Conservative From-Point Visibility

Visibility determination has been an important part of the computer graphics research for several decades. First studies of the visibility were hidden line removal algorithms, and later hidden surface removal algorithms. Today’s visibility determination is mainly concentrated on conservative, object level visibility determination techniques. Conservative methods are used to accelerate the rendering process when some exact visibility determination algorithm is present. The Z-buffer is a typical exact visibility determination algorithm. The Z-buffer algorithm is implemented in practically every modern graphics chip. This thesis concentrates on a subset of conservative visibility determination techniques. These techniques are sometimes called from-point visibility algorithms. They attempt to estimate a set of visible objects as seen from the current viewpoint. These techniques are typically used with real-time graphics applications such as games and virtual environments. Concentration is on the view frustum culling and occlusion culling. View frustum culling discards objects that are outside of the viewable volume. Occlusion culling algorithms try to identify objects that are not visible because they are behind some other objects. Also spatial data structures behind the efficient implementations of view frustum culling and occlusion culling are reviewed. Spatial data structure techniques like maintaining of dynamic scenes and exploiting spatial and temporal coherences are reviewed.