I enjoy experimenting with ChatGPT and similar AI tools. Some are for productivity – like writing code. But there’s also some fun to be had.
ChatGPT’s new 4o image generation model is leaps and bounds better than previous versions. For one, its text rendering capabilities are vastly improved. The images are also more consistent and realistic. It’s very cool (and scary for a society already teetering on edge).
I wanted to see what this new model could do, so I uploaded my headshot photo (below left) and asked AI to create different versions. I also threw in a selfie with my dog for fun. Here’s what happened:
The results were…interesting! The image-generation model is adept at animation styles. I love the Simpsons-inspired image. The more realistic requests became a bit creepy, though. Asking ChatGPT to change my hair and add tattoos also impacted my facial features. That’s not quite what I wanted.
For my next trick, I wanted to create a comic strip based on my photo. I instructed ChatGPT to create the comic and described the contents for each panel.
The results were solid. I had to make a few edits, but ChatGPT got the gist of my instructions.
On the downside, notice the facial inconsistencies between the two comic strips. Be aware of this limitation if you want to use the tool for branding. Even so, it’s much better than I could do by hand or in Photoshop. After all, I struggle to draw stick figures!
The technology isn’t perfect. However, it’s worlds ahead of where it was just a few months ago. I shudder to think of where we’ll be by the end of 2025.