RoboCloth: 500 Fabrics, 80B Observations Break 2003 MERL BRDF Data Reliance
taiyasaki · x · 2026-10-20
Zhen (Colin) Li, an MSc student at SFU's GrUVi Lab under Andrea Tagliasacchi, released RoboCloth, a neural materials dataset and model for cloth.
- Neural materials research has long relied on just 100 BRDFs measured at MERL in 2003; RoboCloth provides 500 cloths × 580 views — roughly 80 billion observations — specifically for cloth materials.
- Project homepage and arXiv paper are public, covering a full material capture pipeline with novel BRDF modeling.
- Lead-author-adjacent researcher taiyasaki amplified the release, noting the work was "much more challenging than it looks."
Related event: RoboCloth: Massive Real-Cloth Dataset With 80 Billion Observations Released(4 posts)→
More from Research
- Interpretability researchers debate whether early monosemantic features persist through RLHF — davidmanheim · 2026-10-20
- MemTensor releases MemT-1.0, a memory-native architecture embedding continuous memory into model computation — MemTensor · 2026-10-20
- Tsinghua and AgiBot release RoboValue, a sim-and-real benchmark for robotic value models — Shengbang Liu · 2026-10-20
- Paper proves transformers forget prompts via power-law decay, and trained models forget faster — Shuyang Xiang · 2026-10-20
- MARS steers multimodal safety using only text-derived refusal directions, no multimodal safety data needed — MHUGLab · 2026-10-20
- Lenovo paper: safety benchmark scores mostly measure labeling policy, not detection capability — Vahid Jebraeeli · 2026-10-20