Research map / 3D vision / 3D generation
DiffWind: Physics-Informed Differentiable Modeling of Wind-Driven Object Dynamics
Yuanhang Lei, Boming Zhao, Zesong Yang, Xingxuan Li, Tao Cheng, Haocheng Peng, Ru Zhang, yang yang, Siyuan Huang, Yujun Shen, Ruizhen Hu, Hujun Bao and 1 more
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 physics-informed LBM direction loss (ablation) | DiffWind · WD-Objects synthetic | PSNR: 51.31 -> 53.24 | Improved |
| add densification of invisible internal points (ablation) | DiffWind · WD-Objects real-world | PSNR: 27.31 -> 27.90 | Improved |
| use MLLM-inferred Young's modulus instead of predefined soft/medium/hard (ablation) | DiffWind w/o L_phys · WD-Objects synthetic | PSNR: 49.83 -> 51.31 (vs medium) | Improved |
| vary predefined Young's modulus soft vs medium (ablation) | DiffWind w/o L_phys · WD-Objects synthetic | PSNR: 49.78 vs 49.83 | No clear effect |
| increase grid resolution 128^3 to 256^3 (ablation) | DiffWind · WD-Objects synthetic | PSNR: 51.31 -> 51.90 | No clear effect |
| decrease grid resolution 128^3 to 64^3 (ablation) | DiffWind · WD-Objects synthetic | PSNR: 51.31 -> 49.08 | Hurt |
| DiffWind vs Deformable-GS on novel view synthesis | DiffWind · WD-Objects synthetic | PSNR: 36.68 -> 47.15 | Improved |
| DiffWind vs Deformable-GS on Pompon real scene | DiffWind · WD-Objects real-world (Pompon) | PSNR: 26.06 vs 26.14 (4D-GS 26.23 higher) | No clear effect |
| DiffWind simulation vs video generation models | DiffWind · Wind simulation human eval | visual quality: 2.84 -> 4.34 | Improved |
Closest papers by meaning
- Seeing the Wind from a Falling Leaf (NeurIPS 2025 · similarity 0.89)
- DecoupledGaussian: Object-Scene Decoupling for Physics-Based Interaction (CVPR 2025 · similarity 0.85)
- Force Prompting: Video Generation Models Can Learn And Generalize Physics-based Control Signals (NeurIPS 2025 · similarity 0.84)
- Unleashing the Potential of Multi-modal Foundation Models and Video Diffusion for 4D Dynamic Physical Scene Simulation (CVPR 2025 · similarity 0.84)
- DreamPhysics: Learning Physics-Based 3D Dynamics with Video Diffusion Priors (AAAI 2025 · similarity 0.84)
- OmniPhysGS: 3D Constitutive Gaussians for General Physics-Based Dynamics Generation (ICLR 2025 · similarity 0.84)
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.
