Recherche avancée

Sur d’autres sites (9388)

• FFProbe generated data don't seem to agree with calculated data using ffmpeg libraries

  18 janvier 2020, par ark1974

  Using ffmpeg library avformat I am trying to check if the ffprobe generated data agrees with the data generated by the library. The code objective is to try to seek to the nearest key frame. When trying to seek at 100 frame or less, the codes returns 0 all the time.
  When trying to seek at 200 frame, the codes returns 4 all the time. But the result ie 4th frame don’t seem to be right. Where am I wrong ? Is my time_base conversion to actual frame faulty ?

  The test result using ffprobe

  Filename = test.mp4  
  
  Duration = 00:00:10.56 
  
  Fps = 25
  
  Total frames = 256   
  
  The key frames pkt_pts_time are at 2.120000  and 0.000000 (using -skip_frame nokey )
  
  Corresponding pkt_duration_time: 0.040000 and 0.040000 (  same, why?)

  Abstract of the code :

  // Objective: seek to the nearest key frame
  
  frameIndex = 200;
  
  int64_t timeBase = (int64_t(pCodecCtx->time_base.num) * AV_TIME_BASE) / int64_t(pCodecCtx->time_base.den);  
  
  int64_t seekTarget = int64_t(frameIndex) * timeBase;
  
  
  
  if (av_seek_frame(pFormatCtx, -1, seekTarget, AVSEEK_FLAG_FRAME | AVSEEK_FLAG_BACKWARD) < 0) return -1;
  
  
  
  //convert the time_base to actual frame
  
  auto time2frame = [&](int64_t tb) {
  
      return tb * int64_t(pCodecCtx->time_base.den) / (int64_t(pCodecCtx->time_base.num) * AV_TIME_BASE);
  
  };
  
  
  
  AVPacket avPacket;
  
  int result = av_read_frame(pFormatCtx, &avPacket);
  
  if (result == 0) {
  
  
  
      auto dts = avPacket.dts;
  
      auto pts = avPacket.pts;
  
      auto idx = avPacket.stream_index;
  
      auto f = time2frame(pts); // expecting the actual frame here
  
      std::cout << dts << pts << idx << f;
  
  }

• Save two video streams in one video file c#, Windows 8 application [on hold]

  19 septembre 2014, par manos

  I am developing a Windows store application which will be able to record video from two video sources (e.g. two usb cameras or usb camera and embeded tablet camera) and then save the two video streams in one video file. So, when the saved video will play, it will show a large video frame (video from camera 1) and inside that, in the top right corner, a smaller video frame (video from camera 2). Basically, i want to render one video on top of the other. I think the process is called picture in picture, but I am not 100% sure. I am able two capture the 2 video streams (same length, same video quality) but I do not know how to proceed after that. How to do the rendering. Do I save the streams into two files and then process those files or do i need to process the streams directly (merge them) and then save to file ?
  I have read that you can do something like that by using ffmpeg libraries. But so far I have not managed to find any c# code to it. Any ideas ?

  Thanks in advance,
  Manos

• avcodec/mpeg4videodec : Use union to save space

  21 mai, par Andreas Rheinhardt

  avcodec/mpeg4videodec : Use union to save space
  At most one of block32 and dpcm_macroblock is used at any given time.
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
  

  • [DH] libavcodec/mpeg4videodec.h