Procedural Generation of Urban Environments Using GIS Data

Sammanfattning

Realistic and scalable urban environments are increasingly important in simulations used for planning, analysis, and training in various contexts, such as urban development, defense simulations, and emergency response. This thesis addresses that need by developing a workflow that procedurally generates Swedish urban areas using GIS data and architectural rules. The work is motivated by the Swedish Defence Research Agency (FOI), which requires detailed and adaptable digital environments for simulation purposes, including future applications in electronic warfare and signal propagation. The proposed workflow begins with building footprint data in GeoPackage format, which is processed in QGIS and enriched using custom Python scripts that add attributes such as roof color and custom identifiers. The data is then imported into CityEngine, where CGA rule files generate 3D buildings based on Swedish architectural styles. The result is a flexible system that supports both high fidelity and performance optimized models. Models are exported in the glTF format for lightweight use or as UDATASMITH scenes for direct visualization in Unreal Engine. The final environment is showcased in a demo simulation, demonstrating the system’s ability to produce visually consistent, scalable, and geospatially accurate urban environments. The results show that procedural generation, informed by regional architectural patterns and GIS data, is a viable method for building realistic simulation environments for national scale applications.