Research map / Image generation / Super-resolution
Time-Aware One Step Diffusion Network for Real-World Image Super-Resolution
Tianyi Zhang, Zheng-Peng Duan, Chun-Le Guo, Peng-Tao Jiang, Bo Li, Ming-Ming Cheng, Chongyi Li
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 Time-Aware VAE Encoder (TAE) (ablation) | TADSR · RealSR | CLIPIQA: 0.6857 -> 0.7283 | Improved |
| add Time-Aware VAE Encoder (TAE) (ablation) | TADSR · DrealSR | PSNR: 27.95 -> 28.39 | Improved |
| add Time-Aware VSD loss (TAVSD) instead of plain VSD (ablation) | TADSR · RealSR | CLIPIQA: 0.6930 -> 0.7283 | Improved |
| add Time-Aware VSD loss (TAVSD) instead of plain VSD (ablation) | TADSR · DrealSR | PSNR: 28.03 -> 28.39 | Improved |
| add both TAE and TAVSD relative to original VAE encoder + VSD baseline (ablation) | TADSR · RealSR | PSNR: 24.39 -> 25.16 | Improved |
| add both TAE and TAVSD relative to baseline (ablation) | TADSR · DrealSR | CLIPIQA: 0.6887 -> 0.7398 | Improved |
| sample student timestep randomly (baseline) instead of fixing it, without TAE or TAVSD (ablation) | TADSR baseline · RealSR | PSNR | No clear effect |
| increase timestep condition ts (ablation) | TADSR · DrealSR | PSNR vs QALIGN: PSNR decreases, QALIGN increases | Mixed |
| use TADSR (one step) instead of OSEDiff, same parameters | SD 2.1-base · RealLR200 | QALIGN: 3.8459 -> 4.2622 | Improved |
| use TADSR instead of OSEDiff on LPIPS | SD 2.1-base · RealSR | LPIPS: 0.2920 -> 0.3168 | Hurt |
Closest papers by meaning
- Unleashing the Power of One-Step Diffusion based Image Super-Resolution via a Large-Scale Diffusion Discriminator (NeurIPS 2025 · similarity 0.92)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.91)
- One Diffusion Step to Real-World Super-Resolution via Flow Trajectory Distillation (ICML 2025 · similarity 0.91)
- Effective Diffusion Transformer Architecture for Image Super-Resolution (AAAI 2025 · similarity 0.91)
- When Latents Forget Pixels: Restoring Fidelity in Diffusion Transformer Super-Resolution (NeurIPS 2026 · similarity 0.90)
- One-Step Effective Diffusion Network for Real-World Image Super-Resolution (NeurIPS 2024 · similarity 0.90)
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.
