Efficient virtual plant data structure for visualization and animation

marc jaeger,ruoxi sun,jinyuan jia,vincent le chevalier
2010-01-01
Abstract:Virtual plants are interesting but costly objects in video games and virtual reality applications. Despite advances on Level of Detail approaches and increasing GPU power, previous work fails in producing convincing and generic vegetation with low bandewidth requirements for web display. The efficiency drops down considering costs of compression and inner structure redundancy is still poorly explored. We propose here a tree graph as a generic output of simulation, easy to implement from any rule based plant generator. This graph allows to efficiently compress similar structures, that appear naturally as a result of procedural definition. Nodes carry minimal attributes, mainly related to age. Upon display requests, nodes orientation and precise geometry are computed using a simple graph exploring approach based on a very small number of parameters,. This exploration allows to generate all past growth stages and dynamic effects such as pruning. We show how this graph can be efficiently generated from L-systems, as well as from a functional-structural plant simulator. We illustrate that, thanks to this encoding and reconstruction scheme, a simple tree simulation can generate a wide range of geometrical plant exports at various ages, and is compatible with classical level of detail rendering approaches for geometric models. © 2010 IADIS.
What problem does this paper attempt to address?