ReVideo: Remake a Video with Motion and Content Control

TL;DR

ReVideo introduces a novel video editing framework that enables precise, localized content and motion modifications using a three-stage training strategy and a spatiotemporal fusion module, advancing the accuracy and flexibility of diffusion-based video editing.

Contribution

The paper presents a new method for coupled content and motion editing in videos, with a three-stage training process and a fusion module, addressing coupling and training imbalance issues.

Findings

Effective local content modification while maintaining motion.

Ability to customize motion trajectories without altering content.

Supports multi-area editing without additional training.

Abstract

Despite significant advancements in video generation and editing using diffusion models, achieving accurate and localized video editing remains a substantial challenge. Additionally, most existing video editing methods primarily focus on altering visual content, with limited research dedicated to motion editing. In this paper, we present a novel attempt to Remake a Video (ReVideo) which stands out from existing methods by allowing precise video editing in specific areas through the specification of both content and motion. Content editing is facilitated by modifying the first frame, while the trajectory-based motion control offers an intuitive user interaction experience. ReVideo addresses a new task involving the coupling and training imbalance between content and motion control. To tackle this, we develop a three-stage training strategy that progressively decouples these two aspects from coarse to fine. Furthermore, we propose a spatiotemporal adaptive fusion module to integrate content and motion control across various sampling steps and spatial locations. Extensive experiments demonstrate that our ReVideo has promising performance on several accurate video editing applications, i.e., (1) locally changing video content while keeping the motion constant, (2) keeping content unchanged and customizing new motion trajectories, (3) modifying both content and motion trajectories. Our method can also seamlessly extend these applications to multi-area editing without specific training, demonstrating its flexibility and robustness.