Hi =) The only place in the Direct3D help file I noticed mention of the TnL HAL is during initializing, so I''m assuming that once you''ve found an TnL HAL device and intitiated Direct3D using it, all the rest of Direct3D API is used as normal, only that it will be TnL accelerated? For example I dont need to call something like DrawIndexPrimitiveTNL(...) or some other thing. Also I''m wondering about the best way to use Vertex Buffers with a TnL card. I read someplace that ProcessVertices is always done in software, so using it with a TnL card is pointless. But then access to already transformed vertices could be very usefull, for frustrum checks, collision checks, etc. So I thought a combination of both would be the best, have a vertex buffer with (for example) a bounding box of some object and have another vertex buffer with the actual object''s vertices, run ProcessVertices on the bounding box and check of frustrum visibility, if viewable then send the actual object''s vertex buffer for rendering, this way I do minimal ProcessVertices and use hardware acceleration when needed. Am I in the right track here? One last question, what is the concept behind Optimize() function, does Direct3D assume no more operations will be done in the VB and arranges the data in a way better suited for rendering? Thanks, Rick

Can''t access it ? does it mean that U can''t transform it by calling ProcessVerex (or something like that) ?