All right, I have a question for the well-informed. I''ve read that OpenGL has lights you can program for rendering shading and such. I''ve also heard of pixel and vertex shaders. And then there''s dot3 bumpmapping. Now, I''ve got dot3 bump-mapping to work, but it is slow because I use homemade algorithms to manipulate the texture coordinates of the normal and light maps based off of some position matrices. Other than using vertex and pixel shaders of high end cards, how can I hardware accelerate my bump-mapping using a card like a GeForce2--possibly converting my home grown algorithms? Please e-mail if anyone can help me at yoale_andrew@hotmail.com.

GL_ARB_texture_env_dot3?

------------------------------
BASIC programmers don''t die, they just GOSUB and don''t return.

That is the dot3 function I use. It works extremely well for my purposes, and it is fast. What kills my program is changing the texture coordinates of the light map in software and not hardware. I read through an article by nvidia on how to do the calcs on hardware, but soon realized I had a GeForce2--not a GeForce3 or better.

you mean the tangentspace transformation per vertex?

well then. think about it:
if your mesh is animated, it will be software transformed (means animated) on your cpu anyways => the additional 3x3matrix transform per vertex doesn''t really hurt much
if your mesh is _NOT_ animated, you could "pretransform" all the normals on the normal map from tangentspace to objectspace => removing the tangenspace matrix per vertex completely. the result is rather simple: object space bumpmapping. only needs the dot3 with the (object space) light vector, wich you only need to calc 1time per object. the rest, you can do with clever texcoordgen matrix setup, if you want..

go check "object space bumpmapping", there are some demos out..

"take a look around" - limp bizkit
www.google.com

although object space bumpmapping is really only any use with infinite lights...