[XviD-devel] GMC/Segmentation/AdaptiveQuant

[Marco Al]

daniel smith wrote:

> MfA pointed this out on doom9's forum:
>
> "It's a paper called "A perceptual model for JPEG applications based on
> classification, texture masking, and luminance masking" BTW. It is totally ad-
> hoc and empirical, or in other words actually useful (Unlike say Watson's
> stuff, which I am ashamed to admit I was totally unfoundedly fond of in the
> past.) It will require DCT coefficients from the original frames, but coded
> right that should only take about 1-2 millisecond extra per frame on a fast
> machine (DCT takes about 250 cycles from cache, memory bandwith wont be the
> limitation ... so with prefetch on a 1+ GHz that should be possible)."
>
> sounds like a useful paper.

It is a more sophisticated form of luminance masking, but it doesnt remove the
need to more accurately guess the rate savings. So as I said there, it is pretty
much tangential to Gruel's intentions. Once the code is there to optimize
quantizers&quantization based on an accurate estimate of rate (or even measured
rate) and a distortion measure you can plug in any distortion measure you want.

Marco