MediaCodecInfo.CodecCapabilities.COLOR_FormatSurface about yasea HOT 4 CLOSED

There seems no difference between MediaCodec and MediaRecorder that way. Actually yasea needs some pre-processing on getting the YUV frames from the camera before sending them into the encoder like cropping, flipping and rotation. For instance, the original video data source in android is a landscape picture and what I want is a portrait one, then I need to rotate and may also flip the whole YUV frame.

COLOR_FormatSurface can really work indeed. But if I do not know the specific color format, how do I make transformation of the YUV frames on pre-processing?

from yasea.

We don't need pre-process frames, at the output we get raw h264 (with COLOR_FormatSurface or another, it does not matter) from mediacodec. I noticed that when use COLOR_FormatSurface - SPS and PPS are not sent. I'm trying to add them (SrsRtmpFlv.writeSampleData()), but it still does not work
https://gist.github.com/Nook2007/92f460c87d8556d35ab5676f40a5e840

from yasea.

Where is your raw h.264 data from? Does it wrapped in the byte buffer returned by getOutputBuffers() of MediaCodec? Typically the SPS and PPS are wrapped in the IDR frame generated by the encoder if you do not miss them.

The gist code seems unreliable. The IDR frame is NOT always the first one from the encoder. See SrsFlvMuxer.java to find out the right way.

After all I am afraid I will not try your case on yasea for I am busy with other issues.

from yasea.

@Nook2007 我的也是H264的祼数据, 希望通过writeSampleData来推流, 没有成功, 请问你怎么解决的呢

from yasea.