Research map / 3D vision / Gaussian splatting
Multi4D: High-Fidelity Dynamic Gaussian Splatting via Multi-Level Competitive Allocation
Rui Wang, Quentin Lohmeyer, Siyu Tang, Mirko Meboldt
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 persistent dynamic subset Gd (ablation) | Multi4D · Neu3D ablation (4 scenes) | PSNR: 32.78 -> 33.92 | Improved |
| add transient subset Gt (ablation) | Multi4D · Neu3D ablation (4 scenes) | PSNR: 32.86 -> 33.92 | Improved |
| add transient subset Gt (ablation) | Multi4D · Neu3D ablation (4 scenes) | dynamic Gaussians: 25 k -> 13 k + 152 k | Hurt |
| use velocity-aware periodic lifting instead of random init (ablation) | Multi4D · Neu3D ablation (4 scenes) | PSNR: 33.22 -> 33.92 | Improved |
| add SSIM diversity loss (ablation) | Multi4D · Neu3D ablation (4 scenes) | PSNR: 33.66 -> 33.92 | Improved |
| add mask-aware utility-based pruning (ablation) | Multi4D · Neu3D ablation (4 scenes) | PSNR: 33.68 -> 33.92 | Improved |
| multi-level decomposition versus monolithic 4DGS | Multi4D · Neu3D ablation (4 scenes) | dynamic Gaussians: 4215 k -> 165 k | Improved |
| multi-level decomposition versus 4DGS on training time | Multi4D · Neu3D | training time: 5.5 hours -> 1.2 hours | Improved |
| restrict semantic features to persistent subset versus TRASE | Multi4D · Neu3D-Mask 4D segmentation | FPS: 21 -> 204 | Improved |
| multi-level decomposition on monocular NeRF-DS versus 4D-primitive methods | Multi4D · NeRF-DS (monocular, mean of 7 scenes) | PSNR: 22.54 -> 23.69 | Improved |
| Multi4D versus E-D3DGS on Neu3D DSSIM | Multi4D · Neu3D (mean of 6 scenes) | DSSIM: 0.026 -> 0.026 | No clear effect |
Closest papers by meaning
- Fully Explicit Dynamic Gaussian Splatting (NeurIPS 2024 · similarity 0.93)
- Swift4D: Adaptive divide-and-conquer Gaussian Splatting for compact and efficient reconstruction of dynamic scene (ICLR 2025 · similarity 0.93)
- Motion Decoupled 3D Gaussian Splatting for Dynamic Object Representation (AAAI 2025 · similarity 0.92)
- Temporal Smoothness-Aware Rate-Distortion Optimized 4D Gaussian Splatting (NeurIPS 2025 · similarity 0.92)
- H3D-DGS: Exploring Heterogeneous 3D Motion Representation for Deformable 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- HAIF-GS: Hierarchical and Induced Flow-Guided Gaussian Splatting for Dynamic Scene (NeurIPS 2025 · similarity 0.91)
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.
