Research map / Image generation / Image restoration
Bringing Old Photos Back to Life
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use triplet-domain latent translation with VAEs, two-stage training, and latent adversarial alignment (ablation) | Old photo restoration · Old photo dataset | BRISQUE: 25.549 -> 23.016 | Helped |
| add two VAEs with KL loss (ablation) | Pix2Pix latent translation · Old photo dataset | BRISQUE: 25.549 -> 23.949 | Helped |
| train VAEs separately before latent mapping (ablation) | VAEs · Old photo dataset | BRISQUE: 23.949 -> 23.396 | Helped |
| add latent adversarial loss (ablation) | VAEs-TS · Old photo dataset | BRISQUE: 23.396 -> 23.016 | Helped |
| add partial nonlocal block (ablation) | Old photo restoration network · DIV2K | FID: 143.62 -> 134.35 | Helped |
| add partial nonlocal block (ablation) | Old photo restoration network · DIV2K | PSNR: 23.14 -> 23.33 | Helped |
| add partial nonlocal block (ablation) | Old photo restoration network · DIV2K | SSIM: 0.68 -> 0.69 | Helped |
| add partial nonlocal block (ablation) | Old photo restoration network · DIV2K | LPIPS: 0.26 -> 0.25 | Helped |
| use triplet-domain latent translation rather than pixel-level supervised translation | Old photo restoration · Real old photos | visual restoration quality | Helped |
| use partial nonlocal block for global-context inpainting (ablation) | Old photo restoration · Real old photos | structured defect restoration | Helped |
Closest papers by meaning
- ReFIR: Grounding Large Restoration Models with Retrieval Augmentation (NeurIPS 2024 · similarity 0.86)
- Self-Supervised Selective-Guided Diffusion Model for Old-Photo Face Restoration (NeurIPS 2025 · similarity 0.86)
- DreamClear: High-Capacity Real-World Image Restoration with Privacy-Safe Dataset Curation (NeurIPS 2024 · similarity 0.85)
- GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration (CVPR 2025 · similarity 0.85)
- Vision-Language Gradient Descent-driven All-in-One Deep Unfolding Networks (CVPR 2025 · similarity 0.84)
- Latent Harmony: Synergistic Unified UHD Image Restoration via Latent Space Regularization and Controllable Refinement (NeurIPS 2025 · similarity 0.84)
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.