Research map / Video / Video generation
Noise Calibration: Plug-and-play Content-Preserving Video Enhancement using Pre-trained Video Diffusion Models
Qinyu Yang, Haoxin Chen, Yong Zhang, Menghan Xia, Xiaodong Cun, Zhixun Su, Ying Shan
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 Noise Calibration (2 iterations) to SDEdit | SDEdit with VideoCrafter · Lavie700 enhancement | MSE_l: 4.3447 -> 2.8039 | Improved |
| add Noise Calibration (2 iterations) to SDEdit | SDEdit with VideoCrafter · Lavie700 enhancement | DOVER: 60.17 -> 62.71 | Improved |
| add Noise Calibration to SDEdit | SDEdit with VideoCrafter · Lavie700 enhancement | CLIP-IQA: 0.4482 -> 0.4387 | Hurt |
| increase calibration iterations from 3 to 10 (ablation) | NC-SDEdit · Lavie700 enhancement | DOVER: 62.45 -> 54.42 | Hurt |
| increase calibration iterations from 1 to 2 (ablation) | NC-SDEdit · Lavie700 enhancement | DOVER: 60.62 -> 62.71 | Improved |
| increase threshold frequency from 0.0 to 1.0 (ablation) | NC-SDEdit (t0=600) · 60 videos | low-frequency L2 (lower better): 4.2997 -> 2.7957 | Improved |
| increase initial denoising step t0 from 500 to 700 (ablation) | NC-SDEdit (nu=1.0) · 60 videos | content consistency vs detail: 2.0706 -> 3.9010 (L2), 0.0641 -> 0.0664 (D_SF) | Mixed |
| add Noise Calibration to MS-Vid2Vid-XL | MS-Vid2Vid-XL · EvalCrafter 700 videos | DOVER: 52.89 -> 57.61 | Improved |
| add Noise Calibration to SDXL-1.0-refiner | SDXL-1.0-refiner · SDXL images from 700 texts | MSE_l: 1.2775 -> 0.8750 | Improved |
| add Noise Calibration to SDEdit, inference cost | SDEdit · inference time | additional inference time: less than 10% additional | No clear effect |
Closest papers by meaning
- How I Warped Your Noise: a Temporally-Correlated Noise Prior for Diffusion Models (ICLR 2024 · similarity 0.90)
- VideoGuide: Improving Video Diffusion Models without Training Through a Teacher's Guide (CVPR 2025 · similarity 0.89)
- Training-Free Adaptive Diffusion with Bounded Difference Approximation Strategy (NeurIPS 2024 · similarity 0.88)
- Spectral Progressive Diffusion for Efficient Image and Video Generation (NeurIPS 2026 · similarity 0.88)
- FreqPrior: Improving Video Diffusion Models with Frequency Filtering Gaussian Noise (ICLR 2025 · similarity 0.88)
- Identifying and Solving Conditional Image Leakage in Image-to-Video Diffusion Model (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.
