A Survey on Long Video Generation: Challenges, Methods, and Prospects

TL;DR

This survey reviews recent advancements in long video generation, highlighting key paradigms, models, datasets, and evaluation metrics, while discussing challenges and future prospects in this rapidly evolving field.

Contribution

It provides the first comprehensive overview and classification of methods, datasets, and evaluation metrics for long video generation, identifying key challenges and future directions.

Findings

Two main paradigms: divide and conquer, and autoregressive models.

Classification of datasets and evaluation metrics used in the field.

Identification of current challenges and future research directions.

Abstract

Video generation is a rapidly advancing research area, garnering significant attention due to its broad range of applications. One critical aspect of this field is the generation of long-duration videos, which presents unique challenges and opportunities. This paper presents the first survey of recent advancements in long video generation and summarises them into two key paradigms: divide and conquer temporal autoregressive. We delve into the common models employed in each paradigm, including aspects of network design and conditioning techniques. Furthermore, we offer a comprehensive overview and classification of the datasets and evaluation metrics which are crucial for advancing long video generation research. Concluding with a summary of existing studies, we also discuss the emerging challenges and future directions in this dynamic field. We hope that this survey will serve as an essential reference for researchers and practitioners in the realm of long video generation.