An integrative workflow for 3D city model versioning

Abstract

3D city models are continuously becoming more popular among practitioners due to the volume and versatility of information they contain, which makes them suitable to be used in various applications. However, there is no mechanism to allow maintaining them updated at the same pace that cities evolve, or when error correction is necessary, eventually diminishing their value. Many cities around the world already possess such models which are mostly used for experimentation and research purposes. Such an example is also the city of Rotterdam, whose 3D city model is not regularly updated and has to be outsourced for that purpose. This thesis investigates into addressing this issue by proposing and implementing an integrative maintenance workflow. The workflow is designed to fulfill what the maintenance needs of a typical municipality are expected to be. Those needs were identified after conducting an analysis of the current situation and collecting information from practitioners within the municipality through interviews. The workflow is a combination of 3D city model versioning and visual editing capabilities with the aim to effectively maintain CityJSON encoded models in an intuitive way. Its implementation includes two prototype software implementations: a versioning component, which is utilized to create a workflow inspired by git flow and allows concurrent maintenance and alternative scenario testing in a non-linear and distributed way, and a visual editing component capable of editing CityJSON encoded 3D city models by extending Blender’s functionality. Following the implementation, the workflow was tested by simulating real world maintenance scenarios. The tests demonstrate the feasibility of maintaining 3D city models with such a workflow and more specifically the suitability of git based workflows. At the same time some key parameters of the versioning mechanism are identified which if tuned properly they can optimize the performance, behavior and robustness of 3D city model versioning. With both components being prototype solutions the workflow is far from operational and there is certainly a lot of space for improvement regarding both components. Utilizing the workflow in practice would be the ideal way for collecting useful feedback. Besides that, there are already extensions of Blender that combined with the visual updating component of the workflow can offer advanced integration of editing and analysis capabilities.