Computer modelling of landscapes

This paper presents two modelling techniques which help to increase the realism of the computer generated images of terrain and vegetation. The issues of realism, as related to modelling of natural objects in the landscape, are concerned with the texture of natural terrain and the richness of vegetation growth patterns. The methods for modelling ground surfaces are basically divided into deterministic (i.e. polynomial) and stochastic methods. As opposed to polynomial methods, which provide for modelling of relatively smooth terrain surfaces, the stochastic methods, based on different classes of stochastic primitives, allow for the modelling of irregular (rough) terrain surfaces. This paper is particularly concerned with one such class of modelling primitives based on Iterated Function Systems. The paper also lays out the background theory and rule specification, together with software means, for the modelling of various vegetation growth patterns.