Research map / Image generation / Super-resolution
VOSR: A Vision-Only Generative Model for Image Super-Resolution
Rongyuan Wu, Lingchen Sun, Zhengqiang Zhang, Xiangtao Kong, Jixin Zhao, Shihao Wang, Lei Zhang
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 |
|---|---|---|---|
| replace the fully unconditional CFG branch with a partially conditioned branch (weakened LR structure, no semantics) (ablation) | VOSR · image super-resolution (trained from scratch) | restoration quality | Improved |
| increase inference guidance scale s (ablation) | VOSR-1.4B-ms · image SR (Figure 4) | LPIPS and MUSIQ: LPIPS rises, MUSIQ drops | Mixed |
| add DINO vision semantic condition via cross-attention to the structural condition | VOSR · image super-resolution | perceptual quality | Improved |
| use recursive-consistency distillation variant instead of shortcut variant for one-step student (ablation) | VOSR one-step distillation · image super-resolution | perceptual quality and fidelity balance | Improved |
| scale VOSR from 0.5B to 1.4B parameters (ablation) | VOSR multi-step · LSDIR val | LPIPS: 0.3984 -> 0.3857 | Improved |
| scale VOSR from 0.5B to 1.4B parameters (ablation) | VOSR multi-step · LSDIR val | NIQE: 4.0466 -> 4.4554 | Hurt |
| distill multi-step VOSR-0.5B into one step | VOSR-0.5B · LSDIR val | LPIPS: 0.3984 -> 0.3888 | No clear effect |
| use vision-only VOSR instead of vision-only ResShift or SinSR | generative SR · RealSR | MUSIQ: 58.4687 -> 70.4718 (multi-step) | Improved |
| use vision-only VOSR instead of ResShift for distortion fidelity | generative SR · RealSR | PSNR: 26.2630 -> 25.2886 | Hurt |
| use VOSR-1.4B-ms instead of T2I-based multi-step SR for speed | inference · 512x512 A100 | time (s): 3.943 -> 2.114 (SeeSR) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- SeeSR — SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution (CVPR 2024) · better on 4 of 5 shared results
- ResShift — ResShift: Efficient Diffusion Model for Image Super-resolution by Residual Shifting (NeurIPS 2023) · better on 1 of 3 shared results
- DiT4SR — DiT4SR: Taming Diffusion Transformer for Real-World Image Super-Resolution (ICCV 2025) · better on 2 of 2 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
- Visual Autoregressive Modeling for Image Super-Resolution (ICML 2025 · similarity 0.91)
- SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution (CVPR 2024 · similarity 0.90)
- The Power of Context: How Multimodality Improves Image Super-Resolution (CVPR 2025 · similarity 0.90)
- TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution (CVPR 2025 · similarity 0.89)
- Pixel-level and Semantic-level Adjustable Super-resolution: A Dual-LoRA Approach (CVPR 2025 · similarity 0.89)
- SRSR: Enhancing Semantic Accuracy in Real-World Image Super-Resolution with Spatially Re-Focused Text-Conditioning (NeurIPS 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.
