Recherche avancée

Sur d’autres sites (4506)

• How does FFMPEG copy work ?

  24 novembre 2017, par Tyler Brooks

  How can I implement the equivalent of this command :

  ffmpeg -i test.h264 -vcodec copy -hls_time 5 -hls_list_size 0 foo.m3u8

  My understanding is that you do something like this (pseudo code for clarity) :

  avformat_open_input(&ctx_in, "test.h264", NULL, NULL);
  
  avformat_find_stream_info(ctx_in, NULL);
  
  
  
  avformat_alloc_output_context2(&ctx_out, NULL, "hls", "foo.m3u8");
  
  strm_out = avformat_new_stream(ctx_out, NULL);
  
  strm_out->time_base = (AVRational){1000000, FRAMERATE};
  
  
  
  strm_out->codecpar->codec_id = AV_CODEC_ID_H264;
  
  strm_out->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
  
  strm_out->codecpar->width = width;
  
  strm_out->codecpar->height = height;
  
  strm_out->codecpar->bit_rate = bitrate;
  
  strm_out->codecpar->codec_tag = 0;
  
  
  
  avcodec_parameters_copy(ctx_out->streams[0]->codecpar,
  
    ctx_in->streams[0]->codecpar);
  
  
  
  avformat_write_header(ctx_out, NULL);
  
  
  
  while(1) {
  
    AVPacket pkt;
  
  
  
    ret = av_read_frame(ctx_in, &pkt);
  
    if (ret < 0)
  
      break;
  
  
  
    pkt.pos = -1;
  
    pkt.stream_index = 0;
  
    pkt.dts = AV_NOPTS_VALUE;
  
    pkt.pts = AV_NOPTS_VALUE;
  
    pkt.pts = SOME_DURATION;
  
  
  
    av_write_frame(ctx_out, &pkt);
  
    av_packet_unref(&pkt);
  
  }
  
  avformat_close_input(&ctx_in);
  
  av_write_trailer(ctx_out);
  
  avformat_free_context(ctx_out);

  I crash on the avformat_find_stream_info() call. I think this is the case because an H264 elemental stream is not really a container (which avformat wants).

  What is the proper way to implement the FFMPEG ’copy’ command of an elemental stream ?

• Using Mencoder to Convert Sequence of Images Sorted Numerically to Video

  7 décembre 2017, par txizzle

  I want to use mencoder to convert a sequence of PNG files to video. The PNG files are named like file1.png, file2.png, ... file1000.png. However, if I run this :

  mencoder mf://*.png -mf fps=25 -ovc copy -oac copy output.avi

  The resulting video will not stitch together the frames in the correct order (correct : 1,2,3,.... incorrect : 1, 10, 100, ...). Answers to a similar question suggest that I can manually input the range of frames to have mencoder use the correct numerical order :

  mencoder mf://file{1..1000}.png -mf fps=25 -ovc copy -oac copy output.avi

  or to manually sort the files with something like ls -v and then save the results to a list (say, list.txt), and then pass that into mencoder :

  mencoder mf://@list.txt -mf fps=25 -ovc copy -oac copy output.avi

  However, my question is : Is there a way to do this in a single command without specifying the last frame ?

  Ideally, I would want an equivalent to mf://file*.png that uses a numerical sorting as opposed to the default sorting. For example, ffmpeg has the option to specify input files like file%d.png, which would order the files numerically.

  Thanks !

• How to install ffmpeg for a django app on heroku ?

  10 janvier 2018, par yndolok

  I’d like to use ffmpeg to extract a frame from a video to use it as a poster. This is my first time deploying an app, let alone on heroku, so I’m not sure how to install ffmpeg on the server.

  I’ve found this build of ffmpeg with instructions to ’vendor’ it into my app, and then adjust my app’s configuration settings / path. What does it mean to vendor something ? I have a feeling it’s a rails thing, because I can’t seem to find an explanation by googling.

  Also, what would be the django equivalent of the instructions to run
  heroku config:set PATH=bin:vendor/ffmpeg/bin:<...> -a yourapp and heroku config:set LD_LIBRARY_PATH=vendor/ffmpeg/lib:/usr/local/lib -a yourapp ?