Android-MediaCodec的视频编码质量

Question

I use a Surface where I draw openGL images on it (and rotate them). For some images, when I encode the video using MediaCodec, the result looks pixelated, or compressed..

Original background picture :

Result encoded :

You can see that on the bottom-left corner, it is pixelated.

Here is the values I use to create the MediaCodec encoder :

    int width = 800;
    int height = 800;
    int bitRate = 10485760; // bps, = 10Mbps (1080p)
    bufferInfo = new MediaCodec.BufferInfo();
    MediaFormat format = MediaFormat.createVideoFormat(GLVideoEncoderImpl.MIME_TYPE, width, height);

    // Set some properties.  Failing to specify some of these can cause the MediaCodec
    // configure() call to throw an unhelpful exception.
    format.setInteger(MediaFormat.KEY_COLOR_FORMAT,
            MediaCodecInfo.CodecCapabilities.COLOR_FormatSurface);
    format.setInteger(MediaFormat.KEY_BIT_RATE, bitRate);
    format.setInteger(MediaFormat.KEY_FRAME_RATE, 30);
    format.setInteger(MediaFormat.KEY_I_FRAME_INTERVAL, 1);
    try {
        mediaCodec = MediaCodec.createEncoderByType(GLVideoEncoderImpl.MIME_TYPE);
    } catch (IOException e) {
        e.printStackTrace();
        UtilsAndroid.throwException("Impossible to create media codec");
    }

    mediaCodec.configure(format, null, null, MediaCodec.CONFIGURE_FLAG_ENCODE);

Is it possible to improve the output quality ? Thanks in advance

Answer 1

For some images, the result looks pixelated, or compressed.

Well, that's what lossy video compression does, especially when you look at stills of the video frames (video compression is nasty, and the perceptual error relies on the fact it's animating so you don't see it).

Things to try:

• increase bitrate at a fixed FPS
• drop the framerate at a fixed bitrate,
• increase the number of I frames (which will reduce error propagation artefacts)
• decrease the number of I frames (I frame take a lot more bits to encode than P frames and B frames, so lots of I frames means worse compression, which means worse quality in a fixed bitrate).

However, note that most video encoders are designed to compress videos of the real world, and the perceptual errors are designed with that in mind. Most tend to be rubbish at compressing simple gradient fills, line art, etc, unless you run them at very high bitrates.