1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-18 03:19:31 +02:00
Commit Graph

110994 Commits

Author SHA1 Message Date
Anton Khirnov
2b3905254e fftools/ffmpeg_enc: use AVFrame.hw_frames_ctx for encoder hw setup
It should be the same as the one that can be extracted from the filter
and does not require access to outside data.
2023-05-28 10:47:59 +02:00
Anton Khirnov
f15b206286 fftools/ffmpeg_hw: move hw_device_setup_for_encode() to ffmpeg_enc
This function is entangled with encoder setup, so it is more encoding
code rather than ffmpeg_hw code. This will allow making more encoder
state private in the future.
2023-05-28 10:47:59 +02:00
Anton Khirnov
760a9bd306 fftools/ffmpeg_hw: move hw_device_setup_for_decode() to ffmpeg_dec
This function is entangled with decoder setup, so it is more decoding
code rather than ffmpeg_hw code. This will allow making more decoder
state private in the future.
2023-05-28 10:47:59 +02:00
Paul B Mahol
e8e4863325 avfilter/af_silenceremove: fix the any/all confusion and change default
'any' stop mode is now correctly implemented for end trimming and is of limited usage.
Extend examples and other documentation of this filter.
2023-05-27 16:04:00 +02:00
Paul B Mahol
8d662adcc9 avfilter/af_silenceremove: speed-up filtering once start-trim is over
No point to call filter_start on samples that will pass them unchanged
once initial trimming stopped.
2023-05-27 11:18:02 +02:00
Paul B Mahol
965f99f904 avfilter/af_silenceremove: add timeline support 2023-05-27 11:18:00 +02:00
Paul B Mahol
7927a96e2b avfilter/af_silenceremove: add commands support 2023-05-27 11:17:58 +02:00
James Almer
e86e2d5921 fftools/ffmpeg_dec: abort if avcodec_send_packet() returns EAGAIN
As the comment in the code mentions, EAGAIN is not an expected value here
because we call avcodec_receive_frame() until all frames have been returned.
avcodec_send_packet() returning EAGAIN means a packet is still buffered, which
hints that the underlying decoder is buggy and not fetching packets as it
should.

An example of this behavior was in the libdav1d wrapper before f209614290,
where feeding it split frames (or individual OBUs) would result in the CLI
eventually printing the confusing "Error submitting packet to decoder: Resource
temporarily unavailable" error message, and just keep going until EOF without
returning new frames.

Signed-off-by: James Almer <jamrial@gmail.com>
2023-05-26 19:47:54 -03:00
Paul B Mahol
5a13b133f8 avfilter/af_silenceremove: add median silence detector 2023-05-27 00:34:37 +02:00
Paul B Mahol
68d0b881de avfilter/silenceremove_template: fix peak detector
The implementation was not working correctly.
2023-05-27 00:34:35 +02:00
Paul B Mahol
29710dd7f2 avfilter/af_silenceremove: speed-up start-only trimming once trim is over 2023-05-26 10:13:37 +02:00
Paul B Mahol
f59d9514da avfilter/silenceremove_template: skip silence start detection after start-only trimming is over 2023-05-26 10:13:37 +02:00
Paul B Mahol
2b5166addf avfilter/af_silenceremove: add real peak detector
Rename old peak detector to more correct name one.
2023-05-26 10:13:37 +02:00
Paul B Mahol
3235de4883 avfilter/af_silenceremove: switch to activate
Do full rewrite, new code is much faster for typical filter usages.
Also previous code was hard to follow and basically very-hard to maintain.
2023-05-26 10:13:37 +02:00
yuanhecai
82278e8749
avutil/la: Add function performance testing
This patch supports the use of the "checkasm --bench" testing feature
on loongarch platform.

Change-Id: I42790388d057c9ade0dfa38a19d9c1fd44ca0bc3
Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:05:21 +02:00
Jin Bo
cb4ae8baee
swscale/la: Add following builtin optimized functions
yuv420_rgb24_lsx
yuv420_bgr24_lsx
yuv420_rgba32_lsx
yuv420_argb32_lsx
yuv420_bgra32_lsx
yuv420_abgr32_lsx
./configure --disable-lasx
ffmpeg -i ~/media/1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo
-pix_fmt rgb24 -y /dev/null -an
before: 184fps
after:  207fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:05:15 +02:00
Lu Wang
4501b1dfd7
swscale/la: Optimize the functions of the swscale series with lsx.
./configure --disable-lasx
ffmpeg -i ~/media/1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo -s 640x480
-pix_fmt bgra -y /dev/null -an
before: 91fps
after:  160fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:05:08 +02:00
yuanhecai
f6077cc666
avcodec/la: Add LSX optimization for h264 qpel.
./configure --disable-lasx
ffmpeg -i 1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo -y /dev/null -an
before: 214fps
after:  274fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:05:01 +02:00
Lu Wang
8815a7719e
avcodec/la: Add LSX optimization for h264 chroma and intrapred.
./configure --disable-lasx
ffmpeg -i 1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo -y /dev/null -an
before: 199fps
after:  214fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:04:56 +02:00
Hao Chen
7845b5ecd6
avcodec/la: Add LSX optimization for loop filter.
Replaced function(LSX is sufficient for these functions):
ff_h264_v_lpf_chroma_8_lasx
ff_h264_h_lpf_chroma_8_lasx
ff_h264_v_lpf_chroma_intra_8_lasx
ff_h264_h_lpf_chroma_intra_8_lasx
ff_weight_h264_pixels4_8_lasx
ff_biweight_h264_pixels4_8_lasx

./configure --disable-lasx
ffmpeg -i 1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo -y /dev/null -an
before: 161fps
after:  199fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:04:43 +02:00
Shiyou Yin
e1b6ecd20a
avcodec/la: add LSX optimization for h264 idct.
loongson_asm.S is LoongArch asm optimization helper.
Add functions:
  ff_h264_idct_add_8_lsx
  ff_h264_idct8_add_8_lsx
  ff_h264_idct_dc_add_8_lsx
  ff_h264_idct8_dc_add_8_lsx
  ff_h264_idct_add16_8_lsx
  ff_h264_idct8_add4_8_lsx
  ff_h264_idct_add8_8_lsx
  ff_h264_idct_add8_422_8_lsx
  ff_h264_idct_add16_intra_8_lsx
  ff_h264_luma_dc_dequant_idct_8_lsx
Replaced function(LSX is sufficient for these functions):
  ff_h264_idct_add_lasx
  ff_h264_idct4x4_addblk_dc_lasx
  ff_h264_idct_add16_lasx
  ff_h264_idct8_add4_lasx
  ff_h264_idct_add8_lasx
  ff_h264_idct_add8_422_lasx
  ff_h264_idct_add16_intra_lasx
  ff_h264_deq_idct_luma_dc_lasx
Renamed functions:
  ff_h264_idct8_addblk_lasx ==> ff_h264_idct8_add_8_lasx
  ff_h264_idct8_dc_addblk_lasx ==> ff_h264_idct8_dc_add_8_lasx

./configure --disable-lasx
ffmpeg -i 1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo -y /dev/null -an
before: 155fps
after:  161fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:04:25 +02:00
Paul B Mahol
90fba27743 avfilter/trim: switch to activate
This also signals EOF to filter inlink so that unnecessary
extra processing that is immediately dropped by this filter
is not done any more.
2023-05-25 19:52:04 +02:00
Paul B Mahol
c6b6356635 avformat/gifdec: switch to using gif parser
Update fate test, more correct as last packet is not truncated.
2023-05-25 19:51:56 +02:00
Paul B Mahol
e7cadd0ee5 avformat/demux: add support to derive timestamps from packet durations for video 2023-05-25 19:51:47 +02:00
Paul B Mahol
a1838c59bb avfilter/af_astats: add abs_peak_count stat 2023-05-25 19:51:39 +02:00
James Almer
97f4263457 avcodec/av1dec: convert to receive_frame()
This removes the overhead of inserting the av1_frame_split bsf as part of the
decoding process.

Signed-off-by: James Almer <jamrial@gmail.com>
2023-05-25 11:21:35 -03:00
Gyan Doshi
944243477b doc/filters: clarify requirements for subtitles wrap_unicode option 2023-05-25 12:40:08 +05:30
Zhao Zhili
43ace8f2bc avfilter/vf_subtitles: add wrap_unicode option
So CJK can be wrapped automatically.

Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
2023-05-25 11:31:54 +08:00
Haihao Xiang
8e2547ebb2 lavc/qsvenc: the height is aligned to 32 for AV1
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
2023-05-25 09:04:13 +08:00
Haihao Xiang
943a42fc54 lavc/qsvenc: make sure continuous allocation
Intel MediaSDK and oneVPL expect continuous allocation for data[i],
however there are mandatory padding bytes between data[i] and data[i+1].
when calling av_frame_get_buffer. This patch removes all extra padding
bytes.

Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
2023-05-25 09:04:13 +08:00
Haihao Xiang
cf79dab8df lavc/qsvenc: avoid data copy if possible
The data copy is unnecessary for packed formats when frame width and
height are aligned

For example:
$ ffmpeg -f lavfi -i testsrc=size=1920x1088 -vf "format=yuyv422" -c:v hevc_qsv -f null -

Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
2023-05-25 09:04:13 +08:00
Haihao Xiang
cfdfd6aa72 lavc/qsvenc: use the right alignment instead of hard coded value
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
2023-05-25 09:04:13 +08:00
Haihao Xiang
2b0fe52967 lavc/qsv: fallback to the default mfx implementation for internal session on Windows
The mfx implementation based on D3D11 is expected for an internal
session on Windows, however sometimes this implemntation is not
supported [1]. A fallback to the default mfx implementation is added in
this patch.

[1] https://github.com/intel/cartwheel-ffmpeg/issues/246

Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
2023-05-25 09:04:13 +08:00
Niklas Haas
0bce5590c3 lavfi/vf_libplacebo: fix crop expr PTS calculation
mix->timestamps is expressed relative to the source timebase, which is
possibly a different timescale from `base_pts`. We can't mix-and-match
here. The only reason this worked in my previous testing was because I
was testing on a source file which had an exactly matching timebase.

Fix it by always using the exact PTS as tagged on the AVFrame.
2023-05-24 14:39:40 +02:00
Niklas Haas
d815584755 lavif/vf_libplacebo: move code (cosmetic)
To eliminate need for forward declaration
2023-05-23 15:06:18 +02:00
Niklas Haas
ee13414444 lavfi/vf_libplacebo: update settings after adding hooks
Failure to do so led to the update s->hooks never getting correctly
propagated to the render params.
2023-05-23 15:05:21 +02:00
Anton Khirnov
4e521e6102 fate/tests/ffmpeg: use -idct simple for fate-ffmpeg-input-r
Makes the test bitexact on non-x86_64.
2023-05-23 13:54:10 +02:00
James Almer
f209614290 avcodec/libdav1d: only return EAGAIN when there are no buffered packets
Fixes decoding packets containing split temporal units, as generated for example
by the av1_frame_split bsf.

Signed-off-by: James Almer <jamrial@gmail.com>
2023-05-22 16:19:10 -03:00
Paul B Mahol
557686ac63 avdevice/opengl_enc: check sscanf return value 2023-05-22 20:08:55 +02:00
Marvin Scholz
86b252ea9d avfilter/vf_tpad: accept hardware frames in clone-only mode
When no drawing is to be performed, tpad can work fine with
hardware frames, so advertise this in the query_formats
callback and ensure the drawing context is never initialised
when just cloning frames.

Reviewed-by: Thilo Borgmann <thilo.borgmann@mail.de>
Reviewed-by: Niklas Haas <git@haasn.dev>
2023-05-22 18:41:22 +02:00
Marvin Scholz
fa7180b827 avfilter/vf_tpad: use enum for start/stop_mode
Reviewed-by: Thilo Borgmann <thilo.borgmann@mail.de>
Reviewed-by: Niklas Haas <git@haasn.dev>
2023-05-22 18:40:42 +02:00
Anton Khirnov
c894f85f72 fftools/ffmpeg_dec: rename decode_video() to video_frame_process()
This function does not do any decoding anymore.
2023-05-22 17:10:44 +02:00
Anton Khirnov
909f5dfae1 fftools/ffmpeg_dec: inline decode_audio() into dec_packet()
The former function is now trivial - it has 3 lines and cannot fail.
2023-05-22 17:10:44 +02:00
Anton Khirnov
959a5decf2 fftools/ffmpeg_dec: deduplicate code in decode_audio/video() 2023-05-22 17:10:44 +02:00
Anton Khirnov
ab3493006f fftools/ffmpeg_dec: merge check_decode_result() into its callers
Not only is this easier to read, this also makes the code shorter.
2023-05-22 17:10:44 +02:00
Anton Khirnov
c08717adae fftools/ffmpeg: reindent after previous commit 2023-05-22 17:10:44 +02:00
Anton Khirnov
748f2999ee fftools/ffmpeg_dec: restructure audio/video decoding loop
It currently emulates the long-removed
avcodec_decode_audio4/avcodec_decode_video2 APIs, which obfuscates the
actual decoding flow. Restructure the decoding calls so that they
naturally follow the new avcodec_send_packet()/avcodec_receive_frame()
design.

This is not only significantly easier to read, but also shorter.
2023-05-22 17:10:44 +02:00
Anton Khirnov
172f901e35 fftools/ffmpeg_dec: deobfuscate subtitle decoding
It is currently handled in the same loop as audio and video, but this
obscures the actual flow, because only one iteration is ever performed
for subtitles.

Also, avoid a pointless packet reference.
2023-05-22 17:10:44 +02:00
Anton Khirnov
dbf1c6f5f1 fftools/ffmpeg: move decoding code to ffmpeg_dec 2023-05-22 17:10:44 +02:00
Anton Khirnov
79c1dde5c3 fftools/ffmpeg: split decoding loop out of process_input_packet()
process_input_packet() contains two non-interacting pieces of nontrivial
size and complexity - decoding and streamcopy. Separating them makes the
code easier to read.
2023-05-22 17:10:44 +02:00