Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
1000+ FPS 4D Gaussian Splatting for Dynamic Scene Rendering
Authors: Yuheng Yuan, Qiuhong Shen, Xingyi Yang, Xinchao Wang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We have extensively tested our proposed model on various dynamic scene datasets including real and synthetic scenes. As shown in Figure 1, 4DGS-1K reduces storage costs by 41% on the Neural 3D Video datasets [2] while maintaining equivalent scene representation quality. Crucially, it enables real-time rasterization speeds exceeding 1,000 FPS. These advancements collectively position 4DGS-1K as a practical solution for high-fidelity dynamic scene modeling without compromising efficiency. |
| Researcher Affiliation | Academia | 1National University of Singapore 2The Hong Kong Polytechnic University EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methodology in prose and diagrams (e.g., Figure 3), but does not include a distinct section or figure explicitly labeled as 'Pseudocode' or 'Algorithm' containing structured, step-by-step instructions. |
| Open Source Code | No | Answer: [No] Justification: The code will be open-sourced upon acceptance. |
| Open Datasets | Yes | Datasets. We utilize two dynamic scene datasets to demonstrate the effectiveness of our method: (1) Neural 3D Video Dataset (N3V) [2]. This dataset consists of six dynamic scenes, and the resolution is 2704 × 2028. For a fair comparison, we align with previous work [1, 40] by conducting evaluations at a half-resolution of 300 frames. (2) D-Ne RF Dataset [19]. This dataset is a monocular video dataset comprising eight videos of synthetic scenes. We choose standard test views that originate from novel camera positions not encountered during the training process. |
| Dataset Splits | Yes | Datasets. We utilize two dynamic scene datasets to demonstrate the effectiveness of our method: (1) Neural 3D Video Dataset (N3V) [2]... For a fair comparison, we align with previous work [1, 40] by conducting evaluations at a half-resolution of 300 frames. (2) D-Ne RF Dataset [19]... We choose standard test views that originate from novel camera positions not encountered during the training process. |
| Hardware Specification | Yes | Implementation Details. Our method is tested in a single RTX 3090 GPU. |
| Software Dependencies | No | The paper describes implementation details, but does not specify version numbers for any software dependencies, libraries, or frameworks used (e.g., 'PyTorch', 'CUDA', 'Python'). |
| Experiment Setup | Yes | After training, we perform the pruning and filtering strategy. Then, we fine-tune 4DGS-1K for 5,000 iterations while disabling additional clone/split operations. For pruning strategy, the pruning ratio is set to 80% on the N3V Dataset, and 85% on the D-Ne RF Dataset. For the temporal filtering, we set the interval t between key-frames to 20 frames on the N3V Dataset. Considering the varying capture speeds on the D-Ne RF dataset, we select 6 key-frames rather than a specific frame interval. |