Co-Director: Agentic Generative Video Storytelling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Song, Yale, Song, Yiwen, Losier, Nick, Hodson, Nathan, Jin, Ye, Zhu, Rhyard, Xu, Yan, Vlasic, Daniel, Claassen, Carina, Leon, Jasmine, LeViet, Khanh G., Chomyn, Zack, Timmons, Joe, Slatkin, Brett, Penberthy, Scott, Pfister, Tomas
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917442268168192
author Song, Yale
Song, Yiwen
Losier, Nick
Hodson, Nathan
Jin, Ye
Zhu, Rhyard
Xu, Yan
Vlasic, Daniel
Claassen, Carina
Leon, Jasmine
LeViet, Khanh G.
Chomyn, Zack
Timmons, Joe
Slatkin, Brett
Penberthy, Scott
Pfister, Tomas
author_facet Song, Yale
Song, Yiwen
Losier, Nick
Hodson, Nathan
Jin, Ye
Zhu, Rhyard
Xu, Yan
Vlasic, Daniel
Claassen, Carina
Leon, Jasmine
LeViet, Khanh G.
Chomyn, Zack
Timmons, Joe
Slatkin, Brett
Penberthy, Scott
Pfister, Tomas
contents While diffusion models generate high-fidelity video clips, transforming them into coherent storytelling engines remains challenging. Current agentic pipelines automate this via chained modules but suffer from semantic drift and cascading failures due to independent, handcrafted prompting. We present Co-Director, a hierarchical multi-agent framework formalizing video storytelling as a global optimization problem. To ensure semantic coherence, we introduce hierarchical parameterization: a multi-armed bandit globally identifies promising creative directions, while a local multimodal self-refinement loop mitigates identity drift and ensures sequence-level consistency. This balances the exploration of novel narrative strategies with the exploitation of effective creative configurations. For evaluation, we introduce GenAD-Bench, a 400-scenario dataset of fictional products for personalized advertising. Experiments demonstrate that Co-Director significantly outperforms state-of-the-art baselines, offering a principled approach that seamlessly generalizes to broader cinematic narratives. Project Page: https://co-director-agent.github.io/
format Preprint
id arxiv_https___arxiv_org_abs_2604_24842
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Co-Director: Agentic Generative Video Storytelling
Song, Yale
Song, Yiwen
Losier, Nick
Hodson, Nathan
Jin, Ye
Zhu, Rhyard
Xu, Yan
Vlasic, Daniel
Claassen, Carina
Leon, Jasmine
LeViet, Khanh G.
Chomyn, Zack
Timmons, Joe
Slatkin, Brett
Penberthy, Scott
Pfister, Tomas
Artificial Intelligence
Multiagent Systems
Multimedia
While diffusion models generate high-fidelity video clips, transforming them into coherent storytelling engines remains challenging. Current agentic pipelines automate this via chained modules but suffer from semantic drift and cascading failures due to independent, handcrafted prompting. We present Co-Director, a hierarchical multi-agent framework formalizing video storytelling as a global optimization problem. To ensure semantic coherence, we introduce hierarchical parameterization: a multi-armed bandit globally identifies promising creative directions, while a local multimodal self-refinement loop mitigates identity drift and ensures sequence-level consistency. This balances the exploration of novel narrative strategies with the exploitation of effective creative configurations. For evaluation, we introduce GenAD-Bench, a 400-scenario dataset of fictional products for personalized advertising. Experiments demonstrate that Co-Director significantly outperforms state-of-the-art baselines, offering a principled approach that seamlessly generalizes to broader cinematic narratives. Project Page: https://co-director-agent.github.io/
title Co-Director: Agentic Generative Video Storytelling
topic Artificial Intelligence
Multiagent Systems
Multimedia
url https://arxiv.org/abs/2604.24842