I’m going to preface this post by saying first that real photographers are the first choice anyone should have, and there’s absolutely no way that AI can replace them. I’m also very skeptical of AI in general. I’m nervous about what it has access to and how much energy it consumes.

That said, I tried an AI headshot site a few months ago just to see what would happen. I don’t totally hate it somehow! It’s not perfect and it needed some tweaks to get something useable. Like all AI, it needs enough training to get the right answer.

On a whim, I tried an AI headshot service I saw others recommended. It’s called The Multiverse.AI. It was $29, what’s to lose? Plus, there was a 10% coupon so I paid $26.10. I’ve spent more than that on other frivolous things (such as weird domains I’ll never use).

To get started, I uploaded 12 pictures of myself that already existed on my phone. I’d just taken a few earlier in the day that I wasn’t excited about, but they represented what I look like. Twelve is the minimum number, they accept up to 50.

Once I uploaded the photos, the AI started working. After about 30 minutes, I received an email to download the results. There were 99 images generated in total.

A lot of the pictures were pretty close to what I look like, but some were way off. Some looked like Natalie Dormer had aged and put on some weight, and I’m very sure I don’t resemble her. Most of them vaguely look like a person who could be related to me. Others I’m afraid of and never want to meet. Looking at you, lower right. 😳

I grabbed about ten that were passable as me, then narrowed them down to my favorite. I found one I liked best, but there was a problem. All the images I uploaded were from more of an angle, so one side of my face didn’t look right behind my glasses. There was a chunk of my head missing.

I first tried a little fix in Photoshop, but I didn’t love it. Then I found the site’s AI tools to allow me to make changes on an image. I used the site’s tools to fix my head, zoom out, generate a slightly better outfit, and smooth the neckline after changing the outfit collar. It isn’t perfect, but I think it works!

Here is the final image that I’ve used for a few external Slack instances, zoomed out, head corrected, and shirt changed. It’s almost me, and me enough for other people looking at me in a tiny Slack avatar.

Could it be better? Yes, absolutely. I could have taken more pictures and fed them in so that the AI had more about me to go on. Additional angles of my face would have been helpful, I expect. Could I have tried more to fine-tune the results afterward? Or edited other results that were close but not quite right? Sure. But this was a on a whim thing that I didn’t want to spend a lot of time on.

I did try the site’s regeneration option after getting my initial results. You’re provided with credits to generate another small set of four headshots. It will use your existing photo data and ask you to enter a prompt like, “me with short bob haircut, slightly smiling, in front of a green background and wearing a blue blouse.” It was totally a fail. That person wasn’t me at all, far worse than the original set of people who weren’t me. But I do want those green glasses.

Do I recommend this service? Yeah, if you need a more professional picture in a pinch and don’t have the means or opportunity to hire an actual photographer, why not. With enough training and editing prompts, you can get something useable for a very low cost. Do I still recommend real headshots with a real photographer? Yep, 100%. There’s no way to replace the eye of a real person who is trained in their art.

Would you try this service? If you’ve tried something like this, how did it go?