Recherche avancée

Sur d’autres sites (6013)

• Revision 0e1068a4bd : Improve two pass VBR accuracy. Adjustments to the GF interval choice and minimu

  26 septembre 2014, par Paul Wilkins

  Changed Paths :
   Modify /vp9/encoder/vp9_encodeframe.c

  
   Modify /vp9/encoder/vp9_firstpass.c

  
   Modify /vp9/encoder/vp9_ratectrl.c

  
  
  Improve two pass VBR accuracy.

  Adjustments to the GF interval choice and minimum boost.
  Adjustment to the calculation of 2 pass worst q.
  Compared to 09/29 head there is metrics hit on derf of
  (-0.123%,-0.191%)

  Compared to the September 29 head and a baseline on
  September 18 baseline the accuracy of the VBR rate control
  measured on the derf set is as follows :-

  Mean error % / Mean abs(error %)
  Sept 18 baseline (-7.0% / 14.76%)
  Sept 29 head (-15.7%, 19.8%)
  This check in (-1.5% / 14.4%)

  The mean undershoot is reduced slightly but the
  worst case overshoot on e.g. harbour/highway is
  increased. This will be addressed in a later patch.

  Change-Id : Iffd9b0ab7432a131c98fbaaa82d1e5b40be72b58

• Fragmented MP4 - problem playing in browser

  12 juin 2019, par PookyFan

  I try to create fragmented MP4 from raw H264 video data so I could play it in internet browser’s player. My goal is to create live streaming system, where media server would send fragmented MP4 pieces to browser. The server would buffer input data from RaspberryPi camera, which sends video as H264 frames. It would then mux that video data and make it available for client. The browser would play media data (that were muxed by server and sent i.e. through websocket) by using Media Source Extensions.

  For test purpose I wrote the following pieces of code (using many examples I found in the intenet) :

  C++ application using avcodec which muxes raw H264 video to fragmented MP4 and saves it to a file :

  #define READBUFSIZE 4096
  
  #define IOBUFSIZE 4096
  
  #define ERRMSGSIZE 128
  
  
  
  #include <cstdint>
  
  #include <iostream>
  
  #include <fstream>
  
  #include <string>
  
  #include <vector>
  
  
  
  extern "C"
  
  {
  
      #include <libavformat></libavformat>avformat.h>
  
      #include <libavutil></libavutil>error.h>
  
      #include <libavutil></libavutil>opt.h>
  
  }
  
  
  
  enum NalType : uint8_t
  
  {
  
      //NALs containing stream metadata
  
      SEQ_PARAM_SET = 0x7,
  
      PIC_PARAM_SET = 0x8
  
  };
  
  
  
  std::vector outputData;
  
  
  
  int mediaMuxCallback(void *opaque, uint8_t *buf, int bufSize)
  
  {
  
      outputData.insert(outputData.end(), buf, buf + bufSize);
  
      return bufSize;
  
  }
  
  
  
  std::string getAvErrorString(int errNr)
  
  {
  
      char errMsg[ERRMSGSIZE];
  
      av_strerror(errNr, errMsg, ERRMSGSIZE);
  
      return std::string(errMsg);
  
  }
  
  
  
  int main(int argc, char **argv)
  
  {
  
      if(argc &lt; 2)
  
      {
  
          std::cout &lt;&lt; "Missing file name" &lt;&lt; std::endl;
  
          return 1;
  
      }
  
  
  
      std::fstream file(argv[1], std::ios::in | std::ios::binary);
  
      if(!file.is_open())
  
      {
  
          std::cout &lt;&lt; "Couldn't open file " &lt;&lt; argv[1] &lt;&lt; std::endl;
  
          return 2;
  
      }
  
  
  
      std::vector inputMediaData;
  
      do
  
      {
  
          char buf[READBUFSIZE];
  
          file.read(buf, READBUFSIZE);
  
  
  
          int size = file.gcount();
  
          if(size > 0)
  
              inputMediaData.insert(inputMediaData.end(), buf, buf + size);
  
      } while(!file.eof());
  
      file.close();
  
  
  
      //Initialize avcodec
  
      av_register_all();
  
      uint8_t *ioBuffer;
  
      AVCodec *codec = avcodec_find_decoder(AV_CODEC_ID_H264);
  
      AVCodecContext *codecCtxt = avcodec_alloc_context3(codec);
  
      AVCodecParserContext *parserCtxt = av_parser_init(AV_CODEC_ID_H264);
  
      AVOutputFormat *outputFormat = av_guess_format("mp4", nullptr, nullptr);
  
      AVFormatContext *formatCtxt;
  
      AVIOContext *ioCtxt;
  
      AVStream *videoStream;
  
  
  
      int res = avformat_alloc_output_context2(&amp;formatCtxt, outputFormat, nullptr, nullptr);
  
      if(res &lt; 0)
  
      {
  
          std::cout &lt;&lt; "Couldn't initialize format context; the error was: " &lt;&lt; getAvErrorString(res) &lt;&lt; std::endl;
  
          return 3;
  
      }
  
  
  
      if((videoStream = avformat_new_stream( formatCtxt, avcodec_find_encoder(formatCtxt->oformat->video_codec) )) == nullptr)
  
      {
  
          std::cout &lt;&lt; "Couldn't initialize video stream" &lt;&lt; std::endl;
  
          return 4;
  
      }
  
      else if(!codec)
  
      {
  
          std::cout &lt;&lt; "Couldn't initialize codec" &lt;&lt; std::endl;
  
          return 5;
  
      }
  
      else if(codecCtxt == nullptr)
  
      {
  
          std::cout &lt;&lt; "Couldn't initialize codec context" &lt;&lt; std::endl;
  
          return 6;
  
      }
  
      else if(parserCtxt == nullptr)
  
      {
  
          std::cout &lt;&lt; "Couldn't initialize parser context" &lt;&lt; std::endl;
  
          return 7;
  
      }
  
      else if((ioBuffer = (uint8_t*)av_malloc(IOBUFSIZE)) == nullptr)
  
      {
  
          std::cout &lt;&lt; "Couldn't allocate I/O buffer" &lt;&lt; std::endl;
  
          return 8;
  
      }
  
      else if((ioCtxt = avio_alloc_context(ioBuffer, IOBUFSIZE, 1, nullptr, nullptr, mediaMuxCallback, nullptr)) == nullptr)
  
      {
  
          std::cout &lt;&lt; "Couldn't initialize I/O context" &lt;&lt; std::endl;
  
          return 9;
  
      }
  
  
  
      //Set video stream data
  
      videoStream->id = formatCtxt->nb_streams - 1;
  
      videoStream->codec->width = 1280;
  
      videoStream->codec->height = 720;
  
      videoStream->time_base.den = 60; //FPS
  
      videoStream->time_base.num = 1;
  
      videoStream->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
  
      formatCtxt->pb = ioCtxt;
  
  
  
      //Retrieve SPS and PPS for codec extdata
  
      const uint32_t synchMarker = 0x01000000;
  
      unsigned int i = 0;
  
      int spsStart = -1, ppsStart = -1;
  
      uint16_t spsSize = 0, ppsSize = 0;
  
      while(spsSize == 0 || ppsSize == 0)
  
      {
  
          uint32_t *curr =  (uint32_t*)(inputMediaData.data() + i);
  
          if(*curr == synchMarker)
  
          {
  
              unsigned int currentNalStart = i;
  
              i += sizeof(uint32_t);
  
              uint8_t nalType = inputMediaData.data()[i] &amp; 0x1F;
  
              if(nalType == SEQ_PARAM_SET)
  
                  spsStart = currentNalStart;
  
              else if(nalType == PIC_PARAM_SET)
  
                  ppsStart = currentNalStart;
  
  
  
              if(spsStart >= 0 &amp;&amp; spsSize == 0 &amp;&amp; spsStart != i)
  
                  spsSize = currentNalStart - spsStart;
  
              else if(ppsStart >= 0 &amp;&amp; ppsSize == 0 &amp;&amp; ppsStart != i)
  
                  ppsSize = currentNalStart - ppsStart;
  
          }
  
          ++i;
  
      }
  
  
  
      videoStream->codec->extradata = inputMediaData.data() + spsStart;
  
      videoStream->codec->extradata_size = ppsStart + ppsSize;
  
  
  
      //Write main header
  
      AVDictionary *options = nullptr;
  
      av_dict_set(&amp;options, "movflags", "frag_custom+empty_moov", 0);
  
      res = avformat_write_header(formatCtxt, &amp;options);
  
      if(res &lt; 0)
  
      {
  
          std::cout &lt;&lt; "Couldn't write container main header; the error was: " &lt;&lt; getAvErrorString(res) &lt;&lt; std::endl;
  
          return 10;
  
      }
  
  
  
      //Retrieve frames from input video and wrap them in container
  
      int currentInputIndex = 0;
  
      int framesInSecond = 0;
  
      while(currentInputIndex &lt; inputMediaData.size())
  
      {
  
          uint8_t *frameBuffer;
  
          int frameSize;
  
          res = av_parser_parse2(parserCtxt, codecCtxt, &amp;frameBuffer, &amp;frameSize, inputMediaData.data() + currentInputIndex,
  
              inputMediaData.size() - currentInputIndex, AV_NOPTS_VALUE, AV_NOPTS_VALUE, 0);
  
          if(frameSize == 0) //No more frames while some data still remains (is that even possible?)
  
          {
  
              std::cout &lt;&lt; "Some data left unparsed: " &lt;&lt; std::to_string(inputMediaData.size() - currentInputIndex) &lt;&lt; std::endl;
  
              break;
  
          }
  
  
  
          //Prepare packet with video frame to be dumped into container
  
          AVPacket packet;
  
          av_init_packet(&amp;packet);
  
          packet.data = frameBuffer;
  
          packet.size = frameSize;
  
          packet.stream_index = videoStream->index;
  
          currentInputIndex += frameSize;
  
  
  
          //Write packet to the video stream
  
          res = av_write_frame(formatCtxt, &amp;packet);
  
          if(res &lt; 0)
  
          {
  
              std::cout &lt;&lt; "Couldn't write packet with video frame; the error was: " &lt;&lt; getAvErrorString(res) &lt;&lt; std::endl;
  
              return 11;
  
          }
  
  
  
          if(++framesInSecond == 60) //We want 1 segment per second
  
          {
  
              framesInSecond = 0;
  
              res = av_write_frame(formatCtxt, nullptr); //Flush segment
  
          }
  
      }
  
      res = av_write_frame(formatCtxt, nullptr); //Flush if something has been left
  
  
  
      //Write media data in container to file
  
      file.open("my_mp4.mp4", std::ios::out | std::ios::binary);
  
      if(!file.is_open())
  
      {
  
          std::cout &lt;&lt; "Couldn't open output file " &lt;&lt; std::endl;
  
          return 12;
  
      }
  
  
  
      file.write((char*)outputData.data(), outputData.size());
  
      if(file.fail())
  
      {
  
          std::cout &lt;&lt; "Couldn't write to file" &lt;&lt; std::endl;
  
          return 13;
  
      }
  
  
  
      std::cout &lt;&lt; "Media file muxed successfully" &lt;&lt; std::endl;
  
      return 0;
  
  }
  
  </vector></string></fstream></iostream></cstdint>

  (I hardcoded a few values, such as video dimensions or framerate, but as I said this is just a test code.)

  ---

  Simple HTML webpage using MSE to play my fragmented MP4

      
  
  
  
  
  
      <video width="1280" height="720" controls="controls">
  
      </video>
  
  
  
  <code class="echappe-js">&lt;script&gt;<br />
  
  var vidElement = document.querySelector('video');<br />
  
  <br />
  
  if (window.MediaSource) {<br />
  
    var mediaSource = new MediaSource();<br />
  
    vidElement.src = URL.createObjectURL(mediaSource);<br />
  
    mediaSource.addEventListener('sourceopen', sourceOpen);<br />
  
  } else {<br />
  
    console.log(&quot;The Media Source Extensions API is not supported.&quot;)<br />
  
  }<br />
  
  <br />
  
  function sourceOpen(e) {<br />
  
    URL.revokeObjectURL(vidElement.src);<br />
  
    var mime = 'video/mp4; codecs=&quot;avc1.640028&quot;';<br />
  
    var mediaSource = e.target;<br />
  
    var sourceBuffer = mediaSource.addSourceBuffer(mime);<br />
  
    var videoUrl = 'my_mp4.mp4';<br />
  
    fetch(videoUrl)<br />
  
      .then(function(response) {<br />
  
        return response.arrayBuffer();<br />
  
      })<br />
  
      .then(function(arrayBuffer) {<br />
  
        sourceBuffer.addEventListener('updateend', function(e) {<br />
  
          if (!sourceBuffer.updating &amp;amp;&amp;amp; mediaSource.readyState === 'open') {<br />
  
            mediaSource.endOfStream();<br />
  
          }<br />
  
        });<br />
  
        sourceBuffer.appendBuffer(arrayBuffer);<br />
  
      });<br />
  
  }<br />
  
  &lt;/script&gt;
  

  ---

  Output MP4 file generated by my C++ application can be played i.e. in MPC, but it doesn’t play in any web browser I tested it with. It also doesn’t have any duration (MPC keeps showing 00:00).

  To compare output MP4 file I got from my C++ application described above, I also used FFMPEG to create fragmented MP4 file from the same source file with raw H264 stream. I used the following command :

  ffmpeg -r 60 -i input.h264 -c:v copy -f mp4 -movflags empty_moov+default_base_moof+frag_keyframe test.mp4

  This file generated by FFMPEG is played correctly by every web browser I used for tests. It also has correct duration (but also it has trailing atom, which wouldn’t be present in my live stream anyway, and as I need a live stream, it won’t have any fixed duration in the first place).

  MP4 atoms for both files look very similiar (they have identical avcc section for sure). What’s interesting (but not sure if it’s of any importance), both files have different NALs format than input file (RPI camera produces video stream in Annex-B format, while output MP4 files contain NALs in AVCC format... or at least it looks like it’s the case when I compare mdat atoms with input H264 data).

  I assume there is some field (or a few fields) I need to set for avcodec to make it produce video stream that would be properly decoded and played by browsers players. But what field(s) do I need to set ? Or maybe problem lies somewhere else ? I ran out of ideas.

  ---

  EDIT 1 :
  As suggested, I investigated binary content of both MP4 files (generated by my app and FFMPEG tool) with hex editor. What I can confirm :

  • both files have identical avcc section (they match perfectly and are in AVCC format, I analyzed it byte after byte and there’s no mistake about it)
  • both files have NALs in AVCC format (I looked closely at mdat atoms and they don’t differ between both MP4 files)

  So I guess there’s nothing wrong with the extradata creation in my code - avcodec takes care of it properly, even if I just feed it with SPS and PPS NALs. It converts them by itself, so no need for me to do it by hand. Still, my original problem remains.

  EDIT 2 : I achieved partial success - MP4 generated by my app now plays in Firefox. I added this line to the code (along with rest of stream initialization) :

  videoStream->codec->time_base = videoStream->time_base;

  So now this section of my code looks like this :

  //Set video stream data
  
  videoStream->id = formatCtxt->nb_streams - 1;
  
  videoStream->codec->width = 1280;
  
  videoStream->codec->height = 720;
  
  videoStream->time_base.den = 60; //FPS
  
  videoStream->time_base.num = 1;
  
  videoStream->codec->time_base = videoStream->time_base;
  
  videoStream->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
  
  formatCtxt->pb = ioCtxt;

• this is not be cleared yet , some MTS video file start with several duplicated frame

  24 octobre 2018, par cool jobs

  this is not be cleared yet ,
  some MTS video file start with several duplicated frame with audio,audio is normal
  how to remove this duplicated frame with audio ?
  but i do not want remove duplicated frames all of video .
  only want to be removed video head( sequential frames only )

  ffmpeg have not sophisticated option for this ??