Will Microsoft ever adopt a new compression method?

Humus said:
Ahh cheers. This is one of the reasons MBX has its own texture compression scheme ;-)

I also noted teh following from that spec:
(5) Is the encoding of the RGB components for DXT1 formats correct in this spec? MSDN documentation does not specify an RGB color for the "transparent" encoding. Is it really black?

RESOLVED: Yes. The specification for the DXT1 format initially required black, but later changed that requirement to a recommendation. All vendors involved in the definition of this specification support black. In addition, specifying black has a useful behavior.

Useful? Surely not!!.

IM(NS)HO, premultiplied alpha (which is effectively what we have here) is a mistake once you have any form of texture filtering! When you get a fully transparent texel, the information content drops from 4 dimensions (ARGB) down to 1 (just alpha) so this means you can't filter the colour channels sensibly around the fully transparent pixels. The result: ugly grey/black halos.
 
Probably useful where you're only going to be applying additive blending to the texture, questionable for anything else...

John.
 
JohnH said:
Probably useful where you're only going to be applying additive blending to the texture, questionable for anything else...

John.
Even then I can't see much point. You can just as easily set the dest blend to 1 and src to alpha and, hey presto, "pre-multiplied alpha". <shrug>

.... unless, of course, the developer wanted to square the alpha I suppose.... :?
 
Anyone knows why S3 is no more part of the Futuremark BETA program?

Maybe it's a repeat of the NV30 scenario : they noticed their Deltachrome performed very badly in 3D Mark and therefore decided to boycott it... :rolleyes:
 
xGL said:
Anyone knows why S3 is no more part of the Futuremark BETA program?

Maybe it's a repeat of the NV30 scenario : they noticed their Deltachrome performed very badly in 3D Mark and therefore decided to boycott it... :rolleyes:

too expensive ;)
 
Simon F said:
Humus said:
Ahh cheers. This is one of the reasons MBX has its own texture compression scheme ;-)

I also noted teh following from that spec:
(5) Is the encoding of the RGB components for DXT1 formats correct in this spec? MSDN documentation does not specify an RGB color for the "transparent" encoding. Is it really black?

RESOLVED: Yes. The specification for the DXT1 format initially required black, but later changed that requirement to a recommendation. All vendors involved in the definition of this specification support black. In addition, specifying black has a useful behavior.

Useful? Surely not!!.
Sure it is. If you don't care about transparency, you can use this encoding to get black without sacrificing one of your two palette colors for that block.
 
OpenGL guy said:
Useful? Surely not!!.
Sure it is. If you don't care about transparency, you can use this encoding to get black without sacrificing one of your two palette colors for that block.
Well I was meaning WRT to translucency but, yes, I can see that it does give you black for free. Now where's that image of the model T Ford I need to compress....? :)
 
Simon F said:
OpenGL guy said:
Useful? Surely not!!.
Sure it is. If you don't care about transparency, you can use this encoding to get black without sacrificing one of your two palette colors for that block.
Well I was meaning WRT to translucency but, yes, I can see that it does give you black for free. Now where's that image of the model T Ford I need to compress....? :)
Just think how great a black hole would look! :D
 
Back
Top