64bit depth buffer depth ok?

D3D9 doesn't have a 64bit depth format defined.

Why do you need a 64bit depth buffer? Are you using it for actual Z, or for storing other information? MRTs can help you in some things, depending on exactly what it is you want to do.

If you're getting Z-fighting artifacts, you're likely doing something wrong. A 32bit Depth-Buffer should provide enough accuracy for any use, when used correctly.

Hope this helps.

yes for actual depth calcs. Is 32 bit covered by cards? I can use that.

Many cards still in use don't support D32. Your best bet is D24X8 or D24S8, which are only 24 bits for depth. It's pretty much your only option, and has been supported for a long, long time.

what bit format is d24s8? is that a 24bit floating point number?

Where do you see a 64bit depth buffer? 32bit at max is all I see.

sorry I didn't see 64bit, (I wasn't at my comp to check). I just assumed. But d24 is 24bit I assume. What cards actually don't support 32bit does anybody know. Perhaps just a couple models so I can get an idea of what hardware would not support 32bit?

Thanks,

Devin

My Radeon X800 doesn't support D32.

My GeForce 6800 appears to support D32 according to DirectX Caps Viewer. An X800 is a reasonably recent ATI card. It's probably safe to assume that all ATI cards or that time, and before do not support D32, if Promit's sample is anything to go by. Given that, I'd be drawing up contingency plans for supporting D24.

Edit: Oops - I was looking at the Ref device. Clearly I haven't touched DirectX in a while... cheers Mastaba.
Just to reiterate - D16 and the D24 variants are all that are supported on my 6800.

[Edited by - Ro_Akira on January 4, 2007 4:34:29 PM]

Quote:

Original post by Ro_Akira My GeForce 6800 appears to support D32 according to DirectX Caps Viewer.

Are you sure you didn't look at the REF device? Mine does not support D32.

my 7600GT doesn't do D32 either.

24 bit are enough for almost any use. There's no reason to look for solutions to things that are not a problem. If you're getting Z-related artifacts, that's when you start looking for solutions.

Hope this helps.

I just checked the caps spreadsheet. Nobody except random SiS, XGI, and 3dlabs cards supports D32. None of the Radeon or GeForce series can do it.

Live search doesn't even register any hits for "64 bit depth buffer" [lol]

Maybe some obscure high-end CAD hardware supports them, but I've never even heard them mentioned in the consumer/gaming space - and I highly doubt they ever will.

32bit depth buffers are a requirement with Direct3D 10 if that's of any use.

hth
Jack

Quote:

Original post by Promit I just checked the caps spreadsheet. Nobody except random SiS, XGI, and 3dlabs cards supports D32. None of the Radeon or GeForce series can do it.

Interestingly, in terms of hardware at least some of the Radeon cards should be able to support D32. IIRC in the Radeon 7x00 era the drivers had an option to enable it.

Thanks for your guys answer. I guess I'll go 32bit for the d3d10 version. But 24bit for now. This is a exotic trick where I will need as much numeric distance between the depth buffer values max range and it's min value range. I'm allowing for 10,000 of 10,000 of 10,000 of 10,000... and so on, different depth values from min to max.

-Devin

perhaps it just needs to be enabled on the cards then?

I don't really know what you're doing, but linearized depth might be helpful if you need to alter how the depth precision behaves.