Hi, Actually there are two question, but first some explenation:
I have to use AVIOContext to read data from stream, and I have to show decoded video from stream as fast it is possible. I'm using h264_cuvid to get hardware acceleration and video frames are rendered in OpenGL. I have two problems: 1. I have to use function set in AVIOContext: int (*read_packet)(void *opaque, uint8_t *buf, int buf_size) But in that case I have to copy data. Is there a way to just set pointer with filled data? Or some other way how to speedup that part of the code? I'm storing data in queue. 2. I have to copy data from AVFrame into GPU, for example: glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, GL_RED, GL_UNSIGNED_BYTE, m_frame->data[0]); Is there a way to ommit that part, cuvid should have already that data in GPU, would be nice to somehow get texture id directly and show it. Regards Zygmunt
_______________________________________________ Libav-user mailing list Libav-user@ffmpeg.org http://ffmpeg.org/mailman/listinfo/libav-user