Reproduce “Autoregressive B-Rep Shape Generation with Parametric Surfaces” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dafei Qin, Rui Xu, Zeyu Shen, Kaichun Qiao, Hongyang Lin, Qixuan Zhang, Huaijin Pi, Lan Xu and 3 more
ParaCAD, an autoregressive framework developed by researchers including those from The University of Hong Kong and ShanghaiTech University, generates Boundary Representation (B-Rep) CAD models directly from 3D point clouds by synthesizing native parametric surfaces. It achieves superior geometric accuracy and semantic fidelity compared to existing methods, enabling downstream CAD operations and manufacturing.
Key insights
- Directly operating on native parametric surfaces and their explicit definitions in a tokenized format preserves geometric fidelity and semantic information better than approximating surfaces.
- Decoupling the prediction of geometric surface parameters from the complex task of explicit topology construction simplifies the generative learning problem.
- A robust, deterministic geometric post-processing step can effectively assemble topologically valid B-Rep models from individually generated surface patches, guided by point cloud coverage and topological constraints.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.