[Ffmpeg-devel] Optimizing encoding when knowing changed area

[Rich Felker]

On Mon, Oct 10, 2005 at 12:33:33AM -0700, Loren Merritt wrote:
> On Mon, 10 Oct 2005, Oleg Oksyuk wrote:
> 
> >Hello everyone,
> >
> >I use ffmpeg to encode graphics animation produced by another program. The
> >contains API interface that allows to get output pixmap in RGB format and
> >changed area, for example (100, 100) - (200, 300).
> >When encoding MPEG using FFMPEG I simply convert changed area from RGB to
> >YUV each 1/30 of second. But often nothing was changed or only some small
> >rectangle was changed. Are there any way to use knowledge about changed 
> >area
> >to optimize encoding?
> 
> While I haven't used them myself, AVCodecContext.me_threshold and 
> mb_threshold look like what you're asking for.

I think he wants to avoid the colorspace conversion for the unchanged
regions.

Rich