Recherche avancée

Sur d’autres sites (10652)

• Nomenclature #4519 : Renommage de terminologie (blacklist / whitelist)

  16 juillet 2020, par RastaPopoulos ♥

  Bordel mais faut vraiment répondre à ça ? Tu utilises un réduction par l’absurde, qui ne démontre strictement rien.

  C’est quoi le rapport entre dire qu’une "liste noire" c’est une liste des choses à bannir, et donc assimiler de manière totalement culturelle, subjective et non-universelle ce qui est noir à ce qui est mal et à bannir (ça ne vaut que dans certaines cultures), avec le fait de décrire factuellement quelque chose de noir (la nuit noire, la chaussure noire, etc) ? RIEN. Aucun fucking rapport.

  Par ailleurs c’est quoi que tu as peur de ne plus pouvoir dire ? Ça t’interdit quoi dans ta vie de faire ça ? Tu te sens trop oppressé de changer ces mots ? En tant que personne blanche sans aucun problème en rapport avec ce sujet, ça t’empêche de formuler quoi comme pensée ?

  Toujours du côté du bâton, dès qu’il y a un mini changement culturel possible à faire à notre modeste échelle…

• How to strip metadata from a video file with reliable/reproducable results ?

  30 mai 2018, par Daniel Quinn

  I’m trying to find a clean way to produce "just the raw data" from a variety of media files. By this I mean, to take a file, test.mp4 and strip all of the metadata/headers off it so I can then generate a hash of the actual video data.

  After a lot of digging on this subject, ffmpeg seems to be my best shot at this, but the command I found to do the metadata stripping, appears to produce different results depending on the version of ffmpeg, which leads me to think that either (a) I’ve got the incantation wrong, or (b) ffmpeg isn’t actually returning just the raw data.

  To test for this, I used jrottenberg’s ffmpeg Docker containers to create a hash of the same file across multiple ffmpeg versions :

  for tag in 4.0-centos 4.0-alpine 3.4-alpine 3.4-centos 3.0-alpine 3.0-centos; do
  
      docker pull jrottenberg/ffmpeg:${tag}
  
      docker run --rm \
  
        -v /data/:/data/ \
  
        -it jrottenberg/ffmpeg:${tag} \
  
        -i /data/test.mp3 \
  
        -map_metadata -1 -c:v copy -c:a copy \
  
        -f mp4 - | md5sum
  
  done

  The result was that every one of these instances had a different hash output (Docker output truncated for clarity) :

  d7e3577ffe65d73240f48842e8d42207  -
  
  da2bda81911d758c877aace6ed3c0025  -
  
  ed24948c1dedf5d53870dfcfe24c7c70  -
  
  3dbb89d812c26711a33ca670403ccc20  -
  
  a4533446d3225e755eb041167e32b279  -
  
  69a51d82acc9987eed8b517a748435eb  -

  So my question : is there a more reliable way to do this ? Did I just miss an option for ffmpeg ? Am I missing something here, or is ffpeg just not a good tool for this job ? If not that, what is a good choice ?

• FFmpeg av_read_frame returns packets from audio stream

  1er août 2018, par yultan

  I am currently trying to learn the FFmpeg API, following this tutorial. However, I already have issues with the first lesson on video decoding. My code is basically the same as the one from the tutorial except I am using C++. My issue is that the video stream does not match the one from the packet returned by av_read_frame.

  The video stream is obtained looping on the available streams until the video stream is found.

  for(int i = 0; i < pFormatCtx->nb_streams; i++) { // nb_streams == 2
  
  
  
      if(pFormatCtx->streams[i]->codec->codec_type==AVMEDIA_TYPE_VIDEO) {
  
          videoStream = i; 
  
          break; // videoStream == 0
  
      }
  
  }

  Then when retrieving the frame data, it seams grabbing the audio channel.

  while(av_read_frame(pFormatCtx, &packet) >= 0) { // read returns 0
  
  
  
      // Is this a packet from the video stream?
  
      if(packet.stream_index == videoStream) { 
  
          //packet.stream_index == 1, which correspond to the audio stream
  
      } 
  
  }

  I have not found examples online where this test is actually failing. Have I miss some way to specify the stream_index that is not in the tutorial ? Maybe the tutorial is not up to date and is doing something wrong ? If so, what is the correct way to extract the frame data ? In case that matters, I am using the latest FFmpeg 4.0.2 build, on Windows 64-bits, compiling with Visual Studio 2017.

  On videos with no sound, the two streams match and I am able to decode and display the frames correctly.