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..
Novel View Synthesis from A Few Glimpses via Test-Time Natural Video Completion
Authors: Yan Xu, Yixing Wang, Stella X. Yu
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate our method on LLFF [30], DL3DV [23], DTU [17], and Mip Ne RF-360 [3] datasets. LLFF consists of 8 forward-facing scenes. Following standard practice [54, 22], we train our model using only 3 input views on this dataset. DL3DV comprises diverse indoor and outdoor scenes captured by humans walking through scenes, exhibiting complex and dynamic camera motions. The Mip-Ne RF 360 dataset consists of real-world indoor and outdoor scenes designed for evaluating novel view synthesis in large, unbounded environments. The rendering quality is assessed using PSNR, SSIM, and LPIPS metrics. To validate the effectiveness of our proposed components in the pseudo-view generation (Sec. 4.1) and the 3D-GS optimization (Sec. 4.2), we conduct an extensive ablation study on DL3DV. |
| Researcher Affiliation | Academia | Yan Xu1 Yixing Wang1 Stella X. Yu1,2 1University of Michigan 2UC Berkeley EMAIL |
| Pseudocode | No | The paper describes the overall workflow and methods using text and figures (Fig. 1 and Fig. 2) and mathematical equations, but it does not include a clearly labeled 'Pseudocode' or 'Algorithm' block. |
| Open Source Code | No | Furthermore, we will release the code upon acceptance. The paper will provide open access to the data and code with instructions to reproduce all experimental results, in the camera ready version upon acceptance. |
| Open Datasets | Yes | We evaluate our method on LLFF [30], DL3DV [23], DTU [17], and Mip Ne RF-360 [3] datasets. |
| Dataset Splits | Yes | Following standard practice [54, 22], we train our model using only 3 input views on this dataset. For the DTU dataset, we follow the protocol from Reg Ne RF [22], using 3 training views across 15 evaluation scenes. DL3DV dataset under 3, 6, and 9 view settings. |
| Hardware Specification | No | The paper mentions 'compute support provided by the NAIRR Pilot under CIS240421' and in the NeurIPS checklist states 'We disclose the the computation resources and computation time we use in supplementary material.' However, specific hardware details like GPU models, CPU models, or memory are not provided in the main paper text. |
| Software Dependencies | No | The main paper text does not specify any software dependencies with version numbers. |
| Experiment Setup | Yes | In each cycle, we train the 3D-GS model for 10K iterations, followed by an update of the pseudo-view images using the video diffusion model. After each pseudo-view update, we reset the learning rate schedule of 3D-GS before starting the next optimization cycle to avoid overfitting. For the uncertainty estimation in Eq. (4), we set the bandwidth parameters to s1 = 100 and s2 = 0.25. The δ in Eq. (5) is fixed at 0.5 across all experiments. The loss weights are configured as follows: w1 = 0.8, w2 = 0.2, w3 = 1.0, w4 = 1.0, w5 = 0.2, and w6 = 1.0. Additional implementation details are provided in supplementary materials. |