mpegvideo: add deprecated flags to the rc_strategy option

Forgotten with the commit which removed support for libxvid_rc.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

dcaenc: fix segfault when attempting to encode with invalid samplerate

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

mpegvideo: remove support for libxvid's RC system

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

avfilter/vf_colorchannelmixer: add planar rgb support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_colorchannelmixer: refactor code

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_colorbalance: fix off by one overflow

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_colorchannelmixer: add slice threading

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/flac_parser: Fix infinite loop

Fixes: crbug/827204
Reported-by: Frank Liberato <liberato@google.com>
Reviewed-by: Frank Liberato <liberato@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_colorbalance: add slice threading

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_colorbalance: add planar rgb support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_convolution: use already available dstride

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/videotoolbox: fix kVTCouldNotFindVideoDecoderErr trying to decode HEVC on iOS

Older iOS devices don't have a hardware HEVC decoder, but the
software decoder offered by VideoToolbox is well-optimized and
performs much better than the ffmpeg decoder.

Signed-off-by: Aman Gupta <aman@tmm1.net>

avcodec/videotoolbox: improve logging of decoder errors

Signed-off-by: Aman Gupta <aman@tmm1.net>

avfilter/vf_colorbalance: add 16bit depth support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/nvenc: move reconfig_encoder call inside of push/pop ctx

Also make it void, it must not fail the encode anyway.

avcodec/nvenc: add A53CC support

Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>

avcodec/nvenc: support dynamic bitrate changes

The patch enables dynamic bitrate through ReconfigureEncoder method
from nvenc API.
This is useful for live streaming in case of network congestion.

Signed-off-by: pkviet <pkv.stream@gmail.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>

avcodec/nvenc: support dynamic aspect ratio change

If there is input like DVB-T streams it can change aspect ratio
on-the-fly, so nvenc should respect this change and change aspect ratio
in encoder.

Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>

avcodec/nvenc: make hw_frames_ctx fully optional

avfilter/vf_libvmaf: the libvmaf filter tried to join on an invalid thread id

The thread id was invalid because it was not initialised
during the calls to init_complex_filtergraph.

This adds a flag to check for initialisation before trying to
peform the join.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Kevin Wheatley <kevin.j.wheatley@gmail.com>

avfilter/vf_lut3d: fix typo

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_edgedetect: add more formats support to canny mode

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_edgedetect: add planes option

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_edgedetect: add canny mode

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_convolution: add column/vertical mode

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_lut3d: add planar rgb support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/mediacodecdec: add workaround for buggy amlogic mpeg2 decoder

I tested the previous mediacodec changes on seven different Android
TV devices, with both mpeg2 and h264 content. All except one worked
as expected. The exception was the MiBox3 running Android 6.0.1,
where playback would freeze on a frame every few seconds. I tested
two other AMLogic devices with newer Android versions that did not
show the same problem. H264 decoding on the MiBox3 was also not affected,
so this workaround applies only to OMX.amlogic.mpeg2.decoder.awesome
on Android API22.

There is a rumor that Xiaomi is planning to release Android Oreo for
the MiBox3, so I will revisit in a few months to confirm whether this
is specific to os/driver version or the chipset used in that device.

Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>

avcodec/mediacodecdec: wait on first frame after input buffers are full

The output_buffer_count==0 special case is no longer required, and
can cause spurious EAGAIN to surface to the user when input buffers
are filled up. Since the caller now knows if the decoder is accepting
new input (via current_input_buffer>=0), let the wait parameter
control whether we block or not.

Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>

avcodec/mediacodecdec: restructure mediacodec_receive_frame

The new logic follows a recommendation by @rcombs to use
dequeueInputBuffer with a timeout of 0 as a way to detect
whether the codec wants more data. The dequeued buffer index is
kept in MediaCodecDecContext until it can be used next.

A similar technique is also used by the Google's official media
player Exoplayer: see MediaCodecRenderer.feedInputBuffer().

Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>

avcodec/mediacodec_wrapper: add helper to fetch SDK_INT

Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
Signed-off-by: Aman Gupta <aman@tmm1.net>

avformat: add vapoursynth wrapper

This can "demux" .vpy files. Autodetection of .vpy scripts is
intentionally not done, because it would be a major security issue. You
need to force the format, for example with "-f vapoursynth" for the
FFmpeg CLI tools.

Some minor code copied from other LGPL parts of FFmpeg.

I did not find a good way to test a few of the more obscure VS features,
like VFR nodes, compat pixel formats, or nodes with dynamic size/format
changes. These can be easily implemented on demand.

Adds SRCNN filter.

Signed-off-by: Pedro Arthur <bygrandao@gmail.com>

avfilter/vf_amplify: add more options for finer filtering

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avformat/hls: don't propagate deprecated "user-agent" AVOption

This code will print a warning if any user agent is set - even if the
API user used the proper non-deprecated "user_agent" option.

This change should not even break anything, because even if the user
sets the deprecated "user-agent" option, http.c copies it to the
"user_agent" option anyway.

avformat/matroskaenc: do not write timebase as framerate

If the API user doesn't set avg_frame_rate, matroskaenc will write the
current timebase as "default duration" for the video track. This makes
no sense, because the "default duration" implies the framerate of the
video. Since the timebase is forced to 1/1000, this will make the
resulting file claim 1000fps.

Drop it and don't write the element. It's optional, so it's better not
to write it if the framerate is unknown.

Strangely does not require FATE changes.

avcodec/h263dec: Document padding_bug_score heuristic used for wrong stuffing

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

vaapi_encode_h264: Take VAAPIEncodeH264Context::sei_needed as an ORed value

We may check priv->sei_needed only when writing SEI messages.

Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>

avfilter/drawutils: add support for full range

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/mpeg4videodec: Split decode_studio_vol_header() out of decode_studiovisualobject()

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mpeg4videodec: Move decode_studiovisualobject() parsing in the branch for visual object parsing

Fixes: runtime error: shift exponent -1 is negative
Fixes: 7510/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5024523356209152
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/yuv4mpegdec: simplify math

This one actually works with hd1080 y4m files when seeking backwards.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_neighbor: add >8 depth suppport

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_convolution: unbreak roberts filter

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_convolution: add horizontal/row mode

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_overlay: exclude nv12/nv21 formats from x86 asm check

They are yet to be supported,

Signed-off-by: Paul B Mahol <onemda@gmail.com>

lavf/dashenc: pass standards compliance value to the internal context

Enables one to test possibly nonstandard formats such as Opus or
FLAC in ISOBMFF, among other things.

This becomes much more useful if output segment format becomes an
option, or if the WebM segment feature gets removed.

lavf/dashenc: require experimental mode to be enabled for WebM

It has not ever been working and has not been validated,
Additionally, mention that the segment file names should be changed
to end with webm instead of m4s, which is utilized for ISOBMFF
fragments.

lavf/dashenc: don't call flush_init_segment before avformat_write_header

Fixes crash when muxing MKV-in-DASH

avfilter/vf_convolution: rewrite so it doesn't use temp buffers

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_overlay: add x86 SIMD

Specifically for yuv444, yuv422, yuv420 format when main stream has no alpha, and alpha
is straight.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

fftools/ffmpeg: properly initialize output stream field order

Fixes stream field order written by avformat_write_header when "top"
option is specified on the command-line.

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

fate/cbs: Always overwrite output in cbs fate tests

Before this, a failed test would keep failing because the output file
is not cleaned up and subsequent runs would refuse to overwrite it.

lavc/cbs: Add tests for VP9

Uses the same mechanism as other codecs - conformance test files are
passed through the metadata filter (which, with no options, reads the
input and writes it back) and the output verified to match the input.

lavc: Add VP9 metadata bitstream filter

Can adjust the colour information.

lavc: Add coded bitstream read/write support for VP9

cbs: Add support for array subscripts in trace output

This makes the trace output for arrays significantly nicer.

cbs_h2645: Simplify representation of fixed values

cbs: Fragment/unit data is always reference counted

Make this clear in the documentation and add some asserts to ensure
that it is always true.

avfilter/af_biquads: add slice threading

Helps with multi-channels audio. Otherwise use threads=1.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/af_biquads: change clipping detection from global to channel

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_neighbor: add slice threading

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_neighbor: simplify code little

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_neighbor: rewrite without using temp memory

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/cbs_h2645: create a reference to the existing buffer when decomposing slice units

Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/cbs_mpeg2: create a reference to the existing buffer when decomposing slice units

Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>

fftools/ffmpeg: fix mixed code and declarations

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avdevice/decklink_dec: unref packets on avpacket_queue_put error

Signed-off-by: Marton Balint <cus@passwd.hu>

avdevice/decklink_dec: do not copy video data

Create a buffer from the data instead and use the buffer destructor to free the
DeckLink frame. This avoids a memcpy of the frame data.

Signed-off-by: Marton Balint <cus@passwd.hu>

Use AV_PIX_FMT_FLAG_ALPHA for detecting transparency where nb_components was used

Temporarily keep the old method for ffmpeg_filters.c choose_pix_fmt and
avfiltergraph.c pick_format() until a paletted pixel format without alpha is
introduced.

Signed-off-by: Marton Balint <cus@passwd.hu>

avutil/pixdesc: add AV_PIX_FMT_FLAG_ALPHA to AV_PIX_FMT_PAL8

Signed-off-by: Marton Balint <cus@passwd.hu>

avcodec/hnm4video: fix palette alpha

Signed-off-by: Marton Balint <cus@passwd.hu>

avcodec/anm: fix palette alpha

Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/qtpalette: parse color table according to the QuickTime file format specs

The specs says that the the first color component in the color array is
not alpha, but simply 0.

Fixes 0 alpha of fate-suite/cvid/catfight-cvid-pal8-partial.mov

Signed-off-by: Marton Balint <cus@passwd.hu>

avfilter/vf_mix: initialize last to silence possible warning

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/mpeg4video_parser: Avoid litteral 0x1B6, use named constant instead

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mpeg4video_parser: Fix incorrect spliting of MPEG-4 studio frames

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/m4vdec: Use the same constant names as libavcodec

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/m4vdec: Fix detection of raw MPEG-4 ES Studio

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

configure: add missing pthreads dependency to v4l2_m2m

Signed-off-by: James Almer <jamrial@gmail.com>

doc/filters: add tmix examples

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_mix: make setting weights more user friendly

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/wavpack: Fix integer overflow in DEC_MED() / INC_MED()

Fixes: runtime error: signed integer overflow: 2147483637 + 128 cannot be represented in type 'int'
Fixes: 6701/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WAVPACK_fuzzer-5358324934508544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/wavpack: Fix integer overflow in wv_unpack_stereo()

Fixes: runtime error: signed integer overflow: 2147483531 + 16384 cannot be represented in type 'int'
Fixes: 6615/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WAVPACK_fuzzer-5165715515506688
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_fieldmatch: add support for >8 bit depth in ppsrc

Also fix leaks while here.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_fieldmatch: switch to activate

Unbreaks ppsrc filtering.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter: add amplify filter

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_overlay: use slice_end in alpha_composite()

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/vc1: fix mquant calculation

In vc1_decode_i_blocks_adv mquant needs to be reset to its default value for
each macroblock, instead of once at the beginning of the slice.

DQPROFILE specifies which macroblocks can have an alternative quantizer step
size. When DQPROFILE specifies edges, the selection is applicable to the edges
of the picture. Slice edges are not selected by DQPROFILE.

Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_fieldmatch: use correct subsamplings for y0a, y1a and startx

Signed-off-by: Paul B Mahol <onemda@gmail.com>

doc/ffmpeg - update benchmark option readout details

-benchmark and -benchmark_all now show real, system and user time

avfilter/vf_overlay: add slice threading

Signed-off-by: Paul B Mahol <onemda@gmail.com>

fftools/ffmpeg: change fps progress log message to show two decimal digits

Useful when transcoding videos at 29.97 fps because delivers a more accurate result for monitoring.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

swresample/arm: rename labels to fix xcode build error

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fftools/ffmpeg: Add system time and real time to benchmarking.

The -benchmark and -benchmark_all options now show user, system, and real time,
instead of just user time.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/libav-merge: remove resolved item

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/proresdec2: add missing frame color metadata

Stolen from removed decoder.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/proresdec2: make long name more descriptive

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec: remove duplicate prores decoder

Removed slower one, couldn't figure out why it is slower.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_convolution: make rdiv set to 0 more useful

Use 0 for signaling that rdiv will be calculated from sum
of all matrix elements.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

hwcontext_vaapi: Fix compilation with libva versions < 1.4.0

The BufferHandle API was added in libva 1.4.0 / VAAPI 0.36.0.