Realistic visualizations of 3D city models require the availability of detailed geometric descriptions of the visible buildings. Especially, if virtual viewpoints from a pedestrian perspective have to be generated, texture images have to be additionally available. For this purpose geo-referenced terrestrial images have to be captured and mapped against the corresponding elements of the building model. By these means the large scale structure and the material of the building faces can be visually represented. In our approach this texture mapping is realized using a programmable graphics processing unit. In contrast to the application of standard viewers, this technique allows to model even complex geometric effects like self-occlusion or lens distortion. This allows for a very fast and flexible on-the-fly generation of façade texture using real world imagery. Our approach was implemented within a project aiming on the real-time visualization of urban landscapes, which will be discussed in the final part of the paper.