DynamiCrafter

by CUHK and Tencent

About the model:

(Text-)Image-to-Video model (16 frames).

Model type:

Video

Model performance class:

Limited

Link to the model:

https://huggingface.co/Doubiiu/DynamiCrafter_1024

Base models:

Stable-Diffusion-2-Base

End model:

DynamiCrafter-1024

End model license:

(undefined)

About the organisation:

Chinese University of Hong Kong, an educational institution, and Tencent, a Chinese conglomerate.

Link to the organisation:

[ "https://www.cuhk.edu.hk/english/index.html", "https://ailab.tencent.com/ailab/en/index/" ]

Model release date:

February 2024

Availability

Base Model Data

Trained on a subset of LAION-5B, an openly available dataset.

https://laion.ai/blog/laion-5b/

End User Model Data

WebVid-10M mentioned in paper

Base Model Weights

Weights available through HuggingFace.

https://huggingface.co/stabilityai/stable-diffusion-2-base

End User Model Weights

https://huggingface.co/Doubiiu/DynamiCrafter_1024/tree/main

Training Code

https://github.com/Doubiiu/DynamiCrafter

Documentation

Code Documentation

https://github.com/Doubiiu/DynamiCrafter

Hardware Architecture

Preprint

https://arxiv.org/abs/2310.12190

Paper

https://dl.acm.org/doi/abs/10.1007/978-3-031-72952-2_23

Modelcard

https://huggingface.co/Doubiiu/DynamiCrafter_1024

Datasheet

Data collection procedures for LAION are outlined in a corresponding paper. No datasheet found for end model.

https://arxiv.org/pdf/2210.08402

Access

Licenses

unspecified

Is this information not up to date?

Contribute here ->

Supported by the Centre for Language Studies and the Dutch Research Council. Website design & development © 2024 by BSTN. This version of the index generated 09 April 2026, website content last updated 11 March 2026.