Recherche avancée

Sur d’autres sites (7237)

• break up ffmpeg passthrough with python

  15 janvier 2023, par Cee Wee

  Currently I am reading rtsp (h265) feed via cv2.VideoCapture and write frames through cv2.VideoWriter. Then I need to shrink the video back into h265 again with ffmpeg command. The process is done in python. The capture part goes into buffer queue and if certain conditions meet from time to time, a set of python codes would create a file and read from the queue into the file via video writer.

  


  I used to save from the rtsp feed into video files directly via ffmpeg as follows.

  


  "ffmpeg -i rtsp ://... -c copy -fps_mode vfr output.mp4"

  


  This take very little processing resources.

  


  I have an idea that if we could split the ffmpeg process above into pipes and capture the intermediate contents there would save substantial processing resources.

  


  My aim is to have something similar to this.
"ffmpeg -i rtsp ://... (pipe video and audio outs) " equivalent to cv2.VideoCapture into buffer queue.

  


  "ffmpeg -i - -c copy -fps_mode vfr output.mp4" equivalent to cv2.VideoWrite from time to time.

  


  Please anyone has ever come across or could share how to implement this.

  


• compat : Use '-' instead of '/' for rc.exe options

  10 février 2023, par Ziemowit Laski

  compat : Use '-' instead of '/' for rc.exe options
  When building FFMPEG in the MSYS environment under Windows, one
  
  must not use forward slashes ('/') for command-line options.  It
  
  appears that the MSYS shell interprets these as absolute paths and
  
  then automatically rewrites them into equivalent Windows paths.  For
  
  example, the '/nologo' switch below gets rewritten to something like
  
  'C :/Program Files/Git/nologo', and this obviously breaks the build.
  
  Thankfully, most M$ tools accept dashes ('-') as well.
  Signed-off-by : Ziemowit Łąski <15880281+zlaski@users.noreply.github.com>
  
  Signed-off-by : Martin Storsjö <martin@martin.st>
  

  • [DH] compat/windows/mswindres

• Creating audio segments of identical duration using FFmpeg

  22 mars 2023, par aradalvand

  I have an MP4 file with one video stream and one audio stream. I want to divide each stream into 2-second segments. For the video, I know that I need to have a keyframe (I-frame) precisely every 2 seconds in order for it be to able to be divided into exactly 2-second segments ; which can be achieved via FFmpeg's force_key_frames option, set to a value like expr:gte(t,n_forced*2).

  


  For the audio stream, however, I can't seem to get the segments to be identical in duration. The force_key_frames option is only applicable to videos, and I'm not sure what the equivalent of that would be for audio streams.

  


  This is the FFmpeg command I'm using :

  


  ffmpeg -i input.mp4 \
    -map 0:a \
    -f segment \
    -segment_time 2 \
    -segment_list output/playlist.m3u8 \
    -segment_format mpegts \
    output/seg-%d.ts


  


  And this is the resulting playlist.m3u8 file :

  


  #EXTM3U
#EXT-X-VERSION:3
#EXT-X-MEDIA-SEQUENCE:0
#EXT-X-ALLOW-CACHE:YES
#EXT-X-TARGETDURATION:3
#EXTINF:2.011433,
seg-0.ts
#EXTINF:2.011422,
seg-1.ts
#EXTINF:1.985300,
seg-2.ts
#EXTINF:2.011433,
seg-3.ts
#EXTINF:1.985300,
seg-4.ts
#EXTINF:2.011433,
seg-5.ts
#EXTINF:1.985300,
seg-6.ts
#EXTINF:2.011433,
seg-7.ts
#EXTINF:2.011422,
seg-8.ts
#EXTINF:1.985311,
seg-9.ts
#EXTINF:2.011433,
seg-10.ts
#EXTINF:1.985300,
seg-11.ts
#EXTINF:2.011433,
seg-12.ts
#EXTINF:1.985300,
seg-13.ts
#EXTINF:2.011433,
seg-14.ts
#EXTINF:1.985300,
seg-15.ts
#EXTINF:2.011433,
seg-16.ts
#EXTINF:2.011422,
seg-17.ts
#EXTINF:1.985300,
seg-18.ts
#EXTINF:2.011433,
seg-19.ts
#EXTINF:1.985300,
seg-20.ts
#EXTINF:2.011433,
seg-21.ts
#EXTINF:1.985300,
seg-22.ts
#EXTINF:2.011433,
seg-23.ts
#EXTINF:2.011422,
seg-24.ts
#EXTINF:1.985311,
seg-25.ts
#EXTINF:2.011433,
seg-26.ts
#EXTINF:1.985300,
seg-27.ts
#EXTINF:2.011433,
seg-28.ts
#EXTINF:1.985300,
seg-29.ts
#EXTINF:0.052244,
seg-30.ts
#EXT-X-ENDLIST


  


  As you can see, the segments aren't precisely 2 seconds.
I don't know much about how audio works under the hood, so I don't understand why FFmpeg doesn't seem to be able to divide the audio into segments that all have precisely the duration I specify.

  


  I would also like to know whether anything can be done (analogous to force_key_frames for videos) to solve this problem.

  


  Thank you.