In our Project 1 we used Ray Tracing render a given image. While we managed to do it, it still took a decent amount of time i.e. function evaluations to complete the process. At UCL Computer Science however, Dr. Jan Kautz has developed a new method for rendering these images, and while we only rendered one stationary object, he has optimized this process for interactive rendering and dynamic scenes.
While the article does not describe the details his method, his paper entitled “Interactive Global Illumination Using Implicit Visiblity” explains the main process. The basic process of rendering is similar to that of radiosity, which involves creating links between surface elements and light sources and transferring energy between them until a solution is reached. However the main difference in this algorithm is that it uses a “hierarchal link structure,” which is computed once at the beginning of rendering. After creating this structure, the algorithm traverses this structure and removes an element if its length is too long, that is, light does not reach it from its parent. Many other simplifications are made, but in the end it draws up energy (light) from the leaf nodes and draws energy bottom up to the parent nodes. Lastly it takes the energy at each node and converts it to outgoing radiance.

He also compares this method with Ray Tracing, stating that the basis of the latter, which involves finding the first point of contact to a surface element, is a costly operation that prevents it from doing the interactive illumination that his new method does.

A more detailed analysis of his new method is discussed in his paper, in addition to examples of his work.

On an unrelated note, the water bunny that we rendered on project 1 is also found in the publications section of his webpage.

http://www.ucl.ac.uk/news/news-articles/0802/08022801

http://www.cs.ucl.ac.uk/staff/j.kautz/publications/PG07_ImplicitVis.pdf

http://www.cs.ucl.ac.uk/staff/J.Kautz/