[英]TextureView with MediaCodec decoder for H264 streams
This is a follow-up question of this question . 这是该问题的后续问题 。
This is my TextureView
code: 这是我的
TextureView
代码:
public class VideoTextureView extends TextureView implements SurfaceTextureListener{
private static final String LOG_TAG = VideoTextureView.class.getSimpleName();
private MediaCodecDecoder mMediaDecoder;
private MediaCodecAsyncDecoder mMediaAsyncDecoder;
public VideoTextureView(Context context, AttributeSet attrs) {
super(context, attrs);
setSurfaceTextureListener(this);
Log.d(LOG_TAG, "Video texture created.");
}
@Override
public void onSurfaceTextureAvailable(SurfaceTexture surface, int width, int height) {
Log.d(LOG_TAG, "Surface Available: " + width + " x " + height);
mMediaDecoder = new MediaCodecDecoder();
mMediaDecoder.Start();
mMediaDecoder.SetSurface(new Surface(getSurfaceTexture()));
}
@Override
public void onSurfaceTextureSizeChanged(SurfaceTexture surface, int width, int height) {
// TODO Auto-generated method stub
}
@Override
public boolean onSurfaceTextureDestroyed(SurfaceTexture surface) {
mMediaDecoder.Stop();
return false;
}
@Override
public void onSurfaceTextureUpdated(SurfaceTexture surface) {
// TODO Auto-generated method stub
}
}
My question - Is my TextureView
implementation okay for rendering H264 streams decoded by MediaCodec
? 我的问题-我的
TextureView
实现可以通过MediaCodec
解码H264流吗? Or do I need to do EGL setup or anything else? 还是我需要进行EGL设置或其他?
Thanks in advance! 提前致谢!
I am currently using a TextureView for rendering multiple streams in one activity using collection view cells on android (Sorry for ios terminology there). 我目前正在使用TextureView在android上使用集合视图单元在一个活动中呈现多个流(抱歉,那里的ios术语)。
It works fine but the issue is when you for example rotate the device there will be surface_destroyed followed by surface_available. 它工作正常,但问题是,例如当您旋转设备时,将出现surface_destroyed,然后是surface_available。 As i see you are correctly stopping and starting your decoder.
如我所见,您正在正确地停止和启动解码器。
One thing i do in my Decoder is: 我在解码器中所做的一件事是:
List<NaluSegment> segments = NaluParser.parseNaluSegments(buffer);
for (NaluSegment segment : segments) {
// ignore unspecified NAL units.
if (segment.getType() != NaluType.UNSPECIFIED) {
// Hold the parameter set for stop/start initialization speed
if (segment.getType() == NaluType.PPS) {
lastParameterSet[0] = segment;
} else if (segment.getType() == NaluType.SPS) {
lastParameterSet[1] = segment;
} else if (segment.getType() == NaluType.CODED_SLICE_IDR) {
lastParameterSet[2] = segment;
}
// add to input queue
naluSegmentQueue.add(segment);
}
}
I hold onto the last parameter sets and last keyframe and on start i fill the naluSegmentQueue with these first off to reduce the delay in video rendering. 我保留了最后一个参数集和最后一个关键帧,并从一开始就填充了naluSegmentQueue,以减少视频渲染的延迟。
My TextureView
implementation is okay as I tried with SurfaceView
too and found same result. 我的
TextureView
实现也可以,因为我也尝试过SurfaceView
并找到了相同的结果。 And as @fadden said - 正如@fadden所说-
EGL setup is only required if you're rendering with GLES.
仅当使用GLES进行渲染时才需要EGL设置。 TextureView combines a SurfaceTexture with a custom View, and does the GLES rendering for you.
TextureView将SurfaceTexture与自定义View结合在一起,并为您执行GLES渲染。 Which is why the View must be hardware-accelerated for TextureView to work.
这就是为什么必须对硬件进行硬件加速以使TextureView起作用的原因。
Thanks to @fadden. 感谢@fadden。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.