Using Stochastic Modeling for Texture Generation
Shinichiro Haruyama, Brian A. Barsky · IEEE Computer Graphics and Applications · 1984
A new computer graphics method that uses stochastic modeling to generate highly realistic random textures has been developed. The stochastic function is applied to the normal vectors instead of the surface position. The fractional Brownian motion (fBm) form of stochastic modeling is extended by allowing different values of the self-similarity parameterhon different recursion levels. The self-similarity parameterhdetermines how fast a stochastic factor decreases as the recursion level becomes deeper. A largehcreates a very smooth texture and a smallhmakes a very rough texture; this means thathcan control the spatial frequency distribution. Using different self-similarity parameters for different recursion levels, the roughness can be adjusted in more detail, and even the size of the wrinkles can be easily controls. Furthermore, the height of the bumps in the texture can be controlled by the standard deviation σ.