OSDN Git Service

coroid/ffmpeg_saccubus.git
12 years agoMerge remote-tracking branch 'qatar/master'
Michael Niedermayer [Mon, 11 Jul 2011 23:42:32 +0000 (01:42 +0200)]
Merge remote-tracking branch 'qatar/master'

* qatar/master:
  changelog: misc typo and wording fixes
  H.264: add filter_mb_fast support for >8-bit decoding
  doc: Remove outdated comments about gcc 2.95 and gcc 3.3 support.
  lls: use av_lfg instead of rand() in test program
  build: remove unnecessary dependency on libs from 'all' target
  H.264: avoid redundant alpha/beta calculations in loopfilter
  H.264: optimize intra/inter loopfilter decision
  mpegts: fix Continuity Counter error detection
  build: remove unnecessary FFLDFLAGS variable
  vp8/mt: flush worker thread, not application thread context, on seek.
  mt: proper locking around release_buffer calls.
  DxVA2: unbreak build after [657ccb5ac75ce34e62bd67f228d9bd36db72189e]
  hwaccel: unbreak build
  Eliminate FF_COMMON_FRAME macro.

Conflicts:
Changelog
Makefile
doc/developer.texi
libavcodec/avcodec.h
libavcodec/h264.c
libavcodec/mpeg4videodec.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
12 years agochangelog: misc typo and wording fixes
Diego Biurrun [Sat, 9 Jul 2011 15:43:53 +0000 (17:43 +0200)]
changelog: misc typo and wording fixes

12 years agoH.264: add filter_mb_fast support for >8-bit decoding
Jason Garrett-Glaser [Mon, 11 Jul 2011 21:46:37 +0000 (14:46 -0700)]
H.264: add filter_mb_fast support for >8-bit decoding

Much faster high bit depth deblocking.

12 years agodoc: Remove outdated comments about gcc 2.95 and gcc 3.3 support.
Diego Biurrun [Sat, 9 Jul 2011 18:15:29 +0000 (20:15 +0200)]
doc: Remove outdated comments about gcc 2.95 and gcc 3.3 support.

12 years agolls: use av_lfg instead of rand() in test program
Mans Rullgard [Fri, 8 Jul 2011 14:01:05 +0000 (15:01 +0100)]
lls: use av_lfg instead of rand() in test program

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agobuild: remove unnecessary dependency on libs from 'all' target
Mans Rullgard [Mon, 11 Jul 2011 19:53:15 +0000 (20:53 +0100)]
build: remove unnecessary dependency on libs from 'all' target

The libs are added to the all target elsewhere, no need to do it
again here.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoH.264: avoid redundant alpha/beta calculations in loopfilter
Jason Garrett-Glaser [Sun, 10 Jul 2011 09:25:53 +0000 (02:25 -0700)]
H.264: avoid redundant alpha/beta calculations in loopfilter

12 years agoH.264: optimize intra/inter loopfilter decision
Jason Garrett-Glaser [Sun, 10 Jul 2011 08:28:50 +0000 (01:28 -0700)]
H.264: optimize intra/inter loopfilter decision

12 years agompegts: fix Continuity Counter error detection
Jindrich Makovicka [Thu, 30 Jun 2011 09:03:15 +0000 (09:03 +0000)]
mpegts: fix Continuity Counter error detection

According to MPEG-TS specs, the continuity_counter shall not be
incremented when the adaptation_field_control of the packet
equals '00' or '10'.

Signed-off-by: Jindrich Makovicka <jindrich.makovicka@nangu.tv>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agobuild: remove unnecessary FFLDFLAGS variable
Mans Rullgard [Sun, 10 Jul 2011 21:01:41 +0000 (22:01 +0100)]
build: remove unnecessary FFLDFLAGS variable

This variable is set to the same value for all directories.
Adding the -L flags directly to LDFLAGS is simpler and achieves
the same thing.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agovp8/mt: flush worker thread, not application thread context, on seek.
Ronald S. Bultje [Mon, 11 Jul 2011 04:23:09 +0000 (21:23 -0700)]
vp8/mt: flush worker thread, not application thread context, on seek.

This prevents a crash when seeking.

12 years agomt: proper locking around release_buffer calls.
Ronald S. Bultje [Mon, 11 Jul 2011 00:28:28 +0000 (17:28 -0700)]
mt: proper locking around release_buffer calls.

This fixes a crash when seeking in some webm files with many
threads (e.g. 8).

12 years agoRevert "lavc: remove a half-working attempt at different defaults for audio/video...
Michael Niedermayer [Mon, 11 Jul 2011 03:46:51 +0000 (05:46 +0200)]
Revert "lavc: remove a half-working attempt at different defaults for audio/video codecs."

This reverts commit 18c007ba37b2c7dee5bd2f1a3eb3bfee9b6b3d26.

12 years agoMerge commit '142e76f1055de5dde44696e71a5f63f2cb11dedf'
Michael Niedermayer [Mon, 11 Jul 2011 02:23:25 +0000 (04:23 +0200)]
Merge commit '142e76f1055de5dde44696e71a5f63f2cb11dedf'

* commit '142e76f1055de5dde44696e71a5f63f2cb11dedf':
  swscale: fix crash with dithering due incorrect offset calculation.
  matroskadec: fix stupid typo (!= -> ==)
  build: remove duplicates from order-only directory prerequisite list
  build: rework rules for things in the tools dir
  configure: fix --cpu=host with gcc 4.6
  ARM: use const macro to define constant data in asm
  bitdepth: simplify FUNC/FUNCC macros
  dsputil: remove ff_emulated_edge_mc macro used in one place
  9/10-bit: simplify clipping macros
  matroskadec: reindent
  matroskadec: defer parsing of cues element until we seek.
  lavc: add support for codec-specific defaults.
  lavc: make avcodec_alloc_context3 officially public.
  lavc: remove a half-working attempt at different defaults for audio/video codecs.
  ac3dec: add a drc_scale private option
  lavf: add avformat_find_stream_info()
  lavc: introduce avcodec_open2() as a replacement for avcodec_open().

Conflicts:
Makefile
libavcodec/utils.c
libavformat/avformat.h
libswscale/swscale_internal.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoMerge branch 'sws_32bit_integration'
Michael Niedermayer [Mon, 11 Jul 2011 02:00:41 +0000 (04:00 +0200)]
Merge branch 'sws_32bit_integration'

* sws_32bit_integration:
  regtests/sws: update checksums for recent changes
  sws: dont mess with XInc when the code needing it isnt used
  sws: Fix chroma init for 32bit buffers.
  swscale: error dithering for 16/9/10-bit to 8-bit.
  swscale: fix overflow in 16-bit vertical scaling.
  swscale: fix crash in 8-bpc bilinear output without alpha.
  swscale: fix 16-bit scaling when output is 8-bits.
  sws: fix non native endian 9-15 bit input with 16bit out
  sws: disable scale16 when int32 is used
  sws: fix rgb -> 16bit
  sws: fix uv overwrite in 32bt
  sws: fix gray16_1
  sws:ix yuv2rgb48_1_c_template()
  sws: fix 16/32 bug from merge
  swscale: for >8bit scaling, read in native bit-depth.
  swscale: fix another yuv range conversion overflow in 16bit scaling. (cherry picked from commit 81cc7d0bd1eab0aa782ff8dd49e087025a42cdee)
  swscale: fix yuv range correction when using 16-bit scaling. (cherry picked from commit e0b8fff6c7a293e35079ba1931bd19372686b3f6)
  swscale: implement >8bit scaling support.

Merged-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoregtests/sws: update checksums for recent changes
Michael Niedermayer [Mon, 11 Jul 2011 01:54:30 +0000 (03:54 +0200)]
regtests/sws: update checksums for recent changes

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
12 years agosws: dont mess with XInc when the code needing it isnt used
Michael Niedermayer [Sun, 10 Jul 2011 23:44:23 +0000 (01:44 +0200)]
sws: dont mess with XInc when the code needing it isnt used

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
12 years agosws: Fix chroma init for 32bit buffers.
Michael Niedermayer [Sun, 10 Jul 2011 13:58:24 +0000 (15:58 +0200)]
sws: Fix chroma init for 32bit buffers.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoswscale: error dithering for 16/9/10-bit to 8-bit.
Ronald S. Bultje [Tue, 5 Jul 2011 19:49:11 +0000 (12:49 -0700)]
swscale: error dithering for 16/9/10-bit to 8-bit.

Based on a somewhat similar idea in FFmpeg's swscale copy.

12 years agoswscale: fix overflow in 16-bit vertical scaling.
Ronald S. Bultje [Fri, 8 Jul 2011 21:57:07 +0000 (14:57 -0700)]
swscale: fix overflow in 16-bit vertical scaling.

We operated on 31-bits, but with e.g. lanczos scaling, values can
add up to beyond 0x80000000, thus leading to output of zeroes. Drop
one bit of precision fixes this.

12 years agoswscale: fix crash in 8-bpc bilinear output without alpha.
Ronald S. Bultje [Fri, 8 Jul 2011 19:28:28 +0000 (12:28 -0700)]
swscale: fix crash in 8-bpc bilinear output without alpha.

We accessed the alpha array even it wasn't used and didn't
exist, hence leading to a NULL pointer segfault.
(cherry picked from commit bf2cba453244a74331238a472fe0e309f116f4d9)

12 years agoswscale: fix 16-bit scaling when output is 8-bits.
Ronald S. Bultje [Fri, 8 Jul 2011 21:52:52 +0000 (14:52 -0700)]
swscale: fix 16-bit scaling when output is 8-bits.

We would use the second half of the U plane buffer, rather than the
V plane buffer, to output the V plane pixels.

12 years agosws: fix non native endian 9-15 bit input with 16bit out
Michael Niedermayer [Sun, 3 Jul 2011 00:46:54 +0000 (02:46 +0200)]
sws: fix non native endian 9-15 bit input with 16bit out

12 years agosws: disable scale16 when int32 is used
Michael Niedermayer [Sun, 3 Jul 2011 00:34:00 +0000 (02:34 +0200)]
sws: disable scale16 when int32 is used

12 years agosws: fix rgb -> 16bit
Michael Niedermayer [Sun, 3 Jul 2011 00:08:45 +0000 (02:08 +0200)]
sws: fix rgb -> 16bit

12 years agosws: fix uv overwrite in 32bt
Michael Niedermayer [Sat, 2 Jul 2011 23:04:51 +0000 (01:04 +0200)]
sws: fix uv overwrite in 32bt

12 years agosws: fix gray16_1
Michael Niedermayer [Sat, 2 Jul 2011 22:33:23 +0000 (00:33 +0200)]
sws: fix gray16_1

12 years agosws:ix yuv2rgb48_1_c_template()
Michael Niedermayer [Sat, 2 Jul 2011 21:30:14 +0000 (23:30 +0200)]
sws:ix yuv2rgb48_1_c_template()

12 years agosws: fix 16/32 bug from merge
Michael Niedermayer [Sat, 2 Jul 2011 00:25:02 +0000 (02:25 +0200)]
sws: fix 16/32 bug from merge

12 years agoswscale: for >8bit scaling, read in native bit-depth.
Ronald S. Bultje [Fri, 1 Jul 2011 00:35:13 +0000 (17:35 -0700)]
swscale: for >8bit scaling, read in native bit-depth.

For 9/10bit, it means we don't have to upscale to 16bit before
actual scaling or pixel format conversion, and thus a performance
gain.

12 years agoMerge branch 'ronalds_buggy_gt8bit_sws'
Michael Niedermayer [Fri, 1 Jul 2011 03:28:13 +0000 (05:28 +0200)]
Merge branch 'ronalds_buggy_gt8bit_sws'

* ronalds_buggy_gt8bit_sws:
  swscale: fix another yuv range conversion overflow in 16bit scaling. (cherry picked from commit 81cc7d0bd1eab0aa782ff8dd49e087025a42cdee)
  swscale: fix yuv range correction when using 16-bit scaling. (cherry picked from commit e0b8fff6c7a293e35079ba1931bd19372686b3f6)
  swscale: implement >8bit scaling support.

Conflicts:
libswscale/swscale.c
libswscale/utils.c
libswscale/x86/swscale_template.c
tests/ref/lavfi/pixdesc
tests/ref/lavfi/pixfmts_copy
tests/ref/lavfi/pixfmts_null
tests/ref/lavfi/pixfmts_scale
tests/ref/lavfi/pixfmts_vflip

Merged-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoDxVA2: unbreak build after [657ccb5ac75ce34e62bd67f228d9bd36db72189e]
Jean-Baptiste Kempf [Sun, 10 Jul 2011 23:13:20 +0000 (01:13 +0200)]
DxVA2: unbreak build after [657ccb5ac75ce34e62bd67f228d9bd36db72189e]

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agohwaccel: unbreak build
Mans Rullgard [Sun, 10 Jul 2011 22:55:06 +0000 (23:55 +0100)]
hwaccel: unbreak build

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoEliminate FF_COMMON_FRAME macro.
Diego Biurrun [Wed, 6 Jul 2011 18:08:30 +0000 (20:08 +0200)]
Eliminate FF_COMMON_FRAME macro.

FF_COMMON_FRAME holds the contents of the AVFrame structure and is also copied
to struct Picture.  Replace by an embedded AVFrame structure in struct Picture.

12 years agoswscale: fix crash with dithering due incorrect offset calculation.
Anton Mitrofanov [Sun, 10 Jul 2011 16:07:43 +0000 (20:07 +0400)]
swscale: fix crash with dithering due incorrect offset calculation.

ptrdiff_t can be 4 bytes, which leads to the next element being 4-byte
aligned and thus at a different offset than intended. Forcing 8-byte
alignment forces equal offset of dither16/32 on x86-32 and x86-64.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agomatroskadec: fix stupid typo (!= -> ==)
Anton Khirnov [Sun, 10 Jul 2011 19:48:23 +0000 (20:48 +0100)]
matroskadec: fix stupid typo (!= -> ==)

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agobuild: remove duplicates from order-only directory prerequisite list
Diego Biurrun [Sun, 10 Jul 2011 19:05:45 +0000 (21:05 +0200)]
build: remove duplicates from order-only directory prerequisite list

This reduces startup latency for make invocations, which is especially
noticeable on systems that are slow or have slow I/O, like Windows.

12 years agobuild: rework rules for things in the tools dir
Mans Rullgard [Sun, 10 Jul 2011 19:26:25 +0000 (20:26 +0100)]
build: rework rules for things in the tools dir

Declaring tools associated with each library in their respective
makefiles allows these tools to easily depend on the correct
prerequisites and link against the libs they need.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoHint why SDL_Init() may have failed.
Carl Eugen Hoyos [Sun, 10 Jul 2011 19:11:37 +0000 (21:11 +0200)]
Hint why SDL_Init() may have failed.

12 years agoMatch VA-API / VDPAU with zlib / bzlib build system documentation.
Carl Eugen Hoyos [Sun, 10 Jul 2011 19:09:55 +0000 (21:09 +0200)]
Match VA-API / VDPAU with zlib / bzlib build system documentation.

12 years agoconfigure: fix --cpu=host with gcc 4.6
Mans Rullgard [Sun, 10 Jul 2011 15:17:55 +0000 (16:17 +0100)]
configure: fix --cpu=host with gcc 4.6

The output from -v with gcc 4.6 has changed such that the search
pattern matches too soon without making it more strict.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoARM: use const macro to define constant data in asm
Mans Rullgard [Sat, 9 Jul 2011 11:37:08 +0000 (12:37 +0100)]
ARM: use const macro to define constant data in asm

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agobitdepth: simplify FUNC/FUNCC macros
Mans Rullgard [Sat, 9 Jul 2011 10:44:07 +0000 (11:44 +0100)]
bitdepth: simplify FUNC/FUNCC macros

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agodsputil: remove ff_emulated_edge_mc macro used in one place
Mans Rullgard [Sat, 9 Jul 2011 10:38:39 +0000 (11:38 +0100)]
dsputil: remove ff_emulated_edge_mc macro used in one place

This macro can cause problems in conjunction with the bitdepth
template expansion.  It was presumably added to keep source
compatibility when high bitdepth support was added.  However,
emulated_edge_mc is a dsputil pointer and should not be called
directly, so there is little reason to keep such a macro.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years ago9/10-bit: simplify clipping macros
Mans Rullgard [Sat, 9 Jul 2011 10:03:01 +0000 (11:03 +0100)]
9/10-bit: simplify clipping macros

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agomatroskadec: reindent
Anton Khirnov [Sat, 9 Jul 2011 06:11:30 +0000 (08:11 +0200)]
matroskadec: reindent

12 years agomatroskadec: defer parsing of cues element until we seek.
Aaron Colwell [Sat, 9 Jul 2011 05:48:43 +0000 (07:48 +0200)]
matroskadec: defer parsing of cues element until we seek.

This decreases startup latency.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agoexamples/muxing.c: remove doxy reference to unexisting file
Stefano Sabatini [Sun, 10 Jul 2011 15:58:35 +0000 (17:58 +0200)]
examples/muxing.c: remove doxy reference to unexisting file

12 years agodoc/examples: add decoding/filtering example program
Stefano Sabatini [Mon, 20 Jun 2011 22:33:37 +0000 (00:33 +0200)]
doc/examples: add decoding/filtering example program

12 years agoexamples/encoding.c: remove unnecessary includes
Stefano Sabatini [Sat, 9 Jul 2011 13:24:54 +0000 (15:24 +0200)]
examples/encoding.c: remove unnecessary includes

Simplify.

12 years agoexamples/encoding.c: use av_image_alloc() for allocating an image buffer
Stefano Sabatini [Fri, 8 Jul 2011 07:52:07 +0000 (09:52 +0200)]
examples/encoding.c: use av_image_alloc() for allocating an image buffer

Simplify.

12 years agographparser: simplify condition in avfilter_graph_parse()
Stefano Sabatini [Sat, 2 Jul 2011 13:37:32 +0000 (15:37 +0200)]
graphparser: simplify condition in avfilter_graph_parse()

Since avfilter_graph_parse() creates the "[in]" inout for the first
unlabelled input pad, it is expected that it will create an "[out]"
inout for last unlabelled output pad, even in the case where it cannot
find any open input pad with that name.

This change removes the check on the existence of an open input pad
named "out", so it simplifies the checked condition while implementing
a more intuitive behavior.

12 years agolavc: add support for codec-specific defaults.
Anton Khirnov [Wed, 8 Jun 2011 06:27:53 +0000 (08:27 +0200)]
lavc: add support for codec-specific defaults.

12 years agolavc: make avcodec_alloc_context3 officially public.
Anton Khirnov [Sat, 18 Jun 2011 11:40:48 +0000 (13:40 +0200)]
lavc: make avcodec_alloc_context3 officially public.

Deprecate avcodec_alloc_context/2.

12 years agolavc: remove a half-working attempt at different defaults for audio/video codecs.
Anton Khirnov [Mon, 6 Jun 2011 16:37:46 +0000 (18:37 +0200)]
lavc: remove a half-working attempt at different defaults for audio/video codecs.

It should be replaced with proper per-codec defaults.

12 years agoac3dec: add a drc_scale private option
Anton Khirnov [Fri, 20 May 2011 20:37:59 +0000 (22:37 +0200)]
ac3dec: add a drc_scale private option

Deprecate corresponding AVCodecContext option.

This is the first test of decoder private options.

12 years agolavf: add avformat_find_stream_info()
Anton Khirnov [Sun, 22 May 2011 17:24:59 +0000 (19:24 +0200)]
lavf: add avformat_find_stream_info()

It supports passing options to codecs.

12 years agolavc: introduce avcodec_open2() as a replacement for avcodec_open().
Anton Khirnov [Sun, 22 May 2011 12:10:49 +0000 (14:10 +0200)]
lavc: introduce avcodec_open2() as a replacement for avcodec_open().

Adds support for decoder-private options and makes setting other options
simpler.

12 years agoMerge remote-tracking branch 'qatar/master'
Michael Niedermayer [Sun, 10 Jul 2011 02:28:50 +0000 (04:28 +0200)]
Merge remote-tracking branch 'qatar/master'

* qatar/master: (22 commits)
  H.264: fix filter_mb_fast with 4:4:4 + 8x8dct
  alsa: limit buffer_size to 32768 frames.
  alsa: fallback to buffer_size/4 for period_size.
  doc: replace @pxref by @ref where appropriate
  mpeg1video: don't abort if thread_count is too high.
  segafilm: add support for videos with cri adx adpcm
  gxf: Fix 25 fps DV material in GXF being misdetected as 50 fps
  libxvid: Add const qualifier to silence compiler warning.
  H.264: improve qp_thresh check
  H.264: use fill_rectangle in CABAC decoding
  H.264: Remove redundant hl_motion_16/8 code
  H.264: merge fill_rectangle into P-SKIP MV prediction, to match B-SKIP
  H.264: faster P-SKIP decoding
  H.264: av_always_inline some more functions
  H.264: Add x86 assembly for 10-bit H.264 predict functions
  swscale: rename uv_off/uv_off2 to uv_off_px/byte.
  swscale: implement error dithering in planarCopyWrapper.
  swscale: error dithering for 16/9/10-bit to 8-bit.
  swscale: fix overflow in 16-bit vertical scaling.
  swscale: fix crash in 8-bpc bilinear output without alpha.
  ...

Conflicts:
doc/developer.texi
libavdevice/alsa-audio.h
libavformat/gxf.c
libswscale/swscale.c
libswscale/swscale_internal.h
libswscale/swscale_unscaled.c
libswscale/x86/swscale_template.c
tests/ref/lavfi/pixdesc
tests/ref/lavfi/pixfmts_copy
tests/ref/lavfi/pixfmts_crop
tests/ref/lavfi/pixfmts_hflip
tests/ref/lavfi/pixfmts_null
tests/ref/lavfi/pixfmts_scale
tests/ref/lavfi/pixfmts_vflip

Merged-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoH.264: fix filter_mb_fast with 4:4:4 + 8x8dct
Jason Garrett-Glaser [Sat, 9 Jul 2011 18:03:26 +0000 (11:03 -0700)]
H.264: fix filter_mb_fast with 4:4:4 + 8x8dct

12 years agoalsa: limit buffer_size to 32768 frames.
Justin Ruggles [Wed, 22 Jun 2011 20:38:20 +0000 (16:38 -0400)]
alsa: limit buffer_size to 32768 frames.

In testing, the file output plugin gave a max buffer size of about 20 million
frames, which is way more than what is really needed and causes a memory
allocation error on my system.

12 years agoalsa: fallback to buffer_size/4 for period_size.
Justin Ruggles [Wed, 22 Jun 2011 19:33:56 +0000 (15:33 -0400)]
alsa: fallback to buffer_size/4 for period_size.

buffer_size/4 is the value used by aplay. This fixes output to null
devices, e.g. writing ALSA output to a file.

12 years agodoc: replace @pxref by @ref where appropriate
Diego Biurrun [Fri, 8 Jul 2011 13:34:56 +0000 (15:34 +0200)]
doc: replace @pxref by @ref where appropriate

12 years agompeg1video: don't abort if thread_count is too high.
Frank Barchard [Wed, 29 Jun 2011 21:13:18 +0000 (14:13 -0700)]
mpeg1video: don't abort if thread_count is too high.

Instead, just decrease it to a valid value and use that.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agocosmetic: reindent of dxva2 zigzag workaround
Joakim Plate [Sat, 9 Jul 2011 15:30:25 +0000 (17:30 +0200)]
cosmetic: reindent of dxva2 zigzag workaround

12 years agoAdded ability to enable workaround for dxva2 decoding using older ATI cards
Joakim Plate [Mon, 4 Jul 2011 22:18:32 +0000 (00:18 +0200)]
Added ability to enable workaround for dxva2 decoding using older ATI cards

The workaround need to be enabled per pci id which can not
be detected inside ffmpeg. So this adds a flag that enabled
the alternate behavior.

12 years agoSupport gsm_ms in caf files.
Carl Eugen Hoyos [Sat, 9 Jul 2011 15:13:13 +0000 (17:13 +0200)]
Support gsm_ms in caf files.

12 years agosegafilm: add support for videos with cri adx adpcm
Matthew Hoops [Tue, 5 Jul 2011 16:21:22 +0000 (12:21 -0400)]
segafilm: add support for videos with cri adx adpcm

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agoexamples/encoding.c: replace "this library" with "libavcodec" in doxy
Stefano Sabatini [Sat, 9 Jul 2011 13:40:01 +0000 (15:40 +0200)]
examples/encoding.c: replace "this library" with "libavcodec" in doxy

So it is more clear the context of the sentence.

12 years agoexamples/encoding.c: remove reference to unexisting file
Stefano Sabatini [Sat, 9 Jul 2011 13:39:16 +0000 (15:39 +0200)]
examples/encoding.c: remove reference to unexisting file

12 years agoexamples/muxing.c: set sample format for encoding
Stefano Sabatini [Fri, 8 Jul 2011 08:05:06 +0000 (10:05 +0200)]
examples/muxing.c: set sample format for encoding

This is required, or avcodec_open() will abort because of invalid
sample format.

12 years agogxf: Fix 25 fps DV material in GXF being misdetected as 50 fps
Tomas Härdin [Thu, 23 Jun 2011 13:59:33 +0000 (15:59 +0200)]
gxf: Fix 25 fps DV material in GXF being misdetected as 50 fps

Set DV packet durations using fields_per_frame.
This requires turning gxf_stream_info into the demuxer's context for access to the value in gxf_packet().
Since MPEG-2 seems to work fine this done only for DV.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agolibxvid: Add const qualifier to silence compiler warning.
Diego Biurrun [Wed, 6 Jul 2011 23:31:04 +0000 (01:31 +0200)]
libxvid: Add const qualifier to silence compiler warning.

libavcodec/libxvidff.c:752: warning: initialization discards qualifiers from pointer target type

12 years agoMerge remote-tracking branch 'qatar/master'
Michael Niedermayer [Sat, 9 Jul 2011 00:06:40 +0000 (02:06 +0200)]
Merge remote-tracking branch 'qatar/master'

* qatar/master: (28 commits)
  mp3enc: write a xing frame containing number of frames in the file
  lavf: update AVStream.nb_frames when muxing.
  ffmpeg: remove unused variables from InputStream.
  doc: update ffmpeg -ar and -ac documentation to reflect reality.
  ffmpeg: remove pointless if (nb_input_files)
  ffmpeg: merge input_files_ts_offset into input_files.
  ffmpeg: merge input_codecs into input_streams.
  ffmpeg: drop AV prefixes from struct names.
  ffmpeg: deprecate loop_input and loop_output options
  gif: add loop private option.
  img2: add loop private option.
  AVOptions: in av_opt_find() don't return named constants unless unit is specified.
  x11grab: replace undocumented nomouse hackery with a private option.
  dict: extend documentation.
  lls: whitespace cosmetics
  docs: Use proper markup for a literal command line option
  docs: Remove a remark that isn't relevant any longer
  docs: Explain how to regenerate import libraries with MSVC tools
  docs: Mention that libraries for MSVC can be built with a cross compiler
  docs: Remove old docs that mention setting up a build environment with lib.exe
  ...

Conflicts:
doc/ffmpeg.texi
doc/general.texi
ffmpeg.c
libavcodec/Makefile
libavcodec/dnxhddata.c
libavformat/mp3enc.c
libavformat/utils.c
libavutil/Makefile
tests/copycooker.sh

Merged-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoH.264: improve qp_thresh check
Jason Garrett-Glaser [Wed, 6 Jul 2011 21:08:30 +0000 (14:08 -0700)]
H.264: improve qp_thresh check

Eliminate redundant check in filter_mb_fast, consider bit depth in calculating qp_thresh.

12 years agoH.264: use fill_rectangle in CABAC decoding
Jason Garrett-Glaser [Wed, 6 Jul 2011 20:25:13 +0000 (13:25 -0700)]
H.264: use fill_rectangle in CABAC decoding

12 years agoH.264: Remove redundant hl_motion_16/8 code
Jason Garrett-Glaser [Wed, 6 Jul 2011 19:26:04 +0000 (12:26 -0700)]
H.264: Remove redundant hl_motion_16/8 code

12 years agoH.264: merge fill_rectangle into P-SKIP MV prediction, to match B-SKIP
Jason Garrett-Glaser [Wed, 6 Jul 2011 14:58:50 +0000 (07:58 -0700)]
H.264: merge fill_rectangle into P-SKIP MV prediction, to match B-SKIP

12 years agoH.264: faster P-SKIP decoding
Jason Garrett-Glaser [Wed, 6 Jul 2011 00:55:14 +0000 (17:55 -0700)]
H.264: faster P-SKIP decoding

Inline the relevant parts of fill_decode_caches into P-SKIP mv prediction to
avoid calling the whole thing.

12 years agoH.264: av_always_inline some more functions
Jason Garrett-Glaser [Wed, 6 Jul 2011 01:08:22 +0000 (18:08 -0700)]
H.264: av_always_inline some more functions

These weren't getting inlined all the time in all gcc versions.

12 years agoH.264: Add x86 assembly for 10-bit H.264 predict functions
Daniel Kang [Sun, 3 Jul 2011 18:56:09 +0000 (14:56 -0400)]
H.264: Add x86 assembly for 10-bit H.264 predict functions

Mainly ported from 8-bit H.264 predict.

Some code ported from x264. LGPL ok by author.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agomake the RELEASE file say git and not 0.7
Michael Niedermayer [Fri, 8 Jul 2011 22:28:16 +0000 (00:28 +0200)]
make the RELEASE file say git and not 0.7

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
12 years agoswscale: rename uv_off/uv_off2 to uv_off_px/byte.
Ronald S. Bultje [Fri, 8 Jul 2011 21:39:04 +0000 (14:39 -0700)]
swscale: rename uv_off/uv_off2 to uv_off_px/byte.

12 years agoswscale: implement error dithering in planarCopyWrapper.
Ronald S. Bultje [Wed, 6 Jul 2011 03:48:13 +0000 (20:48 -0700)]
swscale: implement error dithering in planarCopyWrapper.

Based on a somewhat similar idea in FFmpeg's swscale.

12 years agoswscale: error dithering for 16/9/10-bit to 8-bit.
Ronald S. Bultje [Tue, 5 Jul 2011 19:49:11 +0000 (12:49 -0700)]
swscale: error dithering for 16/9/10-bit to 8-bit.

Based on a somewhat similar idea in FFmpeg's swscale copy.

12 years agoswscale: fix overflow in 16-bit vertical scaling.
Ronald S. Bultje [Fri, 8 Jul 2011 21:57:07 +0000 (14:57 -0700)]
swscale: fix overflow in 16-bit vertical scaling.

We operated on 31-bits, but with e.g. lanczos scaling, values can
add up to beyond 0x80000000, thus leading to output of zeroes. Drop
one bit of precision fixes this.

12 years agoswscale: fix crash in 8-bpc bilinear output without alpha.
Ronald S. Bultje [Fri, 8 Jul 2011 19:28:28 +0000 (12:28 -0700)]
swscale: fix crash in 8-bpc bilinear output without alpha.

We accessed the alpha array even it wasn't used and didn't
exist, hence leading to a NULL pointer segfault.

12 years agoswscale: fix 16-bit horizontal scaling underflow.
Ronald S. Bultje [Fri, 8 Jul 2011 21:54:31 +0000 (14:54 -0700)]
swscale: fix 16-bit horizontal scaling underflow.

When using e.g. lanczos scaling, values can drop below 0, so they
should never be unsigned.

12 years agoswscale: fix 16-bit scaling when output is 8-bits.
Ronald S. Bultje [Fri, 8 Jul 2011 21:52:52 +0000 (14:52 -0700)]
swscale: fix 16-bit scaling when output is 8-bits.

We would use the second half of the U plane buffer, rather than the
V plane buffer, to output the V plane pixels.

12 years agomp3enc: write a xing frame containing number of frames in the file
Anton Khirnov [Mon, 21 Mar 2011 20:02:50 +0000 (21:02 +0100)]
mp3enc: write a xing frame containing number of frames in the file

12 years agolavf: update AVStream.nb_frames when muxing.
Anton Khirnov [Wed, 16 Mar 2011 07:45:14 +0000 (08:45 +0100)]
lavf: update AVStream.nb_frames when muxing.

12 years agoffmpeg: remove unused variables from InputStream.
Anton Khirnov [Wed, 6 Jul 2011 06:35:07 +0000 (08:35 +0200)]
ffmpeg: remove unused variables from InputStream.

12 years agodoc: update ffmpeg -ar and -ac documentation to reflect reality.
Anton Khirnov [Fri, 24 Jun 2011 09:22:15 +0000 (11:22 +0200)]
doc: update ffmpeg -ar and -ac documentation to reflect reality.

12 years agoffmpeg: remove pointless if (nb_input_files)
Anton Khirnov [Fri, 24 Jun 2011 06:54:52 +0000 (08:54 +0200)]
ffmpeg: remove pointless if (nb_input_files)

It's required to be non-zero since
cc58300e30e7aea6acfecd72a04a3886b285cfd8

12 years agoffmpeg: merge input_files_ts_offset into input_files.
Anton Khirnov [Thu, 23 Jun 2011 17:30:15 +0000 (19:30 +0200)]
ffmpeg: merge input_files_ts_offset into input_files.

12 years agoffmpeg: merge input_codecs into input_streams.
Anton Khirnov [Thu, 23 Jun 2011 17:22:19 +0000 (19:22 +0200)]
ffmpeg: merge input_codecs into input_streams.

There's no point in keeping them separate.

12 years agoffmpeg: drop AV prefixes from struct names.
Anton Khirnov [Thu, 23 Jun 2011 17:14:08 +0000 (19:14 +0200)]
ffmpeg: drop AV prefixes from struct names.

Those are reserved for the libs.

12 years agoffmpeg: deprecate loop_input and loop_output options
Anton Khirnov [Sat, 18 Jun 2011 07:49:24 +0000 (09:49 +0200)]
ffmpeg: deprecate loop_input and loop_output options

They were replaced by (de)muxer private options.

12 years agogif: add loop private option.
Anton Khirnov [Sat, 18 Jun 2011 07:47:46 +0000 (09:47 +0200)]
gif: add loop private option.

Deprecate AVFormatContext.loop_output.

12 years agoimg2: add loop private option.
Anton Khirnov [Sat, 18 Jun 2011 07:28:49 +0000 (09:28 +0200)]
img2: add loop private option.

Deprecate AVFormatContext.loop_input.