3 ms·
The Rendering Equation [pdf]
- LolWolf 6y agoThis is great! Thanks for the reference. There are some interesting methods to solve these kinds of "recursion" equations, especially when the integral operator, M in this paper, has the given properties (e.g., spectral radius < 1). For example, if M = A + P, where A and P are two other operators whose inverse, for example, is fast to evaluate, there are a certain number of iterative splitting methods which work extremely well and can exploit this structure for very fast convergence. I'm sure people have looked at this and similar ideas, but I have not seen many references as far as I know.
- amelius 6y agoShouldn't the rendering equation be the same as Maxwell's equations? EDIT: this is actually addressed in the paper. It is an approximation. I don't see a derivation starting from Maxwell's equations, though.
- LolWolf 6y agoYeah, it's essentially a paraxial approximation. You can find derivations (through a few steps) by essentially doing the following reductions: Maxwell's -> Fourier optics (scalar wave phenomena) -> Eikonal equation (rays, but in inhomogeneous media) -> Ray optics (rays, but in almost-everywhere homogeneous media). I think most modern optics textbooks (see, e.g., Fundamentals of Photonics) cover the reductions and the corresponding steps.
- dls2016 6y agoClassic paper. I studied this before returning to grad school due to my job at the time. I eventually studied dispersive PDEs... but no matter where you turn in PDEs you always apply contraction mapping to establish existence of solutions. Was nice to have the ray tracing interpretation floating around.