FurE: 10x Faster 3D Animal Fur Reconstruction Without Animal Datasets
FurE：免用動物毛髮資料集，實現 10 倍加速的 3D 動物毛髮重建技術
Reconstructing realistic and editable animal fur from multi-view images is difficult due to self-occlusion and a lack of dedicated animal fur datasets. FurE overcomes this by optimizing a root-conditioned latent field decoded via a PCA decoder learned from human hair data. It reconstructs the underlying defurred animal body using a surface-constrained Gaussian Frosting representation and part-based priors. This approach generalizes across synthetic and real-world sequences, achieving a 10x training speedup over state-of-the-art dense per-strand optimization methods.
Key points
Zero Animal Datasets
Innovatively utilizes a PCA decoder learned from human-hair strand data, bypassing the extreme scarcity of dedicated animal fur datasets.
10x Training Speedup
Achieves a 10x speedup in strand training compared to current SOTA dense per-strand optimization while preserving fine details.
Gaussian Frosting Defurring
Reconstructs the hidden, defurred animal body by leveraging local fur-thickness cues from a surface-constrained Gaussian Frosting representation.
Highly Editable Groom
Recovers a per-strand, highly editable groom, enabling fine-grained adjustments for artists and animators.
How it works
Why it matters
Traditional 3D animal reconstruction often bypasses editable fur due to extreme data scarcity. FurE demonstrates that human-hair models can successfully transfer to animal fur, eliminating the need for expensive animal datasets. This is highly impactful for VFX, gaming, and VR, allowing artists to generate physics-ready, animatable grooms from simple multi-view images with a 10x speedup.
Who it affects
- AI Developer
- AI Researcher
- Content Creator
How to use it
- 1Realistic animal 3D modeling and groom rigging for VFX and gaming.
- 2Rapid creation of interactive digital pets from real-world photos for VR and metaverse applications.
Limitations & caveats
- Relies on a PCA decoder trained on human hair, which may struggle to reconstruct extremely exotic or non-human-like animal fur structures.
- Requires multi-view image inputs, which remains challenging for single-image reconstruction or highly dynamic sequences with motion blur.
Related
Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiency
自我監督信心訓練：免於刻意「學會停止」即可提升 LLM 推理效率
Researchers found that training reasoning models to predict their own confidence at intermediate steps naturally reduces generated tokens by up to 25% at matched accuracy, without explicitly optimizing for length or stopping.
First-Order Stationarity of Reverse Diffusions: Bridging Optimization and Sampling
逆向擴散的一階駐點性：連結優化與生成採樣的數學機制
This research establishes a first-order optimization theory for diffusion models, proving that SDE-based reverse Langevin diffusions contract Fisher divergences exponentially under strongly convex noising.
Statistical Attribute Alignment for Black-Box Generative AI via Output Post-Processing
黑盒生成式 AI 的統計屬性對齊：透過輸出後處理實現公平與多樣性
This paper introduces post-processing algorithms to align the attribute distribution of black-box generative AI outputs with user-specified targets using a mathematically minimized number of queries.