Sur d’autres sites (13397)

• How to open anullsrc input ?

  5 mai 2022, par dgsgsdgsgdhddhg

  I'm trying :

  


      avformat_open_input(
        &input_context,
        "anullsrc",       
        NULL,
        NULL
    );


  


  That however does not work and input_context remains NULL.

  


  The 3rd argument, NULL, is an optional AVInputFormat.
https://ffmpeg.org/doxygen/trunk/group__lavf__decoding.html#gac05d61a2b492ae3985c658f34622c19d

  


  fmt    If non-NULL, this parameter forces a specific input format. Otherwise the format is autodetected.

  


  I wanted to supply it by first creating the AVInputFormat with av_find_input_format - that however fails to find the format (lavfi), because it seems not to be registered in file ./libavformat/allformats.c even though it exists in ./libavdevice/lavfi.c.

  


  Edit

  


  I might have found an issue. When I do :

  


  $ ffmpeg -v 0 -demuxers | grep lavfi


  


  The output is :

  


   D  lavfi           Libavfilter virtual input device



  


  However, when I do :

  


  main | grep lavfi


  


  Where main is my program which includes :

  


      while ((fmt = av_demuxer_iterate(&i)))                                      
        printf("%s\n", fmt->name);  


  


  I get no output (I confirm that some demuxers are listed if I don't grep for lavfi).

  


  So it seems then, that libavformat.h does not know about lavfi demuxer, whereas ffmpeg does. So it seems I need to fix this issue now.

  


• FFmpeg : AAC muxing into two different files in the same loop : The 2nd file is corrupted [on hold]

  1er septembre 2013, par user2677612

  We are using FFmpeg libraries git-ee94362 libavformat v55.2.100.
  Our purpose is to mux synthetic video and audio-file audio to 2 different files (in the same loop)
  with the the same format and parameters.

  With MP3 input, both M3U8 outputs are OK :

  Output file #0 : Total number of frames is : video = 229, audio = 347.

  Output file #1 : Total number of frames is : video = 229, audio = 347.

  But with AAC input, the 1st M3U8 output is OK, whereas the 2nd one - is not :

  Output file #0 : Total number of frames is : video = 403, audio = 690.

  Output file #1 : Total number of frames is : video = 3, audio = 690.

  It is exactly the same code in the main loop. We do not use the "aac_adtstoasc" filter in case of M3U output.
  By the way, the problem exist both with and without the filter,
  since in case of MP4 output, the 1st output file is OK, whereas the 2nd one - is not, - as well.

  Moreover :

  1) I have implemented the main loop as Windows threads, separated for every output file.

  2) I have separated input files, as well (i.e. use two identical input files).
  The problem remains !

  It seems that video PTS begins to be corrupted (wrong-calculated) for the 2nd (3rd, etc.) output file.

  If somebody is known about such a problem ? If yes, what can be a solution/workaround ?
  Thank you.

• Detecting Successful Conversion with ffmpeg

  22 février 2017, par J M

  I have code that scans a file system for videos files encoded with H.264 and re-encodes them with H.265. It runs pretty much on its own constantly, generating various log files for me to review periodically.

  One thing that I want to further improve is the successful conversion detection. Right now, a file returns as being successfully converted after it meets these criteria/checks :

  1. The output file exists
  2. ffprobe can detect that the output file is in hevc format
  3. The duration of the output file matches that of the input file (within 3 seconds)
  4. The length of the output file is greater than 30 MB (it’s rare that I have a video so short where after conversion it is less than this, usually this happens when an error occurs or conversion terminates early).

  Obviously, this is rather computationally intense as there are many file checks to confirm all of this information. I do this because if the file is detected as successful conversion, the old file is overwritten and the new converted file takes it’s place. I don’t want to overwrite a file because I overlooked a scenario where I think conversion is successful but was in fact not. The files are under a crashplan constant backup, so I don’t lose them, but I also do not go through and review every file.

  So, my basic question is if you see any area of improvement for this detection. My goal is to determine, to my best extent, if after conversion the video remains "playable". So deciding programmatically how/what that means is what I’m attempting to do.

  I can post code if you want it (powershell), but the question seems independent of actual program language choice.