Contents Introduction Overview of Deterministic algorithm Overview of Monte Carlo algorithm Scenes used for comparison Accuracy analysis Results obtained Conclusion Acknowledgments |
We used scenes of two kinds: simple and complex. There is only one simple scene for which we can find accurate theoretical values of illuminance distribution. All other scenes are realistic ones that are presented by several companies (listed below). The scenes represent different kinds of interior.
The first scene used for the comparison is simply an interior of a cube (10x10x10 m) with one point light source at its center (Figure 2). Luminous intensity of the light source is equal to 50000 cd that create illumination level 2000 lux at the nearest point on the wall. The wall material is white with the diffuse reflectivity 2/3 so that an indirect component constitutes a greater part in the full illumination.
Because of symmetry all cube faces are equivalent. We've chosen several points on a cube face where the theoretical results are compared with the results produced by the two algorithms being analyzed. Although the scene is very simple, the exact analytical solution for illuminance distribution in it seems to be impossible. We used a combination of the Monte Carlo simulation and the direct numerical integration of the rendering equation to get accurate estimates of illuminance in the points selected.
A series of interior scenes prepared mainly by Japanese companies is used. They are listed below. Scene size is characterized by the total number of triangles used. When a triangle subdivision was applied (to increase a simulation accuracy by reducing the discretization error) the number of triangles after subdivision is presented.
List of scenes:
Scene name | Created by | Triangles | Description |
APART2 | Integra | 5,000 | An apartment |
DREAM0 | Korea National Housing | 14,000 | A living room |
FLOWER | Integra | 17,000 | A flower in a room |
CG | Matsushita | 32,000 | A bathroom |
HONSHA | CTC | 68,000 | An office |
Simulations run on Convex SPP 1000 supercomputer (4x128M).
Contents Introduction Overview of Deterministic algorithm Overview of Monte Carlo algorithm Scenes used for comparison Accuracy analysis Results obtained Conclusion Acknowledgments |
Copyright 1996 Andrei Khodulev. | [Home] |