Mapping textures on 3d terrains: a hybrid cellular automata approach
MetadataShow full item record
It is a time consuming task to generate textures for large 3D terrain surfaces in computer games, flight simulations and computer animations. This work explores the use of cellular automata in the automatic generation of textures for large surfaces. I propose a method for generating textures for 3D terrains using various approaches - in particular, a hybrid approach that integrates the concepts of cellular automata, probabilistic distribution according to height and Wang tiles. I also look at other hybrid combinations using cellular automata to generate textures for 3D terrains. Work for this thesis includes development of a tool called "Texullar" that allows users to generate textures for 3D terrain surfaces by configuring various input parameters and choosing cellular automata rules. I evaluate the effectiveness of the approach by conducting a user survey to compare the results obtained by using different inputs and analyzing the results. The findings show that incorporating concepts of cellular automata in texture generation for terrains can lead to better results than random generation of textures. The analysis also reveals that incorporating height information along with cellular automata yields better results than using cellular automata alone. Results from the user survey indicate that a hybrid approach incorporating height information along with cellular automata and Wang tiles is better than incorporating height information along with cellular automata in the context of texture generation for 3D meshes. The survey did not yield enough evidence to suggest whether the use of Wang tiles in combination with cellular automata and probabilistic distribution according to height results in a higher mean score than the use of only cellular automata and probabilistic distribution. However, this outcome could have been influenced by the fact that the survey respondents did not have information about the parameters used to generate the final image - such as probabilistic distributions, the population configurations and rules of the cellular automata.