Research map / 3D vision / NeRF and radiance fields
ExBluRF: Efficient Radiance Fields for Extreme Motion Blurred Images
Dongwoo Lee, Jeongtaek Oh, Jaesung Rim, Sunghyun Cho, Kyoung Mu 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 |
|---|---|---|---|
| model blur with a 6-DOF camera trajectory and voxel fields instead of DeblurNeRF 2D kernels on extreme blur | ExBluRF · ExBlur dataset | PSNR: 28.87 -> 30.17 | Improved |
| ExBluRF versus DeblurNeRF on mild real motion blur | ExBluRF · Real Motion Blur | PSNR: 25.49 -> 25.93 | No clear effect |
| ExBluRF versus DeblurNeRF on LPIPS in real blur | ExBluRF · Real Motion Blur | LPIPS: 0.182 -> 0.198 | Hurt |
| apply 2D deblurring (Restormer/NAFNet) before NeRF instead of joint blur modeling | NeRF · Synthetic Motion Blur | PSNR: 27.81 -> 21.15 | Hurt |
| replace neural radiance fields with voxel fields (keep 2D kernel, N=21) (ablation) | DeblurNeRF · ExBlur Camellia | training time (hrs): 27.48 -> 5.72 | Improved |
| replace 2D kernel with camera trajectory blur model on voxel fields (ablation) | ExBluRF · ExBlur Camellia | PSNR: 27.79 -> 28.83 | Improved |
| voxel-based fields keep memory constant as blur samples N grow | ExBluRF · GPU memory | GPU memory (GB): 49.09 -> 6.57 (N=19) | Improved |
| increase Bezier order from 1 to 3 (ablation) | ExBluRF · Synthetic Pool | PSNR: 25.30 -> 29.65 | Improved |
| increase Bezier order from 7 to 9 (ablation) | ExBluRF · Synthetic Pool | PSNR: 30.05 -> 29.92 | Hurt |
| increase trajectory samples N from 5 to 21 (ablation) | ExBluRF · ExBlur Postbox | PSNR: 28.62 -> 30.11 | Improved |
| increase trajectory samples N from 21 to 31 (ablation) | ExBluRF · ExBlur Postbox | PSNR: 30.11 -> 30.23 | No clear effect |
| jointly optimize camera trajectory instead of fixed COLMAP init poses | ExBluRF · Synthetic Factory | absolute trajectory error: 0.0480 -> 0.0141 | Improved |
Later papers that compared with this method
1 later accepted paper reports ExBluRF* as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- PRISM3D: Probabilistic Refinement and Robust Initialization for Physically Consistent Scene Modeling under Extreme Motion Blur (ECCV 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 2 of 2 on most shared results.
- DeblurNeRF — Deblur-NeRF: Neural Radiance Fields From Blurry Images (CVPR 2022) · better on 5 of 6 shared results
- BAD-NeRF — BAD-NeRF: Bundle Adjusted Deblur Neural Radiance Fields (CVPR 2023) · 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
- DyBluRF: Dynamic Neural Radiance Fields from Blurry Monocular Video (CVPR 2024 · similarity 0.90)
- Exploiting Deblurring Networks for Radiance Fields (CVPR 2025 · similarity 0.89)
- DynaMoDe-NeRF: Motion-aware Deblurring Neural Radiance Field for Dynamic Scenes (CVPR 2025 · similarity 0.88)
- Mitigating Motion Blur in Neural Radiance Fields with Events and Frames (CVPR 2024 · similarity 0.87)
- Dynamic Gaussian Splatting from Defocused and Motion-blurred Monocular Videos (NeurIPS 2025 · similarity 0.86)
- GS-Blur: A 3D Scene-Based Dataset for Realistic Image Deblurring (NeurIPS 2024 · similarity 0.86)
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.
