Reproduce “FreeForm: Reduced-Order Deformable Simulation from Particle-Based Skinning Eigenmodes” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Donglai Xiang, Vismay Modi, Rishit Dagli, Ty Trusty, Gilles Daviet, Anka He Chen, Nicholas Sharp, David I.W. Levin
FreeForm builds particle-based deformation bases, avoiding per-object neural-field fitting and volumetric meshes. It uses corrected reproducing kernel particle method interpolation and elastic-energy eigenanalysis to derive skinning weights for reduced simulation. On 20 selected Thingi10K shapes, mean basis-training time is 3.19 seconds versus 121.44 seconds for Simplicits; beam MSE is 4.10e-4 versus 1.53e-3, while robot and 3DGS results are qualitative demonstrations.
Key insights
- RKPM provides an effective mesh-free framework for parameterizing deformation subspaces and formulating elastic energy directly from particles.
- The Hessian matrix for Neo-Hookean elastic energy simplifies significantly when using RKPM-based skinning weights, enabling efficient computation of material-aware Laplace eigenmodes.
- Leveraging skinning eigenmodes derived from the Hessian allows for a principled, faster, and more accurate basis generation process compared to iterative neural field optimization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.