Automatic Generation of Semantic 3D City Models from Conceptual Massing Models

We present a workflow to automatically generate semantic 3D city models from conceptual massing models. In the workflow, the massing design is exported as a Collada file. The auto-conversion method, implemented as a Python library, identifies city objects by analysing the relationships between the geometries in the Collada file. For example, if the analysis shows that a closed poly surface satisfies certain geometrical relationships, it is automatically converted to a building. The advantage of this workflow is that no extra modelling effort is required, provided the designers are consistent in the geometrical relationships while modelling their massing design. We will demonstrate the feasibility of the workflow using three examples of increasing complexity. With the success of the demonstrations, we envision the utoconversion of massing models into semantic models will facilitate the sharing of city models between domain-specific experts and enhance communications in the urban design process.