Recherche avancée

Sur d’autres sites (5361)

• HTML Video Exporting Using MediaRecorder vs ffmpeg.js

  3 octobre 2020, par Owen Ovadoz

  TLDR

  


  Imagine I have one video and one image. I want to create another video that overlays the image (e.g. watermark) at the center for 2 seconds in the beginning of the video and export it as the final video. I need to do this on the client-side only. Is it possible to use MediaRecorder + Canvas or should I resort to using ffmpeg.js ?

  


  Context

  


  I am making a browser-based video editor where the user can upload videos and images and combine them. So far, I implemented this by embedding the video and images inside a canvas element appropriately. The data representation looks somewhat like this :

  


  video: {
  url: 'https://archive.com/video.mp4',
  duration: 34,
},
images: [{
  url: 'https://archive.com/img1.jpg',
  start_time: 0,
  end_time: 2,
  top: 30,
  left: 20,
  width: 50,
  height: 50,
}]


  


  Attempts

  


  

  1. I play the video and show/hide images in the canvas. Then, I can use the MediaRecorder to capture the canvas' stream and export it as a data blob at the end. The final output is as expected, but the problem with this approach is I need to play the video from the beginning to the end for me to capture the stream from the canvas. If the video is 60 seconds, exporting it also takes 60 seconds.

  


  


  


  


  function record(canvas) {
  return new Promise(function (res, rej) {
    const stream = canvas.captureStream();
    const mediaRecorder = new MediaRecorder(stream);
    const recordedData = [];

    // Register recorder events
    mediaRecorder.ondataavailable = function (event) {
      recordedData.push(event.data);
    };
    mediaRecorder.onstop = function (event) {
      var blob = new Blob(recordedData, {
        type: "video/webm",
      });
      var url = URL.createObjectURL(blob);
      res(url);
    };

    // Start the video and start recording
    videoRef.current.currentTime = 0;
    videoRef.current.addEventListener(
      "play",
      (e) => {
        mediaRecorder.start();
      },
      {
        once: true,
      }
    );
    videoRef.current.addEventListener(
      "ended",
      (e) => {
        mediaRecorder.stop();
      },
      {
        once: true,
      }
    );
    videoRef.current.play();
  });
}

  


  


  



  

  2. I can use ffmpeg.js to encode the video. I haven't tried this method yet as I will have to convert my image representation into ffmpeg args (I wonder how much work that is).

  


  


• When I use Fluent-Ffmpeg to access Ffmpeg, there are two different threads but I dont want it

  25 mars 2019, par Ahmet Hakan Billur

  I try to broadcast with rtsp live stream from IP camera on web app that is improved with node.js-jsmpeg([a link]https://www.npmjs.com/package/fluent-ffmpeg !), web socket, html5(canvas).Everything ok that live streaming works but missing frame and high CPU usaged by streaming on web app and I try to reduce so I can intervene ffmpeg with fluent-ffmpeg but when I monitor CPU usaged I can see there 2 different threads following as and look at screenshot of CPU ;
  

  ffmpeg -rtsp_trasport tcp -i rtsp ://10.6.0.225 -f mpeg1video - is worked by jsmpeg and canvas/html5
  index.html

      
  
  
  
      
  
      
  
      
  
      
  
  
  
      <div><canvas width="640" height="360"></canvas></div>
  
      div><canvas width="640" height="360"></canvas>  
  
      <code class="echappe-js">&lt;script type=&quot;text/javascript&quot; src='http://stackoverflow.com/feeds/tag/jsLib/jsmpeg.js'&gt;&lt;/script&gt;
      
  &lt;script type=&quot;text/javascript&quot; src='http://stackoverflow.com/feeds/tag/jsLib/ffmpegUtil.js'&gt;&lt;/script&gt;
  &lt;script type=&quot;text/javascript&quot;&gt;<br />
  
        var canvas = document.getElementById('videoCanvas');<br />
  
        var ws = new WebSocket(&quot;ws://10.6.0.206:9999&quot;)<br />
  
        var player = new jsmpeg(ws, {canvas:canvas, autoplay:true,audio:false,loop: true});<br />
  
      &lt;/script&gt;
  

  other one /usr/bin/ffmpeg -i rtsp ://10.6.0.225 -y out.ts is work by following piece of code in app.js

  Stream = require('node-rtsp-stream');
  
  stream = new Stream({
  
      name: 'name',
  
      streamUrl: 'rtsp://10.6.0.225',
  
      wsPort: 9999
  
  });
  
  
  
  var ffmpeg = require('fluent-ffmpeg');
  
  var proc = new ffmpeg();
  
  
  
  proc
  
  .addInput('rtsp://10.6.0.225')
  
  .on('start', function(ffmpegCommand) {
  
      /// log something maybe
  
      console.log('start-->'+ffmpegCommand)
  
  })
  
  .on('progress', function(data) {
  
      /// do stuff with progress data if you want
  
      console.log('progress-->'+data)
  
  })
  
  .on('end', function() {
  
      /// encoding is complete, so callback or move on at this point
  
      console.log('end-->')
  
  })
  
  .on('error', function(error) {
  
      /// error handling
  
      console.log('error-->'+error)
  
  
  
  })
  
  .output('out.ts')
  
  .run();

  and then I don’t want to get two different ffmpeg command threads in there.
  Does anyone have an idea ?
  Thanks in advice.

• Explicitly set soundManager.preferFlash = true for some demo/API/docs cases

  31 juillet 2011

  m demo/360-player/canvas-visualization-basic.html m demo/360-player/canvas-visualization.html m demo/animation-1/script/animation.js m demo/animation-2a/index.html m demo/animation-2b/script/animation.js m demo/api/index.html m demo/christmas-lights/christmaslights.js m (...)