This web site is no longer maintained and the content may be outdated.
Please visit for up-to-date information.
No upcoming events...

Home / Graduate / M.S. Theses Completed
  Cem Yüksel, 2003    

Thesis Title

Photo-realistic rendering using gradient space projection


Photo-realistic rendering has always been a challenge in computer graphics. Global illumination algorithms, which help us in making realistic lighting simulations, have a key role to achieve this goal. In this thesis we present a projection-based method for calculating global illumination. The projection method is Gradient Space Projection (GSP), which projects each polygon in the scene on a plane tangent to the hemisphere centered on the point of interest. We first discuss GSP in detail, and then propose four sampling methods for using this projection. The first sampling method is geodesic dome sampling, which uses the nodes of a projected geodesic dome as the sampling points. The second method is alternative sampling method, which uses the intersection points of a grid as the sampling points. The other method is random sampling and the last sampling method is distance constrained random sampling, which applies a minimum distance rule to the random sampling point generation. We present the test renderings for comparing both the performance and the quality of the sampling methods. Our results show that the geodesic dome sampling is superior when both performance and quality are considered.The GSP method is a general solution to the global illumination problem. It is based on the same principles as the previous methods, so their acceleration and enhancement methods are also applicable to GSP. Even though the complexity of GSP is linear, its basic operation is a simple perspective projection, which is fast.
Boğaziçi University Department of Computer Engineering
Address: 34342 Bebek, Istanbul, TURKEY
Phone: +90 212 359 4523-24 Fax: +90 212 287 2461
general information:   webmaster: