Swing and Miss

Until I have taken enough photos to supplement my blog posts, I'm finding ChatGPT has been a good source for creating images that work to satisfy my needs. But it doesn't always work out as planned.

For example, there have been a few successful illustrations that I've made from this AI tool: the writer at a computer; me, reading a book, or taking a photo; and even, to some extent, of me kayaking on a river (my face wasn't replicated very well in that one so I opted to cover it with sunglasses).

For Monday's post, I wanted to take an existing photo of me on my spin bike. I took the picture a couple of years ago and have used it in previous blog posts. But now that I've restarted my exercise routine, I wanted to add the image again, only I wanted to create a 3-D cartoon version of it.

So, I turned to Chat GPT and wrote

Create a 3-D cartoon image from the attached photo. Do not include the cords on the floor.

Here's what it produced:

I may have gained a few extra kilos over the winter but I'm not as big as the generated person. And I'm nowhere near like that in the provided photo (why does ChatGPT always make the cartoon version of me bigger than I am??). So, I requested an alteration:

Make the man much less fat.

Okay, that was much better. But the face of the character doesn't look like me, so I uploaded another photo and wrote

Use the attached photo to change the face of the man on the spin bike.

It took several minutes for ChatGPT to produce an image, and here's the result:

Again, that looks nothing like me. Three strikes, and I was out.

Maybe ChatGPT is giving me a subtle hint that I should lay off using it to convert selfie photos into cartoons. DW has hated all of the previous images: "What's wrong with the source photos? I like seeing your real face."

What do you think? Should I give ChatGPT a rest when it comes to generating images, especially of myself? Should I stick to asking it questions that can help with crafting my novel?

Let me know in the Comments section.