FADE: Frequency-Aware Diffusion Model Factorization for Video Editing

Yixuan Zhu; Haolin Wang; Shilin Ma; Wenliang Zhao; Yansong Tang; Lei Chen; Jie Zhou

doi:10.1109/cvpr52734.2025.02647

CVPR2025顶会

FADE: Frequency-Aware Diffusion Model Factorization for Video Editing

Yixuan Zhu, Haolin Wang, Shilin Ma, Wenliang Zhao, Yansong Tang, Lei Chen, Jie Zhou

DOI 出版方

2025年份

1顶会引用

摘要

A black swan swimming in a river beside bushes Motion Edit: swimming → flapping the wings Figure 1. Diverse video editing results of FADE. Our training-free approach, utilizing frequency-aware factorization and modulation, achieves high-fidelity, coherent edits across a variety of video types. FADE handles both appearance and motion adjustments with impressive robustness, ensuring precise alignment with input prompts and maintaining temporal consistency. Best viewed in color.