Research map / 3D vision / Gaussian splatting
VEGS: View Extrapolation of Urban Scenes in 3D Gaussian Splatting using Learned Priors
Sungwon Hwang, Min-Jung Kim, Taewoong Kang, Jayeon Kang, Jaegul Choo
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 VEGS priors (normal and diffusion) and dynamic modeling to 3DGS for extrapolated views | 3DGS · KITTI-360 EVS | FID: 211.8 -> 124.4 | Improved |
| add surface-normal covariance guidance to LiDAR-initialized 3DGS+ (ablation) | 3DGS+ · KITTI-360 EVS | FID: 126.3 -> 124.6 | Improved |
| add diffusion score distillation prior on top of normal prior (ablation) | VEGS · KITTI-360 EVS | FID: 124.6 -> 124.4 | Improved |
| add L_axis covariance axes loss (ablation) | VEGS · KITTI-360 EVS-D | FID: 123.28 -> 122.56 | Improved |
| add both L_axis and L_scale (ablation) | VEGS · KITTI-360 EVS-D | FID: 123.28 -> 121.60 | Improved |
| optimize VEGS vs MARS on conventional test cameras | VEGS · KITTI-360 test cameras | SSIM: 0.814 (MARS) vs 0.812 | No clear effect |
| use VEGS vs 3DGS+ on conventional test cameras | 3DGS+ · KITTI-360 test cameras | PSNR: 23.76 -> 23.71 | No clear effect |
| add dynamic scene modeling, LiDAR init and box optimization to 3DGS (ablation) | 3DGS · KITTI-360 | FID: 211.8 -> 126.3 | Improved |
| add VEGS priors on KITTI | 3DGS+ · KITTI EVS | KID: 0.207 -> 0.090 | Improved |
| evaluate on extrapolated views versus conventional test cameras | MARS · KITTI-360 | FID: PSNR 23.13 on test cams vs FID 131.1 on EVS | Mixed |
Later papers that compared with this method
1 later accepted paper reports VEGS as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- CoRoGS: Contextual Gaussian Splatting for Robust Large-Deviation View Synthesis (CVPR 2026 · better on 1 of 1 shared result)
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.
- Mip-NeRF 360 — Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fields (CVPR 2022) · better on 1 of 1 shared result
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
- EVolSplat: Efficient Volume-based Gaussian Splatting for Urban View Synthesis (CVPR 2025 · similarity 0.90)
- EVPGS: Enhanced View Prior Guidance for Splatting-based Extrapolated View Synthesis (CVPR 2025 · similarity 0.89)
- D$^2$GS: Dense Depth Regularization for LiDAR-free Urban Scene Reconstruction (NeurIPS 2025 · similarity 0.88)
- Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes (CVPR 2025 · similarity 0.87)
- FewViewGS: Gaussian Splatting with Few View Matching and Multi-stage Training (NeurIPS 2024 · similarity 0.87)
- Effective Multi-sensor Conditioning for Street-view Novel-view Synthesis (NeurIPS 2026 · 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.
