Research map / 3D vision / Gaussian splatting
DASH: 4D Hash Encoding with Self-Supervised Decomposition for Real-Time Dynamic Scene Rendering
Jie Chen, Zhangchi Hu, Peixi Wu, Huyue Zhu, Hebei Li, Xiaoyan Sun
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 self-supervised dynamic-static decomposition before the 4D hash encoder (ablation) | DASH · Neural 3D Video | PSNR: 31.65 -> 32.22 | Improved |
| add density control for dynamic Gaussians (ablation) | DASH · Neural 3D Video | PSNR: 31.92 -> 32.22 | Improved |
| add spatio-temporal smooth regularization (ablation) | DASH · Neural 3D Video | PSNR: 31.51 -> 32.22 | Improved |
| add smooth regularization (ablation) | DASH · Neural 3D Video | LPIPS: 0.056 -> 0.050 | Improved |
| add decomposition to Grid4D (Grid4D+dec) | Grid4D · Neural 3D Video | PSNR: 31.38 -> 31.74 | Improved |
| add decomposition to Grid4D (Grid4D+dec) | Grid4D · Technicolor Light Field | PSNR: 32.59 -> 33.04 | Improved |
| add decomposition to Grid4D (Grid4D+dec) | Grid4D · Technicolor Light Field | storage (MB): 319 -> 192 | Improved |
| add decomposition to Grid4D (Grid4D+dec) | Grid4D · Technicolor Light Field | training time (hours): 1.2 -> 2 | Hurt |
| replace Grid4D hash decomposition with full DASH | Grid4D · Technicolor Light Field | PSNR: 32.59 -> 33.94 | Improved |
| use DASH rather than 4DGaussian planes | 4DGaussian · Neural 3D Video (coffee martini) | Gaussian count N: 123.7k -> 106.6k | Improved |
| use DASH rather than 4DGaussian planes | 4DGaussian · Neural 3D Video (coffee martini) | FPS: 136.9 -> 264.3 | 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.
- Grid4D — Grid4D: 4D Decomposed Hash Encoding for High-Fidelity Dynamic Gaussian Splatting (NeurIPS 2024) · better on 6 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
- Grid4D: 4D Decomposed Hash Encoding for High-Fidelity Dynamic Gaussian Splatting (NeurIPS 2024 · similarity 0.91)
- Swift4D: Adaptive divide-and-conquer Gaussian Splatting for compact and efficient reconstruction of dynamic scene (ICLR 2025 · similarity 0.87)
- 4D Gaussian Splatting for Real-Time Dynamic Scene Rendering (CVPR 2024 · similarity 0.86)
- Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting (ICLR 2024 · similarity 0.86)
- Temporal Smoothness-Aware Rate-Distortion Optimized 4D Gaussian Splatting (NeurIPS 2025 · similarity 0.85)
- 1000+ FPS 4D Gaussian Splatting for Dynamic Scene Rendering (NeurIPS 2025 · similarity 0.85)
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.
