🤖 AI Summary
This work proposes a novel paradigm for boundary representation (B-Rep) generation based on composable $k$-cell particles to address the challenges of weak context awareness and poor error recovery arising from the tight coupling between geometry and topology and the complex hierarchical structure in conventional approaches. By leveraging shared latent representations of boundaries, the method achieves geometric coupling while decoupling the traditional hierarchy, enabling unified joint generation of vertices, edges, and faces and eliminating cascading dependencies. Integrated with a multimodal flow-matching framework and explicit local particle representations, the approach supports both unconditional generation and conditional 3D reconstruction from single-view images or point clouds. The resulting CAD models demonstrate superior fidelity, validity, editability, and performance in downstream tasks such as local repair compared to existing methods.
📝 Abstract
Boundary Representation (B-Rep) is the widely adopted standard in Computer-Aided Design (CAD) and manufacturing. However, generative modeling of B-Reps remains a formidable challenge due to their inherent heterogeneity as geometric cell complexes, which entangles topology with geometry across cells of varying orders (i.e., $k$-cells such as vertices, edges, faces). Previous methods typically rely on cascaded sequences to handle this hierarchy, which fails to fully exploit the geometric relationships between cells, such as adjacency and sharing, limiting context awareness and error recovery. To fill this gap, we introduce a novel paradigm that reformulates B-Reps into sets of compositional $k$-cell particles. Our approach encodes each topological entity as a composition of particles, where adjacent cells share identical latents at their interfaces, thereby promoting geometric coupling along shared boundaries. By decoupling the rigid hierarchy, our representation unifies vertices, edges, and faces, enabling the joint generation of topology and geometry with global context awareness. We synthesize these particle sets using a multi-modal flow matching framework to handle unconditional generation as well as precise conditional tasks, such as 3D reconstruction from single-view or point cloud. Furthermore, the explicit and localized nature of our representation naturally extends to downstream tasks like local in-painting and enables the direct synthesis of non-manifold structures (e.g., wireframes). Extensive experiments demonstrate that our method produces high-fidelity CAD models with superior validity and editability compared to state-of-the-art methods.