Inversion-Free Video Style Transfer with Trajectory Reset Attention Control and Content-Style Bridging
This work addresses video style transfer for applications in media and entertainment, offering an incremental improvement over existing methods.
The paper tackled the problem of content leakage and style misalignment in video style transfer by introducing Trajectory Reset Attention Control (TRAC) and a Style Medium concept, resulting in a tuning-free framework that enhances content consistency and reduces computational costs compared to inversion-based methods.
Video style transfer aims to alter the style of a video while preserving its content. Previous methods often struggle with content leakage and style misalignment, particularly when using image-driven approaches that aim to transfer precise styles. In this work, we introduce Trajectory Reset Attention Control (TRAC), a novel method that allows for high-quality style transfer while preserving content integrity. TRAC operates by resetting the denoising trajectory and enforcing attention control, thus enhancing content consistency while significantly reducing the computational costs against inversion-based methods. Additionally, a concept termed Style Medium is introduced to bridge the gap between content and style, enabling a more precise and harmonious transfer of stylistic elements. Building upon these concepts, we present a tuning-free framework that offers a stable, flexible, and efficient solution for both image and video style transfer. Experimental results demonstrate that our proposed framework accommodates a wide range of stylized outputs, from precise content preservation to the production of visually striking results with vibrant and expressive styles.