A Particle-Path based Method for Monte Carlo Density Estimation
Miguel Lastra, Carlos Ureña, Jorge Revelles, Rosana Montes · 2002
Radiosity computation on scenes including objects with a complex geometry, or with a large number of faces and meshes with very different sizes, is very complex. We present a new method (based on the photon maps method 6 ) where density estimation on the tangent plane at each surface point is performed for irradiance computation by using photon paths instead of photon impacts. Other density estimation techniques require locally simple and flat surfaces in order to be able to compute approximations to irradiance. This requirement is not present in this method, because it does not use photon impacts. Thus we claim the algorithm is more independent of geometry. The algorithm requires fast ray-disc intersection tests, so we propose techniques for speeding up this part of the process. We compare this approach with the photon map technique, both qualitatively (in terms of its ability to handle complex geometry), and quantitatively (in terms of time and storage complexity). We obtained results which show that better images are obtained with some extra cost both in time and memory.