Faster MediaStream ImageCapture(在 nodejs 中读取媒体流作为缓冲区)

Faster MediaStream ImageCapture (read media stream as buffer in nodejs)

编辑: 我需要这样做 live。我等不及直播结束了。

我从 Electron 的 desktopCapturer:

得到了 MediaStream 对象
navigator.mediaDevices.getUserMedia({
  audio: false,
  video: {
    mandatory: {
      chromeMediaSource: 'desktop',
      chromeMediaSourceId: source.id,
      minWidth: 800,
      maxWidth: 800,
      minHeight: 800,
      maxHeight: 800,
    },
  },
})
.then((stream) => {

我正在尝试使用 ImageCapture:

获取静止帧的节点 Buffer
  const track = stream.getVideoTracks()[0];
  const capturedImage = new ImageCapture(track);

  capturedImage // This takes 200ms for 1000x1000
    .takePhoto()
    .then(blob => {
      toBuffer(blob, function (err, buffer) { // 1.5 ms
        if (err) throw err;
          // TODO: Do some opencv magic with node buffer
      });
    })
    .catch(error => console.error('takePhoto() error:', error));

但是 takePhoto 需要很长时间。有没有可能使这个过程更快?我能以某种方式直接在 nodejs 中访问 MediaStream 吗?

这就是我最终所做的。它实际上非常高效。

}).then((stream) => {
  const video = document.createElement('video');
  video.srcObject = stream;
  video.onloadedmetadata = () => {
  video.play();
  setInterval(() => {
    const canvas = document.createElement('canvas');
    canvas.getContext('2d').drawImage(video, 0, 0, 800, 800);
    canvas.toBlob(blob => {
      toBuffer(blob, function (err, buffer) {
        if (err) throw err;
        // do some magic with buffer
      });
    });
  }, 40);
};