I realized I never wrote anything about the generic convex object visualization I mentioned earlier. It actually turned out really good, and the algorithm is very simple: 1) Get the support point for each of the positive and negative coordinate axes. 2) Build an octahedron with corners at the six points (like a pyramid pointing up, on top of a pyramid pointing down. This is the starting shape. 3) For each triangle in the current shape: Get support point in outwards normal direction. If support point does not equal (within margin) any of the three corners, split the triangle into three new ones, using the new point. 4) Repeat step three until convergence, or maximum number of iterations. Splitting the triangle into three is not ideal since it encourages long thin triangles and a rather uneven triangle distribution, but in practice it works fairly well. I guess one could look into Loop-style edge splitting instead, but it would require more bookkeeping. I add a small sphere to each su
A game technology blog with focus on physics. I'm the co-creator of mobile games Smash Hit, PinOut, Does not Commute, Granny Smith, Beyondium and Sprinkle. Twitter handle: @tuxedolabs