64bit color in the near future?

[HedsSpaz]

Check out John Carmack's latest .plan http://finger.planetquake.com/qfplan...johnc&id=14274

What do you guys think? Theres some interesting stuff in there. More to the point, what do you guys think the chances are that Matrox will quick rework the G800 a bit to support 64bit color? I mean heck, if they are the first to support it, then they sort of set the defacto standard for it. That'd be pretty damn cool if you ask me.

Ian

[smokva]

You should be a f***** owl to see the difference between 32bit & 64bit color depth

[waxling]

WTF?

Most games dont take advatage of the colour range available now. And like smokva said, who would be able to tell?

[Stringy]

Jez you 2, smokva and waxling should READ what was said rather than just spouting OFF!!

If you read the peice you would know that the extra bits aren't for actual COLOR but are to keep the colors from degrading when Multiple filters or layers are used on one Texture....

and I'll quote his example as to why 64Bit is So valuable to a developer NOW..

The situation becomes much worse when you consider the losses after multiple
operations. As a trivial case, consider having multiple lights on a wall,
with their contribution to a pixel determined by a texture lookup. A single
light will fall off towards 0 some distance away, and if it covers a large
area, it will have visible bands as the light adds one unit, two units, etc.
Each additional light from the same relative distance stacks its contribution
on top of the earlier ones, which magnifies the amount of the step between
bands: instead of going 0,1,2, it goes 0,2,4, etc. Pile a few lights up like
this and look towards the dimmer area of the falloff, and you can believe you
are back in 256-color land.

So 64bit color will make the use of multiple operations much easier and allow much more realism...

it's interesting to say the least...
I'm not sure we are ready for it now, but would be nice if the Vid. Co's would work on this now so when we are ready, the HW will be in place...

Craig

[This message has been edited by Stringy (edited 30 April 2000).]

[Brian R.]

That explains why I see color banding when I use 32-bit color displays in games with 32-bit textures.

[HedsSpaz]

it's interesting to say the least...
I'm not sure we are ready for it now, but would be nice if the Vid. Co's would work on this now so when we are ready, the HW will be in place...

I couldn't agree more. I had pretty much the same thoughts when I read it. Yeah we may not have too much of a use for it right NOW. But you can definately see the benefits of 64bit. And indeed, if graphics companies gett their arses in gear right now then we'll have it available when the time comes that we do need it.

Ian

[Damien Green]

You can indeed see a certain amount of banding in games set at 32bit with 32bit textures. It makes sense for Matrox to be an early adopter of 64bit colour as such a feature would tie in nicely with theie reputation for proving superb image quality.

[isochar]

Would the next-gen cards (g800, nv20, rampage)be able to handle 1024x768x64bit with 60fps+? I'm guessing that you'd need a lot of bandwith for that...

[rubank]

The kind of banding you´re talking about has little to do with colour depth, it´s in the implementation of lighting. You can have 128bit colour, or whatever, it won´t help!

Of course, in theory you´ll always have "banding", regardless of colour depth, that´s the nature of digital sampling.

[Phips]

In a way it does make sense.
If you look at DSP architectures, the accumulator register generaly has more precision than the final result the DSP spits out. DSPs do enormous amount of operations on a data stream: possibly hundreds of multiply-accumulate ''series'' operations per sample.
It's not like having a 64 bits DAC, it's using 64 bits precision internaly.
If you are to do multiple passes per pixel, roundoff errors are going to kill you. The operations Carmack is refering to would take hundreds of individual steps per pixel.
Unfortunatly 64bits floating/integer fast hardware multipliers are costly things to implement in silicon.

[Jutai]

I think one way to get over this and maybe pave the way for 64 bit technology while increasing visual quality and having minimal taxing on the hardware would be a solution like VooDoo and Matrox both have used in their dithered down rendering perhaps if the scene was rendered internally at 64bit it would increase the visual quality of the 32 bit result maybe helping with the above mentioned banding I think this would be a compelling feature for Matrox to look into it would really up the anti to the competition and would increase (if that's possible) the amazing visual quality Matrox is known for it would be nice to see this implimented in the G800 chip i guess people would call it like close to 42 bit rendering or something like that well it sure would look nice
-chris k.