Sur d’autres sites (12803)

• ffmpeg avcodec_send_packet/avcodec_receive_frame memory leak

  23 janvier 2019, par G Hamlin

  I’m attempting to decode frames, but memory usage grows with every frame (more specifically, with every call to avcodec_send_packet) until finally the code crashes with a bad_alloc. Here’s the basic decode loop :

  int rfret = 0;
  
  while((rfret = av_read_frame(inctx.get(), &packet)) >= 0){
  
      if (packet.stream_index == vstrm_idx) {
  
  
  
          //std::cout << "Sending Packet" << std::endl;
  
          int ret = avcodec_send_packet(ctx.get(), &packet);
  
          if (ret < 0 || ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
  
              std::cout << "avcodec_send_packet: " << ret << std::endl;
  
              break;
  
          }
  
  
  
          while (ret  >= 0) {
  
              //std::cout << "Receiving Frame" << std::endl;
  
              ret = avcodec_receive_frame(ctx.get(), fr);
  
              if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
  
                  //std::cout << "avcodec_receive_frame: " << ret << std::endl;
  
                  av_frame_unref(fr);
  
                  // av_frame_free(&fr);
  
                  break;
  
              }
  
  
  
              std::cout << "frame: " << ctx->frame_number << std::endl;
  
  
  
              // eventually do something with the frame here...
  
  
  
              av_frame_unref(fr);
  
              // av_frame_free(&fr);
  
          }
  
      }
  
      else {
  
          //std::cout << "Not Video" << std::endl;
  
      }
  
      av_packet_unref(&packet);
  
  }

  Memory usage/leakage seems to scale with the resolution of the video I’m decoding. For example, for a 3840x2160 resolution video, the memory usage in windows task manager consistently jumps up by about 8mb (1 byte per pixel ??) for each received frame. Do I need to do something besides call av_frame_unref to release the memory ?

  (more) complete code below

  ---

  void AVFormatContextDeleter(AVFormatContext* ptr)
  
  {
  
      if (ptr) {
  
          avformat_close_input(&ptr);
  
      }
  
  }
  
  
  
  void AVCodecContextDeleter(AVCodecContext* ptr)
  
  {
  
      if (ptr) {
  
          avcodec_free_context(&ptr);
  
      }
  
  }
  
  
  
  typedef std::unique_ptr AVFormatContextPtr;
  
  typedef std::unique_ptr AVCodecContextPtr;
  
  
  
  AVCodecContextPtr createAvCodecContext(AVCodec *vcodec)
  
  {
  
      AVCodecContextPtr ctx(avcodec_alloc_context3(vcodec), AVCodecContextDeleter);
  
      return ctx;
  
  }
  
  
  
  AVFormatContextPtr createFormatContext(const std::string& filename)
  
  {
  
      AVFormatContext* inctxPtr = nullptr;
  
      int ret = avformat_open_input(&inctxPtr, filename.c_str(), nullptr, nullptr);
  
      //    int ret = avformat_open_input(&inctx, "D:/Videos/test.mp4", nullptr, nullptr);
  
      if (ret != 0) {
  
          inctxPtr = nullptr;
  
      }
  
  
  
      return AVFormatContextPtr(inctxPtr, AVFormatContextDeleter);
  
  }
  
  
  
  int testDecode()
  
  {
  
      // open input file context
  
      AVFormatContextPtr inctx = createFormatContext("D:/Videos/Matt Chapman Hi Greg.MOV");
  
  
  
      if (!inctx) {
  
          // std::cerr << "fail to avforamt_open_input(\"" << infile << "\"): ret=" << ret;
  
          return 1;
  
      }
  
  
  
      // retrieve input stream information
  
      int ret = avformat_find_stream_info(inctx.get(), nullptr);
  
      if (ret < 0) {
  
          //std::cerr << "fail to avformat_find_stream_info: ret=" << ret;
  
          return 2;
  
      }
  
  
  
      // find primary video stream
  
      AVCodec* vcodec = nullptr;
  
      const int vstrm_idx = av_find_best_stream(inctx.get(), AVMEDIA_TYPE_VIDEO, -1, -1, &vcodec, 0);
  
      if (vstrm_idx < 0) {
  
          //std::cerr << "fail to av_find_best_stream: vstrm_idx=" << vstrm_idx;
  
          return 3;
  
      }
  
  
  
      AVCodecParameters* origin_par = inctx->streams[vstrm_idx]->codecpar;
  
      if (vcodec == nullptr) {  // is this even necessary?
  
          vcodec = avcodec_find_decoder(origin_par->codec_id);
  
          if (!vcodec) {
  
              // Can't find decoder
  
              return 4;
  
          }
  
      }
  
  
  
      AVCodecContextPtr ctx = createAvCodecContext(vcodec);
  
      if (!ctx) {
  
          return 5;
  
      }
  
  
  
      ret = avcodec_parameters_to_context(ctx.get(), origin_par);
  
      if (ret) {
  
          return 6;
  
      }
  
  
  
      ret = avcodec_open2(ctx.get(), vcodec, nullptr);
  
      if (ret < 0) {
  
          return 7;
  
      }
  
  
  
      //print input video stream informataion
  
      std::cout
  
              //<< "infile: " << infile << "\n"
  
              << "format: " << inctx->iformat->name << "\n"
  
              << "vcodec: " << vcodec->name << "\n"
  
              << "size:   " << origin_par->width << 'x' << origin_par->height << "\n"
  
              << "fps:    " << av_q2d(ctx->framerate) << " [fps]\n"
  
              << "length: " << av_rescale_q(inctx->duration, ctx->time_base, {1,1000}) / 1000. << " [sec]\n"
  
              << "pixfmt: " << av_get_pix_fmt_name(ctx->pix_fmt) << "\n"
  
              << "frame:  " << inctx->streams[vstrm_idx]->nb_frames << "\n"
  
              << std::flush;
  
  
  
  
  
      AVPacket packet;
  
  
  
      av_init_packet(&packet);
  
      packet.data = nullptr;
  
      packet.size = 0;
  
  
  
      AVFrame *fr = av_frame_alloc();
  
      if (!fr) {
  
          return 8;
  
      }
  
  
  
      int rfret = 0;
  
      while((rfret = av_read_frame(inctx.get(), &packet)) >= 0){
  
          if (packet.stream_index == vstrm_idx) {
  
  
  
              //std::cout << "Sending Packet" << std::endl;
  
              int ret = avcodec_send_packet(ctx.get(), &packet);
  
              if (ret < 0 || ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
  
                  std::cout << "avcodec_send_packet: " << ret << std::endl;
  
                  break;
  
              }
  
  
  
              while (ret  >= 0) {
  
                  //std::cout << "Receiving Frame" << std::endl;
  
                  ret = avcodec_receive_frame(ctx.get(), fr);
  
                  if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
  
                      //std::cout << "avcodec_receive_frame: " << ret << std::endl;
  
                      av_frame_unref(fr);
  
                      // av_frame_free(&fr);
  
                      break;
  
                  }
  
  
  
                  std::cout << "frame: " << ctx->frame_number << std::endl;
  
  
  
                  // do something with the frame here...
  
  
  
                  av_frame_unref(fr);
  
                  // av_frame_free(&fr);
  
              }
  
          }
  
          else {
  
              //std::cout << "Not Video" << std::endl;
  
          }
  
          av_packet_unref(&packet);
  
      }
  
  
  
      std::cout << "RFRET = " << rfret << std::endl;
  
  
  
      return 0;
  
  }

  Update 1 : (1/21/2019) Compiling on a different machine and running with different video files I am not seeing the memory usage growing without bound. I’ll try to narrow down where the difference lies (compiler ?, ffmpeg version ?, or video encoding ?)

  Update 2 : (1/21/2019) Ok, it looks like there is some interaction occurring between ffmpeg and Qt’s QCamera. In my application, I’m using Qt to manage the webcam, but decided to use ffmpeg libraries to handle decoding/encoding since Qt doesn’t have as comprehensive support for different codecs. If I have the camera turned on (through Qt), ffmpeg decoding memory consumption grows without bound. If the camera is off, ffmpeg behaves fine. I’ve tried this both with a physical camera (Logitech C920) and with a virtual camera using OBS-Virtualcam, with the same result. So far I’m baffled as to how the two systems are interacting...

• Video concatenation puts sound out of sync

  9 août 2019, par mmorin

  (Cross-posted from Video Production, where the question received no answers and may be more technical than usual video production.)

  I have several MOV files from a DSLR camera. I concatenate them with directions from this thread :

  ffmpeg -safe 0 -f concat -i files_to_combine -vcodec copy -acodec copy temp.MOV

  where files_to_combine is :

  file ./DSC_0013.MOV
  
  ...
  
  file ./DSC_0019.MOV

  The result has image and sound in sync for the first clip and is out of sync by fractions of a second in the second clip, and out of sync by around a second for the last clip. It is probably related to this error from the log :

  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dd802200] st: 0 edit list: 1 Missing key frame while searching for timestamp: 1000
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dd802200] st: 0 edit list 1 Cannot find an index entry before timestamp: 1000.
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dd802200] Auto-inserting h264_mp4toannexb bitstream filter

  How can I trim the frames to the available sound stream, then concatenate the two videos ?

  The full log from the ffmpeg command is :

  ffmpeg version 4.1.3 Copyright (c) 2000-2019 the FFmpeg developers
  
    built with Apple LLVM version 10.0.1 (clang-1001.0.46.4)
  
    configuration: --prefix=/usr/local/Cellar/ffmpeg/4.1.3_1 --enable-shared --enable-pthreads --enable-version3 --enable-hardcoded-tables --enable-avresample --cc=clang --host-cflags='-I/Library/Java/JavaVirtualMachines/adoptopenjdk-11.0.2.jdk/Contents/Home/include -I/Library/Java/JavaVirtualMachines/adoptopenjdk-11.0.2.jdk/Contents/Home/include/darwin' --host-ldflags= --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libbluray --enable-libmp3lame --enable-libopus --enable-librubberband --enable-libsnappy --enable-libtesseract --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libx265 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librtmp --enable-libspeex --enable-videotoolbox --disable-libjack --disable-indev=jack --enable-libaom --enable-libsoxr
  
    libavutil      56. 22.100 / 56. 22.100
  
    libavcodec     58. 35.100 / 58. 35.100
  
    libavformat    58. 20.100 / 58. 20.100
  
    libavdevice    58.  5.100 / 58.  5.100
  
    libavfilter     7. 40.101 /  7. 40.101
  
    libavresample   4.  0.  0 /  4.  0.  0
  
    libswscale      5.  3.100 /  5.  3.100
  
    libswresample   3.  3.100 /  3.  3.100
  
    libpostproc    55.  3.100 / 55.  3.100
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dc00e000] Auto-inserting h264_mp4toannexb bitstream filter
  
  Input #0, concat, from 'files_to_combine':
  
    Duration: N/A, start: -0.592000, bitrate: 36888 kb/s
  
      Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661), yuvj420p(pc, smpte170m/bt709/bt470m), 1920x1080, 35352 kb/s, 50 fps, 50 tbr, 50k tbn, 100 tbc
  
      Metadata:
  
        handler_name    : VideoHandler
  
      Stream #0:1(eng): Audio: pcm_s16le (sowt / 0x74776F73), 48000 Hz, stereo, s16, 1536 kb/s
  
      Metadata:
  
        handler_name    : SoundHandler
  
  Output #0, mov, to 'temp.MOV':
  
    Metadata:
  
      encoder         : Lavf58.20.100
  
      Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661), yuvj420p(pc, smpte170m/bt709/bt470m), 1920x1080, q=2-31, 35352 kb/s, 50 fps, 50 tbr, 50k tbn, 50k tbc
  
      Metadata:
  
        handler_name    : VideoHandler
  
      Stream #0:1(eng): Audio: pcm_s16le (sowt / 0x74776F73), 48000 Hz, stereo, s16, 1536 kb/s
  
      Metadata:
  
        handler_name    : SoundHandler
  
  Stream mapping:
  
    Stream #0:0 -> #0:0 (copy)
  
    Stream #0:1 -> #0:1 (copy)
  
  Press [q] to stop, [?] for help
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dd802200] st: 0 edit list: 1 Missing key frame while searching for timestamp: 1000
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dd802200] st: 0 edit list 1 Cannot find an index entry before timestamp: 1000.
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f82dd802200] Auto-inserting h264_mp4toannexb bitstream filter
  
  frame=41886 fps=547 q=-1.0 Lsize= 3789826kB time=00:13:58.75 bitrate=37014.8kbits/s speed=10.9x    
  
  video:3631879kB audio:157123kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.021759%

  Update (1 July 2019)

  I thought that the files had a problem at the beginning or at the end, so I
  trimmed one second from each end, but it still had the sound out of sync :

  FILES=files_to_combine
  
  OUTPUT=show2.MOV
  
  rm $FILES
  
  for i in 3 4 5 6 7 8 9; do
  
      rm ${i}.MOV
  
      duration=$(ffprobe -v 0 -show_entries format=duration -of compact=p=0:nk=1  DSC_001${i}.MOV)
  
      trimmed=$(echo $duration - 1 | bc)
  
      ffmpeg -ss 1 -t $trimmed -i DSC_001${i}.MOV -vcodec copy -acodec copy ${i}.MOV
  
      echo file ./${i}.MOV >> $FILES
  
  done
  
  
  
  rm $OUTPUT
  
  ffmpeg -safe 0 -f concat -i $FILES -vcodec copy -acodec copy $OUTPUT

  When I trim a single file near the end, the sound and video do not seem out of sync :

  ffmpeg -ss 00:09:20 -t 20 -i DSC_0014.MOV -vcodec copy -acodec copy end.MOV

  When I concatenate only 30 seconds from each video, the result seems OK :

  FILES=files_to_combine
  
  OUTPUT=show2.MOV
  
  rm $FILES
  
  for i in 3 4 5 6 7 8 9; do
  
      rm ${i}.MOV
  
      duration=$(ffprobe -v 0 -show_entries format=duration -of compact=p=0:nk=1  DSC_001${i}.MOV)
  
      start=$(echo $duration - 30 | bc)
  
      end=$(echo $duration - 1 | bc)
  
      ffmpeg -ss $start -t $end -i DSC_001${i}.MOV -vcodec copy -acodec copy ${i}.MOV
  
      echo file ./${i}.MOV >> $FILES
  
  done
  
  
  
  rm $OUTPUT
  
  ffmpeg -safe 0 -f concat -i $FILES -vcodec copy -acodec copy $OUTPUT

  This last concatenation gives this error multiple times :

  [mov @ 0x7fc3c7837400] Non-monotonous DTS in output stream 0:0; previous: 9080205, current: 9080200; changing to 9080206. This may result in incorrect timestamps in the output file.

  So I am guessing that the problem is small differences in timestamps that
  accumulate and become more noticeable with longer durations and the
  concatenation of multiple files.

  For reference, the DSLR that shot these clips is a Nikon D3300 and the result
  of ffprobe on one of the files is :

  $ ffprobe DSC_0017.MOV -hide_banner
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7fab70003800] st: 0 edit list: 1 Missing key frame while searching for timestamp: 1000
  
  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7fab70003800] st: 0 edit list 1 Cannot find an index entry before timestamp: 1000.
  
  Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'DSC_0017.MOV':
  
    Metadata:
  
      major_brand     : qt  
  
      minor_version   : 537331968
  
      compatible_brands: qt  niko
  
      creation_time   : 2019-06-12T23:52:37.000000Z
  
    Duration: 00:09:53.58, start: 0.000000, bitrate: 36843 kb/s
  
      Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661), yuvj420p(pc, smpte170m/bt709/bt470m), 1920x1080, 35300 kb/s, 50 fps, 50 tbr, 50k tbn, 100 tbc (default)
  
      Metadata:
  
        creation_time   : 2019-06-12T23:52:37.000000Z
  
      Stream #0:1(eng): Audio: pcm_s16le (sowt / 0x74776F73), 48000 Hz, 2 channels, s16, 1536 kb/s (default)
  
      Metadata:
  
        creation_time   : 2019-06-12T23:52:37.000000Z

  Update (9 August 2019)

  I concatenated the files in iMovie and the sound and image are not as out of sync as with FFMPEG. Maybe iMovie aligns the timestamps at the end of each clip instead of concatenating the audio and image streams separately.

  I ran the concatenation again with the latest ffmpeg 4.1.4_1 on these files and others from the same camera. The audio and image are in sync in one case (the results lasts 46 minutes) out of sync in another (the result lasts 48 minutes).

• What is Google Analytics data sampling and what’s so bad about it ?

  16 août 2019, par Joselyn Khor — Analytics Tips, Development
  What is Google Analytics data sampling, and what’s so bad about it ?

  Google (2019) explains what data sampling is :

  “In data analysis, sampling is the practice of analysing a subset of all data in order to uncover the meaningful information in the larger data set.”[1]

  This is basically saying instead of analysing all of the data, there’s a threshold on how much data is analysed and any data after that will be an assumption based on patterns.

  Google’s (2019) data sampling thresholds :

  Ad-hoc queries of your data are subject to the following general thresholds for sampling :
  [Google] Analytics Standard : 500k sessions at the property level for the date range you are using
  [Google] Analytics 360 : 100M sessions at the view level for the date range you are using (para. 3) [2]

  This threshold is limiting because your data in GA may become more inaccurate as the traffic to your website increases.

  Say you’re looking through all your traffic data from the last year and find you have 5 million page views. Only 500K of that 5 million is accurate ! The data for the remaining 4.5 million (90%) is an assumption based on the 500K sample size.

  This is a key weapon Google uses to sell to large businesses. In order to increase that threshold for more accurate reporting, upgrading to premium Google Analytics 360 for approximately US$150,000 per year seems to be the only choice.

  What’s so bad about data sampling ?

  It’s unfair to say sampled data is to be disregarded completely. There is a calculation ensuring it is representative and can allow you to get good enough insights. However, we don’t encourage it as we don’t just want “good enough” data. We want the actual facts.

  In a recent survey sent to Matomo customers, we found a large proportion of users switched from GA to Matomo due to the data sampling issue.

  The two reasons why data sampling isn’t preferable : 

  1. If the selected sample size is too small, you won’t get a good representative of all the data. 
  2. The bigger your website grows, the more inaccurate your reports will become.

  An example of why we don’t fully trust sampled data is, say you have an ecommerce store and see your GA revenue reports aren’t matching the actual sales data, due to data sampling. In GA you may be seeing revenue for the month as $1 million, instead of actual sales of $800K.

  The sampling here has caused an inaccuracy that could have negative financial implications. What you get in the GA report is an estimated dollar figure rather than the actual sales. Making decisions based on inaccurate data can be costly in this case. 

  Another disadvantage to sampled data is that you might be missing out on opportunities you would’ve noticed if you were given a view of the whole. E.g. not being able to see real patterns occurring due to the data already being predicted. 

  By not getting a chance to see things as they are and only being able to jump to the conclusions and assumptions made by GA is risky. The bigger your business grows, the less you can risk making business decisions based on assumptions that could be inaccurate. 

  If you feel you could be missing out on opportunities because your GA data is sampled data, get 100% accurately reported data. 

  The benefits of 100% accurate data

  Matomo doesn’t use data sampling on any of our products or plans. You get to see all of your data and not a sampled data set.

  Data quality is necessary for high impact decision-making. It’s hard to make strategic changes if you don’t have confidence that your data is reliable and accurate.

  Learn about how Matomo is a serious contender to Google Analytics 360. 

  Now you can import your Google Analytics data directly into your Matomo

  If you’re wanting to make the switch to Matomo but worried about losing all your historic Google Analytics data, you can now import this directly into your Matomo with the Google Analytics Importer tool.

  
  Take the challenge !

  Compare your Google Analytics data (sampled data) against your Matomo data, or if you don’t have Matomo data yet, sign up to our 30-day free trial and start tracking !

  References :

  [1 & 2] About data sampling. (2019). In Analytics Help About data sampling. Retrieved August 14, 2019, from https://support.google.com/analytics/answer/2637192