Recherche avancée

Sur d’autres sites (5900)

• Create one mpeg-dash segment

  19 février 2018, par Yuriy Pryyma

  I have video encoded in mpeg-dash(*.mpd, init segments, video segments)
  Task is to replace one of video segments with segment generated by me. And so video is still can be played.

  I read properties of one segment(combining init segment and video segment) using

  avformat_open_input
  
  avformat_find_stream_info

  1) I have tried to use dash muxer to generate new segment.

  avformat_alloc_output_context2(&avFormatContext, NULL, "dash", filename)

  But the problem is it creates mpd manifest, and other segments. So how to force it to create one segment ?

  2) When I parsed segment I noticed that it had format "QuickTime / MOV".

  avformat_alloc_output_context2(&avFormatContext, NULL, "mov", filename);

  So my another idea was to use "mov" encoder. And set properties same as in initial segment(start time, duration, resolution).
  But here the question is how to remove init header ? Because I already have init segment.

  Are there any other approaches to generate one mpeg-dash segment and how to solve my problems ?

• Play HLS segments through Media source extensions

  11 février 2018, par ler

  I got a list of m4s and init.mp4 from this FFMPEG command

  ffmpeg -i bunny.mp4 -f hls -hls_segment_type fmp4 -c:v copy playlist.m3u8

  I send those chunks using Socketand try to play them through MSE.
  When i send them in this order :

  init.mp4 + playlist0.m4s + playlist1.m4s ...

  They play without any problem, But when i want to start from the chunk number 3 meaning init.mp4 + playlist3.m4s for example i get this error :

  video frame with PTS 0us has negative DTS -80000us after applying timestampOffset, handling any discontinuity, and filtering against append window.

  I want to be able to start from any chunk, currently the only way to play the video is to star by init.mp4 + playlist0.m4s meaning playlist0.m4s because init.mp4 contain just headers of the video, This is the client code i’m using :

  var socket = io();
  
  var video = document.querySelector('video');
  
  var mimeCodec = 'video/mp4; codecs="avc1.64000d,mp4a.40.2"'; // true
  
  if ('MediaSource' in window && MediaSource.isTypeSupported(mimeCodec)) 
  
  {
  
      var mediaSource = new MediaSource;
  
      video.src = URL.createObjectURL(mediaSource);
  
      mediaSource.addEventListener('sourceopen', function () {
  
              var mediaSource = this;
  
              var sourceBuffer = mediaSource.addSourceBuffer(mimeCodec);
  
             sourceBuffer.mode = 'segments'; 
  
              sourceBuffer.addEventListener('updateend', function (_) { video.play().then(function() { }).catch(function(error) { }); });
  
              socket.on('broadcast', function (chunk) {
  
                  downloadData(chunk.uri, function(arrayBuffer) {
  
                      sourceBuffer.appendBuffer(arrayBuffer);
  
                  });
  
              });
  
      });
  
  } else {
  
      console.error('Unsupported MIME type or codec: ', mimeCodec);
  
  }
  
  function downloadData(url, cb) {
  
      var xhr = new XMLHttpRequest;
  
      xhr.open('get', url);
  
      xhr.responseType = 'arraybuffer';
  
      xhr.onload = function () {
  
          cb(new Uint8Array(xhr.response));
  
      };
  
      xhr.send();
  
  }

• Deinterlace video frames

  23 janvier 2018, par Gui

  I am using a Accord.net’s VideoFileReader to open an ’.mts’ (AVCHD File) file from a Sony HandyCam.

  var file = new VideoFileReader();
  
  file.Open(@"pathToFile");
  
  
  
  var frame = file.ReadVideoFrame();

  This appears to open without issue. The resolution properties of the VideFileReader instance are correct, however it identifies the frame rate as being double that of the actual file. Upon inspection it appears to be interlaced via some method.

  

  I would like to traverse video files frame by frame for some simple analysis, where speed is relatively unimportant, however the time of each frame is critical. I am happy to deinterlace frames myself if I had a clear understanding of how I needed to manipulate the frame images.

  From my research on interlacing, I was expecting the resolution of each frame to be half that of the final output, where alternating frames need to be spliced -
  one containing even lines, and the other odd ; this appears not to be the case. All Frames are the full resolution of the video, even the very first frame seems to have two unique images present in it.

  How could I go about recovering a valid frame back from my video on a frame by frame basis (accessing adjacent frames if needed is fine) ? Is this something I have missed, not understood, or overlooked in the Documentation ?

  I’m not necessary looking for a code solution (although that would be great), an explanation of what is going on and what actions I could take will sufficiently address my needs.