I’m trying to implement dual depth shadow mapping as described in http://wwwvis.informatik.uni-stuttgart.de/ger/research/pub/pub2003/eg03-weiskopf.pdf .
I’m running into a problem on the second pass where you use a fragment program to compare the current pixel’s depth value to a previously rendered depth texture (so that you can KIL it if it is equal and get a second-depth map).
It seems that no matter what I do, I cannot get correct values out of my bound depth texture.
Here is some pseudo-code. I am using Mark Harris’ RenderTexture for the depth texture.
// Pass 1 --------------------------
depthBuff1->BeginCapture();
glClear(GL_DEPTH_BUFFER_BIT);
applyLightMatrices( );
renderScene( );
depthBuff1->EndCapture( );
// At this point, so far so good.
// I can render the scene using depthBuff1 as the
// shadowmap and get correct shadowing, albeit
// with the usual artifacts
// Pass 2 ----------------------------
depthBuff2->BeginCapture();
glClear(GL_DEPTH_BUFFER_BIT);
applyLightMatrices( );
depthBuff1->BindDepth( );
enableDepthCompareFragmentProgram( );
renderScene( );
disableDepthCompareFragmentProgram( );
depthBuff2->EndCapture( );
And here is my (test version) fragment program. The real fragment program will do a comparison, but this one has been simplified to just copy the depth texture into the newly cleared depth buffer.
!!ARBfp1.0
TEMP firstDepth;
TEMP tc;
// Get texcoords into [0,1] range
MUL tc, fragment.position, { 0.00125, 0.001721, 1.0, 1.0};
// Lookup shadow texture
TEX firstDepth, tc, texture[0], 2D;
// Copy result to new depth buffer (have also tried firstDepth.z, firstDepth.w, etc)
MOV result.depth.z, firstDepth.x;
Unfortunately this does not work. DepthBuff2 is always blank after running pass 2. Is this the correct way to look up a depth map from the fragment program? Is it possible to have a depth-based render-to-texture enabled and bind another one at the same time?
If it matters I’m on a 6800 ultra with 71.89 driver version.
Not sure what I’m doing wrong here so any help would be appreciated.