Terrain rendering using regular mesh cell

The earlier algorithm of rendering terrain focused on the terrain mesh only in GPU, and tried to send as less triangle as possible to display card. However, the model simplification and multi-level task cause GPU a lot of computation loads. In recent year, with the fast development of GPU capability, the bottleneck of terrain rendering system is CPU. In order to make full use of GPU capability, and package as many scattered triangles as possible to GPU in the same time, i.e. submit more triangles when transfer graphs to render functions, and decrease the transfer number of graph rendering function. Based on the above rules, this paper proposed a new method to render terrain layered structure, the relative location between layers keeps still, which is strict nested structure. The whole mesh structure is composed by the translation and zooming of a mesh cell. Before rendering terrain mesh, this mesh cell is compiled to OpenGL display list. Because the display list is only a number, so the transfer data amount is a little when rendering this mesh cell, which can greatly improve the rendering performance. When this mesh is in GPU, mesh construction in CPU is avoided by sampling height texture to obtain height value during the vertices tinter.