Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xie, Zhifei, Tang, Daniel, Tan, Dingwei, Klein, Jacques, Bissyand, Tegawend F., Ezzini, Saad
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2408.11788
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909293101449216
author Xie, Zhifei
Tang, Daniel
Tan, Dingwei
Klein, Jacques
Bissyand, Tegawend F.
Ezzini, Saad
author_facet Xie, Zhifei
Tang, Daniel
Tan, Dingwei
Klein, Jacques
Bissyand, Tegawend F.
Ezzini, Saad
contents Current video generation models excel at creating short, realistic clips, but struggle with longer, multi-scene videos. We introduce \texttt{DreamFactory}, an LLM-based framework that tackles this challenge. \texttt{DreamFactory} leverages multi-agent collaboration principles and a Key Frames Iteration Design Method to ensure consistency and style across long videos. It utilizes Chain of Thought (COT) to address uncertainties inherent in large language models. \texttt{DreamFactory} generates long, stylistically coherent, and complex videos. Evaluating these long-form videos presents a challenge. We propose novel metrics such as Cross-Scene Face Distance Score and Cross-Scene Style Consistency Score. To further research in this area, we contribute the Multi-Scene Videos Dataset containing over 150 human-rated videos.
format Preprint
id arxiv_https___arxiv_org_abs_2408_11788
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DreamFactory: Pioneering Multi-Scene Long Video Generation with a Multi-Agent Framework
Xie, Zhifei
Tang, Daniel
Tan, Dingwei
Klein, Jacques
Bissyand, Tegawend F.
Ezzini, Saad
Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
Software Engineering
TsingHua University
Current video generation models excel at creating short, realistic clips, but struggle with longer, multi-scene videos. We introduce \texttt{DreamFactory}, an LLM-based framework that tackles this challenge. \texttt{DreamFactory} leverages multi-agent collaboration principles and a Key Frames Iteration Design Method to ensure consistency and style across long videos. It utilizes Chain of Thought (COT) to address uncertainties inherent in large language models. \texttt{DreamFactory} generates long, stylistically coherent, and complex videos. Evaluating these long-form videos presents a challenge. We propose novel metrics such as Cross-Scene Face Distance Score and Cross-Scene Style Consistency Score. To further research in this area, we contribute the Multi-Scene Videos Dataset containing over 150 human-rated videos.
title DreamFactory: Pioneering Multi-Scene Long Video Generation with a Multi-Agent Framework
topic Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
Software Engineering
TsingHua University
url https://arxiv.org/abs/2408.11788