Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918375681163264 |
|---|---|
| author | Wang, Jingkai Tang, Yixin Gong, Jue Li, Jiatong Li, Shu Liu, Libo Lan, Jianliang Liu, Yutong Zhang, Yulun |
| author_facet | Wang, Jingkai Tang, Yixin Gong, Jue Li, Jiatong Li, Shu Liu, Libo Lan, Jianliang Liu, Yutong Zhang, Yulun |
| contents | Diffusion transformer (DiT) architectures show great potential for real-world image super-resolution (Real-ISR). However, their computationally expensive iterative sampling necessitates one-step distillation. Existing one-step distillation methods struggle with Real-ISR on DiT. They suffer from fundamental trajectory mismatch and generate severe grid-like periodic artifacts. To tackle these challenges, we propose StrSR, a novel one-step adversarial distillation framework featuring spectral and trajectory regularization. Specifically, we propose an asymmetric discriminative distillation architecture to bridge the trajectory gap. Additionally, we design a frequency distribution matching strategy to effectively suppress DiT-specific periodic artifacts caused by high-frequency spectral leakage. Extensive experiments demonstrate that StrSR achieves state-of-the-art performance in Real-ISR, across both quantitative metrics and visual perception. The code and models will be released at https://github.com/jkwang28/StrSR . |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_06275 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution Wang, Jingkai Tang, Yixin Gong, Jue Li, Jiatong Li, Shu Liu, Libo Lan, Jianliang Liu, Yutong Zhang, Yulun Computer Vision and Pattern Recognition Diffusion transformer (DiT) architectures show great potential for real-world image super-resolution (Real-ISR). However, their computationally expensive iterative sampling necessitates one-step distillation. Existing one-step distillation methods struggle with Real-ISR on DiT. They suffer from fundamental trajectory mismatch and generate severe grid-like periodic artifacts. To tackle these challenges, we propose StrSR, a novel one-step adversarial distillation framework featuring spectral and trajectory regularization. Specifically, we propose an asymmetric discriminative distillation architecture to bridge the trajectory gap. Additionally, we design a frequency distribution matching strategy to effectively suppress DiT-specific periodic artifacts caused by high-frequency spectral leakage. Extensive experiments demonstrate that StrSR achieves state-of-the-art performance in Real-ISR, across both quantitative metrics and visual perception. The code and models will be released at https://github.com/jkwang28/StrSR . |
| title | Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.06275 |