Early Z / TBDR and shader ops like texm3x2depth or texdepth

ram

Newcomer
Since PS 1.3 the depth value of a pixel can be calculated in the pixel shader rather than beeing fixed to the interpolated value based on the vertices. Therefore, any occlusion detection technique would fail if it is only based on vertex data and therefore has to be deactivated if a buch of triangles is using a PS program containing one of this instructions.

Are these z-value ops very usefull for developers? For what kind of things could these ops being used for beside z-correct BM and depth peeling?
 
I wouldn't really use it for anything other than the things you mentioned...

I'd guess this would be the same for other developers.
 
They are useful in an academic sense but the extra cost means its not used that much at the moment.

Some other things it can be used are non-polygon rendering (draw a quad, solve a ray/sphere equation, viola perfect sphere in your scene), CSG style ops, merging pre-rendered scenes (res-evil) and pin boards (imposters that z-buffer correctly)

The last 2 (which are really the same thing) are the main uses I can see a game using at the moment.
 
Back
Top