Research map / Image generation / Super-resolution
PatchScaler: An Efficient Patch-Independent Diffusion Model for Image Super-Resolution
Yong Liu, Hang Dong, Jinshan Pan, Qingji Dong, Kai Chen, Rongxiang Zhang, Lean Fu, Fei Wang
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 |
|---|---|---|---|
| use texture prompt instead of text prompt (ablation) | PatchScaler · RealSet110 | ManIQA: 0.4849 -> 0.5442 | Improved |
| use texture prompt instead of text prompt (ablation) | PatchScaler · RealSet110 | CLIPIQA: 0.7890 -> 0.8213 | Improved |
| use texture prompt instead of text prompt (ablation) | PatchScaler · RealSet110 | MUSIQ: 66.79 -> 70.97 | Improved |
| add PGS and GRM to StableSR training-free | StableSR · RealSet110 | ManIQA: 0.3856 -> 0.5035 | Improved |
| add PGS and GRM to StableSR training-free | StableSR · RealSet110 | MUSIQ: 62.08 -> 65.54 | Improved |
| add PGS and GRM to StableSR training-free | StableSR · RealSet110 | CLIPIQA: 0.7851 -> 0.7809 | No clear effect |
| add PGS to StableSR | StableSR · x4 512->2048 SR | runtime (s): 35.22 -> 8.53 | Improved |
| use two GPUs for patch groups | PatchScaler · x4 512->2048 SR | runtime (s): 3.41 -> 2.26 | Improved |
| reduce intermediate time steps from [1000,1000,1000] to [400,700,1000] (ablation) | PatchScaler · RealSet110 | quality metrics | No clear effect |
| reduce time steps further below [400,700,1000] (ablation) | PatchScaler · RealSet110 | quality metrics | Hurt |
| reduce sampling steps from [20,20,20] to [8,14,20] (ablation) | PatchScaler · RealSet110 | quality metrics | No clear effect |
| reduce sampling steps further below [8,14,20] (ablation) | PatchScaler · RealSet110 | quality metrics | Hurt |
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.
- ResShift — ResShift: Efficient Diffusion Model for Image Super-resolution by Residual Shifting (NeurIPS 2023) · better on 4 of 4 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
- Solving Diffusion ODEs with Optimal Boundary Conditions for Better Image Super-Resolution (ICLR 2024 · similarity 0.90)
- Taming Diffusion Prior for Image Super-Resolution with Domain Shift SDEs (NeurIPS 2024 · similarity 0.90)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.90)
- Patched Denoising Diffusion Models For High-Resolution Image Synthesis (ICLR 2024 · similarity 0.89)
- BUFF: Bayesian Uncertainty Guided Diffusion Probabilistic Model for Single Image Super-Resolution (AAAI 2025 · similarity 0.89)
- Hierarchical Patch Diffusion Models for High-Resolution Video Generation (CVPR 2024 · 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.
