> That's pretty darn clever. You could sample the mesh via raycasting to get > the heightfield data from it. Anybody want to try it? : ). Thinking some more, if you only allowed negative offsets, then you could do normal cylinder collision tests before you'd need to test the heightfield... -- gl