Final Project Documentation – A Storybook of The Three Body Problem

Project Link

https://github.com/Ellen10155/storybook-

Concept

The concept of this project is to test AI painting and prompt engineering to see their limitations. All of the visual images in this project were generated by AI using DALLE-2, for not only its convenience, but also as an experiment to test how AI could generate paintings that are beyond human imaginations. The prompts are marked on the lower right corner of the images.

Experiment

To think about this project as an experiment, the result that I got was pretty satisfying. Because the cartoon version of The Three Body Problem just came out, and I compared them with the images generated by AI, I think AI-generated paintings looked a bit better aesthetically. However, because the cartoon only released two episodes so far, and hasn’t released the part about the Trisolaris planet (the part that I focused on in my project), I was only able to have a rough comparisons.

Cartoon:

AI:

However, the limitations of the AI was very obvious too. It lacks the ability to produce complicated or detailed prompts, so the information shown in each images are much less than each frames from the cartoon.

Coding:

Since I chose to focus on the visual part and the prompt engineering part of the project, the coding part of this project was very simple. The only challenge that I had was to learn how to use HTML, CSS with Javascript, since I do not really have any experience on this before. I was confused a lot by what sort of function should be utilized in which, and I’m still confused now, because it seems like many functions, for example, to set the canvas, can be both written in HTML and Javascript, but not both.

User Testing

I only got one person to user test my project, and he thinks that even though the visual images are cool, my project looks like a power point slides. So to strengthen and polish the UI design is the next big step that I’m aiming for.

Future Steps

First, I’ll need to complete the story, which is hard, because it’s such a long and detailed book. But if I can do it, it would be really meaningful. Second, I want to add more interactive elements, including some machine learning models to make this storybook more like an interactive game book that is friendly to children. Third, as I mentioned above, to strengthen and polish the UI design.