We present a method to create diffusion-based video models from pretrained Text-to-Image (T2I) models. Recently, AnimateDiff proposed freezing the T2I model while only training temporal layers. We advance this method by proposing a unique architecture, incorporating a mapping network and frame-wise tokens, tailored for video generation while maintaining the diversity and creativity of the original T2I model. Key innovations include novel loss functions for temporal smoothness and a mitigating gradient sampling technique, ensuring realistic and temporally consistent video generation despite limited public video data. We have successfully integrated video-specific inductive biases into the architecture and loss functions. Our method, built on the frozen StableDiffusion model, simplifies training processes and allows for seamless integration with off-the-shelf models like ControlNet and DreamBooth. project page: https://kwonminki.github.io/HARIVO
@article{arxiv.2410.07763,
title = {HARIVO: Harnessing Text-to-Image Models for Video Generation},
author = {Mingi Kwon and Seoung Wug Oh and Yang Zhou and Difan Liu and Joon-Young Lee and Haoran Cai and Baqiao Liu and Feng Liu and Youngjung Uh},
journal= {arXiv preprint arXiv:2410.07763},
year = {2024}
}