How to realise models suitable for efficient visualisation within real time rendering applications is the aim of this text. In particular, it deals with the problem of segmenting laser scanning data and how these are then subjected to a twofold simplification process: one for areas characterised by high detail and one for areas with low or no curvature. Another topic that will be addressed is that of defining the topological rules of low-detail models so that they are suitable for interactive visualisation without any gaps or shading errors. Finally, the theme of texturing for the perceptive improvement of models through “irregular” textures, i.e. suitable for one and only one model, or “regular”, i.e. subject to repetition, will be considered. Here too, the segmentation and hierarchization of the urban layout and the constituent elements of the individual buildings has a priority role as it links architectural reading to computer reading with the aim of the perceptive and visual effectiveness of the digital model.