Precomputed Conservative Visibility

Bachelor-Thesis

Visualization of very complex scenes is a challenging problem in many Computer Graphics applications such as games, virtual reality or urban planning. To efficiently render such complex scenes, many accelleration schemes can be performed, which reduce the amount of primitives being sent to the graphics pipeline. One possibility is to perform occlusion culling by discarding geometry which is not visible from the viewer's position.
We can distinguish between on-the-fly methods, which perform occlusion culling for the current viewpoint and preprocessing approaches, which determine the visibility for given regions in advance.
In particular, the task of the thesis is to implement an occlusing culling algorithm based on precomputed conservative visibility sets and evaluate its performance in terms of memory and computational efficiency.

Advisor(s)