Terrain rendering has heretofore been computed by a CPU and rendered by a combination of CPU and GPU. It is possible to implement a fast terrain renderer which works optimally with current 3D hardware. This is done by using geo-mipmapping which splits the terrain into a set of smaller meshes called patches. Each patch is triangulated view-dependently into one single triangle strip. Special care is taken to avoid gaps and t-vertices between neighboring patches. An arbitrary number of textures can be applied to the terrain which are combined using multiple alpha-blended rendering passes. Since the terrain’s triangulation changes over time, vertex normals cannot be used for lighting. Instead a pre-calculated lightmap is used. In order to reduce popping when a patch switches between two tessellation levels geo-morphing is implemented. As will be pointed out later, this splitting of the terrain into small patches allows some very helpful optimizations.
[1]
Willem H. de Boer.
Fast Terrain Rendering Using Geometrical MipMapping
,
2000,
WWW 2000.
[2]
Carlo H. Séquin,et al.
Visibility preprocessing for interactive walkthroughs
,
1991,
SIGGRAPH.
[3]
Hugues Hoppe.
Smooth view-dependent level-of-detail control and its application to terrain rendering
,
1998,
Proceedings Visualization '98 (Cat. No.98CB36276).
[4]
Steven Skiena,et al.
Optimizing triangle strips for fast rendering
,
1996,
Proceedings of Seventh Annual IEEE Visualization '96.
[5]
John M. Airey,et al.
Increasing update rates in the building walkthrough system with automatic model-space subdivision and potentially visible set calculations
,
1990
.
[6]
Mark A. Duchaineau,et al.
ROAMing terrain: Real-time Optimally Adapting Meshes
,
1997,
Proceedings. Visualization '97 (Cat. No. 97CB36155).