Is it Really Me?

In this week's blog I look at the hot topic of AI headshots. Headshots are a challenge for a lot of individuals and companies alike. For individuals the cost of a professional photographer is prohibitive, and time-consuming. For companies, there's the challenge of constant attrition and new employee arrivals, multiple locations, and the need for consistency. One potential solution is AI. 

There are dozens of purveyors of this service, but five or six have risen to the top in terms of mainstream usage.  The drivers of popularity are combinations of slick and easy self-service software, the variety of backgrounds and apparel options, along with a reasonable price-point.  I looked at one such leader in the market, HeadshotPro (https://www.headshotpro.com/).  At $29 for 1 shot, $39 for 4, it's an approachable platform, with pricing and solutions for enterprise shots (we can't all be wearing the same shirt now, can we!).

HeadshotPro has a nice mobile UX in which you take about 15 selfies: against a white background, facing a window in daylight, and outside, each with variations of "smile", "look serious", "look left" etc.  Once done you can upload additional pictures in the form of additional selfies and/or photos from your camera reel.  The more pictures, the better the output apparently.

2 hours later your headshots are ready.  I had wondered why they needed to send 120 pictures to pick 4, but soon learned that this was no vanity issue a la selecting wedding pics.  100 or so were, how should I put it?... not me.  Nose wrong, wider chin, cleft chin, muscular arms (I wish!!!), overly saggy neck, or just a general feel of "not me".  For those of you who know me well enough to tell, here are some of the less accurate depictions:

Nothing against my somewhat more mature, long-necked, strange-handed, or buff lookalikes, but they are not me.

So it then requires some painstaking selecting to at least get an approximation of a true likeness. Clothing choices are nice, and certainly of a quality and newness lacking in my own wardrobe. And backgrounds vary from red brick wall to blue backdrop, cafe to Paris rooftop.

I'm curious to see how these systems handle enterprise-wide shoots.  Is their access to clothing, skin color, body types etc. sufficiently diverse? Do they ensure no double-ups on apparel? Can the backgrounds be sufficiently different while maintaining a specific corporate "look"? My assumption is probably yes, given how AI works and how it has almost limitless access to imagery.  

So, I will keep you in suspense no longer. After much deliberation, here are the winners:

Let me know what you think? Are they me? Drop me a note with your thoughts!