I’ve been thinking about this for some time now, but never found a way, so maybe someone can help…
Setup: Facing the camera I’m stacking quads front to back with alpha cutouts in their corresponding textures, so that in the renderer one can see parts of the second layer through “holes” in the first layer, and then again through other holes in the second layer down to the third layer and so on.
So there’s only very little portions of the third layer’s texture that can be seen. If I wanted to run my composited image through a pixel shader this wouldn’t be a problem, but wanting different effect settings on each layer I’d have to process each texture individually and it seems to me I’m wasting a lot of computational power in places that can’t even be seen in the final image.
So is there a way to only process those areas of a texture which can actually be seen?
Concepts like depth prepass or usage of depth stencil won’t work I think, as there’s no real geometries but only flat planes with alpha.
If you don’t know what I’m talking about, maybe this helps:
Thanks in advvvvance :)