The Lost Melody: Empirical Observations on Text-to-Video Generation From A Storytelling Perspective

TL;DR

This paper investigates the storytelling capabilities of current text-to-video generation models, highlighting their limitations and proposing an evaluation framework to assess their storytelling quality.

Contribution

It introduces an empirical analysis of storytelling aspects in text-to-video models and proposes a new evaluation framework for this dimension.

Findings

Current models focus on visual fidelity but lack storytelling coherence.

Empirical observations reveal significant storytelling limitations.

An evaluation framework for storytelling in videos is proposed.

Abstract

Text-to-video generation task has witnessed a notable progress, with the generated outcomes reflecting the text prompts with high fidelity and impressive visual qualities. However, current text-to-video generation models are invariably focused on conveying the visual elements of a single scene, and have so far been indifferent to another important potential of the medium, namely a storytelling. In this paper, we examine text-to-video generation from a storytelling perspective, which has been hardly investigated, and make empirical remarks that spotlight the limitations of current text-to-video generation scheme. We also propose an evaluation framework for storytelling aspects of videos, and discuss the potential future directions.