[XviD-devel] Status of GMC

[Christoph Lampert]

On Sun, 3 Nov 2002, peter ross wrote:

> hi,
> 
> gruel,syskin: excellent work on gmc.
> 
> cleaning up the borders (e.g. forcing constant luma/chrom) imho
> isn't a video-codec feature. it can be done at the pre-filter stage.
> avisynth provides AddBorders. transcode/mencoder should have something
> simular.
> 
> >From: Christoph Lampert <chl@math.uni-bonn.de>
> >Anyway, the problem is simply how to keep GMC efficient (once it is at
> >all), if parts of the picture are static (e.g. pure black) and one
> >previously knows about this.
> 
> to improve motion estimation for letterboxed material, why not support
> a range of interest. the user specifies the area where compression is
> most important. for the remainding area a quick inter/inter4v decision
> is performed, local gmc, and maybe even higher quants.


Sorry, I just noticed that my problem had already been solved (I just
didn't implement this, yet): Only the motion vectors of 
"structured" blocks should be used to determine global motion. Others
don't have reliable motion vectors. Pure black can surely not be called
"structure", so black bars will be ignore automatically. 


gruel