Research map / Video / Video segmentation and tracking
Animation from Blur: Multi-modal Blur Decomposition with Motion Guidance
Zhihang Zhong, Xiao Sun, Zhirong Wu, Yinqiang Zheng, Stephen Lin, Imari Sato
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 predicted motion guidance (single sample) vs Jin et al. baseline | blur decomposition · B-Aist++ | PSNR: 17.01 -> 19.97 | Improved |
| sample more guidances and keep best (P1 to P5) | guidance predictor · B-Aist++ | PSNR: 19.97 -> 23.49 | Improved |
| sample more guidances and keep best (P1 to P5) | guidance predictor · GenBlur | PSNR: 23.41 -> 23.61 | Improved |
| use motion guidance from video optical flow instead of BIN | blur decomposition · B-Aist++ (video) | PSNR: 22.84 -> 24.03 | Improved |
| use motion guidance from video optical flow instead of BIN | blur decomposition · GenBlur (video) | PSNR: 24.82 -> 27.04 | Improved |
| add motion guidance input during training (ablation) | decomposition network · B-Aist++ | training convergence | Improved |
| add two-stage coarse-and-refinement pipeline at same model size (ablation) | decomposition network · B-Aist++ | PSNR: gain of 1 dB | Improved |
| quantize guidance into 4 directions instead of 2 (ablation) | motion guidance · B-Aist++ (validation loss) | validation loss | Improved |
| quantize guidance into 8 or 16 directions instead of 4 (ablation) | motion guidance · B-Aist++ (validation loss) | validation loss | No clear effect |
| add inverse-direction samples as augmentation | decomposition network training · GenBlur and B-Aist++ | diversity/dependency | Improved |
Closest papers by meaning
- Optical Flow Estimation from a Single Motion-blurred Image (AAAI 2021 · similarity 0.89)
- Parametric Object Motion From Blur (CVPR 2016 · similarity 0.88)
- Dynamic Video Deblurring Using a Locally Adaptive Blur Model (TPAMI 2017 · similarity 0.88)
- Motion Blur Decomposition with Cross-shutter Guidance (CVPR 2024 · similarity 0.88)
- Dynamic Gaussian Splatting from Defocused and Motion-blurred Monocular Videos (NeurIPS 2025 · similarity 0.87)
- Motion-adaptive Separable Collaborative Filters for Blind Motion Deblurring (CVPR 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.
