Quote:Original post by Yann L
As AndyTX mentioned, current MSAA is just not usable with any form of deferred shading, unless there somehow was an efficient way to access the pre-resolved samples for all involved components, with full MRT support. We're not there yet, I'm afraid.
I looked into the support in D3D10 more and indeed you can access the pre-resolved data but unfortunately I'm not sure that deferred shading can make use of this efficiently. The problem is that you can't get access to any sort of compression flag to indicate whether all of the samples are the same, which means you are stuck evaluating the BRDF for all of the samples anyways... why not just super-sample?
Quote:Original post by Yann L
However, the visual impact of having completely dynamic area soft shadows everywhere is definitely worth it :)
Sounds really nice - any chance we can see some screenshots or get some info on the techniques that you're using?
Quote:Original post by Yann L
Oh, and could ATI *PLEASE* fix the two trillion depth buffer related bugs in their current OpenGL driver already ? Working around their bugs is a real PITA.
Haha, you ran into this too? The other major annoyance on ATI is the poor performance of MRT in general. In a deferred shading demo that I wrote a while back even with huge resolutions and lots of geometry it was *still* faster on ATI to render the buffers one by one, whereas on NVIDIA the crossover is legitimately at something like ~1000 polygons.
Quote:Original post by Yann L
Anyway, cheers guys, and thanks for your suggestions !
Sorry they couldn't be any more useful :( Good luck with your project though!