Image Gallery

All images are Copyright (c) University of Manchester,
and may not be reproduced without permission.

Perceptually-Driven Radiosity (1995-1998)

The following images were generated using the VRad radiosity system, which was developed during the course of my Ph.D. research. The radiosity solutions shown here were generated using a perceptually-adaptive algorithm, that ensured that only features of the radiosity solution which are visible to a human observer were computed to any great accuracy. Those features that were not visible were only approximated using a much faster technique. This enabled radiosity solutions to be generated much more quickly than before.

S. Gibson, R.J. Hubbold, "A Perceptually-Driven Parallel Algorithm for Efficient Radiosity Simulation", IEEE Transactions on Visualization and Computer Graphics, 6(3), July-Sept 2000, pp. 220-235.

S. Gibson, "Efficient Radiosity Simulation using Perceptual Metrics and Parallel Processing". Ph.D. Thesis, Department of Computer Science, University of Manchester, September 1998.

S. Gibson, R.J. Hubbold, "Perceptually-Driven Radiosity", Computer Graphics Forum, 16(2), June 1997, pp. 129-140.

These images were generated from a model of the Advanced Interfaces Laboratory, and show both artificial and natural illumination.
The model in these images is from part of a processing plant. It is illuminated using a hemispherical skylight model.
The illumination in these images was generated using a discontinuity meshing algorithm.
These images show how ray-tracing can be used to add specular effects to a view-independent radiosity solution as a post-process. The atrium model is from the University of Aizu.

The model used to generate these images contains over 1.2 million polygons. Solution time was 17 hours for 2000 iterations of a progressive radiosity algorithm.

Finally, these are images from a radiosity solution of the University of Manchester's Computer Science Department. Portalling and light-mapping was used to allow real-time navigation through the model on a Silicon Graphics O2 or PC equipped with a 3DFx Voodoo2 graphics card.