Research map / 3D vision / Gaussian splatting
ReflFlow: Learning Geometry-Guided Ray Tracing for Dynamic Specular Reconstruction
Jiachen Tao, Junyi Wu, Haoxuan Wang, Zongxin Yang, Dawen Cai, Yan Yan
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 time-conditioned residual network for Gaussians (ablation) | ReflFlow · NeRF-DS ablation (4 scenes) | PSNR: 16.29 -> 18.97 | Improved |
| add residual correction network on dynamic environment Gaussian (ablation) | ReflFlow · NeRF-DS ablation (4 scenes) | PSNR: 18.97 -> 20.25 | Improved |
| add geometry-aligned normal loss Lnorm (ablation) | ReflFlow · NeRF-DS ablation (4 scenes) | PSNR: 20.25 -> 22.12 | Improved |
| add temporal-consistent normal supervision Ltc-norm (NormalCrafter) (ablation) | ReflFlow · NeRF-DS ablation (4 scenes) | PSNR: 22.12 -> 25.23 | Improved |
| add temporal-consistent normal supervision Ltc-norm (ablation) | ReflFlow · NeRF-DS ablation (4 scenes) | LPIPS: 0.2149 -> 0.1536 | Improved |
| ReflFlow versus SpectroMotion | ReflFlow · NeRF-DS (mean of 7 scenes) | PSNR / SSIM / LPIPS: 24.17 -> 24.91 PSNR; 0.8522 -> 0.8880 SSIM; 0.1778 -> 0.1471 LPIPS | Improved |
| ReflFlow versus 4DGS on HyperNeRF dataset | ReflFlow · HyperNeRF | PSNR: 24.89 -> 22.47 | Hurt |
| ReflFlow versus SpectroMotion on HyperNeRF dataset | ReflFlow · HyperNeRF | LPIPS: 0.3295 -> 0.3106 | Improved |
| ReflFlow versus static specular method EnvGS | ReflFlow · NeRF-DS (mean of 7 scenes) | PSNR: 20.21 -> 24.91 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- SpectroMotion — SpectroMotion: Dynamic 3D Reconstruction of Specular Scenes (CVPR 2025) · better on 6 of 6 shared results
- NeRF-DS — NeRF-DS: Neural Radiance Fields for Dynamic Specular Objects (CVPR 2023) · better on 4 of 6 shared results
- EnvGS — EnvGS: Modeling View-Dependent Appearance with Environment Gaussian (CVPR 2025) · better on 3 of 3 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
- SpecTRe-GS: Modeling Highly Specular Surfaces with Reflected Nearby Objects by Tracing Rays in 3D Gaussian Splatting (CVPR 2025 · similarity 0.92)
- SpectroMotion: Dynamic 3D Reconstruction of Specular Scenes (CVPR 2025 · similarity 0.91)
- MaterialRefGS: Reflective Gaussian Splatting with Multi-view Consistent Material Inference (NeurIPS 2025 · similarity 0.91)
- Reflective Gaussian Splatting (ICLR 2025 · similarity 0.90)
- Multi-times Monte Carlo Rendering for Inter-reflection Reconstruction (NeurIPS 2024 · similarity 0.89)
- Spec-Gaussian: Anisotropic View-Dependent Appearance for 3D Gaussian Splatting (NeurIPS 2024 · 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.
