[XviD-devel] GME and Block ME

[Christoph Lampert]

On Tue, 22 Apr 2003, Michael Militzer wrote:
> I wonder why do we perform this coarse MEAnalysis at all? Couldn't we simply do 
> a full and precise ME step including halfpel instead? So in case we'd decide 
> the current frame should become a P-frame, we would be finished already without 
> any additional work needed. In case we code the frame as b-frame, we can store 
> the results and then don't need to perform forward ME anymore when coding the b-
> frame.

Possible and perhaps the best choice. But we still would have to do parts
of forward ME, because IIRC median predictors are different in B-frames
(reset at left boundary).

gruel