Research map / 3D vision / Face and shape
SNUG: Self-Supervised Neural Dynamic Garments
Igor Santesteban, Miguel A. Otaduy, Dan Casas
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use SNUG's physics-based self-supervised training instead of PBNS | Garment deformation model · AMASS test set | strain loss: 0.111 -> 0.064 | Improved |
| Use SNUG's physics-based self-supervised training instead of PBNS | Garment deformation model · AMASS test set | bending loss: 0.007 -> 0.004 | Improved |
| Use SNUG's physics-based self-supervised training instead of PBNS | Garment deformation model · AMASS test set | gravity loss: 0.044 -> 0.028 | Improved |
| Use SNUG's physics-based self-supervised training instead of PBNS | Garment deformation model · AMASS test set | inertia loss: 0.0035 -> 0.0034 | Improved |
| Add bending loss (ablation) | SNUG · Ground-truth simulated test set | mean-curvature error: 17.3 -> 2.7 | Improved |
| Add strain loss (ablation) | SNUG · Ground-truth simulated test set | mean-curvature error: 19.1 -> 2.7 | Improved |
| Add gravity loss (ablation) | SNUG · Ground-truth simulated test set | mean-curvature error: 7.5 -> 2.7 | Improved |
| Add inertia loss (ablation) | SNUG · Ground-truth simulated test set | mean-curvature error: 2.8 -> 2.7 | Improved |
| Initialize GRU hidden states with sampled noise rather than zeros | SNUG recurrent regressor · Long motion sequences | generalization to longer sequences | Improved |
| Train on 3-frame subsequences rather than longer subsequences | SNUG recurrent regressor · Dynamic garment deformation | dynamic deformation quality | Improved |
Later papers that compared with this method
1 later accepted paper reports SNUG as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- CaPhy: Capturing Physical Properties for Animatable Human Avatars (ICCV 2023 · better on 4 of 4 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TailorNet — TailorNet: Predicting Clothing in 3D as a Function of Human Pose, Shape and Garment Style (CVPR 2020) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning 3D Garment Animation from Trajectories of A Piece of Cloth (NeurIPS 2024 · similarity 0.92)
- Self-supervised Garment Dynamics with Persistent Wrinkles (ECCV 2026 · similarity 0.92)
- TailorNet: Predicting Clothing in 3D as a Function of Human Pose, Shape and Garment Style (CVPR 2020 · similarity 0.91)
- NeuralGarSim: Geometry-agnostic Garment Simulation with Neural Fields (ECCV 2026 · similarity 0.90)
- Analyzing Clothing Layer Deformation Statistics of 3D Human Motions (ECCV 2018 · similarity 0.90)
- HOOD: Hierarchical Graphs for Generalized Modelling of Clothing Dynamics (CVPR 2023 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
