Recherche avancée

Sur d’autres sites (6597)

• Encoding with ffmpeg : video quality drops

  21 mai 2014, par Adi Ulici

  I’m trying to encode a set of pictures into a video with ffmpeg. I’m new to this stuff, but I managed to get it work. I have only one problem : the first second or two of video look good but as the time goes the video quality keeps dropping and dropping. At the end (it’s a video of about 16 seconds) the quality is really bad and I can’t understand why.

  I’m using AV_CODEC_ID_MPEG1VIDEO as video codec (frankly, it’s the only one I could make work) and here is a sample of my code :

  c = avcodec_alloc_context3(codec);
  
  if (!c) {
  
      fprintf(stderr, "Could not allocate video codec context\n");
  
      exit(1);
  
  }
  
  
  
  c->bit_rate = 400000;
  
  c->width = width;
  
  c->height = height;
  
  struct AVRational time_base = {1, 25};
  
  c->time_base = time_base;
  
  c->gop_size = 10;
  
  c->max_b_frames = 1;
  
  c->pix_fmt = AV_PIX_FMT_YUV420P;
  
  
  
  if (codec_id == AV_CODEC_ID_H264)
  
      av_opt_set(c->priv_data, "preset", "slow", 0);
  
  
  
  // Open the codec
  
  if (avcodec_open2(c, codec, NULL) < 0) {
  
      fprintf(stderr, "Could not open codec\n");
  
      exit(1);
  
  }
  
  
  
  fopen_s(&f, filename, "wb");
  
  if (!f) {
  
      fprintf(stderr, "Could not open %s\n", filename);
  
      exit(1);
  
  }
  
  
  
  frame = av_frame_alloc();
  
  if (!frame) {
  
      fprintf(stderr, "Could not allocate video frame\n");
  
      exit(1);
  
  }
  
  frame->format = c->pix_fmt;
  
  frame->width  = c->width;
  
  frame->height = c->height;
  
  
  
  ret = av_image_alloc(frame->data, frame->linesize, c->width, c->height,
  
                       c->pix_fmt, 32);
  
  if (ret < 0) {
  
      fprintf(stderr, "Could not allocate raw picture buffer\n");
  
      exit(1);
  
  }
  
  
  
  struct SwsContext *frameConvertContext = sws_getContext(width, height,
  
        PIX_FMT_RGB24,
  
        c->width, c->height,
  
        c->pix_fmt,
  
        SWS_LANCZOS | SWS_ACCURATE_RND, NULL, NULL, NULL);
  
  
  
  for (i = 0; i < framesNr; i++) {
  
      // Read an image
  
      std::stringstream filename;
  
      filename << "input/image-" << (i+1) << ".jpg";
  
      Image* img;
  
      img = Image::fromJpeg((char*) filename.str().c_str());
  
      int srcSliceY[1] = { img->getWidth() * 3 };
  
      const uint8_t* imgbuf[1] = { (uint8_t*) img->getData(sizeof(uint8_t)) };
  
  
  
      av_init_packet(&pkt);
  
      pkt.data = NULL;
  
      pkt.size = 0;
  
  
  
      fflush(stdout);
  
      frameConvertContext = sws_getCachedContext(frameConvertContext, width, height,
  
        PIX_FMT_RGB24,
  
        c->width, c->height,
  
        c->pix_fmt,
  
        SWS_LANCZOS | SWS_ACCURATE_RND, NULL, NULL, NULL);
  
  
  
      sws_scale(frameConvertContext, imgbuf, srcSliceY, 0, img->getHeight(), frame->data, frame->linesize);
  
  
  
      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(".");
  
          fwrite(pkt.data, 1, pkt.size, f);
  
          av_free_packet(&pkt);
  
      }
  
  
  
      // Free the memory
  
      delete img;
  
  }

  Any tips ? Thanks a lot !

• ffmpeg seeking drops consistently two frames from the start of segments

  27 janvier 2018, par Fernando D'Andrea

  I’m trying to script-cut a 24 fps video and it gets close to the right thing, except for two missing frames at the beginning of every segment I cut with seeking.

  I’ve found out that index-frames (keyframes) could become an issue for seeking, so I first create an intermediate file with every frame as a index frame using this line :

  ffmpeg -i input.mov -c:v libx264 -g 1 -c:a copy intermediate.mp4

  Then I proceed to make a cut with this line. Notice I’m drawing some text on top of it :

  ffmpeg -hide_banner -y -i intermediate.mp4 -ss 00:00:02.083333333333 -to 00:00:02.875000000000 -vf drawtext=fontfile=DroidSans.ttf:fontsize=48:text="segment2":fontcolor=yellow:box=1:boxborderw=16:boxcolor=black@0.5:x=w/32:y=h/18 output.mp4

  No matter the length of the segment, the first two frames get dropped.

  I’m using python do control ffmpeg.

  Any ideas ?

  ---

  Edit :

  Running ffprobe on one of the generated segments with ffprobe -hide_banner Nixing_0010_0010.mp4 says the following :

  Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'Nixing_0010_0010.mp4':

  Metadata:
   major_brand     : isom
   minor_version   : 512
   compatible_brands: isomiso2avc1mp41
   encoder         : Lavf57.82.101
 Duration: 00:00:02.11, start: 0.000000, bitrate: 4717 kb/s
   Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 1920x1080 [SAR 1:1 DAR 16:9], 4751 kb/s, 24 fps, 24 tbr, 12288 tbn, 48 tbc (default)
   Metadata:
     handler_name    : VideoHandler
     timecode        : 00:00:00:00
   Stream #0:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 2 kb/s (default)
   Metadata:
     handler_name    : SoundHandler
   Stream #0:2(eng): Data: none (tmcd / 0x64636D74), 0 kb/s
   Metadata:
     handler_name    : TimeCodeHandler
     timecode        : 00:00:00:00

• lavu/adler32 : Fix doxy group definition, take two

  16 juin 2013, par James Almer

  lavu/adler32 : Fix doxy group definition, take two
  The description if for the function, not the group.
  Signed-off-by : James Almer <jamrial@gmail.com>
  
  Signed-off-by : Michael Niedermayer <michaelni@gmx.at>
  

  • [DH] libavutil/adler32.h