Research map / Image generation / Editing and style transfer
Old Photo Restoration via Deep Latent Space Translation
Ziyu Wan, Bo Zhang, Dong Chen, Pan Zhang, Dong Chen, Fang Wen, Jing Liao
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 partial nonlocal block (ablation) | triplet translation network · DIV2K synthetic | FID: 143.62 -> 134.35 | Improved |
| add partial nonlocal block (ablation) | triplet translation network · DIV2K synthetic | PSNR: 23.14 -> 23.33 | Improved |
| add partial nonlocal block (ablation) | triplet translation network · DIV2K synthetic | LPIPS: 0.26 -> 0.25 | Improved |
| use VAE latent space instead of Pix2Pix (ablation) | translation network · Real old photos | Wasserstein distance: 1.837 -> 1.048 | Improved |
| add two-stage training of VAEs (ablation) | VAE latent translation · Real old photos | Wasserstein distance: 1.048 -> 0.765 | Improved |
| add latent adversarial loss (ablation) | VAE latent translation · Real old photos | Wasserstein distance: 0.765 -> 0.581 | Improved |
| add latent adversarial loss (ablation) | VAE latent translation · Real old photos | BRISQUE: 23.396 -> 23.016 | Improved |
| use hierarchical injection instead of single 256x256 injection (ablation) | face enhancement network · Face enhancement | FID: 15.653 -> 13.175 | Improved |
| use hierarchical injection instead of single 256x256 injection (ablation) | face enhancement network · Face enhancement | PSNR: 24.938 -> 25.282 | Improved |
| increase single injection resolution from 16x16 to 128x128 (ablation) | face enhancement network · Face enhancement | FID: 24.177 -> 14.236 | Improved |
| inject only at the highest 256x256 scale rather than 128x128 (ablation) | face enhancement network · Face enhancement | FID: 14.236 -> 15.653 | Hurt |
| add joint training of face enhancement with translation network (ablation) | face enhancement · Real old photos | visual quality | Improved |
Closest papers by meaning
- Bringing Old Photos Back to Life (CVPR 2020 · similarity 0.98)
- Self-Supervised Selective-Guided Diffusion Model for Old-Photo Face Restoration (NeurIPS 2025 · similarity 0.87)
- Latent Harmony: Synergistic Unified UHD Image Restoration via Latent Space Regularization and Controllable Refinement (NeurIPS 2025 · similarity 0.87)
- ReFIR: Grounding Large Restoration Models with Retrieval Augmentation (NeurIPS 2024 · similarity 0.87)
- DreamClear: High-Capacity Real-World Image Restoration with Privacy-Safe Dataset Curation (NeurIPS 2024 · similarity 0.86)
- GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration (CVPR 2025 · 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.
