Research map / Video / Video generation
DNI: Dilutional Noise Initialization for Diffusion Video Editing
Sunjae Yoon, Gwanhyeong Koo, Ji Woo Hong, Chang D. Yoo
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 |
|---|---|---|---|
| apply DNI noise dilution to existing video editors | TAV, FateZero, Video-P2P, TokenFlow · DAVIS/TGVE non-rigid editing | CLIP text-video: TAV 22.6 -> 27.6; TF 21.7 -> 25.9 | Improved |
| apply DNI to rigid editing | TokenFlow · DAVIS/TGVE rigid editing | SSIM: 0.677 -> 0.733 | Improved |
| apply DNI to FateZero fidelity (non-rigid LPIPS) | FateZero · DAVIS/TGVE non-rigid editing | LPIPS: 0.173 -> 0.168 | No clear effect |
| use adaptive spectral filter instead of Gaussian low-pass filters to extract visual noise (ablation) | DNI noise disentanglement · visual noise extraction | PSNR: 13.2 -> 18.4 | Improved |
| use only the noise branch (drop visual branch) (ablation) | DNI · qualitative thumbs-up prompt | fidelity | Hurt |
| use visual branch without dilution (ablation) | DNI · qualitative thumbs-up prompt | non-rigid edit quality | Hurt |
| raise rigid mask scaler alpha above 0.7 (ablation) | DNI · rigid editing sensitivity | textual alignment | Hurt |
| increase non-rigid mask scaler beta (ablation) | DNI · non-rigid editing sensitivity | edit synthesis | Improved |
Closest papers by meaning
- FADE: Frequency-Aware Diffusion Model Factorization for Video Editing (CVPR 2025 · similarity 0.86)
- Schedule Your Edit: A Simple yet Effective Diffusion Noise Schedule for Image Editing (NeurIPS 2024 · similarity 0.86)
- MotionEditor: Editing Video Motion via Content-Aware Diffusion (CVPR 2024 · similarity 0.86)
- NaRCan: Natural Refined Canonical Image with Integration of Diffusion Prior for Video Editing (NeurIPS 2024 · similarity 0.85)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.85)
- How I Warped Your Noise: a Temporally-Correlated Noise Prior for Diffusion Models (ICLR 2024 · similarity 0.85)
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.
