Boundary representation (B-Rep) generation is a fundamental task in computer-aided design, yet the direct synthesis of high-fidelity and structurally valid B-Reps remains a major challenge. Existing deep generative methods suffer from two forms of brittleness: representation brittleness, caused by padding noise and feature contamination in the latent space, and generation brittleness, stemming from sequential error propagation and a train-inference mismatch due to non-differentiable validity enforcement. We propose HiFi-BRep, a novel framework that addresses these limitations through two synergistic contributions. First, a topology-aware encoder constructs a high-fidelity latent representation by eliminating padding via learnable queries and preventing feature contamination with topology-guided attention. Second, a single-stage decoder jointly predicts geometry and topology in parallel, embedding core manifold constraints as a differentiable learning objective. This design ensures mutual guidance between geometry and topology while avoiding cascaded errors. Extensive experiments show that HiFi-BRep significantly outperforms state-of-the-art methods in both structural validity and geometric fidelity, providing a robust solution for high-quality B-Rep synthesis. Code and models are publicly available at https://github.com/1nnoh/HiFi-BRep.
Figures & tables
Figure 1: HiFi-BRep generates B-Rep models by jointly decoding geometry and topology from a high-fidelity latent space in a single stage, with key validity constraints embedded as differentiable objectives rather than deferred to post-processing. Benefiting from this unified representation, our approach supports both unconditional synthesis and generation conditioned on various inputs, such as class labels, point clouds, or images.
Figure 2: Input B-Rep formulation. We construct unified primitive features by decomposing the shape into geometric and topological components. Specifically, we derive the face feature by summing the embeddings of the bounding box Fp and Bézier control grid Fz . Similarly, the edge feature is formed by summing the embeddings of the bounding box Ep , control points Ez , and explicit endpoints Ev . The global structural connectivity is explicitly encoded via the edge-face incidence matrix.
Figure 3: Overview of HiFi-BRep . (a) Face and edge tokens undergo per-stream self-attention and cross-stream attention masked by edge-face incidence (Topo-Mask), then learnable queries pool them into a fixed-length latent. (b) The decoder first predicts face and edge counts to build hard padding masks, then updates learnable face and edge queries. Stacked DecBiBlocks then decode topology-aware geometry sequences. Geometry heads regress primitive parameters, while a topology solver head predicts adjacency with row-wise softmax and two-peak targets.
Dataset
Method
COV ( ↑ )
MMD-CD ( ↓ )
JSD ( ↓ )
Novel ( ↑ )
Unique ( ↑ )
Compilability (%, ↑ )
Valid (%, ↑ )
DeepCAD
DeepCAD
76.67
1.09
0.77
93.80
89.79
88.46
68.20
BRepGen
47.03
1.51
3.12
99.72
99.18
20.91
20.76
BrepDiff
45.03
1.32
2.39
99.27
99.10
63.69
63.69
DTGBrepGen
73.50
1.06
0.98
99.79
99.33
92.48
43.20
HiFi-BRep (ours)
70.40
1.05
1.72
99.81
99.15
90.38
72.20
ABC
BRepGen
34.73
2.08
5.72
99.68
99.05
20.77
20.19
Table 1: Unconditional generation on DeepCAD and ABC. Best is bold , second-best is underlined . MMD-CD and JSD are ×100 (DeepCAD convention) .
Figure 4: Unconditional generations on DeepCAD. Columns (a–e) show DeepCAD, BRepGen, DTGBrepGen, BrepDiff, and HiFi-BRep , respectively. Red boxes highlight typical artifacts observed in baselines—open shells or missing faces around holes/fillets, non-manifold junctions, and degenerate thin plates/rods—while the rightmost column also marks a rare failure of our method for transparency. Overall, HiFi-BRep yields more coherent solids with consistent edge–face incidence and fewer topology errors, consistent with its higher Validity.
Figure 5: Reconstruction validity by face count on DeepCAD. Validity remains stable across common counts and ≥61.5% in high–face-count bins, evidencing robust generalization from the encoder.
Table 3: Runtime comparison (seconds per shape) on the DeepCAD dataset. Results are averaged over 1000 runs
Figure 6: Typical failure modes. (a) Trimming disagreement / missing patch : face count is correct, but decoded loops fail to form a valid trimmed region, so the kernel drops the face. (b) Junction inconsistency / non-manifold edge : T-junctions or duplicated segments after consolidation break manifold incidence. (c) Degenerate geometry / sliver face : ill-conditioned control points yield near-zero-area or self-intersecting patches.
Appendix figures & tables9 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 7: Retrieval-based novelty check. For each generated shape (left in each triplet), we show its two nearest neighbors from the training set retrieved by Chamfer Distance (CD) and Light Field Distance (LFD). Despite proximity under both metrics, the generated results display distinct geometric features, supporting that HiFi-BRep does not simply memorize training examples.
Figure 8: Class-conditioned generation on Furniture. Colors indicate categories (ten classes). Samples from different categories are interleaved, showing class-specific traits and intra-class diversity.
Figure 9: Point cloud-conditioned generation on CADNet40. Given a 2,048 point input, the model produces solids that match the global shape while allowing small variations.
Figure 10: Partial point cloud-conditioned generation. The model generates plausible completions for missing regions while respecting visible geometry.
Figure 11: Sketch-conditioned generation. Given a sketch, the model generates solids that respect the sketch while maintaining topological consistency.
Figure 12: Single-view conditioned generation. For each input image, we present two generated candidates, at least one of which consistently aligns with the conditioning view while capturing the overall shape with minor variations.
Figure 13: Multi-view conditioned generation. With several views, the outputs align across viewpoints and preserve more fine details.
Figure 14: Unconditional generations on ABC. Columns (a–d) show BRepGen, DTGBrepGen, BrepDiff, and HiFi-BRep respectively. Red boxes highlight typical artifacts observed in baselines, including open shells or missing faces around holes and fillets, non-manifold junctions, and degenerate sliver plates or rods.
max shared
count
0
847
1
142020
2
40735
3
1830
4
509
5
110
Appendix
Table 4: Maximum shared edges per face pair in ABC (counts). Left: before splitting periodic faces/edges. Right: after splitting. The distribution is long tailed, so multiple shared edges are common.
Boundary representation (B-rep) is the standard 3D modeling format in CAD systems, encoding both geometric primitives and topological connectivity. Despite its prevalence, deep generative modeling of valid B-rep structures remains challenging due to the intricate interplay between discrete topology and continuous geometry. In this paper, we propose HiDiGen, a hierarchical generation framework that decouples geometry modeling into two stages, each guided by explicitly modeled topological constraints. Specifically, our approach first establishes face-edge incidence relations to define a coherent topological scaffold, upon which face proxies and initial edge curves are generated. Subsequently, multiple Transformer-based diffusion modules are employed to refine the geometry by generating precise face surfaces and vertex positions, with edge-vertex adjacencies dynamically established and enforced to preserve structural consistency. This progressive geometry hierarchy enables the generation of more novel and diverse shapes, while two-stage topological modeling ensures high validity. Experimental results show that HiDiGen achieves strong performance, generating novel, diverse, and topologically sound CAD models.
Shurui Liu, Weide Chen, Ancong Wu
School of Computer Science and Engineering, Sun Yat-sen University, China · School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen University, China
Generating a boundary representation (B-rep) conditioned on a single image requires faithful reconstruction of geometry, valid topology, and support for complex shapes. We present UniBRep, a geometry-first framework that adapts a pretrained image-to-3D model to generate a feature mesh as a unified intermediate representation. Its surface provides a geometric scaffold, while spatially aligned learned features encode face-separation cues for topology recovery. Dual decoder branches generate the geometry and face-separation features; a geometry- and feature-guided construction pipeline then fits parametric surfaces, recovers boundary curves and connectivity, and assembles an explicit B-rep using a CAD kernel. Recovering topology from mesh regions avoids predefined architectural face-count limits, allowing face count to scale with shape complexity. On the standard DeepCAD benchmark, UniBRep produces valid B-reps for 80.49% of inputs and reduces face Chamfer distance from 0.1096 to 0.0345 relative to CADDreamer. In a matched comparison, UniBRep also outperforms the HoLa public demo across all reported metrics. Further evaluations demonstrate scalability to high-complexity shapes beyond the standard 30-face range, generalization to objects outside the CAD training distribution, and qualitative transfer to real photographs.
Haiyang Ying, Allen Tu, Jiaye Wu +2
University of Maryland, College Park, Maryland, USA · Amazon, Bellevue, Washington, USA
Multi-stage boundary representation (B-Rep) generation leverages intermediate wireframes to synthesize CAD models. However, geometric and topological risks in these wireframes -- such as self-intersections, edge collapses, and disconnected vertices -- can propagate to invalid final B-Reps. Mitigating such failures by retraining large generative models is computationally prohibitive. We propose Wireframe Detection and Repair (WDR), a training-free framework that intervenes at the intermediate wireframe stage to improve downstream B-Rep validity. WDR features a Geometric-Topology Anomaly Detector (GTAD) that combines parallel VLM-based coarse screening with geometric and topological detectors to predict downstream invalidity risk and route generation to dedicated branches. An Energy-Guided Geometric-Topology Repair (EGGTR) module then performs detector-triggered guided regeneration through geometry and topology branches. By scaling test-time computation via Energy-Guided Resampling and training-free guidance for diffusion models, WDR can be integrated into autoregressive and diffusion pipelines without retraining. Extensive experiments demonstrate consistent improvements in kernel-checked validity while largely retaining the measured diversity and distributional quality of synthesized CAD models. The code will be made publicly available upon acceptance.
Jingyu Wu, Youcheng Cai, Tengyu Luo +1
University of Science and Technology of China, Hefei, China