\n");
	61	exit(5);
	62	}
	63
	64
	65	FILE* fh_ref = fopen(argv[1],"rb");
	66	FILE* fh_cmp = fopen(argv[2],"rb");
	67
	68	int width = atoi(argv[3]);
	69	int height = atoi(argv[4]);
	70
	71	uint8_t* yp_ref = (uint8_t)malloc(widthheight);
	72	uint8_t* yp_cmp = (uint8_t)malloc(widthheight);
	73
	74	double mse_y=0.0, ssim_y=0.0;
	75	int nFrames=0;
	76
	77	for (;;)
	78	{
	79	if (fread(yp_ref,1,widthheight,fh_ref) != widthheight) {
	80	break;
	81	}
	82	if (fread(yp_cmp,1,widthheight,fh_cmp) != widthheight) {
	83	break;
	84	}
	85
	86	if (feof(fh_ref)) break;
	87	if (feof(fh_cmp)) break;
	88
	89	fprintf(stderr,"yuv-distortion processing frame %d\r",nFrames+1);
	90
	91	fseek(fh_ref,width*height/2,SEEK_CUR);
	92	fseek(fh_cmp,width*height/2,SEEK_CUR);
	93
	94	double curr_mse_y = MSE(yp_ref, width, yp_cmp, width, width, height);
	95	mse_y += curr_mse_y;
	96
	97	double curr_ssim_y = ssim(yp_ref, yp_cmp, width, height);
	98	ssim_y += curr_ssim_y;
	99
	100	printf("%4d %f %f\n",nFrames,PSNR(curr_mse_y),curr_ssim_y);
	101
	102	nFrames++;
	103	}
	104
	105	printf("total: %f %f\n",PSNR(mse_y/nFrames),ssim_y/nFrames);
	106	fprintf(stderr,"\n");
	107
	108	fclose(fh_ref);
	109	fclose(fh_cmp);
	110
	111	return 0;
	112	}