MLV-Edit: Towards Consistent and Highly Efficient Editing for Minute-Level Videos

TL;DR

MLV-Edit is a flow-based, training-free framework that enables consistent and efficient minute-level video editing by addressing temporal coherence and structural drift through segment-wise processing and specialized modules.

Contribution

It introduces a novel divide-and-conquer approach with Velocity Blend and Attention Sink modules to improve long-duration video editing quality without training overhead.

Findings

Outperforms state-of-the-art in temporal stability

Reduces flickering and boundary artifacts

Maintains semantic fidelity in long videos

Abstract

We propose MLV-Edit, a training-free, flow-based framework that address the unique challenges of minute-level video editing. While existing techniques excel in short-form video manipulation, scaling them to long-duration videos remains challenging due to prohibitive computational overhead and the difficulty of maintaining global temporal consistency across thousands of frames. To address this, MLV-Edit employs a divide-and-conquer strategy for segment-wise editing, facilitated by two core modules: Velocity Blend rectifies motion inconsistencies at segment boundaries by aligning the flow fields of adjacent chunks, eliminating flickering and boundary artifacts commonly observed in fragmented video processing; and Attention Sink anchors local segment features to global reference frames, effectively suppressing cumulative structural drift. Extensive quantitative and qualitative experiments demonstrate that MLV-Edit consistently outperforms state-of-the-art methods in terms of temporal stability and semantic fidelity.