[FFmpeg-soc] [soc]: r461 - rv40/rv40.c

[Kostya]

On Tue, Jul 17, 2007 at 10:22:14AM -0700, Mike Melanson wrote:
> Kostya wrote:
> > On Tue, Jul 17, 2007 at 07:14:59PM +0200, kostya wrote:
> >> Author: kostya
> >> Date: Tue Jul 17 19:14:59 2007
> >> New Revision: 461
> >>
> >> Log:
> >> Use luma and chroma quantizer tables in dequantizing
> > 
> > Now many intraframes in many files will be decoded. They will be
> > decoded with artifacts but at least close to original.
> 
> Are the artifacts the result of no post processing, or due to parts of
> the decoding algorithm yet to be determined?

The second one, mostly because of some flaw in 16x16 prediction and
quantizing is not done perfectly too.
The worst part is that some files (like test.rmvb or spygames-2MB.rm)
use undetermined (for now) VLC set for many intra frames.
 
> 	Thanks...
> -- 
> 	-Mike Melanson
> http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-soc