Procedurally generated realistic virtual rural worlds
Manually creating virtual rural worlds is often a difficult and lengthy task for artists, as plant species selection, plant distributions and water networks must be deduced such that they realistically reflect the environment being modelled. As virtual worlds grow in size and complexity, climates vary on the terrain itself and a single ecosystem is no longer sufficient to realistically model all vegetation. Consequentially, the task is only becoming more difficult for these artists. Procedural methods are extensively used in computer graphics to partially or fully automate some tasks and take some of the burden off the user. Input parameters for these procedural algorithms are often unintuitive, however, and their impact on the final results, unclear. This thesis proposes, implements, and evaluates an approach to procedurally generate vegetation and water networks for realistic virtual rural worlds. Rather than placing these to reflect the environment being modelled, the work-flow is mirrored and the user models the environment directly by specifying the resources available. These intuitive input parameters are subsequently used to configure procedural algorithms and determine suitable vegetation, plant distributions and water networks. By design, the placeable plant species are configurable so any type of environment can be modelled at various levels of detail. The system has been tested by creating three ecosystems with little effort on the part of the user.