Multi-Resolution Foliage Faithful to the Visual and Botanic Concepts

Foliage, the first element to be perceived and the first clue to distinguish the species determines the final shape and silhouette of the tree. However, a large number of isolated polygons in the foliage model challenge the real-time rendering of dense trees. In this paper, a novel GPU-based foliage simplification algorithm faithful to visual and botanic concepts is presented for extracting the multi-resolution model of the tree foliage. Different from the geometric foliage simplification approaches, we introduce the information-theoretic metric to measure the leaf visibility from a set of external viewpoints and determine the best leaf pair for collapsing. Moreover, leaves in the foliage are clustered into clouds of cells faithful to the branching structure of the tree. With the help of the cell density and distance factor of the observer to tree model, the level of detail (LoD) foliage models close to the original appearance in terms of visual and geometric criterion are parallel produced in real time. Our approach is suitable for applications which require exact time complexity but also high visual quality.