How To Generate Professional AI Headshots (and a Few Reasons To Think Twice Before Doing It): Hands-On

AI photo-enhancing technologies promise to make it easier than ever to get professional-looking photos–but can leave users wanting in execution. Two technologies in particular, Artisse and Remini, allow users to create highly detailed self portraits by uploading selfies and letting the AI generate new photos with the person as the focus. While these AI photographers can be useful for a new LinkedIn headshot, legal concerns and faulty images for some users could deter adoption.

Artisse’s claim to fame is its approval from high-profile celebrities, showcased in photos like this one featuring singer/songwriter Henry Lau.

Artisse’s CEO sees the technology as “revolutionizing professional domains like modeling, advertising, and journalism” by leveling the playing field for access to high-quality images. 

But the question of who these images would belong to remains unanswered. A U.S. court in Washington D.C. recently ruled that AI-generated art created without any human input cannot be copyrighted under U.S. law. The ruling is the first in the country over legal protections for AI-generated art, and the judge presiding over the case emphasized that copyright law depends on “human authorship.” 

While this ruling may not apply to AI-generated photos, it still sets a precedent for future rulings over who can claim the copyright to an AI-generated photograph. Copyright laws usually assign human photographers the exclusive right to their work, including displaying, reproducing, and distributing it. But with no photographer, it's unclear who would have the right to distribute AI generated photographs if given copyright.

Artisse collects, uses, stores and manages user data, and states that if users incorporate any of its images, designs, or filters, Artisse retains "all rights, title and ownership" to their content and any content derived from it. As stated above, it's unclear whether that's enforceable, but even if it's not, it does raise privacy concerns.

To test out Artisse, download the app on iOS or Android. The AI’s expertise lies in depicting photo-realistic faces in self-portraits from 15-30 selfies. 

Unfortunately, after I uploaded some images, Artisse got glitchy. The app promised a wait time of 30-40 minutes to train the AI. Over two hours after submitting photos, I was still looking at the same screen.

When I finally passed the training stage, I had the option to generate a photo based on a prompt or a preloaded photo, with up to 20 personalization factors.

I chose to generate a photo based on Artisse's library of pictures, but the end result wasn't quite me.

Remini loaded more quickly than Artisse, but to disappointing results. This AI photo-generating and enhancing app, also available on iOS and Android, requires credit card information before use. I committed to a free one-week trial, with a membership fee of $9.99 per week (over $520 a year) that would kick in automatically if I don’t cancel in time.  

Remini also needed at least 8 selfies. I uploaded 10. It took the app about 20 minutes to generate 6 images. None of the images would be able to pass for me on a professional platform.

After using two popular AI photo generators, my personal experience reflects that the technology might not be ready for everyone to use yet. This deficiency could reflect a lack of diversity among the training data. Taken in conjunction with the copyright issues that are still being worked out in courts, I think I'll stick to human photographers for now.