Research map / 3D vision / 3D generation
MotionPhysics: Learnable Motion Distillation for Text-Guided Simulation
Miaowei Wang, Jakub Zadrożny, Oisin Mac Aodha, Amir Vaxman
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 range-constrained prompts for LLM initialization instead of PhysFlow initialization (ablation) | MotionPhysics · Average over scenes | OC x10^-2: 18.04 -> 18.18 | Improved |
| add LMD optimization after our initialization (ablation) | MotionPhysics · Average over scenes | OC x10^-2: 18.05 -> 18.18 | Improved |
| use LMD loss instead of SDS loss (ablation) | MotionPhysics · Average over scenes | OC x10^-2: 18.13 -> 18.18 | Improved |
| use MotionPhysics instead of PhysFlow | PhysFlow · Average over scenes | ECMS: 13.07 -> 11.37 | Improved |
| use MotionPhysics instead of PhysFlow | PhysFlow · Average over scenes | OC x10^-2: 17.96 -> 18.18 | Improved |
| use MotionPhysics instead of PhysDreamer | PhysDreamer · Average over scenes | ECMS: 27.48 -> 11.37 | Improved |
| use MotionPhysics instead of OmniPhysGS | OmniPhysGS · Average over scenes | optimization time: ~9 h -> 18.39 min | Improved |
| use MotionPhysics instead of PhysFlow | PhysFlow · User study Real World | physical realism preference: 66.13% | Improved |
| use median-value initialization of constrained ranges (ablation) | MotionPhysics · Hat scene | qualitative | Hurt |
| compare MotionPhysics with baselines on objective metrics in Droplet scene | MotionPhysics · Droplet scene | CLIPSIM and ECMS | Mixed |
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.
- OmniPhysGS — OmniPhysGS: 3D Constitutive Gaussians for General Physics-Based Dynamics Generation (ICLR 2025) · better on 3 of 4 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
- DreamPhysics: Learning Physics-Based 3D Dynamics with Video Diffusion Priors (AAAI 2025 · similarity 0.92)
- PhysSplat: Efficient Physics Simulation for 3D Scenes via MLLM-Guided Gaussian Splatting (ICCV 2025 · similarity 0.89)
- PhysCtrl: Generative Physics for Controllable and Physics-Grounded Video Generation (NeurIPS 2025 · similarity 0.89)
- OmniPhysGS: 3D Constitutive Gaussians for General Physics-Based Dynamics Generation (ICLR 2025 · similarity 0.88)
- Unleashing the Potential of Multi-modal Foundation Models and Video Diffusion for 4D Dynamic Physical Scene Simulation (CVPR 2025 · similarity 0.88)
- DeformMaster: An Interactive Physics-Neural World Model for Deformable Objects from Videos (NeurIPS 2026 · 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.
