BrepForge: Factorized B-rep Synthesis via Wireframe Composition and Boundary-Conditioned Surface Instantiation

2026-05-19

AI Summary

This work addresses the long-standing challenge in learning-driven boundary representation (B-rep) modeling, where the strong coupling between topology and geometry hinders simultaneous achievement of topological completeness and geometric fidelity. To overcome this, the authors propose a factorized disentanglement strategy that decomposes B-rep generation into two stages: first, a face-aware autoregressive model constructs a topological wireframe with complete vertex-edge-face (V-E-F) connectivity; second, surface patches are precisely instantiated under strict boundary geometric constraints. By exploiting the asymmetry between the high-entropy decisions in wireframe generation and the strong geometric constraints on surface boundaries, this approach achieves, for the first time, accurate co-modeling of topology and geometry. Experiments demonstrate that the method significantly outperforms existing techniques in both geometric complexity and topological validity, reliably producing high-quality B-rep models.

Abstract

Boundary representation (B-rep) is the de facto standard for modern CAD, yet learning-based B-rep synthesis remains challenging due to the tight coupling between discrete topology and continuous geometry. We observe a fundamental asymmetry in B-reps: while wireframe composition involves high-entropy structural decisions, the interior surface geometry is largely constrained by its boundary loops. Motivated by this observation, we propose BrepForge, a generative framework that factorizes B-rep synthesis into two stages: wireframe composition and boundary-conditioned surface instantiation. In the first stage, a face-aware autoregressive model serializes the wireframe into structured sequences that explicitly encode hierarchical Vertex-Edge-Face (V-E-F) connectivity, yielding a topologically complete scaffold. In the second stage, precise surface geometries are instantiated by incorporating learning-free geometric priors derived from boundaries, transforming the complex synthesis task into a structured refinement process. This factorized approach ensures both topological integrity and geometric precision, effectively addressing the inherent complexities of B-rep modeling. Extensive experiments demonstrate that BrepForge outperforms existing baselines with superior geometric complexity and topological validity.

Problem

• B-rep synthesis
• topology-geometry coupling
• CAD modeling
• generative design

Innovation

• B-rep synthesis
• wireframe composition
• boundary-conditioned surface instantiation
• topological integrity
• geometric priors