RIFLEx: A Free Lunch for Length Extrapolation in Video Diffusion Transformers
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866918116285480960 |
|---|---|
| author | Zhao, Min He, Guande Chen, Yixiao Zhu, Hongzhou Li, Chongxuan Zhu, Jun |
| author_facet | Zhao, Min He, Guande Chen, Yixiao Zhu, Hongzhou Li, Chongxuan Zhu, Jun |
| contents | Recent advancements in video generation have enabled models to synthesize high-quality, minute-long videos. However, generating even longer videos with temporal coherence remains a major challenge and existing length extrapolation methods lead to temporal repetition or motion deceleration. In this work, we systematically analyze the role of frequency components in positional embeddings and identify an intrinsic frequency that primarily governs extrapolation behavior. Based on this insight, we propose RIFLEx, a minimal yet effective approach that reduces the intrinsic frequency to suppress repetition while preserving motion consistency, without requiring any additional modifications. RIFLEx offers a true free lunch--achieving high-quality 2x extrapolation on state-of-the-art video diffusion transformers in a completely training-free manner. Moreover, it enhances quality and enables 3x extrapolation by minimal fine-tuning without long videos. Project page and codes: https://riflex-video.github.io/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_15894 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | RIFLEx: A Free Lunch for Length Extrapolation in Video Diffusion Transformers Zhao, Min He, Guande Chen, Yixiao Zhu, Hongzhou Li, Chongxuan Zhu, Jun Computer Vision and Pattern Recognition Recent advancements in video generation have enabled models to synthesize high-quality, minute-long videos. However, generating even longer videos with temporal coherence remains a major challenge and existing length extrapolation methods lead to temporal repetition or motion deceleration. In this work, we systematically analyze the role of frequency components in positional embeddings and identify an intrinsic frequency that primarily governs extrapolation behavior. Based on this insight, we propose RIFLEx, a minimal yet effective approach that reduces the intrinsic frequency to suppress repetition while preserving motion consistency, without requiring any additional modifications. RIFLEx offers a true free lunch--achieving high-quality 2x extrapolation on state-of-the-art video diffusion transformers in a completely training-free manner. Moreover, it enhances quality and enables 3x extrapolation by minimal fine-tuning without long videos. Project page and codes: https://riflex-video.github.io/. |
| title | RIFLEx: A Free Lunch for Length Extrapolation in Video Diffusion Transformers |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2502.15894 |