RGBench/garment_assets

RGBench Garment Assets — Research Subset

A research subset of the garment asset library from Real Garment Benchmark (RGBench).

• 📄 Paper: arXiv:2511.06434
• 🌐 Project page: rgbench.github.io
• 📊 Companion dataset: RGBench-Cloth-Sim2Real-v1 — real-world capture data with ground-truth point clouds (CC-BY 4.0)

Contents

This repository holds a research subset of the RGBench industrial garment asset library.

Format	Contents
OBJ	Garment geometry, usable in any 3D toolchain
Style3D	Native garment format carrying garment construction data and measured physical parameters

These are source assets, released as authored and without further processing.

Physical parameters

Measured physical parameters — stretch stiffness, bending stiffness, area density, and thickness — are carried in the Style3D-format files and can be read with Style3D Studio. Reference values per fabric category are also tabulated in the paper.

Scripts for mesh and physical-parameter parsing, and for post-processing these assets for use in physics simulators, will be released on GitHub.

Full asset library

The models here are a subset of RGBench's complete industrial garment asset library (4,000+ models). For access, or to discuss research collaboration and data licensing, contact wanghmin@gmail.com

Additional garment assets

Other garment datasets that work well alongside RGBench. These are not redistributed here — please obtain them from their original sources and follow their respective terms.

• ClothesNet (Zhou et al., 2023) — https://sites.google.com/view/clothesnet The garment collection described in our paper also draws on ClothesNet meshes, which are available directly from the official site.
• DexGarmentLab — https://huggingface.co/datasets/wayrise/DexGarmentLab (Apache-2.0) USD-format garment assets built on ClothesNet models — a convenient route if you need simulator-ready garments for Isaac Sim.
• GarmageSet (Style3D) — https://huggingface.co/datasets/Style3D/GarmageSet Please cite Li et al., GarmageNet, ACM TOG 2025.

License

The assets in this repository are provided under the custom RGBench Project-Specific Non-Commercial Research License. This is not an OSI-approved open-source license — use, redistribution, and commercial applications are restricted. Please review the full license before downloading or using the assets.

Other RGBench components are released under different, more permissive licenses: the Sim2Real capture dataset is CC-BY 4.0, and the RGBench code is MIT.

Citation

@misc{hu2025realgarmentbenchmarkrgbench,
  title         = {Real Garment Benchmark (RGBench): A Comprehensive Benchmark for Robotic Garment Manipulation featuring a High-Fidelity Scalable Simulator},
  author        = {Wenkang Hu and Xincheng Tang and Yanzhi E and Yitong Li and Zhengjie Shu and Wei Li and Huamin Wang and Ruigang Yang},
  year          = {2025},
  eprint        = {2511.06434},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  url           = {https://arxiv.org/abs/2511.06434}
}