VSTAR: Generative Temporal Nursing for Longer Dynamic Video Synthesis

TL;DR

VSTAR introduces a novel approach for generating longer, more dynamic videos from text prompts by dynamically controlling temporal content through video synopsis prompting and attention regularization, overcoming limitations of existing models.

Contribution

The paper proposes VSTAR, a method that enhances long video synthesis by dynamically adjusting temporal dynamics using LLM-generated prompts and attention regularization, a novel combination in T2V synthesis.

Findings

VSTAR produces longer, more visually appealing videos.

It effectively controls temporal dynamics over existing models.

The approach improves the alignment of visual change with text prompts.

Abstract

Despite tremendous progress in the field of text-to-video (T2V) synthesis, open-sourced T2V diffusion models struggle to generate longer videos with dynamically varying and evolving content. They tend to synthesize quasi-static videos, ignoring the necessary visual change-over-time implied in the text prompt. At the same time, scaling these models to enable longer, more dynamic video synthesis often remains computationally intractable. To address this challenge, we introduce the concept of Generative Temporal Nursing (GTN), where we aim to alter the generative process on the fly during inference to improve control over the temporal dynamics and enable generation of longer videos. We propose a method for GTN, dubbed VSTAR, which consists of two key ingredients: 1) Video Synopsis Prompting (VSP) - automatic generation of a video synopsis based on the original single prompt leveraging LLMs, which gives accurate textual guidance to different visual states of longer videos, and 2) Temporal Attention Regularization (TAR) - a regularization technique to refine the temporal attention units of the pre-trained T2V diffusion models, which enables control over the video dynamics. We experimentally showcase the superiority of the proposed approach in generating longer, visually appealing videos over existing open-sourced T2V models. We additionally analyze the temporal attention maps realized with and without VSTAR, demonstrating the importance of applying our method to mitigate neglect of the desired visual change over time.