skip to main content
Modeling global diffuse illumination for image synthesis
Publisher:
  • University of Texas at Austin
  • Computer Science Dept. Taylor Hall 2.124 Austin, TX
  • United States
Order Number:UMI Order No. GAX92-12495
Bibliometrics
Skip Abstract Section
Abstract

Rapid developments in the design of algorithms for rendering globally illuminated scenes have taken place in the past five years. Net energy methods such as radiosity algorithms have become effective at computing the energy balance for scenes containing diffusely reflecting objects. Such methods first break up a scene description into a large set of elements, or possibly several levels of elements. Energy transfers among these elements are then determined using a variety of means. While much progress has been made in the design of energy transfer algorithms, little or no attention has been paid to the proper generation of the mesh of surface elements.

This dissertation presents a technique for adaptively creating a mesh of surface elements as the energy transfers are computed. The method allows large numbers of small elements to be placed at parts of the scene where the most active energy transfers occur without requiring that other parts of the scene be subdivided needlessly to the same degree. As a result, the computational effort in the energy transfer computations can be concentrated where it has the most effect.

Since the sources of direct and indirect illumination in the scene are polygonal elements, the effects of light sources with finite area must be computed. Most methods simplify the problem by approximating the area source with a collection of point sources. We present an object space algorithm to model illumination from polygonal light sources analytically. The result is a collection of smooth-shaded polygonal facets that may be rendered from any viewing position. Binary Space Partitioning trees are used to compute umbra and penumbra boundaries efficiently. Fast analytic techniques are developed for illumination calculations. Numerical optimization methods ensure that the shading function is sampled finely enough to find all significant illumination gradations. Illumination calculations are optimized to concentrate computational effort on parts of the scene where they are most needed.

Contributors
  • The University of Texas at Austin

Index Terms

  1. Modeling global diffuse illumination for image synthesis

    Recommendations