Video Generation Models: A Survey of Post-Training and Alignment

Chaoyu Li^1,†,✉️, Xiaoyi Gu^2,†, Yogesh Kulkarni¹, Eun Woo Im¹, Mohammadmahdi Honarmand³, Zeyu Wang⁴, Juntong Song⁵, Fei Du⁶, Xilin Jiang⁷, Kexin Zheng⁸, Tianzhi Li⁹, Fei Tao⁵, Pooyan Fazli^1,✉️

¹Arizona State University · ²Twitch · ³Stanford University · ⁴eBay · ⁵NewsBreak · ⁶Microsoft · ⁷Columbia University · ⁸University of Southern California · ⁹Carnegie Mellon University

^†Equal contribution. ^✉️Corresponding author: Chaoyu Li , Pooyan Fazli.

---

[!IMPORTANT] We welcome your help in improving the repository and paper. Please feel free to submit a pull request to:

• Add a relevant paper not yet included.
• Suggest a more suitable category.
• Update the information.
• Ask for clarification about any content.

---

🔥 News

• [2026.06.09] Our survey paper has been accepted to TMLR 🚀.
• [2026.06.09] Added papers accepted to ICML 2026.
• [2026.06.09] Added papers accepted to CVPR 2026.
• [2026.03.23] Added papers accepted to ICLR 2026.
• [2026.02.23] The v1 survey is now published! We have also initialized the repository.

🎯 Motivation

While large-scale pretraining has significantly improved video generation models, aligning them with human intent, physical constraints, and deployment requirements remains a major challenge. As a result, post-training and alignment techniques have rapidly become a central research focus in modern video generation.

In our survey paper, we systematically review recent progress in supervised fine-tuning, self-training and distillation, preference-based optimization, and inference-time alignment methods.

This repository serves as a companion resource to the survey.
It aims to:

• 📚 Curate recent papers on post-training and alignment for video generation
• 📊 Collect datasets and evaluation benchmarks related to post-training and alignment

We hope this repository provides researchers and practitioners with a convenient, up-to-date reference for exploring the evolving landscape of post-training and alignment in video generation models.

📌 Citation

If you find our paper or this resource helpful, please consider cite:

@article{li2026video,
  title={Video Generation Models: A Survey of Post-Training and Alignment},
  author={Chaoyu Li and Xiaoyi Gu and Yogesh Kulkarni and Eun Woo Im and Mohammadmahdi Honarmand and Zeyu Wang and Juntong Song and Fei Du and Xilin Jiang and Kexin Zheng and Tianzhi Li and Fei Tao and Pooyan Fazli},
  journal={Transactions on Machine Learning Research},
  issn={2835-8856},
  year={2026},
  url={https://openreview.net/forum?id=YlUEWLESIu},
}

📚 Content

• Awesome Video Generation Post Training
  • Datasets
  • Benchmarks
  • Conference Papers
  • arXiv Papers

---

Datasets

Benchmarks

Conference Papers

arXiv Papers

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🧑‍💻 Contributors

👏 Thanks to these contributors for this excellent work！

✉️ Contact

For questions, suggestions, or collaboration opportunities, please feel free to reach out:

✉️ Email: chaoyuli@asu.edu, pooyan@asu.edu

✨ Star History