Virtual 3D city models represent spatial and spatio-temporal information about real or imaginary urban spaces. They consist of a multitude of heterogeneous data that include geometry, structure, usage, and domain-specific thematic information. In a sense, virtual 3D city models are perfect use cases for computer graphics research. However, they also exhibit limits of today?s technology.

This talk will illustrate some of these issues. In particular, a schema for decoupling model complexity from rendering complexity is presented that enables visualization of complex models even on lightweight devices such as smart phones and tablets. The talk also sketches fundamental usability issues such as photorealistic versus non-photorealistic rendering, cognitive load of depictions, and questions about their three-dimensionality.