HARDNESS DATA SYNTHESIS FOR HEIGHT-FIELD BASED LANDSCAPE MODELS
Korneliusz K. Warszawski, Sławomir S. Nikiel, Tomasz Zawadzki · 2012
Landscape modelling is a key element of synthetic environments in virtual reality systems such as: military training systems, virtual reconstruction of cultural heritage, digital entertainment and game development. Most popular and simplest to implement data structures for terrain models are based on the height-field representation. Unfortunately, such models do not contain information about hardness of terrain or it susceptibility for erosion forces in different landscape areas. It is possible to derive hardness data from actual geological records, but for virtual environments we propose the fully automated synthesis of hardness information for terrain models. Our method stems from classical Poisson Faulting algorithm, that was originally used by Mandelbrot and Voss to model ragged landscapes and whole planets. The proposed technique can help to generate hardness data under the height-field base terrain, with regard to different geological materials. The algorithm ensures that those materials are not randomly displaced but form clusters throughout the entire virtual terrain. In addition, the technique can be simply parallelized and implemented in CUDA programming environment. Obtained data structure can be used for further synthesis of eroded landscapes or can work as a test bed for comparison of different geological erosion simulating some models.