A Simple Spectral Approach to Stochastic Modelling for Natural Objects

Abstract

Stochastic modeling has been widely used in computer graphics to depict natural objects or phenomena. Various techniques are available, depending on what object is to be represented and what extent of realism to be achieved. This paper presents a simple approach to stochastic spectral synthesis for producing a large range of natural objects and scenes at low computational cost. By considering a wide class of power spectra, the method can provide twodimensional stochastic models for various objects, involving terrains, clouds and sea waves. Moreover, some functional operations and parameters for the models are introduced, which can make the models more flexible in describing a three-dimensional complex scene.