Recherche avancée

Sur d’autres sites (9061)

• Correct Use Of avcodec_encode_video2() Flush

  5 février 2016, par mFeinstein

  I have a camera sending pictures to a callback function and I want to make a movie with this pictures using FFmpeg. I have followed the decoding_encoding example here but am not sure how to use the got_output for flushing the encoder and getting the delayed frames.

  1. Shoud I encode all my camera’s pictures when they arrive, and later when I want to stop capturing and close the video, I do the flush loop ?

  Or

  2. Should I do the flush periodically, let’s say, every 100 pictures received ?

  My video capture program could be running for hours, so I am worried about how this delayed frames work in memory consumption, if they stack up there until the flush, this could take all my memory.

  ---

  This is the encoding performed by the example, it makes 25 dummy Frames for 1 second of video, and later, in the end, it loop through avcodec_encode_video2() looking for got_output for delayed frames :

  /////  Prepare the Frame, CodecContext and some aditional logic.....
  
  
  
  /* encode 1 second of video */
  
  for (i = 0; i < 25; i++) {
  
      av_init_packet(&pkt);
  
      pkt.data = NULL;    // packet data will be allocated by the encoder
  
      pkt.size = 0;
  
      fflush(stdout);
  
      /* prepare a dummy image */
  
      /* Y */
  
      for (y = 0; y < c->height; y++) {
  
          for (x = 0; x < c->width; x++) {
  
              frame->data[0][y * frame->linesize[0] + x] = x + y + i * 3;
  
          }
  
      }
  
      /* Cb and Cr */
  
      for (y = 0; y < c->height/2; y++) {
  
          for (x = 0; x < c->width/2; x++) {
  
              frame->data[1][y * frame->linesize[1] + x] = 128 + y + i * 2;
  
              frame->data[2][y * frame->linesize[2] + x] = 64 + x + i * 5;
  
          }
  
      }
  
      frame->pts = i;
  
      /* encode the image */
  
      ret = avcodec_encode_video2(c, &pkt, frame, &got_output);
  
      if (ret < 0) {
  
          fprintf(stderr, "Error encoding frame\n");
  
          exit(1);
  
      }
  
      if (got_output) {
  
          printf("Write frame %3d (size=%5d)\n", i, pkt.size);
  
          fwrite(pkt.data, 1, pkt.size, f);
  
          av_free_packet(&pkt);
  
      }
  
  }
  
  /* get the delayed frames */
  
  for (got_output = 1; got_output; i++) {
  
      fflush(stdout);
  
      ret = avcodec_encode_video2(c, &pkt, NULL, &got_output);
  
      if (ret < 0) {
  
          fprintf(stderr, "Error encoding frame\n");
  
          exit(1);
  
      }
  
      if (got_output) {
  
          printf("Write frame %3d (size=%5d)\n", i, pkt.size);
  
          fwrite(pkt.data, 1, pkt.size, f);
  
          av_free_packet(&pkt);
  
      }
  
  }
  
  
  
  /////  Closes the file and finishes.....

• ffmpeg vs mediainfo for getting thumbnail and metadata

  14 juillet 2012, par Spottsworth

  I need to extract the metadata information from a video file as well as a thumbnail for that file. For this I have tried ffmpeg and the metadata (such as duration, resolution, codecs, creation_time etc) is seen on the stdout. If I need to use these I should parse the stdout and extract the metadata I need.

  I've also read about the MediaInfo utility which also delivers metadata. I'm not sure if it can deliver thumbnails. What I also know is that MediaInfo does not use ffmpeg under the hood.

  I was wondering if anyone has a working knowledge of both ffmpeg and MediaInfo and with respect to the requirement I mentioned above, whether someone could suggest which of the two is a better suited.

  Memory footprint comparison of the two would also be great.

• Does RTMP relay server need GPU ? (NodeMediaServer)

  25 août 2020, par Hung Nguyen

  I have a curious about GPU/CPU specs of an RTMP relay server (only push) with NodeMediaServer, that use ffmpeg under the hood.

  


  I think I will receive an RTMP stream from a device and immediately push it to a CDN (that CDN will handle transcode part). So my server doesn't need to handle anything, just a relay server.

  


  So does it need to use GPU in that case ? does a cheap cloud compute (without GPU) is enough for me ?

  


  Thank you so much