I am using a heightmap-based terrain with a camera view that is predominantly top-down. I''d like to know an efficient method of mouse selection to select portions (i.e., individual tris) of the terrain. I am dividing my terrain into 16k tri patches, however, that is still far too many tris to evaluate when using ray-tri intersection tests. When I shoot the ray down the current camera view, how can I perform collision tests on the fewest tris possible? I am using quadtrees for spacial queries for the purpose of object-object collision detection. That works well because the number of objects in each quad is relatively small. However, I don''t see how this method can work with terrain. Can anyone point me in the right direction? Is there a more efficient mouse selection techqniue given my environment?