Research map / 3D vision / Gaussian splatting
PhysConvex: Physics-Informed Dynamic Convex Fields for Reconstruction and Simulation
Dan Wang, Xinrui Cui, Serge Belongie, Ravi Ramamoorthi
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 boundary-driven convex primitives with reduced-order simulation instead of Gaussian-based Vid2Sim | Vid2Sim · GSO dynamic reconstruction | PSNR: 28.57 -> 30.00 | Improved |
| Same change on system identification | Vid2Sim · GSO system identification | MAE log(E): 0.51 -> 0.37 | Improved |
| Same change on future state prediction | Vid2Sim · GSO future state prediction | PSNR: 25.07 -> 25.71 | Improved |
| Same change on future state FoVVDP | Vid2Sim · GSO future state prediction | FoVVDP: 7.770 -> 7.683 | No clear effect |
| Increase convex points K from 3 to 6 (ablation) | PhysConvex · GSO dynamic reconstruction | PSNR: 24.90 -> 30.00 | Improved |
| Increase K from 6 to 8 (ablation) | PhysConvex · GSO dynamic reconstruction | PSNR: 30.00 -> 30.19 | No clear effect |
| Increase K from 6 to 8 (ablation) | PhysConvex · GSO future state prediction | PSNR: 25.71 -> 25.38 | No clear effect |
| Vary reduced DOFs M from 10 to 8 or 12 (ablation) | PhysConvex (K=6) · GSO dynamic reconstruction | PSNR: 30.00 vs 29.96 / 29.88 | No clear effect |
| Use convex primitives instead of Gaussian/NeRF baselines | PAC-NeRF · GSO | Avg training time: 51 min -> 6 min | Improved |
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.
- Vid2Sim — Vid2Sim: Realistic and Interactive Simulation from Video for Urban Navigation (CVPR 2025) · better on 4 of 5 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
- PhysGaussian: Physics-Integrated 3D Gaussians for Generative Dynamics (CVPR 2024 · similarity 0.88)
- PhysCtrl: Generative Physics for Controllable and Physics-Grounded Video Generation (NeurIPS 2025 · similarity 0.88)
- MonoPhysics: Estimating Geometry, Appearance, and Physical Parameters from Monocular Videos (NeurIPS 2026 · similarity 0.88)
- DeformMaster: An Interactive Physics-Neural World Model for Deformable Objects from Videos (NeurIPS 2026 · similarity 0.88)
- PhysX-3D: Physical-Grounded 3D Asset Generation (NeurIPS 2025 · similarity 0.88)
- Vid2Sim: Generalizable, Video-based Reconstruction of Appearance, Geometry and Physics for Mesh-free Simulation (CVPR 2025 · similarity 0.88)
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.
