[Ffmpeg-devel] ScreenKast exporting

Guillaume POIRIER gpoirier
Mon Aug 28 22:08:36 CEST 2006


Bram Biesbrouck a ?crit :

> Here's the concept: the recording is stored as zlib-compressed rect-updates, 
> that get decompressed on the fly and drawn to an OpenGL canvas, together with 
> text-balloons, commentboxes, etc. Currently, the ffmpeg export-code grabs 
> that (entire) OpenGL canvas every time something changes, but this is too 
> costly, because often, only small areas get updated.

Well, I suppose you could reuse the motion estimation routines from 
FFmpeg... but I imagine you'd need to have some video-compression 
background to make some sense out of them.

If you grep through the source tree, you should get some of these 
routines. I imagine that you'd just have to look at which functions 
calls them to piece by piece get a grasp of what has to be done to run a 
simple motion compensation algorithm.

You may also wanna dig the ml for a patch that added H.264 encoding support.
I'd give you an idea of what's needed to build an encoder, though H.264 
isn't the easiest codec when you're a beginner.


More information about the ffmpeg-devel mailing list