Procedural generation of landscapes

Abstract

[en] The goal of this work is to study, develop, analyse and compare different techniques and methodologies to generate and visualize realistic landscapes in real time. These techniques will mainly follow the principles of procedural content generation, meaning the landscape will be generated pseudo-randomly and during run time. We need to emphasize the term pseudo-randomly since, as we will see in the work, procedural-generated content ultimately follows patterns, and we will try to use this pseudo-randomness in order to customize the landscapes. Landscapes will have terrain, plants, water and clouds. Terrain and plants will be generated using fractal techniques, and water and clouds will use different types of noise. The work includes mathematical foundations from the fields of Algebra and Geometry to formalize the processes of modelling, rendering and visualising scenes in Computer Graphics and from probability to understand how computers create the randomness used in procedural generation. This will enable us to formalize various techniques and methodologies, which will allow us to have a better understanding of how and why the techniques work.