Is this cubic shadow mapping a special technique because from what I have seen it doesn’t seem to be very similar to the NVidia presentations. Does anyone know how it is working? The source code from the demo is directx and at the moment I don’t know much about the way shadow maps are working so it is pretty hard for me to follow the demo’s algorithm…
I will take a look at the demo when I get a chance. My guess is that its basically like doing standard 2d shadow mapping but from all directions (basically in world space with a cube map. Which is great if the light and the scene don’t change. But if the do you may be stuck updated the cube map a lot.
Unfortunately the 3 pages of NVidia presentation about that are just a rough overview so it won’t help you much if you don’t know how exactly this cubic shadow mapping is working (well, at least that’s the case for me ).
The cubic shadow mapping is really looking good in the demo I have found but rendering the scene 6 times per light is of course pretty expensive. And I have seen that shadow maps have more problems than I thought at the beginning. So probably shadow volumes are still the best way to go if you have everything shadowing everything.