Interactive Design of Urban Spaces using Geometrical and Behavioral Modeling Carlos A. Vanegas 1 Daniel G. Aliaga 1 Bedřich Beneš 1 Paul A. Waddell 2 1 Purdue University 2 University of California, Berkeley Abstract The main contribution of our work is in closing the loop between behavioral and geometrical modeling of cities. Editing of urban design variables is performed intuitively and visually using a graphical user interface. Any design variable can be constrained or changed. The design process uses an iterative dynamical system for reaching equilibrium: a state where the demands of behavioral modeling match those of geometrical modeling. 3D models are generated in a few seconds and conform to plausible urban behavior and urban geometry. Our framework includes an interactive agent-based behavioral modeling system as well as adaptive geometry generation algorithms. We demonstrate interactive and incremental design and editing for synthetic urban spaces spanning over 200 square kilometers. Keywords: interactive, editing, 3D models, urban spaces. CR Categories: I.3 [Computer Graphics], I.3.3 [Picture/Image Generation], I.3.5 [Computational Geometry and Object Modeling], I.3.6 [Methodology and Techniques]. 1. INTRODUCTION We present a framework for intuitive and interactive design of 3D geometric models of large, complex, and realistic urban spaces. An urban space is a collection of architectural structures arranged into buildings, parcels, blocks, and neighborhoods interconnected by roads. The key notion behind our approach is to close the loop between behavioral modeling and geometrical modeling of urban spaces. We model the design and editing process as a dynamical system using a set of functions that describe the change the variable values. Our system produces models resembling existing cities, and is useful for a variety of applications ranging from games and movies to urban planning and emergency management. Previous research in urban modeling can be divided into the areas of geometrical modeling and behavioral modeling: the first is purely computer graphics oriented (e.g., [Parish and Muller 2001, Wonka et al. 2003, Mueller et al. 2006, Aliaga et al. 2008, Chen et al. 2008]), and the second lies outside this research domain (e.g., [Alkheder et al. 2008, Waddell 2002]). The results of urban behavioral modeling are intended for decision-making regarding urban policies in current and future urban areas. In general, however, behavioral simulation models use limited and fixed 2D geometric features (e.g., grid cells or parcels) and are computationally too expensive to run at interactive rates. Some research has been performed in feeding the output of a behavioral modeling system into a geometrical modeling system producing 2D layouts or 3D models that change over time (e.g., [Honda et al. 2004, Vanegas et al. 2009, Weber et al. 2009]). However, the focus is not on designing and editing a new urban model, but rather on computing changes (e.g., growth) over time to a provided model. Figure 1. Urban Model Design. This example city is incrementally generated in a two-step process. First, based on designer input, the system creates a low-density town in a valley by the coast (a). Then, the designer replaces the office buildings (b) with high-rises (c), and constrains the downtown of the existing city and the forest area around it (d). The system increases the population as a result of the larger number of jobs and locates the population in accessible land outside the valley, creating new roads, parcels, and buildings, while leaving the original downtown unchanged (e, f). f terrain jobs a b c d e