RE: Misc Codec Questions....

"Sebastien St-Laurent" <[email protected]> Tue, 21 May 2002 09:41:28 -0700
Newsgroups gmane.comp.multimedia.ogg.tarkin.devel
Message-ID <002b01c200e6$5b5991f0$0100000a@olivert>
Hi again...

>> In essence it's for a XBox/GameCube game. The free solutions on that
>> platform aren't so good at the moment and the only viable 3rd party
solution
>> is Bink but the codec costs 6500$ per-title/per-platform. I'v always had
>> some interests in Audio/Video compression so i'v decided to try and write
my
>> own on my spare time.
><snip>
>> As for royalty issues... I'm not exactly sure what you're talking
about...
>
>Well reference source code for MPEG-4 is available, and shouldnt be a huge
task >to port ... of course the licensing fees are once
>again up in the air (sigh ... patents promote innovation my ass) but the
>initial now retracted terms of $.25 per unit and $.00033
>per minute of video seem reasonable for your applications.

Eeeekk... This is worse than a fixed price liscense like Bink offers.
Imagine the logistic nightmare of dealing with such a scheme (assuming they
don't change the rates on us in the middle). Plus, i personally like the
experimenting involved in developping my own codec ;)

>> Not that i necessarily want to use mo-comp, however anything along the
lines
>> of 3D Wavelets is generally too expensive memory wise because you need a
GOP
>> big enough to take advanage of it.
>
>Im not a fan of 3D wavelets at all ... just wanted to indicate that
multiscale >motion compensation ignores the fact that motion
>boundaries are only really present at the finest scale, compensating a
wavelet >who's footprint crosses the motion boundary seems
>wrong. Even ignoring that wavelets are not translation invariant.

I understand the issue inherent with wavelets and mocomp. In a perfect
world, maybe some kind of wavelet space image wraping would work good but
this is fairly expensive and probably not applicable to a real-time
algorithm.

Sebastien St-Laurent


--- >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.