UrbanMod: Text-to-3D Modeling for Accelerated City Architecture Planning

Abstract Urban planning involves intricate processes traditionally reliant on visual and spatial representation. With advancements in technology, bridging text and 3D modeling can enhance these processes. We present UrbanMod, a pioneering framework designed for converting textual descriptions of urban settings into intricate 3D models. This approach utilizes cutting-edge natural language processing to accurately capture architectural features and spatial configurations from written input. By facilitating swift prototyping, UrbanMod enhances visualization capabilities for city layouts. The framework employs generative modeling techniques to deliver detailed and flexible 3D representations. Validation through various case studies underscores UrbanMod's capacity to shorten planning durations and foster better engagement among stakeholders in urban design.