Controllable Generation of Large-Scale 3D Urban Layouts with Semantic and Structural Guidance

2025-09-28

AI Summary

Existing urban layout generation methods suffer from either geometric discontinuity (in image-based approaches) or semantic incompleteness (in graph-based approaches), hindering simultaneous scalability and controllability in large-scale 3D urban modeling. To address this, we propose a novel graph-structured representation that unifies geometric continuity with parcel-level semantics: we extend 2D parcel graphs into semantically enriched 3D layout graphs by introducing weighted topological edges and building-height embeddings. Furthermore, we design a semantic-conditioned graph neural network that jointly models parcel topology, geometric constraints, and user-specified semantic labels. Experiments demonstrate that our method generates layouts exhibiting both geometric plausibility and semantic coherence across large-scale urban scenes. It significantly improves modeling scalability and interactive controllability, making it suitable for applications in urban planning, virtual scene synthesis, and game development.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Graph-based Machine Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

Urban modeling is essential for city planning, scene synthesis, and gaming. Existing image-based methods generate diverse layouts but often lack geometric continuity and scalability, while graph-based methods capture structural relations yet overlook parcel semantics. We present a controllable framework for large-scale 3D vector urban layout generation, conditioned on both geometry and semantics. By fusing geometric and semantic attributes, introducing edge weights, and embedding building height in the graph, our method extends 2D layouts to realistic 3D structures. It also enables users to directly control the output by modifying semantic attributes. Experiments show that it produces valid, large-scale urban models, offering an effective tool for data-driven planning and design.

Problem

• Generating geometrically continuous large-scale 3D urban layouts
• Integrating semantic and structural guidance for urban modeling
• Enabling user control over semantic attributes in layout generation

Innovation

• Fuses geometric and semantic attributes for urban layouts
• Introduces edge weights and building height in graph
• Enables direct user control via semantic attribute modification