Path Planning for Natural Phenomena

The ability to create su ciently detailed natural phenomena is important for building virtual environments, unfortunately this is a laborious task. Although various procedural techniques are available, new techniques that can ease the burden on digital artists are highly sought after. The use of path planning techniques that originate from arti cial intelligence is a viable avenue for the generation of models resembling natural phenomena characterized by their dendritic nature such as plants, coral, trees and lightning. The technique involves nding the least-cost paths through a weighted graph from a single source to a number of destination points. This work presents a new method of informing the behaviour of the path planner through images in order to build models that emulate certain familiar objects. The e ectiveness of the method is shown in three examples, two for trees and one for coral.