Is this possible??

Ok, bump mapping, its a pain to say the least. But here is my theoretical question. Lets say I made a texture. Set the texture to grey scale, and used that scale to build a hight map. You know white is highest, black is lowest. (something like this, but with varing shades of grey)

Then with texture points I create a triangle mesh, connecting each point to the next. (Maybe not useing every point, maybe every other or every fourth)(Creating a hight map)Sorry if im being redunant, im not sure how to say this. Then using each face of the triangles created, create a normal for each face, or maybe for each vertex, i supposs it dosnt realy matter. Now I would have a fair repressentation of the lighting equivilant of Bumps on the texture. Now say i created a LOW polymodel, and textured that model with said texture. Here is the BIG question. Is there ANY way possible to use just the normals created from the hightmap on the model created. The reason I wish to do this is simple. If I could create a hightmap in this fashion, and then not have to use the verts, but simply use the normals on another model, i think it would be a fast effiecent method of bump mapping, more or less. I know this probly wont work, but if anyone has any idea of how it might, or any similar idea that will work. Please let me know. Bump Mapping math is really kicking my a**.


if I understood it right this is more or less what is done with Q2 & Q3 models. They have a hires Model with 250000 verts and then reduce it to 5000 or so and then use the normals of hires models.

Search the forum for “Quake” or “md3”.
There was a discussion on this lately.



What you describe is very similar to tangent space bump mapping. NVIDIA has a lot of articles about it on their developer web site.


Thanks for the info, i will try my best to find this info. I knew it was possible.

Well, OK, i tried for a while today, and i still have NO idea how to do this. Anyone know any specifics???

[This message has been edited by dabeav (edited 06-04-2002).]