Hi! After reading about HOMs in RTR 2nd. ed. and in the forums here at GameDev.net, I decided to implement it myself. I've then written a software-renderer, that renders my occluders to a Z-buffer. Now I have a few questions: What is the most efficient method to downsample my Z-buffer, that is building the Z-pyramid? After I have build my Z-pyramid, how should I in the most efficent way, test if an object is occluded? Should I check for every projected bounding-volume pixel etc.? Thanks in advance! Best regards, Roquqkie ps. Thanks Yann for sharing your precious information with us!