Re: Misc Codec Questions....

Holger Waechtler wrote:
> 
> --- Sebastien St-Laurent <[email protected]> wrote:
> >
> > Out of curiosity (and laziness i guess since i could check the code).
> > What
> > approach does Tarkin take so far to code the wavelet bits?
> 
> Binary runlength encoding in bitplanes, the current code uses parallel
> runlength encoders. Runlengths are written out by a modified Golomb
> coder. Should be free of IP issues but not very efficient because we
> don't take any advantage of spatial correlation, there is no context
> sensitive entropy coding.

Check this out, if you haven't already:
http://www.cs.tut.fi/~loktem/publications.html

The author of this page has invented an alternative entropy-coder
which has better performance than arithmetic coding in his tests.
I can't find any info on legal issues on his pages, though.

By the way, I have tried to improve the prediction in the Golomb coder
by looking at the previous scanline of the bitplane.
I could only get about 3% better compression with my test images ...
The prediction error rate was still at an average around 50% in many
tests. Maybe I did something wrong... nah, well.

/ Johan

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to '[email protected]'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.