Inverse Procedural Modeling for 3D Urban Models
Architectural models and urban spaces are important in computer graphics, computer vision, and virtual environments; but creating realistic buildings needs time and domain expertise. We developed a generalized proceduralization framework to convert existing 3D models into procedural representations. We first conduct shape processing algorithms to exploit similarities, then use grammar discovery approaches to extract grammars. We also proposed new applications as completion for reconstruction, procedural generation, and interactive synthesis tools.