Inverse Procedural Modeling of Facade Layouts

Participants : Weiming Dong, Bin Wang, Dong-Ming Yan, Hua-Liang Xie, Jean-Claude Paul.

We want to address the following open research problem: How can we generate a deterministic shape grammar that explains a given facade layout? An approximate dynamic programming framework will tackle this problem. The proposed solution contributes to the compression of urban models, architectural analysis, and the generation of shape grammars for large-scale urban modeling. As a major contribution of this work we want to formulate the inverse procedural modeling problem for facade layouts as a smallest grammar problem. We also want to propose an automatic algorithm to derive a shape grammar for a given facade layout. In this work, we will assume segmented and labeled facade layouts as input and do not derive the shape grammars directly from photographs. The joint optimization of segmentation and grammar extraction remains an aspirational goal for this work.

Architecture Design

Participants : Jean-Claude Paul, Bin Wang, Weiming Dong, Lin Li, Yan Kong, Yong Zhang, Fan Tang, Fuzhang Wu, Cui-Gong Wang.

In cooperation with UC Berkeley - Department of Architecture

We want to propose a method for automated generation of architectural models for computer graphics applications. Our focus is not only on the building layout: the internal organization of spaces within the building, but also the Architectural composition of volumes, roofs and facades. We focus on the generation of various types of buildings: residences, schools, museums, hospitals, civic enters, office buildings. Our work builds on grammar-based procedural modeling, inverse procedural modeling and composition rules, especially symmetry and scaling, and interactivity. Moreover, we consider the architecture design process as an iterative trial-and-error process that requires significant expertise and learning by doing.