Research map / 3D vision / 3D generation
PhysHanDI: Physics-Based Reconstruction of Hand-Deformable Object Interactions
Jihyun Lee, Changmin Lee, Donghwan Kim, Tae-Kyun Kim
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 |
|---|---|---|---|
| add inverse-physics hand refinement (ablation) | PhysHanDI · PhysTwin-full, single-view future prediction | Chamfer distance: 42.8 -> 33.5 | Improved |
| add inverse-physics hand refinement (ablation) | PhysHanDI · PhysTwin-full, single-view future prediction | Hand CD: 7.57 -> 7.17 | Improved |
| drive simulation with dense MANO hand vertices instead of ~30 sparse control points | PhysTwin · PhysTwin-dense, reconstruction & resimulation | CDdyn (mm): 10.78 -> 8.32 | Improved |
| drive simulation with dense MANO hand vertices instead of sparse control points | PhysTwin · PhysTwin-dense, future prediction | CDdyn (mm): 16.32 -> 14.35 | Improved |
| drive simulation with dense MANO hand vertices instead of sparse control points | PhysTwin · DenseHDI, future prediction | PSNR: 19.76 -> 19.75 | No clear effect |
| use dense hand reconstruction under perturbed tracking signals | PhysTwin · PhysTwin-dense, perturbed tracking | CD (mm): 9.60 -> 5.56 | Improved |
| use dense hand reconstruction under perturbed controller | PhysTwin · PhysTwin-full, perturbed controller | Track Err.: PhysTwin 1.25 vs ours 1.29 | Hurt |
| use dense hand points for contacts instead of sparse points | spring-mass topology · PhysTwin-dense | RRD virtual springs: 2.63 -> 0.35 | Improved |
| increase initial max connected nodes dmax to 50 (with delta 0.002) (ablation) | PhysHanDI zero-order optimization · double_stretch_sloth sequence | CD: 4.44 -> 6.15 | Hurt |
| set initial connection radius delta to 0.001 instead of 0.002 (ablation) | PhysHanDI zero-order optimization · double_stretch_sloth sequence | CD: 4.44 -> 6.81 | Hurt |
| add hand reconstruction and refinement stages | PhysHanDI pipeline · per-frame runtime | seconds per frame: object recon zero-order 21.22 -> 17.63 faster, hand recon adds 27.38 | Hurt |
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.
- PhysTwin — PhysTwin: Physics-Informed Reconstruction and Simulation of Deformable Objects from Videos (ICCV 2025) · better on 15 of 16 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
- HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video (CVPR 2024 · similarity 0.89)
- EasyHOI: Unleashing the Power of Large Models for Reconstructing Hand-Object Interactions in the Wild (CVPR 2025 · similarity 0.89)
- Hand-Centric Motion Refinement for 3D Hand-Object Interaction via Hierarchical Spatial-Temporal Modeling (AAAI 2024 · similarity 0.87)
- Learning Explicit Contact for Implicit Reconstruction of Hand-Held Objects from Monocular Images (AAAI 2024 · similarity 0.87)
- Re-HOLD: Video Hand Object Interaction Reenactment via adaptive Layout-instructed Diffusion Model (CVPR 2025 · similarity 0.87)
- DiffGrasp: Whole-Body Grasping Synthesis Guided by Object Motion Using a Diffusion Model (AAAI 2025 · similarity 0.87)
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.
