“Kandinsky As You Preferred” by Le-Zhou, Wang, Wu and Zhang
Conference:
Type(s):
Title:
- Kandinsky As You Preferred
Session/Category Title: Art & Design
Presenter(s)/Author(s):
Abstract:
?Kandinsky As You Preferred? introduces a large text-to-image model personalization method – Semantic Injection. Through working with a Kandinsky expert, the authors establish a semantic descriptive guideline and a text-to-image dataset of the Kandinsky Bauhaus style and apply the Semantic Injection method to obtain an Artist Model, empowering users to create preferred Kandinsky content in a deterministically controllable manner.
References:
[1]
Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. LoRA: Low-Rank Adaptation of Large Language Models. arXiv:2106.09685 (Oct. 2021). http://arxiv.org/abs/2106.09685 arXiv:2106.09685 [cs].
[2]
Wassily Kandinsky. 1977. Concerning the Spiritual in Art. Dover Publications, Inc.
[3]
Wassily Kandinsky. 1979. Point and Line to Plane. Courier Corporation.
[4]
Jon McCormack, Oliver Bown, Alan Dorin, Jonathan McCabe, Gordon Monro, and Mitchell Whitelaw. 2014. Ten Questions Concerning Generative Computer Art. Leonardo 47, 2 (April 2014), 135?141. https://doi.org/10.1162/LEON_a_00533
[5]
Kang Zhang and Jinhui Yu. 2013. Generating abstract paintings in kandinsky style. In SIGGRAPH Asia 2013 Art Gallery. 1?6.
[6]
Kang Zhang and Jinhui Yu. 2016. Generation of Kandinsky art. Leonardo 49, 1 (2016), 48?54.