02-14-2013 06:59 PM
I'm looking at performing post-processing on playing videos - playing a video in my app and adding a specific effect to the playing media file before it gets displayed over the HDMI interface. Whilst I could use ffmpeg for video input etc, filter/modify the buffers, and then write these to the screen render buffer, I would prefer to use the built-in APIs for opening and playing files.
Is there any way to either a) inject my own filter into the processing graph for the video, or b) access the screen/buffer before between it being written by the mm-renderer or similar and when that screen/buffer gets posted to the screen, to allow me to modify the buffer?
For example, can I set the output device for the mm-renderer as a buffer, and then rather than posting that buffer to the screen, my code instead can apply the effect to the buffer (or an intermediate buffer) and then blit that onto the visible screen and post it myself.