How do you set up a depth buffer to read from? I tried creating a texture, setting it to a D3DUSAGE_DEPTHSTENCIL and then setting the depthStencilSurface to it's surface. This worked for the render process (z Buffer functioned normally), but when I use tex2D look ups on the depth map, it returns garbage. So, I thought I'd try copying the depth buffer to a standard texture surface, but this proves troublesome as StretchRect requires similar formats. The first method seems like the way to go.. is there something special I need to do to be able to use tex2D on a depth buffer formatted surface?

I see. Well.. that's balls.

Well, can I copy the depth buffer somewhere where it can be read?

I'm rendering to a normal/depth map for edge testing, but if I don't include the shadow data in that pass I'll have to render the entire screen geometry again. I could use a non-deferred shadow map, but that wouldn't allow me to use PCF to blend the shadows.

Any other ideas? :/

Edit: Is there any way render more than 4 floats in one pass? Ideally I'd like to be able to do all the deferred maps in one go. I know you can render to multiple targets, but I assume that's just a copy? Or can you specify color[n] to set it to a different target?

Quote:Original post by MJP
You can put whatever data you want in each target, by using color[n]. This is very common for deferred rendering where you'll write out depth, normals, diffuse material, specular material, etc. in one pass.

Cool, yeah.. I'm checking this out now.

It seems that my render targets are sharing alpha values, however.

If color0.a == 1 then render target 2 gets color1, if not it blends in with the back buffer. This will be a problem for the decal pass, as decals are most certainly going to need alpha...

I'm really not sure what's going on with this method of rendering (Well, I get the theory but there's a lot of new semantics), so if you could direct me to a tutorial, or something, that'd help a lot.