PDA

View Full Version : Does software stencil force a software rasterization path?



Robbo
05-01-2002, 11:36 PM
I was thinking about using stencil buffer for some application but can't quite get my head around its use - particularly as most cards have limited hardware support.

Will the software path for stencil force software path for all rasterization too? Otherwise, the scene would surely have to be rendered once in hardware and then again in software with all the floating point miss-matches associated with these different paths.

zeckensack
05-02-2002, 04:09 AM
Stencil stuff is a per fragment operation. I don't see any way to do that in software while keeping hardware accelerated rendering.

So my - hopefully educated - guess is:
Yip.

davepermen
05-02-2002, 05:36 AM
most cards have limited support..

well..
geforces have it
radeons have it

if you want to use it on older hardware it will be too slow for any use anyways, just because of fillrate.. http://www.opengl.org/discussion_boards/ubb/smile.gif

Robbo
05-02-2002, 06:08 AM
When you say limited support, you are talking certain pixel formats\resolutions?

witcomb
05-02-2002, 06:32 AM
He means that only a few video cards support hardware accelerated stencil buffers.

Neil

Humus
05-02-2002, 10:59 AM
Which isn't quite true. Hardware stencil has been in since TNT,Rage128,Voodoo4,G400.

Robbo
05-02-2002, 11:21 AM
So - let me get things straight. If I'm using GeForce 2MX, GeForce 3 and GeForce 4MX, I will have hardware stencil.

Someone posted about that stencil test is a `per fragment' operation. I expected it to be a `per pixel' operation. Which of these is correct? (I was under the impression that fragments can contain one or more pixels - at least my alpha test code shows this to be the case).

LordKronos
05-02-2002, 11:45 AM
Fragment and pixel are pretty much the same thing. They can be different in some circumstances. For instance, when using antialiasing you get multiple fragments per pixel.