Generative AI for 2D Character Animation

Google Research
TL;DR: In this pilot project, we teamed up with artists to develop new workflows for 2D animation while producing a short educational cartoon. We identified several workflows to streamline the animation process, bringing the artists' vision to the screen more effectively.

Introduction

We produced a short educational cartoon using novel generative AI workflows. Educational videos are under-served relative to entertainment videos, and educators don't generally have the skill or talent to produce animations that teach real concepts. We adopted a 2D style because it can be more forgiving of the kind of geometric inconsistencies that generative AI sometimes produces, which limit the perceived quality of realistic or 3D-style imagery. Our narrative occurs in a fantasy world with some non-human characters, giving further flexibility for variable appearance.


Workflows

Previous generative animation projects have focused on automating story simulation and visual design, but in this work, our goal is to accelerate the process of character animation. We identified several promising workflows, bringing artists' vision to the screen more effectively.


Episode Scripts


Acknowledgements

Thanks to the Not-So-Supervillains creative and production crew: Jason Mayland, Kelly McNutt, Woei Lee, Gabe Liu, David Andrews, Eloise Fassler, Kavitha Gopalakrishnan, HOPR. Voice talent: Kory Mathewson, Warren Reid, Kiara Lee, Andru Anderson. Brain trust: Kory Mathewson, Cassidy Curtis, Alonso Martinez, Anna Kipnis. PM: Thomas Iljic.

BibTeX


    @misc{guajardo2024,
      title={Generative AI for 2D Character Animation},
      author={Jaime Guajardo and Ozgun Bursalioglu and Dan B Goldman},
      eprint={2405.11098},
      archivePrefix={arXiv},
      primaryClass={cs.GR},
      year={2024}
    }