Abstract

Maxwell’s equations completely define the behavior of electromagnetic radiation, according to the electric and magnetic fields. Visible light is electromagnetic radiation, and therefore its behavior also follows the same laws. For a perfect physically-based simulation of the illumination of a scene, the ideal algorithm would solve these equations. However this would be unpractical even for the simplest scene due to very high and unreasonable computation times and the need to model the scene in terms of electric and magnetic fields.

In Computer Graphics, even in the field of physically-based rendering, simplified and more intuitive models are used. These simplifications enable a more practical simulation, with understandable and intuitive modeling based on light sources, geometry, material parameters and participating media. Algorithms that simulate the illumination on those scenes find an adequate trade-off between accuracy and rendering time. Still, the potential results of a simplified system will always be a subset of all the potential optic phenomena that are owed to Maxwell equations. By simplifying the simulation, even though rendering time decreases and scene modeling becomes more intuitive, limitations on the possible simulated scenes arise.

However, these limitations can be overcome. More complex scene models that are able to represent scenes that include new phenomena without losing intuitiveness can be designed. New algorithms that are able to simulate a broader spectra of phenomena can be developed. Actually, there is plenty of previous work that tries to focus into these phenomena that are related to the nature of light, considering complex physically based models and creating new algorithms to simulate these new scenes. But, where is the limit? How close can we get to the complete simulation of light scattering without getting to unreasonable rendering times? How complex can a physically-based model become without losing intuitiveness and practicality? How close can we get to Maxwell’s equations? These are the questions that this PhD is trying to answer by trying to get to these limits, going further into the physical simulation of light scattering in scenes while keeping the modeling of scenes as intuitive and simple as possible.

Files

• Thesis [pdf - 32.7 MB]
• Slides [odp - 38.5 MB]

BibTeX