Authors:
Sid'Ali Kalem
and
Assia Kourgli
Affiliation:
USTHB, Algeria
Keyword(s):
Large Terrain, Rendering, LOD, View-dependent, GPU, Wavelet, QuadTree.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Real-Time Rendering
;
Rendering
Abstract:
The goal of the following paper is to point out an alternative approach to the adaptive triangulation problem. A new technique of terrain rendering which uses wavelet transform to select appropriate LOD is described. This technique is a region-based multi-resolution approach that partitions the terrain into tiles that can be processed independently. The algorithm organizes the heightmap into a QuadTree of nodes and computes maximum world-space errors for each node. World-space errors are then calculated at preprocess step. As the datasets of realistic terrains are usually huge, we suggest using the multi-resolution wavelet decomposition to localize the position of the maximum world-space error estimated and limit the region of research inside the node. It permits to choose the appropriate resolution of the regular grid that will represent the node at run time. By this way, computation load on the CPU is greatly reduced.