Recherche avancée

Sur d’autres sites (11254)

• Does this FFMPEG command create videos compatible with ABR playback on video.js player ?

  16 mars 2018, par alchemical

  We’re working on a site that will allow users to upload videos. It then encodes these videos into 4 bitrates. The initial version just lets the user manually select which of the 4 version they would like to play. However, the final version should allow for adaptive bitrate ABR using video.js player. Ideally, we should not need to re-encode the videos again for them to stream with ABR.

  I’m trying to understand if the current FFMPEG statement will support ABR :

  -i \"0\" -i \"1\" -threads 5 -filter_complex \"
  [0:v]scale=-2 :2[bg] ;[1:v]scale=-2:40*2/1080[watermark_scaled] ;[bg]
  [watermark_scaled]overlay=x=main_w-overlay_w-(10*
  2/1080):10*2/1080\" -acodec copy -vcodec libx264 -b:v
  3 \"4\"

• Adjust PTS and DTS before mp4 video creation

  17 mars 2018, par Cristiano

  I’m retrieving raw h264 compressed frames from usb camera to create an mp4 video. This is my simple code :

  for(int i = 0; i<120;i++)
  
  {
  
      AVPacket pkt;
  
      av_init_packet(&pkt);
  
      ret = av_read_frame(inputFormatCtx, &pkt);
  
  
  
      pkt.pts = pkt.dts = i;
  
      pkt.pts = av_rescale_q(pkt.pts, inputStream->time_base, outputStream->time_base);
  
      pkt.dts = av_rescale_q(pkt.dts, inputStream->time_base, outputStream->time_base);
  
  
  
      ret = av_interleaved_write_frame(outputFormatCtx, &pkt);
  
      av_packet_unref(&pkt);
  
  } 
  
  ret = av_write_trailer(outputFormatCtx);

  This works well. Now I would like to store these AVPackets to create the video in a second moment.
  I changed my code in this way

  for(int i = 0; i<120;i++){
  
  AVPacket pkt;
  
  av_init_packet(&pkt);
  
  ret = av_read_frame(inputFormatCtx, &pkt);
  
  packets.push_back(pkt);
  
  } 
  
  
  
  vector<avpacket>::reverse_iterator it;
  
  int j = 0;
  
  for(it = packets.rbegin(); it != packets.rend(); it++){
  
      AVPacket n = (*it);
  
      n.pts = n.dts = j;
  
      j++;
  
      n.pts = av_rescale_q(n.pts, inputStream->time_base, outputStream->time_base);
  
      n.dts = av_rescale_q(n.dts, inputStream->time_base, outputStream->time_base); 
  
      ret = av_interleaved_write_frame(outputFormatCtx, &amp;n);
  
      av_packet_unref(&amp;n);
  
  }
  
  
  
  ret = av_write_trailer(outputFormatCtx);
  
  </avpacket>

  The resulting video is not so fluid and I used ffprobe to see more details.
  These are the first three frame generating with the first block of code.

  [FRAME]
  
  media_type=video
  
  stream_index=0
  
  key_frame=1
  
  pkt_pts=0
  
  pkt_pts_time=0.000000
  
  pkt_dts=0
  
  pkt_dts_time=0.000000
  
  best_effort_timestamp=0
  
  best_effort_timestamp_time=0.000000
  
  pkt_duration=512
  
  pkt_duration_time=0.033333
  
  pkt_pos=48
  
  pkt_size=12974
  
  width=1920
  
  height=1080
  
  pix_fmt=yuv420p
  
  sample_aspect_ratio=N/A
  
  pict_type=I
  
  coded_picture_number=0
  
  display_picture_number=0
  
  interlaced_frame=0
  
  top_field_first=0
  
  repeat_pict=0
  
  [/FRAME]
  
  [FRAME]
  
  media_type=video
  
  stream_index=0
  
  key_frame=0
  
  pkt_pts=512
  
  pkt_pts_time=0.033333
  
  pkt_dts=512
  
  pkt_dts_time=0.033333
  
  best_effort_timestamp=512
  
  best_effort_timestamp_time=0.033333
  
  pkt_duration=512
  
  pkt_duration_time=0.033333
  
  pkt_pos=13022
  
  pkt_size=473
  
  width=1920
  
  height=1080
  
  pix_fmt=yuv420p
  
  sample_aspect_ratio=N/A
  
  pict_type=P
  
  coded_picture_number=1
  
  display_picture_number=0
  
  interlaced_frame=0
  
  top_field_first=0
  
  repeat_pict=0
  
  [/FRAME]
  
  [FRAME]
  
  media_type=video
  
  stream_index=0
  
  key_frame=0
  
  pkt_pts=1024
  
  pkt_pts_time=0.066667
  
  pkt_dts=1024
  
  pkt_dts_time=0.066667
  
  best_effort_timestamp=1024
  
  best_effort_timestamp_time=0.066667
  
  pkt_duration=512
  
  pkt_duration_time=0.033333
  
  pkt_pos=13495
  
  pkt_size=511
  
  width=1920
  
  height=1080
  
  pix_fmt=yuv420p
  
  sample_aspect_ratio=N/A
  
  pict_type=P
  
  coded_picture_number=2
  
  display_picture_number=0
  
  interlaced_frame=0
  
  top_field_first=0
  
  repeat_pict=0
  
  [/FRAME]

  While these are the same frames created using the second block of code.

  [FRAME]
  
  media_type=video
  
  stream_index=0
  
  key_frame=1
  
  pkt_pts=14848
  
  pkt_pts_time=0.966667
  
  pkt_dts=14848
  
  pkt_dts_time=0.966667
  
  best_effort_timestamp=14848
  
  best_effort_timestamp_time=0.966667
  
  pkt_duration=512
  
  pkt_duration_time=0.033333
  
  pkt_pos=757791
  
  pkt_size=65625
  
  width=1920
  
  height=1080
  
  pix_fmt=yuv420p
  
  sample_aspect_ratio=N/A
  
  pict_type=I
  
  coded_picture_number=58
  
  display_picture_number=0
  
  interlaced_frame=0
  
  top_field_first=0
  
  repeat_pict=0
  
  [/FRAME]
  
  [FRAME]
  
  media_type=video
  
  stream_index=0
  
  key_frame=0
  
  pkt_pts=15360
  
  pkt_pts_time=1.000000
  
  pkt_dts=15360
  
  pkt_dts_time=1.000000
  
  best_effort_timestamp=15360
  
  best_effort_timestamp_time=1.000000
  
  pkt_duration=512
  
  pkt_duration_time=0.033333
  
  pkt_pos=823416
  
  pkt_size=29642
  
  width=1920
  
  height=1080
  
  pix_fmt=yuv420p
  
  sample_aspect_ratio=N/A
  
  pict_type=P
  
  coded_picture_number=60
  
  display_picture_number=0
  
  interlaced_frame=0
  
  top_field_first=0
  
  repeat_pict=0
  
  [/FRAME]
  
  [FRAME]
  
  media_type=video
  
  stream_index=0
  
  key_frame=1
  
  pkt_pts=30208
  
  pkt_pts_time=1.966667
  
  pkt_dts=30208
  
  pkt_dts_time=1.966667
  
  best_effort_timestamp=30208
  
  best_effort_timestamp_time=1.966667
  
  pkt_duration=512
  
  pkt_duration_time=0.033333
  
  pkt_pos=1546454
  
  pkt_size=66021
  
  width=1920
  
  height=1080
  
  pix_fmt=yuv420p
  
  sample_aspect_ratio=N/A
  
  pict_type=I
  
  coded_picture_number=117
  
  display_picture_number=0
  
  interlaced_frame=0
  
  top_field_first=0
  
  repeat_pict=0
  
  [/FRAME]

  I immediately noticed that pkt_pts_time and other fields are not starting from 0 and are not increasing linearly (respect to the 30 fps that I established). It is possible to obtain the same video writing the AVPackets in a second moment as I’m doing ? Thank you.

• Past duration 0.750206 too large when capturing my screen

  2 avril 2018, par it_is_a_literature

  I want to capture my screen with ffmpeg

  xrandr
  
  Screen 0: minimum 320 x 200, current 1920 x 1080, maximum 8192 x 8192
  
  ffmpeg -f x11grab -framerate 30 -threads 0 -s 1920 x 1080 -i :0 /tmp/capture.mp4

  Error info :

  [x11grab @ 0x556025373ae0] Stream #0: not enough frames to estimate rate; consider increasing probesize
  
  Past duration 0.750206 too large     435kB time=00:00:01.00 bitrate=3564.5kbits/s dup=19 drop=0 speed=0.382x    
  
  Past duration 0.749168 too large

  Try other ffmpeg command.

  ffmpeg -f x11grab -video_size 1360x768 -framerate 30 -i :0.0 -f pulse -i default -preset ultrafast -crf 18 -pix_fmt yuv420p out.mkv
  
  
  
  [x11grab @ 0x55db2f72fc00] Thread message queue blocking; consider raising the thread_queue_size option (current value: 8)
  
  [pulse @ 0x55db2f7374e0] Thread message queue blocking; consider raising the thread_queue_size option (current value: 8)
  
  [output stream 0:1 @ 0x55db2f7784e0] 100 buffers queued in output stream 0:1, something may be wrong.
  
  [libvorbis @ 0x55db2f753720] Queue input is backward in time
  
  [matroska @ 0x55db2f750760] Non-monotonous DTS in output stream 0:1; previous: 3411, current: 3272; changing to 3411. This may result in incorrect timestamps in the output file.

  How to fix it ?