Research map / 3D vision / Gaussian splatting
ParticleGS: Learning Neural Gaussian Particle Dynamics from Videos for Prior-free Physical Motion Extrapolation
Jinsheng Quan, Qiaowei Miao, Yichao Xu, Zizhuo Lin, Ying Li, Wei Yang, Zhihui Li, Yawei Luo
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 Neural ODE evolver instead of MLP (ablation) | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 34.551 -> 36.471 | Improved |
| use Neural ODE evolver instead of MLP (ablation) | ParticleGS · Dynamic Indoor Scene Dataset | extrapolation PSNR: 27.455 -> 31.103 | Improved |
| add physical (Rodrigues) decoding (ablation) | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 35.343 -> 36.471 | Improved |
| add physical (Rodrigues) decoding (ablation) | ParticleGS · Dynamic Indoor Scene Dataset | extrapolation PSNR: 28.653 -> 31.103 | Improved |
| add progressive training (ablation) | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 35.438 -> 36.471 | Improved |
| add progressive training (ablation) | ParticleGS · Dynamic Indoor Scene Dataset | extrapolation LPIPS: 0.218 -> 0.133 | Improved |
| add online neighborhood regularization (ablation) | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 35.986 -> 36.471 | Improved |
| add online neighborhood regularization (ablation) | ParticleGS · Dynamic Indoor Scene Dataset | extrapolation PSNR: 30.232 -> 31.103 | Improved |
| use F=8 dynamic fields instead of F=1 (ablation) | ParticleGS · Dynamic Indoor Scene Dataset | extrapolation PSNR: 28.026 -> 31.103 | Improved |
| increase dynamic fields from 8 to 16 (ablation) | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 36.471 -> 36.552 | No clear effect |
| use factorized encoding instead of per-Gaussian encoding (ablation) | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 36.391 -> 36.471 | Improved |
| ParticleGS vs velocity-field FreeGave | ParticleGS · Dynamic Object Dataset | extrapolation PSNR: 33.632 -> 36.471 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- FreeGave — FreeGave: 3D Physics Learning from Dynamic Videos by Gaussian Velocity (CVPR 2025) · better on 10 of 10 shared results
- TRACE — TRACE: 5D Temporal Regression of Avatars With Dynamic Cameras in 3D Environments (CVPR 2023) · better on 4 of 6 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
- Learning 3D Particle-based Simulators from RGB-D Videos (ICLR 2024 · similarity 0.90)
- FreeGave: 3D Physics Learning from Dynamic Videos by Gaussian Velocity (CVPR 2025 · similarity 0.88)
- FreeTimeGS: Free Gaussian Primitives at Anytime Anywhere for Dynamic Scene Reconstruction (CVPR 2025 · similarity 0.87)
- NeuMA: Neural Material Adaptor for Visual Grounding of Intrinsic Dynamics (NeurIPS 2024 · similarity 0.87)
- DEL: Discrete Element Learner for Learning 3D Particle Dynamics with Neural Rendering (NeurIPS 2024 · similarity 0.87)
- Fully Explicit Dynamic Gaussian Splatting (NeurIPS 2024 · 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.
