Image-Free Timestep Distillation via Continuous-Time Consistency with Trajectory-Sampled Pairs

Efficient Timestep Distillation Without Real Images

Bao Tang, Shuai Zhang, Yueting Zhu, Jijun Xiang, Xin Yang, Li Yu, Wenyu Liu, Xinggang Wang^📧

Huazhong University of Science and Technology (HUST)

(📧 corresponding author: xgwang@hust.edu.cn)

📰 News

• [Upcoming] Training code will be released soon.
• [2025.11.25] We’ve released our paper on arXiv.

📝 Introduction

Timestep distillation is an effective approach for improving the generation efficiency of diffusion models. The Consistency Model (CM), as a trajectory-based framework, demonstrates significant potential due to its strong theoretical foundation and high-quality few-step generation. Nevertheless, current continuous-time consistency distillation methods still rely heavily on training data and computational resources, hindering their deployment in resource-constrained scenarios and limiting their scalability to diverse domains. To address this issue, we propose Trajectory-Backward Consistency Model (TBCM), which eliminates the dependence on external training data by extracting latent representations directly from the teacher model's generation trajectory. Unlike conventional methods that require VAE encoding and large-scale datasets, our self-contained distillation paradigm significantly improves both efficiency and simplicity. Moreover, the trajectory-extracted samples naturally bridge the distribution gap between training and inference, thereby enabling more effective knowledge transfer. Empirically, TBCM achieves 6.52 FID and 28.08 CLIP scores on MJHQ-30k under one-step generation, while reducing training time by approximately 40% compared to Sana-Sprint and saving a substantial amount of GPU memory, demonstrating superior efficiency without sacrificing quality. We further reveal the diffusion-generation space discrepancy in continuous-time consistency distillation and analyze how sampling strategies affect distillation performance, offering insights for future distillation research.

📊 Results

⭐ Acknowledgements

This work is built upon the Sana series (Sana, Sana 1.5, Sana-Sprint). We sincerely thank the authors for their wonderful works and contributions to the community.

📄 Citation

If you find TBCM useful, please consider giving us a star 🌟 and citing it as follows:

@misc{tang2025tbcm,
      title={Image-Free Timestep Distillation via Continuous-Time Consistency with Trajectory-Sampled Pairs}, 
      author={Bao Tang and Shuai Zhang and Yueting Zhu and Jijun Xiang and Xin Yang and Li Yu and Wenyu Liu and Xinggang Wang},
      year={2025},
      eprint={2511.20410},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2511.20410}, 
}