i need some debunking

[fisix]

in this thread

http://forum.doom9.org/showthread.php?s=&threadid=36500

somebody offers a 'hacked' version of huffyuv that supposedly is much faster than the original.

i can't check this right now, could someone take a look at it and see if it really is faster, and still is a real implementation of huff?

thanks,

-fisix

[Zarxrax]

I seems that it only works faster in CCE.

I think it would be worthwhile however for someone to try speeding it up a little, maybe add some MMX or SSE or something...

[-h]

I started work on a lossless codec a while back now, based loosely on huffyuv's algorithm but achieving better compression. This led me to read over huffyuv's source code, and i can vouch for it being *very* fast code (I spent an hour optimising an mmx median calculation only to arrive at the same code as huffyuv's).

Of course it doesn't natively support the YV12 colourspace, which would speed things up in terms of bytes processed (25% fewer bytes in YV12 versus YUY2) and algorithm complexity (perhaps 20% faster thanks to planar access).

-h

[Zarxrax]

Ah, cool. I cant wait on your codec :D