FurE Delivers Fast, Instance-Specific 3D Animal Fur from Multiview Images
A new method reconstructs editable animal fur from calibrated views without animal-fur data, using a compact code and local thickness cues.
- Publication
- arXiv
- Stage
- Preprint
- What we read
- Summary of the paper
- Authors
- Srinjay Sarkar, Prakhar Kaushik, Soumava Paul, Alan Yuille
- Universities and research institutions
- Not yet supplied in verified metadata; the Brief does not guess.
What they did and found
FurE reconstructs per-strand fur for animals from calibrated multi-view images. It uses a defurred body surface with a compact latent code decoded into strand geometry, guided by Gaussian rendering and part priors. The method achieves about a 10x speedup over prior dense strand optimization and works on real-world data.
Why it matters
This approach makes realistic, editable fur assets more accessible without large animal-fur datasets. It helps teams balance faster generation with strand-level detail in practice.
FurE assigns strand roots on a defurred body and uses a compact, learned code to shape each strand. The system decodes these codes into strands and renders them with cylindrical Gaussians, trained against multi-view photos without animal-specific fur data.
The method emphasizes speed and realism but notes limits in rapid motion, heavy occlusion, and multilayer coats, with results tied to specific animals rather than broad generalization.
Original sources · 1
- FurE: Efficient Instance-Specific 3D Fur Reconstruction without Animal-Fur Datasets ↗arXiv · 2026-09-28
Check the original paper for its authors, methods, version and access terms.
What remains uncertain
Limited to calibrated multi-view images of relatively static animals; not yet shown for dynamic motion or multilayer coats.
What is your take?
Ask a question, add useful context or share a different perspective. Keep the conversation respectful and grounded.