Recherche avancée

Sur d’autres sites (7531)

• fate/mxf : Fix d10-user-comments test

  4 décembre 2020, par Andreas Rheinhardt

  fate/mxf : Fix d10-user-comments test
  The mxf_d10 muxer is very picky regarding the input it accepts :
  
  The only video accepted is MPEG-2 with absolutely constant bitrate,
  
  i.e. all packets need to have exactly the same size ; and only a few
  
  bitrates are accepted.
  The sample file used did not abide by this : Writing the first packet
  
  (a video packet) errors out and afterwards an audio packet from the
  
  muxing queue has been written. That's all besides metadata (which this
  
  test is about). The FFmpeg cli returned an error, but said error has
  
  been ignored by the md5 test.
  This commit changes the test to actually send a compliant stream to the
  
  muxer, so that it does not error out ; furthermore, the test is changed
  
  to explicitly check the metadata instead of it only being implicitly
  
  included in the md5 checksum. The compliant stream is created by our
  
  encoder at runtime.
  Finally, the test now also covers writing user-specified
  
  product/company/version identification.
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
  

  • [DH] tests/fate/mxf.mak
  • [DH] tests/ref/fate/mxf-d10-user-comments

• ffmpeg video segment conflicts with frame_skip setpts filters

  30 janvier, par Jeff Hansen

  My goal is to record an rtsp stream into 5 second segments, but I want those segments to only include every 5th frame and be as compact as possible. Essentially, the resulting video will be a 1 second video that shows 5 seconds of realtime content.

  


  I have found that ffmpeg has video segment muxing that works really well. I have also found that I can apply a -filter:v select=not(mod(n\,{frame_skip})),setpts=PTS/{frame_skip} filter to only keep every 5th frame and also ensure that the resulting mp4 is 1 second instead of 5 seconds. This time reduction saves space, AND it also ensures that there aren't duplicate frames stored and displayed in the mp4.

  


  This is the segment muxing command that works :

  


  sec_overlap = 1
segment_time = 5
ffmpeg_command = (
    f'ffmpeg -rtsp_transport tcp -i "{rtsp_url}" '
    f'-c:v libx264 -preset ultrafast -tune zerolatency -an '
    f'-vf "scale=iw/10:ih/10" '
    f'-f segment '
    f'-segment_time {segment_time} '
    f'-reset_timestamps 1 '
    f'-segment_time_delta {sec_overlap} '
    f'-strftime 1 "{directory}\{cam_label}_%Y-%m-%d-%H-%M-%S.mp4"'
)


  


  And this is the the filter command that successfully skips every 5th frame and shrinks the video length by 5x :

  


  sec_overlap = 1
frame_skip = 5
segment_time = 5 #5 / frame_skip**2
ffmpeg_command = (
    f'ffmpeg -rtsp_transport tcp -i "{rtsp_url}" '
    f'-c:v libx264 -preset ultrafast -tune zerolatency -an '
    f"-filter:v '"
        f"select=not(mod(n\,{frame_skip})),"
        f"setpts=PTS/{frame_skip},"
        f"fps={frame_skip*frame_skip}"
    f"' "
    f'-f segment '
        f'-segment_time {segment_time} '
        f'-reset_timestamps 1 '
        f'-segment_time_delta {sec_overlap} '
        f'-strftime 1 '
    f'"{directory}\{cam_label}_%Y-%m-%d-%H-%M-%S.mp4"'
)


  


  THIS IS MY PROBLEM. The resulting video ends up being a 10 second duration video that covers 50 SECONDS of realtime rtsp video footage. It appears that adjusting the PTS parameter, causes the segment_time calculation to record much longer.

  


  Does anyone know how to get 1sec videos of 5sec coverage while still applying this video filter ?

  


• x86 : AVX-512 pixel_sad

  7 mai 2017, par Henrik Gramner

  x86 : AVX-512 pixel_sad
  Covers all variants : 4x4, 4x8, 4x16, 8x4, 8x8, 8x16, 16x8, and 16x16.
  

  • [DH] common/pixel.c
  • [DH] common/x86/pixel.h
  • [DH] common/x86/sad-a.asm
  • [DH] encoder/slicetype.c
  • [DH] tools/checkasm.c