Two years ago, AI image generation was so bad you didn't even need the six fingers to give it away. You could just tell. So I stopped using it in early 2024.

But then 2 weeks ago, OpenAI released GPT Image 2.0, their best image generation model yet. And people on X started turning photos of themselves into MS Paint drawings. I had to try it.

This week I tried… creating an MS Paint version of my wife and I using ChatGPT

My first impression of GPT Image 2.0 is just: wow.

From what I tested, it's way better at following specific instructions, rendering text, and creating images that actually look real. It’s not perfect, but noticeably better.

The MS Paint trend on X is the perfect way to see what it can do.

X post demonstrating how a portrait can be recreated as a deliberately scribbly MS Paint drawing.

I used the prompt from the tweet and attached a picture of Grace (my wife) and I to see how it would look.

Tyler and Grace standing together outside beneath trees.AI-generated MS Paint-style drawing of Tyler and Grace, based on their photo.

And another of Grace being Grace:

Grace posing among colourful tulips.AI-generated MS Paint-style drawing of Grace among tulips.

I’m impressed that it even got her ring. If I’m being honest, I don’t even think I can draw that badly, that well.

I'm planning to start using ChatGPT a lot more for visuals for my GradSimple blog posts. I'll keep you posted on how that goes.

Now go try the prompt yourself:

  • Redraw the attached image in the most clumsy, scribbly, and utterly pathetic way possible. Use a white background, and make it look like it was drawn in MS Paint with a mouse. It should be vaguely similar but also not really, kind of matching but also off in a confusing, awkward way, with that low-quality pixel-by-pixel feel that really emphasizes how ridiculously bad it is. Actually, you know what, whatever, just draw it however you want.