How to Generate Astonishing Animations with ChatGPT

New vision-based generative models are continually improving, with software such as Midjourney, Leonardo AI, and more recently DALL·E 3 generating incredible artworks. But there is actually another, simple, yet elegant way to quickly generate visualizations that are animated or interactive with large language models (LLMs) such as GPT-4.

Don’t believe me? Take a look at the effects I generated in just a few minutes while chatting with ChatGPT:

Additionally, the thumbnail was generated with ChatGPT.

How, you ask? The answer might surprise you.

Just ask it to generate HTML and JavaScript effects

If you are not familiar with HTML or JavaScript, don’t worry, you can still generate effects like this. If you are familiar with them, on the other hand, it might surprise you that these two basic components of web development could be such well-suited tools for ChatGPT and similar AI tools to generate effects.

The reason is simple. In a single HTML file, you can easily generate a self-contained example utilizing JavaScript, HTML canvas graphics, and even CSS.

The main obstacle for LLMs to do things on their own is the lack of programmatic interfaces to programs. In other words, ChatGPT and other apps can guide you through various steps to perform tasks, but ideally, if you want to generate a complete solution, you should be able to do so in a single file containing text.