Skip to main content
It looks like you're using Internet Explorer 11 or older. This website works best with modern browsers such as the latest versions of Chrome, Firefox, Safari, and Edge. If you continue with this browser, you may see unexpected results.

Storying Exhibition

by students of Contemporary Issues

Artist Name: Zhenhao Sun

Title: Training AI study Chinese calligraphy from Lantingji Xu ('Preface to the Poems Collected from the Orchid Pavilion') and Nebulas

Medium: Digital image study by AI and Google Images

Year: 2020


Artist Statement:

I chose to work on this short animation for the title "story" because I believe every single Chinese word is a story. From the most famous Chinese calligraphy piece "Lantingji xu", AI can be well-trained with the calligraphy basic, thus putting in the existing model of nebula images can enhance the accuracy of the new model and merge the two in its learning process. 

I was inspired by the artwork "a book from the sky" by Bing Xu. If we want to create a new word, certainly it is possible but the size of this work is extremely big. And also the idea of creating a new word to ordinary people is stupid. In today's artificial intelligence environment, recreating a new word is much easier. Although the word itself is still meaningless, it is still inviting people to give it a new meaning. 


However, this is the judgment made by artificial intelligence in the face of human complex information. It has a sense of awkwardness, but the simplicity of the machine has a feeling of Japanese haiku and Chinese ancient poetry, which is composed of some basic words. 
In China, many complicated things are summarized in four-character idioms, which is actually a classification. Chinese culture is actually a kind of symbolization. I think this work presents my understanding of my culture as well.


Technical process:

Using Runway ML, StyleGan2: Training an AI model with a dataset of images captured from one of the most famous Chinese calligraphy piece “兰亭序 (Lantingji Xu). While using an existing model of nebula images to enhance the accuracy of the new model and merging the two in its learning process. After the calligraphy model is trained, I go through randomly generated vectors of AI’s calligraphy to select and download the image code for my initial character. Finally, the initial image code is referenced through a network address and using StyleGan2 animation files, random calligraphy is generated remotely via the internet from the very first image code. To be more clear, all these words you can see from this animation are not real Chinese words, it is a new creation from Image Study by AI studying from Lantingji Xu and Nebulas pictures I found online.