NV40 doesnt support gamma-correct FSAA,does it?

[engall]

Ditto,
2years ago R300 support gamma-correct and 6x RGMA FSAA
Does NV40 support these features?

 

[Xmas]

No, it does not.
What is RGMA? I think you mean 6x SGMS.

 

[engall]

No, it does not.
What is RGMA? I think you mean 6x SGMS.

Rotated Grid Multiple Sample=RGMS

 

[Xmas]

Where's the A?
6x isn't "rotated grid", it's "sparse grid".

 

[T2k]

Gamma corrected AA: AFAIK ATI patented it.

 

[engall]

Where's the A?
6x isn't "rotated grid", it's "sparse grid".

???

 

[DemoCoder]

Patented? For christ sakes. I hope that's a patent on the hardware implementation details and not the very idea itself.

GCAA has been in the literature since atleast 1989 (oldest PDF I have on it)

 

[Xmas]

Where's the A?
6x isn't "rotated grid", it's "sparse grid".

???

You wrote RGMA in your first post. And 6x isn't called "rotated grid" because unlike 2x and 4x, it doesn't look like a rotated ordered pattern.

 

[engall]

Where's the A?
6x isn't "rotated grid", it's "sparse grid".

???

You wrote RGMA in your first post. And 6x isn't called "rotated grid" because unlike 2x and 4x, it doesn't look like a rotated ordered pattern.

Sorry about “Aâ€￾ !
Thanks for your input.

 

[Remi]

Gamma corrected AA: AFAIK ATI patented it.

Remember: a patents is a claim for an original invention. The actual reality of the recognition of the validity of the claim is judged by a judge, should a conflict arise.

 

[Simon F]

Gamma corrected AA: AFAIK ATI patented it.

Remember: a patents is a claim for an original invention. The actual reality of the recognition of the validity of the claim is judged by a judge, should a conflict arise.

In theory, that is true, but my understanding is that it will be generally be assumed to be valid and will be very costly to challenge in such a manner.


However, on that note it could be argued it was obvious. I have Occam code that did gamma correct down scaling (for all intents and purposes, identical to the filter of AA) that I wrote in 1990/1.

 

[Tim]

Gamma corrected AA: AFAIK ATI patented it.

AFAIK, they patented their implementation not the concept of gamma corrected AA. You can do gamma corrected AA if you want to, you just can't do it the same way as Ati.

And just because something is patented, it doe not mean that the patent will hold up.

 

[K.I.L.E.R]

So if nVidia wants to have GC they will have to go into a lawsuit? :?

That's aweful, can't nV use another algorithm?

 

[DemoCoder]

Ah, the good ole days of the multitexturing patent.

 

[Tim]

I don't seem to able to find the patent, anyone got the patent-number?

 

[pcchen]

Why should FSAA be gamma corrected? Isn't it better for back buffers to be in linear space?

 

[Simon F]

Why should FSAA be gamma corrected? Isn't it better for back buffers to be in linear space?

Only if you can afford 12+ bits per colour channel.

 

[Ailuros]

I'm using a CRT which I believe I've callibrated close to the optimum and from all the gamma tests I used so far it seems that I'm on spot with the usual 2.2 gamma value too.

Gamma correction does make a difference, but I don't think it can be easily seen on screenshots. Is the difference huge? I wouldn't say so. Then again so far I was able to compare only NV's and ATI's 2x MSAA modes in real time; to compare ordered with sparse grid for 4xAA is senseless.

Does gamma correction make any sense on LCD monitors on the other?

 

[pcchen]

I think it's enough if the RAMDAC has a 12 bits look-up table? Of course, DVI-D is limited to 8 bits...

 

[GameCat]

If the bacbuffer is in linear space, 8 bits per channel gives too little precision in the low ranges/dark colours. If the backbuffer is linear like you proposed, I don't see what the DAC precision has to do with anything since the transform will be linear?