Abstract:
As the GPU's processing power has been improving much faster than the CPU's. the terrain rendering algorithms have evolved to use the graphics hardware as much as possible. One of the recently developed GPU-based Level of Detail (LOD) algorithm is Continuous View Dependent Adaptive LOD using hardware tessellation. In this study, Continuous View Dependent Adaptive LOD using hardware tessellation is enhanced using three additional methods. First method is determining pixel-based LOD using occlusion query. Because of hidden surface culling, render time is reduced. Second method is determining pixel-based LOD using occlusion query that newly developed by using OpenGL and CUDA interoperability. Œird method is extension of second method that includes quad-tree based query for each terrain node. Œird method aims to increase rendering quality of partial rendered terrain node.