Recherche avancée

Sur d’autres sites (9526)

• ffserver "dimensions not set" when loading stream

  29 mai 2013, par GreenGiant

  I am live streaming a webcam from my raspberry pi using avconv (ffmpeg "replacement")

  avconv -f video4linux2 -v debug -r 5 -s 176x144 -i /dev/video0 -vcodec mjpeg http://192.168.0.3:8090/feed1.ffm

  to my local network OSX machine (for testing) running ffserver

  Port 8090
  
  BindAddress 0.0.0.0
  
  MaxHTTPConnections 2000
  
  MaxClients 1000
  
  MaxBandwidth 10000
  
  CustomLog -
  
  NoDaemon
  
  
  
  <feed>
  
     File feed1.ffm
  
     FileMaxSize 20M
  
     ACL allow 192.168.0.10
  
  </feed>
  
  
  
  <stream>
  
     Feed feed1.ffm
  
     Format mjpeg
  
     NoAudio
  
     VideoQMin 1
  
     VideoQMax 10
  
     VideoSize 176x144
  
     VideoFrameRate 5
  
  </stream>

  When I start avconv it appears to be streaming to ffserver fine :

  Output #0, ffm, to 'http://192.168.0.3:8090/feed1.ffm':
  
    Metadata:
  
      encoder         : Lavf55.0.1
  
      Stream #0.0, 0, 1/1000000: Video: mjpeg, yuvj420p, 320x240, 1/5, q=2-31, 200 kb/s, 1000k tbn, 5 tbc
  
  Stream mapping:
  
    Stream #0:0 -> #0:0 (rawvideo -> mjpeg)
  
  Press ctrl-c to stop encoding
  
  frame=  108 fps= 18 q=21.7 size=     688kB time=21.60 bitrate= 260.9kbits/s 

  And the ffserver status page shows the stream

  

  However when I load http://localhost:8090/test.mjpeg in VLC it doesn't play and ffserver spits out :

  Sat May 25 17:25:34 2013 dimensions not set
  
  Sat May 25 17:25:34 2013 Error writing output header
  
  Sat May 25 17:25:34 2013 127.0.0.1 - - [GET] "/test.mjpeg HTTP/1.1" 200 66

  I've tried so many different configurations and settings, I'm at a loss to what is causing that error !

  Thank you

• How to seek by msec with ffmpeg ?

  26 mai 2014, par Srv19

  I am trying to seek in video by milliseconds with ffmpeg. I have been trying to use code from this question, which uses avformat_seek_file (i use it with -1 for stream number and AVSEEK_FLAG_ANY flag).

  After that is called, i try to read next frames, that is :

  if (av_read_frame(fmt_ctx, &pkt) >= 0)
  
  {
  
      int ret = 0;
  
  
  
      if (pkt.stream_index == video_stream_idx) {
  
          /* decode video frame */
  
          ret = avcodec_decode_video2(video_dec_ctx, frame, got_frame, &pkt);
  
          if (ret < 0) {
  
              fprintf(stderr, "Error decoding video frame\n");
  
              return ret;
  
          }
  
  //do something with frame
  
  }

  However, the frame->pts of retrieved frame always holds the time of the frame that was immediatly after last frame that was read before seeking.

  Edit : In spite of frame->pts forming unbroken sequence, seeking does occur. For some bizarre reason next frame i read is the first one. In fact, after i run :

     int got_frame = 0;
  
     do
  
     if (av_read_frame(fmt_ctx, &pkt) >= 0) {
  
         decode_packet_ro(&got_frame, 0);
  
         av_free_packet(&pkt);
  
     }
  
     else
  
     {
  
         read_cache = true;
  
         pkt.data = NULL;
  
         pkt.size = 0;
  
         break;
  
     }
  
     while(!got_frame || this->frame->pts*av_q2d(video_dec_ctx->time_base) * 1000 < tsms);

  next frame i read is always the first one.

• Get frame time in ffmpeg

  30 mai 2013, par Srv19

  I am trying to make a little video player that has seek bar (with ffmpeg, of course). For that i need function that will, using data from frame and/or packet, get me current time in the video that should be set in seek slider.

  It should work like this :

  my_time = get_cur_time()
  
  seek(my_time + 10)
  
  assert(my_time+10 == get_cur_time())
  
  seek(my_time - 10)
  
  assert(my_time-10 == get_cur_time())

  I do understand thatffmpeg does not support precise seeking, so equality here means "something reasonably cloae).

  What code have i used for this thus far :

  frame_time = frame->pts*av_q2d(video_dec_ctx->time_base) * 1000;

  where frame is AVFrame and video_dec_ctx is AVCodecContext.

  And for seeking :

     int fn = ffmpeg::av_rescale(tsms,fmt_ctx->streams[video_stream->index]->time_base.den,
  
                                 fmt_ctx->streams[video_stream->index]->time_base.num);
  
     int frame = fn/1000;
  
  
  
     printf("\t avformat_seek_file to %d\n",frame);
  
     int flags = AVSEEK_FLAG_FRAME;
  
     if (frame < this->frame->pts)
  
         flags |= AVSEEK_FLAG_BACKWARD;
  
     if(ffmpeg::av_seek_frame(fmt_ctx,video_stream->index,frame,flags))
  
     {
  
         printf("\nFailed to seek for time %d",frame);
  
        return false;
  
     }
  
  
  
     avcodec_flush_buffers(video_dec_ctx);
  
     int got_frame = 0;
  
     do
  
     if (av_read_frame(fmt_ctx, &pkt) >= 0) {
  
         decode_packet_ro(&got_frame, 0);
  
         av_free_packet(&pkt);
  
     }
  
     else
  
     {
  
         read_cache = true;
  
         pkt.data = NULL;
  
         pkt.size = 0;
  
         break;
  
     }
  
     while(!(got_frame && this->frame->pts >= frame));

  The code does forward seeking passably, but after any attempt of backward seeking my second assertion fails. After seeking to previous position, my method of getting time does not return position less that one before seeking. That causes my seek slider to work grossly incorrectly.