Ah, what you''re talking about is a NORMAL MAP (i.e. normals encoded as RGB values). Usually you create that from a height map by treating each pixel in a source height map texture as a vertex for a triangle. The x and y values for the vertices are simply the x and y positions and the z values are specified by the values in the height map. You then just use the same method of finding the normal as you would for any triangle (e.g. cross product), then encode the resultant value as a colour.

As for how to generate them:

1) There is a PhotoShop plugin available on the nVidia site IIRC which generates normal maps from height maps.

2) D3DX has a function to do it too: D3DXComputeNormalMap()

3) Finally you''ll find source code to do it in samples on the nVidia and other IHV websites. The OpenGL code is just as applicable to D3D as long as you''re consistent about which space the normals are in/which elements of the vector maps to R, G and B.

--
Simon O''Connor
Creative Asylum Ltd
www.creative-asylum.com

So you already have a normalalised normal vector per pixel (e.g. you''re doing something like collapsing geometric detail into bump maps) ?. If so here''s a snippet of code:

nmap[i*width+j].r = (BYTE)(128 + 127*nx);
nmap[i*width+j].g = (BYTE)(128 + 127*ny);
nmap[i*width+j].b = (BYTE)(128 + 127*nz);


--
Simon O''Connor
Creative Asylum Ltd
www.creative-asylum.com