From what I've learned from GL programming, for skinned meshes at least, it might involve multiplying vert data by 'x' amount of parent bone matrices to get a world position, and do that every frame for every vert in a mesh. That's a LOT of math per mesh. It should be easier to implement with MD2s though, as they're just vert deformation data, and that could be looked up from ram.
Non key frames (tweened/interpolated frames) would need a little extra math.
Maybe hardware shaders/DX9 can do skinned meshes quickly but I doubt DX7 can.
Anyone know any better?
Non key frames (tweened/interpolated frames) would need a little extra math.
Maybe hardware shaders/DX9 can do skinned meshes quickly but I doubt DX7 can.
Anyone know any better?