Hi ! I'm in the process of implementing hierarchical occlusion maps. I have my Z-hierarchi and I now need to test wether an object is occluded or not. This should be done by projecting the objects bounding box into the Z-hierarchi, but from here it gets a little blurry. Should I: Check each pixel in the projected area and then skip the object if none of the tests is true and continue down the Z-hierarchi if some of the pixel-tests is true? It seems to me that's a lot of pixel-processing >:) Best summer-regards, Roquqkie

Basically yes, this is what should be done, the process is pixel intensive but
you may also use a scaled version for your rasterizer and smd instructions for inner loop like checking z value in a single row for 2floats at time or 4 if you use different packing.