Research map / Image generation / Image restoration
Diffusion Image Prior
Hamadi Chihaoui, Paolo Favaro
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 |
|---|---|---|---|
| raise k_min from 50 to 100 (ablation) | DIIP · Non-uniform deformation | PSNR: 22.18 -> 23.45 | Improved |
| raise k_min from 50 to 100 (ablation) | DIIP · Water-drop removal | PSNR: 22.48 -> 23.78 | Improved |
| raise k_min from 100 to 150 (ablation) | DIIP · Non-uniform deformation | PSNR: 23.45 -> 23.52 | No clear effect |
| raise k_min from 100 to 150 (ablation) | DIIP · Water-drop removal | PSNR: 23.78 -> 23.82 | No clear effect |
| lower epsilon from 0.005 to 0.001 (ablation) | DIIP · Denoising | PSNR: 27.25 -> 28.37 | Improved |
| lower epsilon from 0.005 to 0.001 (ablation) | DIIP · JPEG de-artifacting | PSNR: 22.38 -> 25.29 | Improved |
| lower epsilon from 0.001 to 0.0005 (ablation) | DIIP · Denoising | PSNR: 28.37 -> 28.14 | Hurt |
| lower epsilon from 0.001 to 0.0005 (ablation) | DIIP · JPEG de-artifacting | PSNR: 25.29 -> 25.02 | Hurt |
| use ground-truth-based optimal stopping instead of self-supervised criterion | DIIP · Denoising | PSNR: 28.37 -> 28.63 | Improved |
| use ground-truth-based optimal stopping instead of self-supervised criterion | DIIP · Water-drop removal | PSNR: 23.45 -> 23.73 | Improved |
| replace untrained CNN (DIP) with frozen pretrained diffusion model optimized over input noise | DIP · CelebA denoising | PSNR: 25.81 -> 28.37 | Improved |
| use DIIP instead of DreamClean | DreamClean · JPEG de-artifacting | PSNR: 23.92 -> 25.29 | Improved |
Closest papers by meaning
- Blind Image Restoration via Fast Diffusion Inversion (NeurIPS 2024 · similarity 0.90)
- Reconciling Stochastic and Deterministic Strategies for Zero-shot Image Restoration using Diffusion Model in Dual (CVPR 2025 · similarity 0.90)
- Taming Generative Diffusion Prior for Universal Blind Image Restoration (NeurIPS 2024 · similarity 0.90)
- A Diffusion Model with State Estimation for Degradation-Blind Inverse Imaging (AAAI 2024 · similarity 0.89)
- A Modular Conditional Diffusion Framework for Image Reconstruction (NeurIPS 2024 · similarity 0.89)
- Improving Diffusion-Based Image Restoration with Error Contraction and Error Correction (AAAI 2024 · similarity 0.89)
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.
