Research map / 3D vision / NeRF and radiance fields
NeRF-DS: Neural Radiance Fields for Dynamic Specular Objects
Zhiwen Yan, Chen Li, Gim Hee Lee
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 |
|---|---|---|---|
| Condition the color branch on observation-space position and surface normal | HyperNeRF-based NeRF-DS · Dynamic specular dataset (Mean) | LPIPS: .192 -> .135 | Improved |
| Add a mask-guided deformation field (ablation) | NeRF-DS · Dynamic specular dataset (Mean) | LPIPS: .153 -> .135 | Improved |
| Add surface-aware dynamic NeRF conditioning (ablation) | NeRF-DS · Dynamic specular dataset (Mean) | LPIPS: .142 -> .135 | Improved |
| Delay observation-space position encoding for 50k iterations, then increase to 4 for 50k (ablation) | NeRF-DS color branch · Sheet | LPIPS: 0.120 -> 0.115 | Improved |
| Remove observation-space position from the color branch (ablation) | NeRF-DS · Sheet | LPIPS: 0.115 -> 0.120 | Hurt |
| Sharpen mask-rendering weights with a Gaussian standard deviation schedule from 1 to 0.1 over 30k iterations (ablation) | NeRF-DS mask rendering · Sheet | LPIPS: 0.126 -> 0.115 | Improved |
| Use canonical-space normals directly rather than warping them into observation space (ablation) | NeRF-DS · Sheet | LPIPS: 0.115 -> 0.119 | Hurt |
| Calculate normals directly in observation space rather than warp canonical-space normals (ablation) | NeRF-DS · Sheet | LPIPS: 0.115 -> 0.117 | Hurt |
| Use NeRF-DS on non-specular dynamic scenes | HyperNeRF dataset · HyperNeRF dataset (Mean) | PSNR: 23.1 -> 22.8 | Mixed |
Later papers that compared with this method
9 later accepted papers report NeRF-DS as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.
- ReflFlow: Learning Geometry-Guided Ray Tracing for Dynamic Specular Reconstruction (ICML 2026 · better on 4 of 6 shared results)
- SpectroMotion: Dynamic 3D Reconstruction of Specular Scenes (CVPR 2025 · better on 1 of 1 shared result)
- HAIF-GS: Hierarchical and Induced Flow-Guided Gaussian Splatting for Dynamic Scene (NeurIPS 2025 · better on 1 of 1 shared result)
- TimeFormer: Capturing Temporal Relationships of Deformable 3D Gaussians for Robust Reconstruction (ICCV 2025 · better on 1 of 1 shared result)
- SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes (CVPR 2024 · better on 1 of 3 shared results)
- DN-4DGS: Denoised Deformable Network with Temporal-Spatial Aggregation for Dynamic Scene Rendering (NeurIPS 2024 · better on 3 of 3 shared results)
- D-MiSo: Editing Dynamic 3D Scenes using Multi-Gaussians Soup (NeurIPS 2024 · better on 0 of 2 shared results)
- MotionGS: Exploring Explicit Motion Guidance for Deformable 3D Gaussian Splatting (NeurIPS 2024 · better on 2 of 2 shared results)
1 more not listed.
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.
- Ref-NeRF — Ref-NeRF: Structured View-Dependent Appearance for 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
- D-NeRF: Neural Radiance Fields for Dynamic Scenes (CVPR 2021 · similarity 0.92)
- Ref-NeRF: Structured View-Dependent Appearance for Neural Radiance Fields (CVPR 2022 · similarity 0.91)
- Ced-NeRF: A Compact and Efficient Method for Dynamic Neural Radiance Fields (AAAI 2024 · similarity 0.91)
- A General Implicit Framework for Fast NeRF Composition and Rendering (AAAI 2024 · similarity 0.90)
- Rad-NeRF: Ray-decoupled Training of Neural Radiance Field (NeurIPS 2024 · similarity 0.89)
- NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wild (CVPR 2024 · similarity 0.89)
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.
