Live Range Visibility Constraints for Adaptive Terrain Visualization
Xiaohong Bao, Renato Pajarola, Michael Shafae · 2004
Although there is a remarkable pace in the advance of computational resources and storage for real-time visualization the immensity of the input data continues to outstrip any advances. The task for interactively visualizing such a massive terrain is to render a triangulated mesh using a view-dependent error tolerance, thus intelligently and perceptually managing the scene’s geometric complexity. At any particular instance in time (i.e. displayed frame), this level-of-detail (LOD) terrain surface consists of a mesh composed of hundreds of thousands of dynamically selected triangles. The triangles are selected using the current time-step’s view parameters and the view-dependent error tolerance. Massive terrain data easily exceeds main memory storage capacity such that out-of-core rendering must be performed. This further complicates the triangle selection and terrain rendering owing to tertiary storage’s relatively poor performance.