oesTextureId = TextureHelper.createOESTextureObject()
surfaceTexture = SurfaceTexture(oesTextureId)
val surface = Surface(surfaceTexture)
mediaPlayer.setSurface(surface)

更新播放进度：

override fun onFrameAvailable(surfaceTexture: SurfaceTexture?) {
    glSurfaceView.requestRender()
    val position = mediaPlayer.currentPosition
    videoSeekBar.progress = position
    playTimeView.text = calculateTime(position)
默认情况下，绘制的视频会填满整个GLSurfaceView的大小，通常都会有视频播放拉伸的问题，因为视频的宽高比和容器的宽高比不一致，我们可以通过设置模型矩阵来调整GL顶点：
override fun onVideoSizeChanged(mp: MediaPlayer?, width: Int, height: Int) {
   videoSeekBar.max = mediaPlayer.duration
   sx = 1f * width / renderWidth
   sy = 1f * height / renderHeight
override fun onDrawFrame(gl: GL10?) {
    surfaceTexture?.updateTexImage()
    surfaceTexture?.getTransformMatrix(transform)
    Matrix.setIdentityM(mvpMatrix, 0)
    Matrix.scaleM(mvpMatrix, 0, sx, sy, 1f)
    cameraRender?.drawTexture(mvpMatrix, transform, oesTextureId)
MediaPlayer解出的视频帧转为纹理后，我们就可以通过GL操作纹理做很多有趣的东西了，比如添加水印，添加滤镜效果等，这里简单的将rgb转为gray
Gray = 0.2989 * R + 0.5870 * G + 0.1140 * B
完整的顶点着色器：
void main() {
    vTextureCoord = (u_TextureMatrix * a_TextureCoordinate).xy;
    gl_Position = a_Position * u_MvpMatrix;
片元着色器：
void main() {
    vec4 color = texture2D(u_TextureSampler, vTextureCoord);
    float gray = 0.299 * color.r + 0.587 * color.g + 0.114 * color.b;
    vec4 newColor = vec4(gray, gray, gray, color.a);
    gl_FragColor = mix(color, newColor, identity);
Demo中，在点击播放按钮的时候才开始更新GLSurfaceView，为了避免启动后界面一片黑，我们在GLSurfaceView上盖一层ImageView来展示一帧图像，开始播放后就隐藏这个ImageView，获取视频缩略图：
val mediaMetadataRetriever = MediaMetadataRetriever()
mediaMetadataRetriever.setDataSource(
       assetFileDescriptor.fileDescriptor,
       assetFileDescriptor.startOffset,
       assetFileDescriptor.length
val bitmap = mediaMetadataRetriever.getFrameAtTime(0)
if (bitmap != null) {
     videoThumbnailView.visibility = View.VISIBLE
     videoThumbnailView.setImageBitmap(bitmap)
mediaMetadataRetriever.release()

https://github.com/sifutang/video.git

作者：雪月青