Research map / Video / Video segmentation and tracking
Moment-Reenacting: Inverse Motion Degradation With Cross-Shutter Guidance
Xiang Ji, Guixu Lin, Zhengwei Yin, Jiancheng Zhao, Yinqiang Zheng
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 shutter alignment and aggregation (SAA) (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 27.96 -> 31.31 | Improved |
| use dual-stream instead of single-stream concatenation (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 30.33 -> 31.31 | Improved |
| add temporal positional encoding (TPE) (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 31.06 -> 31.15 | Improved |
| add region-adaptive motion distillation (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR / tOF: 31.15 -> 31.25; tOF 0.45 -> 0.31 | Improved |
| add motion-residue self-prompt (ablation) | Moment-Reenacting · realBR (x9 latent frames) | tOF: 0.45 -> 0.37 | Improved |
| add distillation masks to distillation (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 31.00 -> 31.31 | Improved |
| replace image-level flow-difference prompt with GT-informed prompt (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 31.31 -> 22.66 | Hurt |
| replace image-level prompt with learnable prompt (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 31.31 -> 31.16 | Hurt |
| use dual Blur-RS input instead of two blurred frames (IFED) | IFED · realBR (x9) | PSNR: 24.74 -> 30.84 | Improved |
| use dual Blur-RS input instead of two RS frames (RIFE) | RIFE · realBR (x9) | PSNR: 24.36 -> 30.67 | Improved |
| add random spatial misalignment of RS view (Shift-8) (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 31.31 -> 31.36 | No clear effect |
| add low-light noise to RS view (Noise-300) (ablation) | Moment-Reenacting · realBR (x9 latent frames) | PSNR: 31.31 -> 30.86 | Hurt |
Closest papers by meaning
- Motion Blur Decomposition with Cross-shutter Guidance (CVPR 2024 · similarity 0.91)
- Bringing Rolling Shutter Images Alive with Dual Reversed Distortion (ECCV 2022 · similarity 0.91)
- "UniINR: Event-guided Unified Rolling Shutter Correction, Deblurring, and Interpolation" (ECCV 2024 · similarity 0.91)
- Rolling Shutter Inversion: Bring Rolling Shutter Images to High Framerate Global Shutter Video (TPAMI 2022 · similarity 0.90)
- Rolling Shutter Correction with Intermediate Distortion Flow Estimation (CVPR 2024 · similarity 0.88)
- Single Image Rolling Shutter Removal with Diffusion Models (AAAI 2025 · 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.
