Research map / 3D vision / Face and shape
FreeForm: Reduced-Order Deformable Simulation from Particle-Based Skinning Eigenmodes
Donglai Xiang, Vismay Modi, Rishit Dagli, Ty Trusty, Gilles Daviet, Anka He Chen, Nicholas Sharp, David I.W. Levin
Read the paperCheck 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 Hessian loss instead of random-z elastic loss (grid sampling) (ablation) | RKPM skinning weights · Beam Drop (m=16) | normalized MSE: 1.24e-02 -> 4.45e-04 | Improved |
| use Hessian loss instead of random-z elastic loss (grid sampling) (ablation) | RKPM skinning weights · Beam Twist (m=16) | normalized MSE: 7.29e-03 -> 3.49e-05 | Improved |
| use uniform grid instead of random integration points in training (Hessian loss) (ablation) | RKPM skinning weights · Beam Drop (m=16) | normalized MSE: 4.86e-03 -> 4.45e-04 | Improved |
| use uniform grid instead of random integration points in training (random-z loss) (ablation) | RKPM skinning weights · Beam Drop (m=16) | normalized MSE: 1.58e-02 -> 1.24e-02 | Improved |
| replace iterative optimization with generalized eigenanalysis (ablation) | Hessian loss, grid sampling · Beam (m=32) | training time (s): 145.96 -> 3.93 | Improved |
| replace iterative optimization with generalized eigenanalysis (ablation) | Hessian loss, grid sampling · Beam Drop (m=16) | normalized MSE: 4.45e-04 -> 4.10e-04 | No clear effect |
| replace neural field and random-z loss with RKPM eigenmodes | Simplicits · Thingi10K (Fix Side, m=32) | normalized MSE: 8.97e-03 -> 6.87e-03 | Improved |
| replace neural field and random-z loss with RKPM eigenmodes | Simplicits · Thingi10K (m=32) | training time (s): 121.44 -> 3.19 | Improved |
| replace neural field and random-z loss with RKPM eigenmodes | Simplicits · Simready (Pull Boundary, m=32) | normalized MSE: 8.83e-04 -> 4.16e-04 | Improved |
| increase number of skinning functions m from 6 to 32 | Ours · Beam Bend | normalized MSE: 7.80e-03 -> 2.93e-06 | Improved |
| increase number of skinning functions m from 16 to 32 | Ours · Beam Twist | normalized MSE: 3.46e-06 -> 6.64e-06 | Hurt |
| use random instead of grid test-time integration points (ablation) | Ours · Beam Bend (m=16) | normalized MSE: 4.10e-04 -> 9.42e-04 | Hurt |
Closest papers by meaning
- PIE-NeRF: Physics-based Interactive Elastodynamics with NeRF (CVPR 2024 · similarity 0.86)
- PhysRig: Differentiable Physics-Based Skinning and Rigging Framework for Realistic Articulated Object Modeling (ICCV 2025 · similarity 0.84)
- PhysConvex: Physics-Informed Dynamic Convex Fields for Reconstruction and Simulation (ECCV 2026 · similarity 0.84)
- PhysGaussian: Physics-Integrated 3D Gaussians for Generative Dynamics (CVPR 2024 · similarity 0.83)
- Neural Physical Simulation with Multi-Resolution Hash Grid Encoding (AAAI 2024 · similarity 0.82)
- Vid2Sim: Generalizable, Video-based Reconstruction of Appearance, Geometry and Physics for Mesh-free Simulation (CVPR 2025 · similarity 0.82)
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.
