Research map / Image generation / Image restoration
DifFace: Blind Face Restoration With Diffused Error Contraction
Zongsheng Yue, Chen Change Loy
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 diffusion prior to SwinIR estimator (ablation) | SwinIR · CelebA-Test BFR | FID-G: 93.73 -> 20.29 | Improved |
| add diffusion prior to SwinIR estimator (ablation) | SwinIR · CelebA-Test BFR | PSNR: 24.32 -> 23.44 | Hurt |
| add diffusion prior to SRCNN estimator (ablation) | SRCNN · CelebA-Test BFR | FID-G: 108.68 -> 44.83 | Improved |
| use SwinIR instead of SRCNN as diffused estimator (ablation) | DifFace · CelebA-Test BFR | FID-G: 44.83 -> 20.29 | Improved |
| increase DDIM eta from 0 to 1.0 (ablation) | DifFace · CelebA-Test BFR | LPIPS: 0.475 -> 0.443 | Improved |
| increase DDIM eta from 0 to 1.0 (ablation) | DifFace · CelebA-Test BFR | PSNR: 23.49 -> 23.28 | Hurt |
| increase DDIM eta from 0 to 0.75 (ablation) | DifFace · CelebA-Test BFR | FID-G: 23.51 -> 19.04 | Improved |
| reduce sampling steps from 250 to 20 (N=100 to 8) (ablation) | DifFace · CelebA-Test BFR | FID-G: 20.29 -> 42.54 | Hurt |
| reduce sampling steps from 250 to 50 (N=100 to 20) (ablation) | DifFace · CelebA-Test BFR | FID-F: 48.98 -> 54.32 | Hurt |
| reduce sampling steps from 250 to 100 (N=100 to 40) (ablation) | DifFace · CelebA-Test BFR | PSNR: 23.44 -> 23.41 | No clear effect |
| optimization step with gamma > 0 for inpainting (ablation) | DifFace · CelebA-HQ inpainting | coherence | Improved |
| use diffusion-based method vs GAN-based for inpainting | DifFace · CelebA-HQ inpainting | FID-G: 34.73 -> 24.45 | Improved |
Later papers that compared with this method
9 later accepted papers report DifFace as a baseline in their results tables; 8 of them report a better number on most of the results they share with it.
- Extreme Face Super-Resolution through Identity Fitting and Decoupling (ECCV 2026 · better on 1 of 1 shared result)
- Bridging Information Asymmetry: A Hierarchical Framework for Deterministic Blind Face Restoration (TPAMI 2026 · better on 1 of 1 shared result)
- FaceMe: Robust Blind Face Restoration with Personal Identification (AAAI 2025 · better on 3 of 3 shared results)
- Self-Supervised Selective-Guided Diffusion Model for Old-Photo Face Restoration (NeurIPS 2025 · better on 3 of 3 shared results)
- Posterior-Mean Rectified Flow: Towards Minimum MSE Photo-Realistic Image Restoration (ICLR 2025 · better on 2 of 3 shared results)
- DynFaceRestore: Balancing Fidelity and Quality in Diffusion-Guided Blind Face Restoration with Dynamic Blur-Level Mapping and Guidance (ICCV 2025 · better on 2 of 3 shared results)
- InstaRevive: One-Step Image Enhancement via Dynamic Score Matching (ICLR 2025 · better on 2 of 2 shared results)
- $InterLCM$: Low-Quality Images as Intermediate States of Latent Consistency Models for Effective Blind Face Restoration (ICLR 2025 · better on 0 of 1 shared result)
1 more not listed.
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- VQFR — VQFR: Blind Face Restoration with Vector-Quantized Dictionary and Parallel Decoder (ECCV 2022) · better on 1 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OSDFace: One-Step Diffusion Model for Face Restoration (CVPR 2025 · similarity 0.88)
- FaceMe: Robust Blind Face Restoration with Personal Identification (AAAI 2025 · similarity 0.87)
- ReF-LDM: A Latent Diffusion Model for Reference-based Face Image Restoration (NeurIPS 2024 · similarity 0.87)
- Learning Degradation-unaware Representation with Prior-based Latent Transformations for Blind Face Restoration (CVPR 2024 · similarity 0.86)
- Blind Face Restoration under Extreme Conditions: Leveraging 3D-2D Prior Fusion for Superior Structural and Texture Recovery (AAAI 2024 · similarity 0.86)
- $InterLCM$: Low-Quality Images as Intermediate States of Latent Consistency Models for Effective Blind Face Restoration (ICLR 2025 · 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.
