Generation of Textures and Geometric Pseudo-Urban Models With the Aid of IFS
Xavier Marsault · PsycEXTRA Dataset · 2003
Geometric and functional modelling of cities has become a growing field of interest, raised by the development and democratisation of computers being able to support high-demanding graphics in real time. Actually, more and more applications concentrate on creating virtual environments. ARIA has been working for two years, within the DEREVE project (DER, 2000), on pseudo-urban textures and geometric models generation, by means of fractal or parametric methods. This paper explains our attempt to capture inner coherence of urban shapes and morphologies, by fractal analysis of 2D textures (top view + height) of real and synthetic city maps. The basic ideas lean on autosimilarity detection, fractal coding of regions, and processing with Iterated Function Systems (IFS). We introduce a genetic-like approach, allowing interpolation, alteration and fusion of different urban models, and leading to global or local synthesis of new shapes. Finally, a 3D reconstruction tool has been developed for converting textures to volumes in VRML, simplified enough for real time wanderings, and enhanced by some automatically generated garbage dump and decorated elements. Programs and graphic interface are developed with C++ and QT libraries.