OmniTransfer: All-in-one Framework for Spatio-temporal Video Transfer

TL;DR

OmniTransfer is a versatile framework that enhances spatio-temporal video transfer by leveraging multi-view information, enabling high-quality, flexible, and generalizable video generation without relying on task-specific priors.

Contribution

It introduces a unified approach with three key designs to improve appearance consistency, temporal control, and task adaptability in video transfer tasks.

Findings

Outperforms existing methods in appearance and temporal transfer tasks.

Matches pose-guided methods in motion transfer without pose information.

Establishes a new paradigm for flexible, high-fidelity video generation.

Abstract

Videos convey richer information than images or text, capturing both spatial and temporal dynamics. However, most existing video customization methods rely on reference images or task-specific temporal priors, failing to fully exploit the rich spatio-temporal information inherent in videos, thereby limiting flexibility and generalization in video generation. To address these limitations, we propose OmniTransfer, a unified framework for spatio-temporal video transfer. It leverages multi-view information across frames to enhance appearance consistency and exploits temporal cues to enable fine-grained temporal control. To unify various video transfer tasks, OmniTransfer incorporates three key designs: Task-aware Positional Bias that adaptively leverages reference video information to improve temporal alignment or appearance consistency; Reference-decoupled Causal Learning separating reference and target branches to enable precise reference transfer while improving efficiency; and Task-adaptive Multimodal Alignment using multimodal semantic guidance to dynamically distinguish and tackle different tasks. Extensive experiments show that OmniTransfer outperforms existing methods in appearance (ID and style) and temporal transfer (camera movement and video effects), while matching pose-guided methods in motion transfer without using pose, establishing a new paradigm for flexible, high-fidelity video generation.