Advances in 3D scanning technologies have enabled the practical creation of meshes with hundreds of millions of polygons. Traditional algorithms for display, simplification, and progressive transmission of meshes are impractical for data sets of this size. We describe a system for representing and progressively displaying these meshes that combines a multiresolution hierarchy based on bounding spheres with a rendering system based on points. A single data structure is used for view frustum culling, backface culling, level-of-detail selection, and rendering. The representation is compact and can be computed quickly, making it suitable for large data sets. Our implementation, written for use in a large-scale 3D digitization project, launches quickly, maintains a user-settable interactive frame rate regardless of object complexity or camera position, yields reasonable image quality during motion, and refines progressively when idle to a high final image quality. We have demonstrated the system on scanned models containing hundreds of millions of samples.

From a game developers point of view terrain rendering is much more than just the real-time display of a landscape. More accurately, the latter is only one part of what we call the terrain rendering pipeline. The most important stages of this pipeline are the design of the terrain geometry, the real-time display of the terrain, natural texturing, dynamic lighting, dynamic shadowing, and the enrichment of the terrain with organic features. We intend to survey each stage of this terrain rendering pipeline by giving an example from the actual multi-award winning DX8 game AquaNox. In this way we illustrate the data flow through each stage of the pipeline, effectivley giving a report on the current state of the art in terrain rendering.

Mesh simplification algorithms play an important role in computer graphics. In particular, view-dependent simplification methods are utilized widely to reduce the geometric complexity of large outdoor scenes. The so called continuous level of detail technique, for instance, is prevalent in the area of terrain rendering. In this paper we apply the latter technique to the more general volumetric case. We propose a volume rendering algorithm, which constructs a hierarchical tetrahedral mesh from regular volume data in a view-dependent fashion. The popping effect which generally arises from view-dependent algorithms is addressed by a new fast method for the smooth interpolation of the mesh hierarchy. We demonstrate the performance of our algorithm by displaying clouds in real-time. Here, the application of the pre-integration technique allows for a wide range of cloud appearance and guarantees accurate compositing. Additionally, we show how to utilize our algorithm for the accelerated display of ground fog in terrain rendering scenarios.

Triangulated Irregular Network (TIN) is a high-effect means expressing the terrain surface in the computer, compared with GRID model, TIN model can use less sampling to approach the terrain surface precisely, this is why we study Visualization of TIN model. The data structure of TIN model is complex, therefore to realize View-dependence LOD representation of TIN is difficult. General method is to partition the TIN, every tile adopt the different resolution according to the distance from viewpoint.Quadtree domain decomposition is adopted on TIN in this paper, and this is no longer simple rectangular partition to realize View-dependence LOD representation of TIN, but Through building a quadtree structure for tiling of TIN, and combining quadtree based LOD methods to realize controlling over the tiling of TIN effectively, this method can combines adjoining tile sometimes, the high limit cuts down the triangle quantity in the LOD mesh. This algorithm has combined TIN and quadtree structure, and combines quadtree based LOD methods and TIN based triangle mesh simplification methods to realize the view-dependence lever-of-detail controlling over the TIN model. Research indicates that this algorithm cuts down the triangle quantity in the LOD mesh and doesn’t lose visual effect, a leverof-detail mesh, similar with Grid and the TIN hybrid structure, has been acquired. 1.

plain terrain with NDVI-based vegetation The area of terrain rendering has seen great improvements both in rendering performance and image quality. The latest terrain rendering algorithms efficiently utilize the capabilities of actual programmable graphics hardware in order to achieve real-time visualizations of large terrain models. Despite these advances, a remotely sensed dataset will not look realistic unless the terrain model also accounts for the vegetation. We present an extension of the C-LOD technique that is able to enrich large GIS scenes with forestal detail by rendering trees and bushes in a view-dependent way. The placement of the trees is derived automatically from real NDVI data. We also present a NDVI based texture splatting method in order to render the underlying grass layer in a realistic way.