Research map / 3D vision / NeRF and radiance fields
SUDS: Scalable Urban Dynamic Scenes
Haithem Turki, Jason Y. Zhang, Francesco Ferroni, Deva Ramanan
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 |
|---|---|---|---|
| use the three-branch representation and unlabeled depth, flow, and feature signals | SUDS · City-1M | PSNR: 16.70 -> 21.67 | Improved |
| use SUDS instead of PNF | KITTI image reconstruction · KITTI | PSNR: 27.48 -> 28.31 | Improved |
| use SUDS instead of NSG | novel-view synthesis · KITTI 75% | SSIM: 0.673 -> 0.797 | Improved |
| use SUDS instead of NSG | novel-view synthesis · VKITTI2 75% | LPIPS: 0.317 -> 0.150 | Improved |
| add the depth loss (ablation) | SUDS · KITTI and VKITTI2 novel-view synthesis | PSNR: 22.74 -> 22.95 | Improved |
| add the optical flow loss (ablation) | SUDS · KITTI and VKITTI2 novel-view synthesis | PSNR: 22.18 -> 22.95 | Improved |
| add the warping and flow-related loss terms (ablation) | SUDS · KITTI and VKITTI2 novel-view synthesis | PSNR: 17.53 -> 22.95 | Improved |
| add the appearance embedding (ablation) | SUDS · KITTI and VKITTI2 novel-view synthesis | PSNR: 22.54 -> 22.95 | Improved |
| use occlusion weights in the warping loss (ablation) | SUDS · KITTI and VKITTI2 novel-view synthesis | PSNR: 22.56 -> 22.95 | Improved |
| use separate static, dynamic, and far-field branches (ablation) | SUDS · KITTI and VKITTI2 novel-view synthesis | PSNR: 19.73 -> 22.95 | Improved |
Later papers that compared with this method
6 later accepted papers report SUDS as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- Periodic Vibration Gaussian: Dynamic Urban Scene Reconstruction and Real-time Rendering (IJCV 2026 · better on 3 of 3 shared results)
- AD-GS: Object-Aware B-Spline Gaussian Splatting for Self-Supervised Autonomous Driving (ICCV 2025 · better on 3 of 3 shared results)
- S3R-GS: Streamlining the Pipeline for Large-Scale Street Scene Reconstruction (ICCV 2025 · better on 1 of 1 shared result)
- PaReNeRF: Toward Fast Large-scale Dynamic NeRF with Patch-based Reference (CVPR 2024 · better on 12 of 14 shared results)
- Multi-Level Neural Scene Graphs for Dynamic Urban Environments (CVPR 2024 · better on 9 of 9 shared results)
- RoDUS: Robust Decomposition of Static and Dynamic Elements in Urban Scenes (ECCV 2024 · better on 1 of 4 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
- Multi-Level Neural Scene Graphs for Dynamic Urban Environments (CVPR 2024 · similarity 0.92)
- MatrixCity: A Large-scale City Dataset for City-scale Neural Rendering and Beyond (ICCV 2023 · similarity 0.92)
- A Large-Scale Outdoor Multi-Modal Dataset and Benchmark for Novel View Synthesis and Implicit Scene Reconstruction (ICCV 2023 · similarity 0.91)
- Grid-Guided Neural Radiance Fields for Large Urban Scenes (CVPR 2023 · similarity 0.90)
- Dynamic 3D Gaussian Fields for Urban Areas (NeurIPS 2024 · similarity 0.89)
- BLiRF: Bandlimited Radiance Fields for Dynamic Scene Modeling (AAAI 2024 · similarity 0.88)
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.
