AutoBrep: Autoregressive B-Rep Generation with Unified Topology and Geometry

2025/12/02 by Xiang Xu, Pradeep Kumar Jayaraman, Xu, Xiang +9 · 4 citations
Engineering · #3D Shape Modeling and Analysis #Adjacency list #Boundary (topology) #Boundary representation #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Inference #Manufacturing Process and Optimization #Representation (politics) #Scalability #Solid modeling #Topology (electrical circuits) #Topology Optimization in Engineering #Tree traversal

paper · pdf · doi:10.48550/arxiv.2512.03018

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2025/12/02 · openalex created_date 2025/12/04 · openalex updated_date 2026/08/05

Abstract

The boundary representation (B-Rep) is the standard data structure used in Computer-Aided Design (CAD) for defining solid models. Despite recent progress, directly generating B-Reps end-to-end with precise geometry and watertight topology remains a challenge. This paper presents AutoBrep, a novel Transformer model that autoregressively generates B-Reps with high quality and validity. AutoBrep employs a unified tokenization scheme that encodes both geometric and topological characteristics of a B-Rep model as a sequence of discrete tokens. Geometric primitives (i.e., surfaces and curves) are encoded as latent geometry tokens, and their structural relationships are defined as special topological reference tokens. Sequence order in AutoBrep naturally follows a breadth first traversal of the B-Rep face adjacency graph. At inference time, neighboring faces and edges along with their topological structure are progressively generated. Extensive experiments demonstrate the advantages of our unified representation when coupled with next-token prediction for B-Rep generation. AutoBrep outperforms baselines with better quality and watertightness. It is also highly scalable to complex solids with good fidelity and inference speed. We further show that autocompleting B-Reps is natively supported through our unified tokenization, enabling user-controllable CAD generation with minimal changes. Code is available at https://github.com/AutodeskAILab/AutoBrep.

Citations

• CAD-Llama: Leveraging Large Language Models for Computer-Aided Design Parametric 3D Model Generation
• CMT: A Cascade MAR with Topology Predictor for Multimodal Conditional CAD Generation
• HoLa: B-Rep Generation using a Holistic Latent Representation
• DTGBrepGen: A Novel B-rep Generative Model through Decoupling Topology and Geometry
• CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD Editing
• NeuroNURBS: Learning Efficient Surface Representations for 3D Solids
• CAD-MLLM: Unifying Multimodality-Conditioned CAD Generation With MLLM
• Deep Compression Autoencoder for Efficient High-Resolution Diffusion Models
• Img2CAD: Conditioned 3D CAD Model Generation from Single Image with Structured Visual Geometry
• GenCAD: Image-Conditioned Computer-Aided Design Generation with Transformer-Based Contrastive Representation and Diffusion Priors
• Img2CAD: Reverse Engineering 3D CAD Models from Images through VLM-Assisted Conditional Factorization
• Split-and-Fit: Learning B-Reps via Structure-Aware Voronoi Partitioning
• BrepGen: A B-rep Generative Diffusion Model with Structured Latent Geometry
• Finite Scalar Quantization: VQ-VAE Made Simple
• Hierarchical Neural Coding for Controllable CAD Model Generation
• GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints
• D²CSG: Unsupervised Learning of Compact CSG Trees with Dual Complements and Dropouts
• SkexGen: Autoregressive Generation of CAD Construction Sequences with Disentangled Codebooks
• FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
• SolidGen: An Autoregressive Model for Direct B-rep Synthesis
• DeepCAD: A Deep Generative Network for Computer-Aided Design Models
• RoFormer: Enhanced Transformer with Rotary Position Embedding
• CAPRI-Net: Learning Compact CAD Shapes with Adaptive Primitive Assembly
• Fusion 360 Gallery: A Dataset and Environment for Programmatic CAD Construction from Human Design Sequences
• Sketch2CAD: Sequential CAD Modeling by Sketching in Context
• UCSG-Net -- Unsupervised Discovering of Constructive Solid Geometry Tree
• Jukebox: A Generative Model for Music
• PolyGen: An Autoregressive Generative Model of 3D Meshes
• GLU Variants Improve Transformer
• BSP-Net: Generating Compact Meshes via Binary Space Partitioning
• Root Mean Square Layer Normalization
• Write, Execute, Assess: Program Synthesis with a REPL
• The Curious Case of Neural Text Degeneration
• ABC: A Big CAD Model Dataset For Geometric Deep Learning
• CSGNet: Neural Shape Parser for Constructive Solid Geometry
• Neural Discrete Representation Learning
• Attention Is All You Need
• A Learning Algorithm for Continually Running Fully Recurrent Neural Networks

Cited by

• BrepLLM: Enabling Large Language Models to Understand Boundary Representations
• CADENA: Stepwise CAD Reverse Engineering
• AIMold: An Autonomous AI-based Pipeline for Complex Mold Design
• Towards Valid B-Rep Generation: Training-Free Wireframe Anomaly Detection and Repair

Related

• BrepGPT: Autoregressive B-rep Generation with Voronoi Half-Patch
• SolidGen: An Autoregressive Model for Direct B-rep Synthesis
• DTGBrepGen: A Novel B-rep Generative Model through Decoupling Topology and Geometry
• Bringing Attention to CAD: Boundary Representation Learning via Transformer
• GraphBrep: Learning B-Rep in Graph Structure for Efficient CAD Generation