Recherche avancée

Sur d’autres sites (5765)

• avformat/matroskaenc : Don't override samplerate for CodecDelay

  29 août 2022, par Andreas Rheinhardt

  avformat/matroskaenc : Don't override samplerate for CodecDelay
  Opus can be decoded to multiple samplerates (namely 48kHz, 24KHz,
  
  16Khz, 12 KHz and 8Khz) ; libopus as well as our encoder wrapper
  
  support these sample rates. The OpusHead contains a field for
  
  this original samplerate. Yet the pre-skip (and the granule-position
  
  in the Ogg-Opus mapping in general) are always in the 48KHz clock,
  
  irrespective of the original sample rate.
  Before commit c3c22bee6362737cf290929b7f31df9fb88da983, our libopus
  
  encoder was buggy : It did not account for the fact that the pre-skip
  
  field is always according to a 48kHz clock and wrote a too small
  
  value in case one uses the encoder with a sample rate other than 48kHz ;
  
  this discrepancy between CodecDelay and OpusHead led to Firefox
  
  rejecting such streams.
  In order to account for that, said commit made the muxer always use
  
  48kHz instead of the actual sample rate to convert the initial_padding
  
  (in samples in the stream's sample rate) to ns. This meant that both
  
  fields are now off by the same factor, so Firefox was happy.
  Then commit f4bdeddc3cab807e43e0450744dfe9a45661e1d7 fixed the issue
  
  in libopusenc ; so the OpusHead is correct, but the CodecDelay is
  
  still off*. This commit fixes this by effectively reverting
  
  c3c22bee6362737cf290929b7f31df9fb88da983.
  * : Firefox seems to no longer abort when CodecDelay and OpusHead
  
  are off.
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
  

  • [DH] libavformat/matroskaenc.c
  • [DH] libavformat/version.h

• Using ffmpeg to "re-time" an RTSP stream ?

  29 mars 2021, par Tom S

  I recently set up a birdbox in my garden, and have been using a cheapie camera that I bought, along with ffmpeg on a raspberry pi to stream from the camera to YouTube.

  


  But YouTube keeps telling me that I am sending data too fast - multiple seconds of video per second, and at that point the stream seems to stall and I just get a buffering spinner on YouTube and have to restart ffmpeg to restore the stream.

  


  This is happening very frequently, infact even more frequently now than when I originally got everything up and running for some reason... So much so that Ive had to setup ffmpeg to run as a service, and restart it once an hour via cron. But even this is not enough and now it seems that it can go bonkers within 10-15 minutes of a restart - almost like the more often I restart the quicker it gets...

  


  I dont know where the issue is, or how to find it (does ffmpeg simply restream "verbatim" what ever the camera supplies ?), but I wonder if it might be possible to have ffmpeg re-time the stream so that I only send to YouTube the 25fps that it is happy with (i.e. dropping any frames in excess of this) ?

  


  I have key frames set to 1 second on the camera.

  


  Thanks !

  


  ---

  


  Here is my service file which contains the command that I use to start ffmpeg :

  


  [Unit]
Description=BirbCam 1 streaming service
After=network.target
StartLimitIntervalSec=0

[Service]
Type=simple
Restart=always
RestartSec=1
User=root
# Uncomment one ExecStart line below, then run `sudo systemctl daemon-reload` before restart
# With audio
ExecStart=/usr/local/bin/ffmpeg -rtsp_transport tcp -i rtsp://172.25.144.35:554/user=admin&password=&channel=1&stream=0.sdp?real_stream -tune zerolatency -vcodec libx264 -pix_fmt + -c:v copy -c:a aac -strict experimental -f flv rtmp://a.rtmp.youtube.com/live2/xxxx-xxxx-xxxx-xxxx-xxxx

# Without audio
# ExecStart=/usr/local/bin/ffmpeg -f lavfi -i anullsrc -rtsp_transport tcp -i rtsp://172.25.144.35:554/user=admin&password=&channel=1&stream=0.sdp?real_stream -tune zerolatency -vcodec libx264 -pix_fmt + -c:v copy -c:a aac -strict experimental -f flv rtmp://a.rtmp.youtube.com/live2/xxxx-xxxx-xxxx-xxxx-xxxx

[Install]
WantedBy=multi-user.target


  


• FFMPEG - RADEON - VAAPI - Alfa channel overlay

  25 janvier, par Alexandre Leitão

  Ok I got to my limit so decided to ask for help. I'm running a python app in a docker container. It works fine with VAAPI and hardware acceleration(AMD Radeon GPU) till I use filter complex. This is my command

  


  ffmpeg -v error -y -hwaccel vaapi -vaapi_device /dev/dri/renderD128 -stream_loop -1 -f concat -i video_list.txt -stream_loop -1 -f concat -i overlay_list1.txt -stream_loop -1 -f concat -i overlay_list.txt -filter_complex "[0:v][1:v]overlay=x=0:y=0:shortest=1[out];[out][2:v]overlay=x=0:y=0:shortest=1[out1]" -map "[out1]" -map 1:a -c:v h264_vaapi -c:a aac -b:a 192k -t 1800 text.mp4


  


  video_list.txt contains 2 mp4 videos h264 (30sec each)
overlay_list1.txt contains 2 mov video (alfa channel overlays 2 minutes each)
overlay_list.txt contains 1 mov video (alfa channel overlay 25 minutes long)

  


  

  • the idea is loop video_list.txt to -t (1800 at the moment)

  


  • Loop overlay_list1.txt over it till -t

  


  • and loop overlay.txt above everything

  


  


  Audio only comes from overlay_list1.txt videos
What I get is this output

  


  Impossible to convert between the formats supported by the filter 'Parsed_overlay_1' and the filter 'auto_scale_2'
[fc#0 @ 0x5fadddb56540] Error reinitializing filters!
[fc#0 @ 0x5fadddb56540] Task finished with error code: -38 (Function not implemented)
[fc#0 @ 0x5fadddb56540] Terminating thread with return code -38 (Function not implemented)
[vost#0:0/h264_vaapi @ 0x5fadddb3f400] Could not open encoder before EOF
[vost#0:0/h264_vaapi @ 0x5fadddb3f400] Task finished with error code: -22 (Invalid argument)
[vost#0:0/h264_vaapi @ 0x5fadddb3f400] Terminating thread with return code -22 (Invalid argument)
[out#0/mp4 @ 0x5fadddc3f140] Nothing was written into output file, because at least one of its streams received no packets.


  


  I tried everything and couldn't fix... the last thing I read is I suppose to use hwupload and hwdownload on filter complex but I couldn't understand how to do it.

  


  Any help is welcome guys
thank you and happy new year to y'all