Research map / Image generation / Super-resolution
XPSR: Cross-modal Priors for Diffusion-based Image Super-Resolution
Qu Yunpeng, Kun Yuan, Kai Zhao, Qizhi Xie, Jinhua Hao, Ming Sun, Chao Zhou
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 high-level (and keep low-level) priors to semantic conditioning (ablation) | XPSR · DRealSR | FID: 195.37 -> 164.68 | Improved |
| add low-level priors (keep high-level) (ablation) | XPSR · DRealSR | MANIQA: 0.5505 -> 0.5713 | Improved |
| use parallel Semantic-Fusion Attention instead of serial cross-attention (ablation) | XPSR · DRealSR | SSIM: 0.6834 -> 0.7220 | Improved |
| add pixel and latent Degradation-Free Constraint (ablation) | XPSR · DRealSR | SSIM: 0.6979 -> 0.7220 | Improved |
| add negative prompts at inference (ablation) | XPSR · DRealSR | MUSIQ: 61.70 -> 67.84 | Improved |
| add negative prompts at inference (ablation) | XPSR · DRealSR | SSIM: 0.7355 -> 0.7220 | Hurt |
| use SD-v2.1 instead of SD-v1.5 as base model (ablation) | XPSR · DRealSR | MUSIQ: 67.84 -> 63.79 | Hurt |
| degrade LR input with JPEG compression or Gaussian blur before MLLM prompting (ablation) | XPSR · DRealSR | FID: 164.68 -> 175.77 (Gaussian blur) | Hurt |
| use larger LLaVA-13b for prompts (ablation) | XPSR · RealSR | SSIM: 0.6870 -> 0.6920 | No clear effect |
| use XPSR instead of GAN-based and diffusion baselines | XPSR · DIV2K-Val | PSNR: 24.42 (BSRGAN) vs 22.80 (XPSR) | 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.
- SeeSR — SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution (CVPR 2024) · better on 8 of 9 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
- Taming Diffusion Prior for Image Super-Resolution with Domain Shift SDEs (NeurIPS 2024 · similarity 0.92)
- RAP-SR: RestorAtion Prior Enhancement in Diffusion Models for Realistic Image Super-Resolution (AAAI 2025 · similarity 0.90)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.89)
- When Latents Forget Pixels: Restoring Fidelity in Diffusion Transformer Super-Resolution (NeurIPS 2026 · similarity 0.89)
- TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution (CVPR 2025 · similarity 0.88)
- Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution (AAAI 2025 · similarity 0.88)
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.
