Research map / 3D vision / Face and shape
Neural Shape Deformation Priors
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace ARAP with learned local deformation priors | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 5.568 -> 0.752 | Helped |
| Replace ARAP with learned local deformation priors | Neural Shape Deformation Priors · DeformingThing4D S2 | ℓ₂ × 0.001: 9.794 -> 0.795 | Helped |
| Replace ARAP with learned local deformation priors | Neural Shape Deformation Priors · Deformation Transfer | ℓ₂ × 0.001: 5.145 -> 2.495 | Helped |
| Use local deformation fields instead of a single global field (ablation) | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 2.970 -> 0.752 | Helped |
| Use local deformation fields instead of a single global field (ablation) | Neural Shape Deformation Priors · DeformingThing4D S2 | ℓ₂ × 0.001: 2.973 -> 0.795 | Helped |
| Use continuous deformation fields rather than 3D U-Net grid-based fields (ablation) | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 3.590 -> 0.752 | Helped |
| Use attention-based feature querying rather than Euclidean-distance interpolation (ablation) | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 1.184 -> 0.752 | Helped |
| Canonicalize source poses before target deformation (ablation) | Neural Shape Deformation Priors · DeformingThing4D S2 | ℓ₂ × 0.001: 0.969 -> 0.795 | Helped |
| Replace PointNet++ encoder with point transformer encoder (ablation) | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 0.886 -> 0.752 | Helped |
| Increase input point sampling from 5000 to 7500 (ablation) | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 0.752 -> 0.732 | Helped |
| Increase input point sampling from 5000 to 7500 (ablation) | Neural Shape Deformation Priors · DeformingThing4D S2 | ℓ₂ × 0.001: 0.795 -> 0.789 | Helped |
| Add Gaussian source-mesh noise with standard deviation 0.0050 (ablation) | Neural Shape Deformation Priors · DeformingThing4D S1 | ℓ₂ × 0.001: 0.752 -> 0.851 | Hurt |
Closest papers by meaning
- Implicit-ARAP: Efficient Handle-Guided Neural Field Deformation via Local Patch Meshing (NeurIPS 2025 · similarity 0.90)
- Reduced Representation of Deformation Fields for Effective Non-rigid Shape Matching (NeurIPS 2022 · similarity 0.90)
- 4Deform: Neural Surface Deformation for Robust Shape Interpolation (CVPR 2025 · similarity 0.90)
- Implicit Neural Surface Deformation with Explicit Velocity Fields (ICLR 2025 · similarity 0.89)
- Neural Implicit Shape Editing using Boundary Sensitivity (ICLR 2023 · similarity 0.89)
- Small Steps and Level Sets: Fitting Neural Surface Models with Point Guidance (CVPR 2024 · similarity 0.86)
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.