ffmpeg.git
6 years agolavc: add AV prefix to codec ids.
Anton Khirnov [Sun, 5 Aug 2012 08:36:55 +0000 (10:36 +0200)]
lavc: add AV prefix to codec ids.

6 years agobuild: fix library installation on cygwin
Yaakov Selkowitz [Mon, 6 Aug 2012 23:32:48 +0000 (00:32 +0100)]
build: fix library installation on cygwin

This installs libraries using the proper names and locations,
generates an import lib for the DLL, and drops no longer needed
linker flags.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agompc8: add a flush function
Justin Ruggles [Mon, 6 Aug 2012 17:06:25 +0000 (13:06 -0400)]
mpc8: add a flush function

Ensures that the next frame decoded after seeking will be decoded as a
keyframe.

6 years agompc8: set packet duration and stream start time instead of tracking frames
Justin Ruggles [Mon, 6 Aug 2012 16:38:57 +0000 (12:38 -0400)]
mpc8: set packet duration and stream start time instead of tracking frames

Fixes mpc8 timestamps.

6 years agonuv: K&R formatting cosmetics
Diego Biurrun [Mon, 6 Aug 2012 13:49:51 +0000 (15:49 +0200)]
nuv: K&R formatting cosmetics

6 years agobuild: generalise rules and variable settings for av* programs
Mans Rullgard [Sat, 4 Aug 2012 18:04:08 +0000 (19:04 +0100)]
build: generalise rules and variable settings for av* programs

This simplifies adding extra flags for individual programs
and also allows more than one object file per program.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agonuv: check RTjpeg header for validity
Janne Grunau [Mon, 6 Aug 2012 11:59:04 +0000 (13:59 +0200)]
nuv: check RTjpeg header for validity

CC: libav-stable@libav.org
6 years agoRevert "nuv: check per-frame header for validity."
Janne Grunau [Mon, 6 Aug 2012 11:50:51 +0000 (13:50 +0200)]
Revert "nuv: check per-frame header for validity."

The check is bogus since the nuv frameheader is already skipped
and the (decompressed) RTjpeg header is checked.

This reverts commit f6afacdb3b708720c9fb85984b4f7fdbca2b2036.

CC: libav-stable@libav.org
6 years agoimc: remove unused field IMCContext.one_div_log2
Mans Rullgard [Sun, 5 Aug 2012 21:36:09 +0000 (22:36 +0100)]
imc: remove unused field IMCContext.one_div_log2

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoimc: fix size of a memset()
Mans Rullgard [Sun, 5 Aug 2012 21:32:20 +0000 (22:32 +0100)]
imc: fix size of a memset()

IMCContext was changed from an array to a pointer in 66b84e4,
but this memset() was not updated.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoimc: remove empty if() block
Mans Rullgard [Sun, 5 Aug 2012 17:32:03 +0000 (18:32 +0100)]
imc: remove empty if() block

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agofate: simplify variable setting filter.mak
Mans Rullgard [Sun, 5 Aug 2012 11:06:19 +0000 (12:06 +0100)]
fate: simplify variable setting filter.mak

This removes some needless indirection and duplication.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agolavf: Declare an AVRational struct without a struct literal
Martin Storsjö [Sun, 5 Aug 2012 19:55:21 +0000 (22:55 +0300)]
lavf: Declare an AVRational struct without a struct literal

At this place, the normal way of initializing a struct works
fine, there's no need for a struct literal.

Signed-off-by: Martin Storsjö <martin@martin.st>
6 years agox86: h264_idct: Rename x264_add8x4_idct_sse2 --> h264_add8x4_idct_sse2
Diego Biurrun [Tue, 31 Jul 2012 22:17:43 +0000 (00:17 +0200)]
x86: h264_idct: Rename x264_add8x4_idct_sse2 --> h264_add8x4_idct_sse2

6 years agorational: add av_inv_q() returning the inverse of an AVRational
Mans Rullgard [Sun, 29 Jul 2012 13:58:53 +0000 (14:58 +0100)]
rational: add av_inv_q() returning the inverse of an AVRational

This allows simplifying a few expressions.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agodpx: Make start offset unsigned
Derek Buitenhuis [Sat, 4 Aug 2012 23:41:20 +0000 (19:41 -0400)]
dpx: Make start offset unsigned

Some corrupted files would end up with a negative offset,
and segfault.

Fixes bug #177.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agolavfi: properly signal out-of-memory error in ff_filter_samples
Reinhard Tartler [Sat, 4 Aug 2012 13:24:15 +0000 (15:24 +0200)]
lavfi: properly signal out-of-memory error in ff_filter_samples

Found with a clang-scan report on http://fate.libav.org/csa/

6 years agocosmetics: Fix a few switched periods and linebreaks
Derek Buitenhuis [Sat, 4 Aug 2012 19:40:35 +0000 (15:40 -0400)]
cosmetics: Fix a few switched periods and linebreaks

Based on a patch by Piotr Bandurski.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agozerocodec: Fix memleak in decode_frame
Derek Buitenhuis [Sat, 4 Aug 2012 02:13:43 +0000 (22:13 -0400)]
zerocodec: Fix memleak in decode_frame

If there was a failure inflating, or reinitializing
the zstream, the current frame's buffer would be lost.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agozerocodec: Cosmetics
Derek Buitenhuis [Sat, 4 Aug 2012 01:41:24 +0000 (21:41 -0400)]
zerocodec: Cosmetics

Be consistent with error messages and code formatting.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agolavr: fix handling of custom mix matrices
Justin Ruggles [Sun, 29 Jul 2012 23:22:19 +0000 (19:22 -0400)]
lavr: fix handling of custom mix matrices

Adds some validation for changing parameters after setting the matrix and
fixes a bug in the conversion path setup.

6 years agofate: force pix_fmt in lagarith-rgb32 test
Mans Rullgard [Sat, 4 Aug 2012 11:24:23 +0000 (12:24 +0100)]
fate: force pix_fmt in lagarith-rgb32 test

This makes big and little endian systems use the same output
format.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agofate: add tests for lagarith lossless video codec.
Ronald S. Bultje [Fri, 3 Aug 2012 03:45:29 +0000 (20:45 -0700)]
fate: add tests for lagarith lossless video codec.

Based on patch by Oana Stratulat <oanaandreeastratulat@gmail.com>.

6 years agoARMv6: vp8: fix stack allocation with Apple's assembler
Mans Rullgard [Thu, 2 Aug 2012 22:18:08 +0000 (23:18 +0100)]
ARMv6: vp8: fix stack allocation with Apple's assembler

In the GNU assembler, a relational expression, bizarrely, has the
value -1 if true, whereas in Apple's it is +1.  This patch makes
sure the correct expression is used in both cases.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoARM: vp56: allow inline asm to build with clang
Mans Rullgard [Thu, 2 Aug 2012 21:53:47 +0000 (22:53 +0100)]
ARM: vp56: allow inline asm to build with clang

The clang integrated assembler does not support pre-UAL syntax,
while gcc requires pre-UAL syntax for ARM code.  A patch[1] for
clang to support the old syntax as well has been ignored since
January.

This patch chooses the syntax appropriate for each compiler,
allowing both to build the code.  Notably, this change allows
building for iphone with the latest Apple Xcode update.

[1] http://llvm.org/bugs/show_bug.cgi?id=11855

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agofft: 3dnow: fix register name typo in DECL_IMDCT macro
Ronald S. Bultje [Fri, 3 Aug 2012 21:59:16 +0000 (14:59 -0700)]
fft: 3dnow: fix register name typo in DECL_IMDCT macro

Signed-off-by: Diego Biurrun <diego@biurrun.de>
6 years agox86: dct32: port to cpuflags
Diego Biurrun [Wed, 1 Aug 2012 17:28:08 +0000 (19:28 +0200)]
x86: dct32: port to cpuflags

6 years agox86: build: replace mmx2 by mmxext
Diego Biurrun [Sun, 8 Jul 2012 16:42:12 +0000 (18:42 +0200)]
x86: build: replace mmx2 by mmxext

Refactoring mmx2/mmxext YASM code with cpuflags will force renames.
So switching to a consistent naming scheme beforehand is sensible.
The name "mmxext" is more official and widespread and also the name
of the CPU flag, as reported e.g. by the Linux kernel.

6 years agoRevert "wmapro: prevent division by zero when sample rate is unspecified"
Ronald S. Bultje [Fri, 3 Aug 2012 19:06:38 +0000 (12:06 -0700)]
Revert "wmapro: prevent division by zero when sample rate is unspecified"

This reverts commit 36936080239ba9e162f9f5d0f722ce579cb606cf. It was
already applied; no idea why it didn't error out while re-applying it.

6 years agowmapro: prevent division by zero when sample rate is unspecified
Sean McGovern [Thu, 2 Aug 2012 19:37:28 +0000 (15:37 -0400)]
wmapro: prevent division by zero when sample rate is unspecified

This fixes Bugzilla #327:

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agolagarith: fix color plane inversion for YUY2 output.
Ronald S. Bultje [Fri, 3 Aug 2012 03:46:08 +0000 (20:46 -0700)]
lagarith: fix color plane inversion for YUY2 output.

6 years agolagarith: pad RGB buffer by 1 byte.
Ronald S. Bultje [Fri, 3 Aug 2012 03:46:09 +0000 (20:46 -0700)]
lagarith: pad RGB buffer by 1 byte.

For left HFYU prediction, we predict from the buffer buf+1 using 8- or
16-byte reads. This means that aligning the buffer by 16 bytes is in
itself not sufficient, because if the width itself is 16- or 8-byte
aligned, the buffer will not be padded, and thus a read of size 16 at
buf+1 will overflow boundaries at the right edge. Padding the buffer by
1 byte is sufficient to not overflow its boundaries.

Fixes bug 342.

6 years agodsputil: make add_hfyu_left_prediction_sse4() support unaligned src.
Ronald S. Bultje [Fri, 3 Aug 2012 14:58:26 +0000 (16:58 +0200)]
dsputil: make add_hfyu_left_prediction_sse4() support unaligned src.

This makes add_hfyu_left_prediction_sse4() handle sources that are not
16-byte aligned in its own function rather than by proxying the call to
add_hfyu_left_prediction_ssse3(). This fixes a crash on Win64, since the
sse4 version clobberes xmm6, but the ssse3 version (which uses MMX regs)
does not restore it, thus leading to XMM clobbering and RSP being off.

Fixes bug 342.

6 years agovc1dec: Remove separate scaling function for interlaced field MVs
Mashiat Sarker Shakkhar [Fri, 3 Aug 2012 14:53:36 +0000 (20:53 +0600)]
vc1dec: Remove separate scaling function for interlaced field MVs

The scaling process for obtaining direct MVs from co-located field MVs
is the same for interlaced field and progressive pictures.

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
6 years agovc1dec: Invoke edge_emulation regardless of MV precision
Mashiat Sarker Shakkhar [Fri, 3 Aug 2012 14:53:35 +0000 (20:53 +0600)]
vc1dec: Invoke edge_emulation regardless of MV precision

In VC-1 interlaced field pictures, chroma motion vectors can extend beyond
picture boundary even if luma vectors are bounded. The problem shows up
only for hpel interpolated MVs, and may be due to the way motion vectors
are scaled / cropped.

Thanks to Konstantin Shishkov for suggesting the fix. This fixes
long-known segfaults in MC-VC1.ts from videolan streams archive.

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
6 years agox86: Use consistent 3dnowext function and macro name suffixes
Diego Biurrun [Wed, 1 Aug 2012 13:31:43 +0000 (15:31 +0200)]
x86: Use consistent 3dnowext function and macro name suffixes

Currently there is a wild mix of 3dn2/3dnow2/3dnowext.  Switching to
"3dnowext", which is a more common name of the CPU flag, as reported
e.g. by the Linux kernel, unifies this.

6 years agog723_1: scale output as supposed for the case with postfilter disabled
Kostya Shishkov [Thu, 2 Aug 2012 17:34:53 +0000 (19:34 +0200)]
g723_1: scale output as supposed for the case with postfilter disabled

6 years agog723_1: increase excitation storage by 4
Kostya Shishkov [Thu, 2 Aug 2012 17:15:51 +0000 (19:15 +0200)]
g723_1: increase excitation storage by 4

Fixed codebook mode in 5300 rate may write up to SUBFRAME_LEN + 4 and
that is considered normal by the reference decoder. Without that additional
padding it might overwrite first elements of LPC history.

6 years agog723_1: fix upper bound parameter from inverse maximum autocorrelation
Kostya Shishkov [Thu, 2 Aug 2012 05:54:16 +0000 (07:54 +0200)]
g723_1: fix upper bound parameter from inverse maximum autocorrelation

6 years agog723_1: make scale_vector() behave like the reference
Kostya Shishkov [Thu, 2 Aug 2012 05:50:49 +0000 (07:50 +0200)]
g723_1: make scale_vector() behave like the reference

6 years agog723_1: fix off-by-one error in normalize_bits()
Kostya Shishkov [Thu, 2 Aug 2012 05:48:08 +0000 (07:48 +0200)]
g723_1: fix off-by-one error in normalize_bits()

6 years agog723_1: save/restore excitation with offset to store LPC history
Kostya Shishkov [Thu, 2 Aug 2012 05:42:44 +0000 (07:42 +0200)]
g723_1: save/restore excitation with offset to store LPC history

The same buffer with saved data is used later in LPC reconstruction, so
it should have some head space for LPC history.

6 years agowmapro: prevent division by zero when sample rate is unspecified
Sean McGovern [Thu, 2 Aug 2012 19:37:28 +0000 (15:37 -0400)]
wmapro: prevent division by zero when sample rate is unspecified

This fixes Bugzilla #327:

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
6 years agox86: proresdsp: improve SIGNEXTEND macro comments
Diego Biurrun [Sun, 15 Jul 2012 22:01:00 +0000 (00:01 +0200)]
x86: proresdsp: improve SIGNEXTEND macro comments

6 years agox86: h264dsp: K&R formatting cosmetics
Diego Biurrun [Tue, 31 Jul 2012 11:18:20 +0000 (13:18 +0200)]
x86: h264dsp: K&R formatting cosmetics

6 years agoLICENSE: Document all GPL files
Derek Buitenhuis [Thu, 2 Aug 2012 16:09:09 +0000 (16:09 +0000)]
LICENSE: Document all GPL files

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agox86: fft: fix imdct_half() for AVX
Ronald S. Bultje [Thu, 2 Aug 2012 17:15:46 +0000 (12:15 -0500)]
x86: fft: fix imdct_half() for AVX

Some calculations were changed in b6a3849 to use mmsize, which was not correct
for the AVX version, which uses INIT_YMM and therefore has mmsize == 32.

Fixes Bug 341.

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
6 years agortmppkt: Add missing libavcodec/bytestream.h include.
Alex Rønne Petersen [Thu, 2 Aug 2012 16:24:01 +0000 (16:24 +0000)]
rtmppkt: Add missing libavcodec/bytestream.h include.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agortmp: add functions for reading AMF values
Jordi Ortiz [Wed, 1 Aug 2012 09:25:19 +0000 (11:25 +0200)]
rtmp: add functions for reading AMF values

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
6 years agovc1dec: remove useless #include simple_idct.h
Mans Rullgard [Wed, 1 Aug 2012 16:15:42 +0000 (17:15 +0100)]
vc1dec: remove useless #include simple_idct.h

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agodct-test: always link with aandcttab.o
Mans Rullgard [Wed, 1 Aug 2012 13:32:19 +0000 (14:32 +0100)]
dct-test: always link with aandcttab.o

This allows building dct-test even if aandcttab.o is not pulled in
by any enabled codec.  The DCT with which these tables are used does
not use them directly, so building it without the tables is possible.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agovp8: pack struct VP8ThreadData more efficiently
Mans Rullgard [Wed, 1 Aug 2012 13:01:08 +0000 (14:01 +0100)]
vp8: pack struct VP8ThreadData more efficiently

Reordering the members in this struct reduces the holes required
to maintain alignment.  With this order, the only remaining, and
unavoidable, hole is 3 bytes following left_nnz.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agox86: remove libmpeg2 mmx(ext) idct functions
Mans Rullgard [Wed, 1 Aug 2012 12:16:23 +0000 (13:16 +0100)]
x86: remove libmpeg2 mmx(ext) idct functions

These functions are not faster than other mmx implementations on
any hardware I have been able to test on, and they are horribly
inaccurate.  There is thus no reason to ever use them.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoeamad: Use dsputils instead of a custom bswap16_buf
Derek Buitenhuis [Wed, 1 Aug 2012 19:22:50 +0000 (19:22 +0000)]
eamad: Use dsputils instead of a custom bswap16_buf

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agoCanopus Lossless decoder
Derek Buitenhuis [Wed, 1 Aug 2012 17:46:53 +0000 (17:46 +0000)]
Canopus Lossless decoder

At the moment it only does BGR24, but I plan to add the rest after.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
6 years agodca: Switch dca_sample_rates to avpriv_ prefix; it is used across libs
Diego Biurrun [Wed, 1 Aug 2012 09:12:08 +0000 (11:12 +0200)]
dca: Switch dca_sample_rates to avpriv_ prefix; it is used across libs

6 years agoARM: use =const syntax instead of explicit literal pools
Mans Rullgard [Tue, 31 Jul 2012 23:59:35 +0000 (00:59 +0100)]
ARM: use =const syntax instead of explicit literal pools

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoARM: use standard syntax for all LDRD/STRD instructions
Mans Rullgard [Tue, 31 Jul 2012 22:58:58 +0000 (23:58 +0100)]
ARM: use standard syntax for all LDRD/STRD instructions

The standard syntax requires two destination registers for
LDRD/STRD instructions.  Some versions of the GNU assembler
allow using only one with the second implicit, others are
more strict.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agofft: port FFT/IMDCT 3dnow functions to yasm, and disable on x86-64.
Ronald S. Bultje [Wed, 1 Aug 2012 03:23:39 +0000 (20:23 -0700)]
fft: port FFT/IMDCT 3dnow functions to yasm, and disable on x86-64.

64-bit CPUs always have SSE available, thus there is no need to compile
in the 3dnow functions. This results in smaller binaries.

6 years agodct-test: allow to compile without HAVE_INLINE_ASM.
Ronald S. Bultje [Wed, 1 Aug 2012 03:30:29 +0000 (20:30 -0700)]
dct-test: allow to compile without HAVE_INLINE_ASM.

6 years agox86/dsputilenc: bury inline asm under HAVE_INLINE_ASM.
Ronald S. Bultje [Thu, 26 Jul 2012 03:30:19 +0000 (20:30 -0700)]
x86/dsputilenc: bury inline asm under HAVE_INLINE_ASM.

6 years agodca: Move tables used outside of dcadec.c to a separate file.
Diego Biurrun [Tue, 31 Jul 2012 18:09:23 +0000 (20:09 +0200)]
dca: Move tables used outside of dcadec.c to a separate file.

6 years agodca: Rename dca.c ---> dcadec.c
Diego Biurrun [Tue, 31 Jul 2012 18:00:35 +0000 (20:00 +0200)]
dca: Rename dca.c ---> dcadec.c

This will allow adding dca.c with tables used from other files.

6 years agox86: h264dsp: Remove unused variable ff_pb_3_1
Diego Biurrun [Tue, 31 Jul 2012 15:12:51 +0000 (17:12 +0200)]
x86: h264dsp: Remove unused variable ff_pb_3_1

6 years agoapetag: change a forgotten return to return 0
Anton Khirnov [Tue, 31 Jul 2012 19:32:25 +0000 (21:32 +0200)]
apetag: change a forgotten return to return 0

6 years agompc8: return more meaningful error codes.
Anton Khirnov [Mon, 30 Jul 2012 05:29:34 +0000 (07:29 +0200)]
mpc8: return more meaningful error codes.

6 years agompc: return more meaningful error codes.
Anton Khirnov [Sun, 29 Jul 2012 20:19:35 +0000 (22:19 +0200)]
mpc: return more meaningful error codes.

6 years agowv,mpc8: don't return apetag data in packets.
Anton Khirnov [Mon, 30 Jul 2012 05:28:35 +0000 (07:28 +0200)]
wv,mpc8: don't return apetag data in packets.

6 years agortmp: do not warn about receiving metadata packets
Samuel Pitoiset [Thu, 26 Jul 2012 19:10:09 +0000 (21:10 +0200)]
rtmp: do not warn about receiving metadata packets

They are managed in get_packet()

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
6 years agox86: h264dsp: Adjust YASM #ifdefs
Diego Biurrun [Tue, 31 Jul 2012 10:32:29 +0000 (12:32 +0200)]
x86: h264dsp: Adjust YASM #ifdefs

This fixes compilation with YASM disabled.

6 years agox86: yadif: Mark mmxext optimizations as such
Diego Biurrun [Tue, 31 Jul 2012 10:17:02 +0000 (12:17 +0200)]
x86: yadif: Mark mmxext optimizations as such

The yadif mmx optimizations contain the pmaxsw and pmaxub mmxext
instructions, causing sigills on CPUs that do not support mmxext.

6 years agoh264: convert loop filter strength dsp function to yasm.
Ronald S. Bultje [Sat, 28 Jul 2012 17:11:00 +0000 (10:11 -0700)]
h264: convert loop filter strength dsp function to yasm.

This completes the conversion of h264dsp to yasm; note that h264 also
uses some dsputil functions, most notably qpel. Performance-wise, the
yasm-version is ~10 cycles faster (182->172) on x86-64, and ~8 cycles
faster (201->193) on x86-32.

6 years agoImprove descriptiveness of a number of codec and container long names
Diego Biurrun [Tue, 24 Jul 2012 01:23:48 +0000 (03:23 +0200)]
Improve descriptiveness of a number of codec and container long names

6 years agoavformat: Drop pointless "format" from container long names
Diego Biurrun [Tue, 24 Jul 2012 21:51:41 +0000 (23:51 +0200)]
avformat: Drop pointless "format" from container long names

6 years agoswscale: bury one more piece of inline asm under HAVE_INLINE_ASM.
Ronald S. Bultje [Thu, 26 Jul 2012 02:27:03 +0000 (19:27 -0700)]
swscale: bury one more piece of inline asm under HAVE_INLINE_ASM.

6 years agowv: K&R formatting cosmetics
Anton Khirnov [Sat, 28 Jul 2012 10:42:25 +0000 (12:42 +0200)]
wv: K&R formatting cosmetics

6 years agoconfigure: Add missing descriptions to help output
Diego Biurrun [Fri, 27 Jul 2012 11:01:04 +0000 (13:01 +0200)]
configure: Add missing descriptions to help output

6 years agoh264_ps: declare array of colorspace strings on its own line.
Ronald S. Bultje [Sun, 29 Jul 2012 04:45:24 +0000 (21:45 -0700)]
h264_ps: declare array of colorspace strings on its own line.

6 years agofate: amix: specify f32 sample format for comparison
Mans Rullgard [Sun, 29 Jul 2012 17:11:27 +0000 (18:11 +0100)]
fate: amix: specify f32 sample format for comparison

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agotiny_psnr: support 32-bit float samples
Mans Rullgard [Sun, 29 Jul 2012 17:10:36 +0000 (18:10 +0100)]
tiny_psnr: support 32-bit float samples

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoeamad/eatgq/eatqi: call special EA IDCT directly
Mans Rullgard [Sun, 29 Jul 2012 12:09:10 +0000 (13:09 +0100)]
eamad/eatgq/eatqi: call special EA IDCT directly

These decoders use a special non-MPEG2 IDCT.  Call it directly
instead of going through dsputil.  There is never any reason
to use a regular IDCT with these decoders or to use the EA IDCT
with other codecs.

This also fixes the bizarre situation of eamad and eatqi decoding
incorrectly if eatgq is disabled.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoeamad: remove use of MpegEncContext
Mans Rullgard [Fri, 27 Jul 2012 16:00:38 +0000 (17:00 +0100)]
eamad: remove use of MpegEncContext

There is no sense in pulling in this monster struct just for
a handful of fields.  The code does not call any functions
expecting an MpegEncContext.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agompegvideo: remove unnecessary inclusions of faandct.h
Mans Rullgard [Sun, 29 Jul 2012 12:20:54 +0000 (13:20 +0100)]
mpegvideo: remove unnecessary inclusions of faandct.h

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agoaf_asyncts: avoid overflow in out_size with large delta values
Justin Ruggles [Sat, 28 Jul 2012 17:12:38 +0000 (13:12 -0400)]
af_asyncts: avoid overflow in out_size with large delta values

6 years agoaf_asyncts: add first_pts option
Justin Ruggles [Sat, 28 Jul 2012 16:57:05 +0000 (12:57 -0400)]
af_asyncts: add first_pts option

This allows for padding/trimming at the start of stream. By default, no
assumption is made about the first frame's expected pts, so no padding or
trimming is done.

6 years agoflvdec: remove spurious use of stream id
Luca Barbato [Thu, 26 Jul 2012 14:18:11 +0000 (16:18 +0200)]
flvdec: remove spurious use of stream id

We match streams by codec id now.

6 years agolavf: deprecate r_frame_rate.
Anton Khirnov [Tue, 26 Jun 2012 11:10:01 +0000 (13:10 +0200)]
lavf: deprecate r_frame_rate.

According to its description, it is supposed to be the LCM of all the
frame durations. The usability of such a thing is vanishingly small,
especially since we cannot determine it with any amount of reliability.
Therefore get rid of it after the next bump.

Replace it with the average framerate where it makes sense.

FATE results for the wtv and xmv demux tests change. In the wtv case
this is caused by the file being corrupted (or possibly badly cut) and
containing invalid timestamps. This results in lavf estimating the
framerate wrong and making up wrong frame durations.
In the xmv case the file contains pts jumps, so again the estimated
framerate is far from anything sane and lavf again makes up different
frame durations.

In some other tests lavf starts making up frame durations from different
frame.

6 years agolavf: round estimated average fps to a "standard" fps.
Anton Khirnov [Fri, 27 Jul 2012 12:04:07 +0000 (14:04 +0200)]
lavf: round estimated average fps to a "standard" fps.

6 years agolavf: use dts difference instead of AVPacket.duration in find_stream_info()
Anton Khirnov [Thu, 28 Jun 2012 13:49:51 +0000 (15:49 +0200)]
lavf: use dts difference instead of AVPacket.duration in find_stream_info()

AVPacket.duration is mostly made up and thus completely useless, this is
especially true for video streams.
Therefore use dts difference for framerate estimation and
the max_analyze_duration check.

The asyncts test now needs -analyzeduration, because the default is 5
seconds and the audio stream in the sample appears at ~10 seconds.

6 years agoavf: introduce nobuffer option
Luca Barbato [Thu, 26 Jul 2012 23:35:12 +0000 (16:35 -0700)]
avf: introduce nobuffer option

Useful in cases where a significant analyzeduration is
still needed, while minimizing buffering before output.

An example is processing low-latency streams where all
media types won't necessarily come in if the
analyzeduration is small.

Additional changes by Josh Allmann <joshua.allmann@gmail.com>

Signed-off-by: Anton Khirnov <anton@khirnov.net>
6 years agofate: make yadif tests consistent across systems
Mans Rullgard [Sat, 28 Jul 2012 17:15:56 +0000 (18:15 +0100)]
fate: make yadif tests consistent across systems

MMX-enabled systems by default use some dsputil functions differing
from the C versions.  Adding these flags ensures accurate ones are
used everywhere.

Signed-off-by: Mans Rullgard <mans@mansr.com>
6 years agovf_hqdn3d: support 9 and 10bit colordepth
Loren Merritt [Fri, 27 Jul 2012 01:42:05 +0000 (01:42 +0000)]
vf_hqdn3d: support 9 and 10bit colordepth

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agovf_hqdn3d: reduce intermediate precision
Loren Merritt [Thu, 26 Jul 2012 22:51:15 +0000 (22:51 +0000)]
vf_hqdn3d: reduce intermediate precision

11% faster on penryn, 7% on sandybridge, 5% on bulldozer
Negligible change to output.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agovf_hqdn3d: simplify and optimize
Loren Merritt [Thu, 26 Jul 2012 22:51:14 +0000 (22:51 +0000)]
vf_hqdn3d: simplify and optimize

14% faster on penryn, 2% on sandybridge, 9% on bulldozer

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agofactor identical ff_inplace_start_frame out of two filters
Loren Merritt [Thu, 26 Jul 2012 22:51:12 +0000 (22:51 +0000)]
factor identical ff_inplace_start_frame out of two filters

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agovf_hqdn3d: cosmetics
Loren Merritt [Thu, 26 Jul 2012 22:51:13 +0000 (22:51 +0000)]
vf_hqdn3d: cosmetics

Change code style to match the rest of libav.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agoavprobe/avconv: fix tentative declaration compile errors on MSVS.
Ronald S. Bultje [Thu, 26 Jul 2012 04:23:23 +0000 (04:23 +0000)]
avprobe/avconv: fix tentative declaration compile errors on MSVS.

6 years agoh264_idct_10bit: port x86 assembly to cpuflags.
Ronald S. Bultje [Sat, 28 Jul 2012 15:01:12 +0000 (08:01 -0700)]
h264_idct_10bit: port x86 assembly to cpuflags.

6 years agox86inc: clip num_args to 7 on x86-32.
Loren Merritt [Fri, 27 Jul 2012 16:33:41 +0000 (09:33 -0700)]
x86inc: clip num_args to 7 on x86-32.

This allows us to unconditionally set the cglobal num_args
parameter to a bigger value, thus making writing yasm code
even easier than before.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
6 years agox86inc: sync to latest version from x264.
Ronald S. Bultje [Fri, 27 Jul 2012 03:26:17 +0000 (20:26 -0700)]
x86inc: sync to latest version from x264.