Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution

Jingkai Wang, Yixin Tang, Jue Gong, Jiatong Li, Shu Li, Libo Liu, Jianliang Lan, Yutong Liu, Yulun Zhang, "Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution", arXiv, 2026

🔥🔥🔥 News

• 2026-03-06: This repo is released.

---

Abstract: 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. Experiments demonstrate that StrSR achieves state-of-the-art performance in Real-ISR, across both quantitative metrics and visual perception.

---

⚒️ TODO

• Release code and pretrained models
• Datasets
• Models
• Testing
• Training
• Acknowledgements

🔗 Contents

• Results
• Citation

🔎 Results

We achieved state-of-the-art performance on synthetic and real-world datasets.

 Quantitative Comparisons (click to expand)

Results in Table 1 on synthetic dataset (DIV2K-Val) from the main paper.

Results in Table 2 on real-world dataset (RealSR) from the main paper.

Results in Table 3 on real-world dataset (RealLQ250) from the main paper.

 Visual Comparisons (click to expand)

Results in Figure 7 on synthetic dataset (DIV2K-Val) from the main paper.

Results in Figure 8 on real-world dataset (RealSR) from the main paper.

Results in Figure 9 on real-world dataset (RealLQ250) from the main paper.

📎 Citation

If you find the code helpful in your research or work, please cite the following paper(s).

@article{wang2026strsr,
    title={Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution},
    author={Jingkai Wang and Yixin Tang and Jue Gong and Jiatong Li and Shu Li and Libo Liu and Jianliang Lan and Yutong Liu and Yulun Zhang},
    journal={arXiv preprint 2603.06275},
    year={2026}
}

💡 Acknowledgements

[TBD]