Dynamic Terrain Generation Based on Multifractal Techniques

The work, described in this paper, covers three topics: real-world terrain analysis, synthetic terrain generation based upon the results of the analysis stage, and dynamic terrain generation. Each topic involves the theory of multifractals. First, a summary of four known fractal-based techniques for terrain generation is presented including fractional Brownian motion, Mid-point displacement, Iterated Function Systems and the multifractal formalism. The multifractal formalism was chosen for both the analysis of real-world terrain data, as well as the generation of synthetic terrain surfaces. The authors implemented a multifractal terrain analysis algorithm that captures terrain characteristics of real-world data, into five parameters. These parameters were put into a multifractal terrain generation algorithm that produced synthetic terrain with features similar to those in the terrain that was analysed. Also, an algorithm for zooming in on synthetic terrain was developed. Finally, an application was developed that generates terrain dynamically.