| title | Enhancing Train-Free Infinite-Frame Generation for Consistent Long Videos |
|---|---|
| authors | X. Feng, J. Zhu, M. Wu, C. Chen, F. Mao, H. Guo, J. Wu, X. Chu, K. Huang |
| date | 2026-05-18 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.18233 |
| affiliation | AGI Lab |
| upvotes | 58 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Without incurring significant computational overhead, train-free long video generation aims to enable foundation video generation models to produce longer videos. Frame-level autoregressive frameworks, e.g., FIFO-diffusion, offer the advantage of generating infinitely long videos with constant memory consumption. However, the mismatch between training and inference, coupled with the challenge of maintaining long-term consistency, limits the effective utilization of foundation models. To mitigate these concerns, we propose MIGA, a novel infinite-frame long video generation method. Firstly, we propose an effective two-stage alignment mechanism that mitigates the training-inference gap by reducing the excessive noise span fed to the model. We then introduce an innovative dual consistency enhancement mechanism, where the self-reflection approach corrects early high-noise frames and the long-range frame guidance approach leverages later low-noise frames with broad coverage to steer generation, jointly improving temporal consistency. Extensive experiments on VBench and NarrLV demonstrate the state-of-the-art performance of MIGA. Our project page is available at https://xiaokunfeng.github.io/miga_homepage/. |
cite
@misc{feng2026enhancing,
title={Enhancing Train-Free Infinite-Frame Generation for Consistent Long Videos},
author={X. Feng and J. Zhu and M. Wu and C. Chen and F. Mao and H. Guo and J. Wu and X. Chu and K. Huang},
year={2026},
eprint={2605.18233},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.18233}
}