<--back--   home   --next-->

The shading model for lit-tensors is based on the Blinn-Phong shading model. Its not important to go through every factor here, but there are three terms: ambient, diffuse, and specular. We should note that the directiffonal parts, the diffuse and specular terms, both depend on a dot product with N, the surface normal. How do you do that in lit-lines, where there is really a whole family of normals? Instead of having to find some particular normal to plug into the shading equation, you can use the tangent direction T, and then by Pythagorus's theorem you can express the dot product with N in terms of T. With a tensor, if its anisotropic, the third eigenvalue will be small, and the direction of the first two eigenvectors can be treated as analogous to tangent directions. In the linear case, only e_1 matters, and in the planar case, e_1 and e_2 matter equally. So how do we vary the contributions of the two eigenvectors based on the anisotropy?