Reimplement stream probe try #2
Signed-off-by: Michael Niedermayer <email@example.com>
Partial rewrite stream probe code.Fix issue2312.Signed-off-by: Michael Niedermayer <firstname.lastname@example.org>
Dont be so picky on decoding pps in extradata.Fixes issue2517
Fix yuvj420p scaling artefact, issue1108.
Correctly implement: commit c0ec9918b066c6bfca8613ac75b589bd3e405971 Author: Måns Rullgård <email@example.com> Date: Tue Aug 24 17:47:05 2010 +0000 Remove global mm_flags variable
Originally committed as revision 24909 to svn://svn.ffmpeg.org/ffmpeg/trunk...
Add sign off recommandition and definition of meaning.Signed-off-by: Michael Niedermayer <firstname.lastname@example.org>
Reword make fate patch checklist entry
Add or elaborate on some documentation comments in avcodec.h.
patch checklist:git format-patchSigned-off-by: Michael Niedermayer <email@example.com>
patch checklist:uggest fate instead of regression testsSigned-off-by: Michael Niedermayer <firstname.lastname@example.org>
patch checklist:Remove pointless checkheaders comment, its just a waste of timeSigned-off-by: Michael Niedermayer <email@example.com>
lavf: inspect more frames for fps when container time base is coarse
As per issue2629, most 23.976fps matroska H.264 files are incorrectlydetected as 24fps, as the matroska timestamps usually have onlymillisecond precision.
Fix that by doubling the amount of timestamps inspected for frame rate...
Merge remote branch 'qatar/master'
Revert "configure: use dlltools instead of lib.exe" Reason, is this was requested by the windows experts / seems dlltool causes alot of problems
This reverts commit ec10a9ab461b26b96eff7bbbb8623f42d8ee04ad.
Fixed-point MDCT with 32-bit unscaled output
Signed-off-by: Mans Rullgard <firstname.lastname@example.org>
yadif:Fix assert() failure
dpx encoder: support 10-bit packing method
dpx decoder: read sample aspect ratio
H264: fix slice end error detectionSigned-off-by: Michael Niedermayer <email@example.com>
h264: change a few comments into error messagesSigned-off-by: Michael Niedermayer <firstname.lastname@example.org>
lavc: deprecate rate_emu
It's not used anywhere and apparrently hasn't been for a very long time.
lavc: mark hurry_up for removal on next major bump
It has been deprecated for about five years, skip_idct/skip_frame shouldbe used instead.
parser: mark av_parser_parse() for removal on next major bump
Fixes build with the bump, since the prototype is already marked forremoval.
lavc: add missing audioconvert includes
jvdec: don't use deprecated CODEC_TYPE_*/PKT_FLAG_KEY
Fix stream mapping regression
drawtext: extend/clarify examples in the filter description
drawtext: fix wrong reference in boxcolor option description
drawtext: add documentation for the shadow parameters
drawtext: allow negative shadow offset values
Timeout TCP open() after 5 seconds.
Fix compilation with --disable-vdpau
Make a plausibility check when setting channel layout for TrueHD.
TrueHD supports more channels than FFmpeg, so a valid samplecould set the channel layout to a value that represents lesschannels than the sample actually consists of.
Match the documentation of -s with the current implementation.
ac3enc: ARM optimised ac3_compute_matissa_size
ac3: armv6 optimised bit_alloc_calc_bap
fate: simplify fft test rules
avio: document avio_alloc_context.
lavf: make compute_chapters_end less picky.
In particular, now it assumes thata) chapters are chronologically orderedb) chapters have the same timebasesc) duration of the stream is knownand asserts if any of these is not met.
Make it properly deal with harsher conditions....
Fix doxy-style for comment ("///>" should be "///<" instead).
sierravmd: fix Indeo3 videos
Signed-off-by: Anton Khirnov <email@example.com>
Allow selecting VDPAU, XvMC via get_format for MPEG-2
To avoid further special-casing, dummy AVHWAccels are addedfor VDPAU.It should be possible to move VDPAU completely to AVHWAccel later.
FFT: simplify fft8()
Part of the fft8() function corresponds to the BUTTERFLIES macro,so use it.
Another aspect ratio fix try. This leaves the setdar addition at the end (preferred by people).
Move flv format #defines to common header so they can be used by the encoder too.
fate: add fixed-point fft/mdct tests
Fixed-point support in fft-test
Increase ffmpeg video encoding buffer size to permit encoding of DPX images
The header generated by the DPX encoder is 1664 bytes.
Add .dpx to the list of supported file extensions for the image2 muxer
DPX image encoder
Cleanup flv keyframe index reading and add some saftey checks.
Fix undefined behavior in ffv1 with insane widths.The new tables is large enough to prevent this together with our image size checks.
ape: check that number of seektable entries is equal to number of frames
Merge remote-tracking branch 'newdev/master'
Insert setdar at the begin of a filter chain instead the end.This fixes -vf setsar
Revert "Remove support for stripping executables"
People did not like this change.
This reverts commit e0be794a71be25a350f68287ee291b4942bcb7d8.
make image2 muxer/demuxer support sequence of individual rawvideo files
img2: stop using CODEC_ID_RAWVIDEO to trigger processing of independant y/u/v image files
This is requireed, as img2 will be modified soon to support .raw image file sequences.
mpegts: propagate avio EOF in read_packet()
Allows distinguishing between EOF and IO error in read_packet return code.
Signed-off-by: Jindrich Makovicka <firstname.lastname@example.org>Signed-off-by: Anton Khirnov <email@example.com>
ALSA: fix a stupid bad use of av_freep.
It caused a segfault when channel reordering was in use and the last chunkof data was not silent.Spotted and diagnosed by Carl Eugen Hoyos.
configure: Initial support for --target-os=symbian
Signed-off-by: Martin Storsjö <firstname.lastname@example.org>
Fixed-point FFT and MDCT
Include dependencies for test programs
This makes sure the various *-test programs are correctly rebuiltwhen necessary.
ac3enc: simplify sym_quant()
These expressions are equivalent since levels is always odd, andoverflow is impossible due to the constraints set by the assert().
flvdec: read index stored in the 'keyframes' tag.
'keyframes' metatag is not part of the standard, it is justconvention to use such kind of metatag information for indexing.Structure is following, it allows to have it inconsistent:keyframes: times (array):...
mov: Add support for zero-sized stsc runs.
A zero sized stsc run doesn't make a lot of sense but the spec does notprohibit them and MPlayer VLC demuxers support them.
Revert "Do no modify terminal parameters using termios.h"
This reverts commit cb48e245e6e770f146220fac0a8bd4dc1a5e006c.
Reason being we like pressing "q" to quit ffmpeg.
In ipod/mov/mp4 muxer, always write esds descriptor length using 4 bytes,ipod shuffle doesn't support anything else.
rtsp: Use GET_PARAMETER for keep-alive for generic RTSP servers
According to the RFC, GET_PARAMETER should be used forthis, and according to a report from Tim Ouellette,OPTIONS doesn't work for keeping the connection alive for someservers. Also, live555 uses GET_PARAMETER for this purpose....
gradfun: add notice from the MPlayer manual
Add notice telling that the filter is designed for playback only.
documentation: add setdar and setsar description to filters.texi
graphparser: add a NULL check on the argument passed to strstr
Fix crash in create_filter() which occurrs if a scale filter with noargs is provided.
Fix typo from e3a84542.
Fix dnxhd custom options, add AVClass in private context
In mov muxer, fix yuv range in avid atoms used by dnxhd.
mlp_parse.c: set AVCodecContext channel_layout
The channel layout isn't getting set for mlp and truehd audio.
APIChanges: mark the place where 0.6 was branched.
Remove a memory allocation and the associated memcpy.
h264: permit h264 decoder to be compiled when svq3 decoder is disabled
SWSCALE: Increase maximim with from 5120 to 21504
Heres a benchmark
VOFW 5120 VOFW 21504 Note10438.8 10344.2 timex ffmpeg -y -i tulip2.mp4 tulip2.yuv25611.9 24256.9 timex ffmpeg -y -itulip2.mp4 -s 1272x724 tulip2.yuv24485.7 26006.6...
A few points for the changelog
vaapi: allow build against older VA-API 0.31.
vaapi: require version >= 0.32.0.
vaapi: move configure check down.
Fix compilation with win32 threads.after 'unbreak avcodec_thread_init'
Also use alsa_reorder_s16_out_51() for AV_CH_LAYOUT_5POINT1.
Fix buffer overflow checks to avoid integer overflows.
avio: make get_checksum() internal.
avio: move ff_crc04C11DB7_update() from avio.h -> avio_internal.h
avio: make init_checksum() internal.
Add MxPEG decoder
Add support for picture_ptr field in MJpegDecodeContext
Move MJPEG's input buffer preprocessing in separate public function
Support reference picture defined by bitmask in MJPEG's SOS decoder
sndio bug fix
Since the code already supports both little- and big-endianaudio for recording, do not fail just because the endianness is notwhat we expect.
Signed-off-by: Ronald S. Bultje <email@example.com>