Domain specific methods for procedural modeling of historical architecture

This thesis is concerned with the question of how to efficiently model and represent historic buildings in the computer. Since detailed 3D modeling can often require considerable amounts of effort, automation is a natural goal to strive for. This automation can be provided by procedural modeling. Common procedural modeling techniques excel at the generation of a vast amount of simple buildings for entire virtual cities. While simple box-shaped buildings can be easily described, for more complex buildings such as complex historic buildings procedural techniques can be used only sparely. Virtually generated buildings and cities are increasingly demanded by virtual worlds, movies, and video games. Modeling them in detail requires a huge amount of resources and historic buildings are one part that is not well covered today. Historic buildings have different parts that need different modeling techniques. This thesis investigates some of those parts and tries to find new answers on how to model them procedurally. The introduced modeling techniques comprise a technique to model complex roof landscapes of historic cities, a technique to procedurally model the geometry of round building parts, and a technique to capture the forms of ornamental decorations in historic buildings. Further it compares domain specific methods from software design to procedural modeling techniques and draws comparisons there. The basis for all the research forms a common programming language that is designed for procedural modeling.Doctor of Philosoph