SYNTHESIS OF NON-REGULAR ARCHITECTURAL FORMS

Abstract

This thesis presents an interactive system for synthesizing non-regular architectural forms that features a growth-based approach for rapidly constructing architectural models in the large scale, while complementing it with intelligent form completion for rapid design synthesis on the small scale. Our growth-based strategy for large scale synthesis allows the incorporation of different behavioral growth routines to generate building units meeting the requirements of view and shape variation. In maximizing views, we exploit the capability of the GPU to achieve good speedup of the computation. For designing in the smaller scale, our system supports fast generation of design plans, rapid form completion and synthesis. Besides optimizing these requirements, our approach exhibits emerging behavior with outputs having complex but useful inter-lockings. These emergent features suggest clever utilization of space as, for example, suspended open spaces, mid-air gardens and ventilation corridors that are of good architectural value. On the whole, our work advances interactive generative design with objective measures that are hard, if not impossible, to achieve manually. It contributes towards rapid prototyping and in taking design of forms beyond monotonous cookie-cutter architecture.