Research map / Image generation / Super-resolution
Bridging Fidelity-Reality with Controllable One-Step Diffusion for Image Super-Resolution
Hao Chen, Junyang Chen, Jinshan Pan, Jiangxin Dong
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 region-adaptive generative prior activation (RGPA) to base (ablation) | CODSR base · DrealSR | CLIPIQA+: 0.5038 -> 0.5109 | Improved |
| add RGPA to base (ablation) | CODSR base · DrealSR | LPIPS: 0.2906 -> 0.2914 | No clear effect |
| add LQFM alone to base (ablation) | CODSR base · DrealSR | MUSIQ: 65.89 -> 65.90 | No clear effect |
| add LQFM on top of RGPA (ablation) | Base w/ RGPA · DrealSR | FID: 133.02 -> 128.34 | Improved |
| add text-matching guidance (TMG) on top of RGPA and LQFM (ablation) | Base w/ RGPA&LQFM · DrealSR | CLIPIQA+: 0.5213 -> 0.5589 | Improved |
| add TMG on top of RGPA and LQFM (ablation) | Base w/ RGPA&LQFM · DrealSR | LPIPS: 0.2902 -> 0.2919 | Hurt |
| add adaptive noise to LQ image instead of latent (ablation) | CODSR base · DrealSR | CLIPIQA+: 0.5038 -> 0.5020 | No clear effect |
| use standard Gaussian noise in latent instead of adaptive noise (ablation) | CODSR base · DrealSR | PSNR (dB): 28.39 -> 28.04 | Hurt |
| replace SFT modulation with element-wise addition in LQFM (ablation) | Base w/ RGPA&LQFM · DrealSR | CLIPIQA+: 0.5213 -> 0.4956 | Hurt |
| modulate latent z_L instead of intermediate U-Net feature (ablation) | Base w/ RGPA&LQFM · DrealSR | CLIPIQA+: 0.5213 -> 0.5065 | Hurt |
| replace VSD loss with GAN loss in stage two (ablation) | CODSR · DrealSR | CLIPIQA+: 0.5589 -> 0.5009 | Hurt |
| increase forward timestep t_s (more adaptive noise) (ablation) | CODSR · DrealSR | PSNR vs MUSIQ | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FaithDiff — FaithDiff: Unleashing Diffusion Priors for Faithful Image Super-resolution (CVPR 2025) · better on 1 of 1 shared result
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
- Taming Diffusion Prior for Image Super-Resolution with Domain Shift SDEs (NeurIPS 2024 · similarity 0.90)
- When Latents Forget Pixels: Restoring Fidelity in Diffusion Transformer Super-Resolution (NeurIPS 2026 · similarity 0.90)
- The Power of Context: How Multimodality Improves Image Super-Resolution (CVPR 2025 · similarity 0.90)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.90)
- Pixel-level and Semantic-level Adjustable Super-resolution: A Dual-LoRA Approach (CVPR 2025 · similarity 0.90)
- RAP-SR: RestorAtion Prior Enhancement in Diffusion Models for Realistic Image Super-Resolution (AAAI 2025 · 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.
