The pitch is genuinely appealing. Upload fifteen photos from your phone, pay thirty-five dollars, and two hours later you have a hundred professional headshots without booking a photographer, taking an afternoon off, or standing in front of a grey backdrop being told to relax.
Sometimes that is exactly what happens. Often the result is a stranger in a suit: your face, technically, but with the skin of a mannequin and eyes that are looking at nothing. The difference between those two outcomes is largely under your control, and it has less to do with which tool you pick than the marketing suggests.
What AI headshots for real estate agents actually cost
Prices reported across 2026 comparisons, in US dollars:
| Tool | Price | Headshots | Turnaround |
|---|---|---|---|
| HeadshotPro | $29–$59 | 30–70 | 2 hours, faster on higher tiers |
| BetterPic | $35–$79 | 20–120 | About 2 hours |
| Aragon AI | $35–$75 | 40–100 | Not published |
| ProPhotos AI | $39–$59 | Not published | Not published |
| Secta Labs | Around $49 | Around 200 | About 2 hours |
| Canva | Free tier | 2 credits per 24h | Not published |
Compare that to a local photographer, who will typically run $200 to $500 for a session in most US markets, and the appeal is obvious.
But the sticker price is the wrong number. What matters is cost per usable headshot, and usable means you would put it on a for-sale sign. A hundred images at thirty-five dollars is worthless if two are usable and neither quite looks like you. Forty images where six are genuinely good is a bargain. Nobody advertises that ratio, because it depends almost entirely on what you uploaded.
Check the billing model before you pay. Some of these are one-time purchases and some quietly enrol you in a subscription, which is a distinction worth two minutes on the checkout page.
Why the results look plastic
Three failure modes account for almost everything people complain about.
Skin that has been smoothed into vinyl. The most reported problem, and the most recognisable. Pores, fine lines and the small asymmetries of a real face are what tell a viewer they are looking at a person. Remove them and something registers as wrong before the viewer can say why. Aragon in particular draws repeated complaints about plastic skin, though every tool does it at some settings.
Dead eyes. Catchlights, the small reflections of a light source in the pupil, are what make eyes look alive. Generated eyes often have none, or have them in inconsistent places between images. It is subtle and it is the single strongest tell.
Someone who is nearly you. A slightly different jaw, a nose that is not quite right, ears that changed shape. This is what happens when the uploaded photos were too similar, too few, or all taken in the same light: the model has not seen enough of your face to reconstruct it, so it fills the gaps with an average one.
That last one is the important one for agents. A headshot that is not quite you is worse than no headshot, because the client meets you at the door and something does not match.
The upload rules that decide the outcome
Every one of these tools says the same thing in its documentation and almost nobody reads it. Realism depends far more on your input photos than on which product you bought.
- Vary everything. Different days, different lighting, different clothes, different angles, different backgrounds. Fifteen photos from one afternoon in one shirt gives the model one version of you.
- Face the camera, mostly. A majority looking straight ahead, some three-quarter, a couple in profile. Not all one thing.
- Nothing on your face. No sunglasses, no hats, hair off your forehead in at least half of them.
- Recent. Photos from four years and two haircuts ago pull the result away from how you look now.
- Alone in frame. Cropping someone else out of a group shot leaves artefacts that end up in the output.
- Neutral expressions too. Upload some where you are not smiling. If every input is a broad smile, every output is, and you will want one that is not.
- Good light, not perfect light. Daylight near a window beats a harsh overhead bulb. Phone photos are fine.
Twenty minutes gathering the right photos changes the result more than paying for the premium tier.
How to judge the ones you get back
You will be handed sixty images and your instinct will be to pick the most flattering. That is the wrong criterion. Work through these instead:
- View it at sign size. Shrink it to the size of a face on a yard sign and look at it from across the room. Most tells vanish at that scale, which is where most people will see it.
- Then view it at full size. Because it will also appear on your website at 400 pixels, where the vinyl skin shows.
- Check the hands and the collar. If hands are in frame, look hard. Collars, lapels, buttons and jewellery are where generation errors survive.
- Zoom into the eyes. Are there catchlights, and are they in the same place in both eyes.
- Show it to someone who knows you. Not asking whether it looks good. Asking whether it looks like you. The answer arrives before they finish thinking about it.
If two or three survive that, you have what you needed. You were never going to use sixty.
When to just hire a photographer
An honest answer, given that this article could easily have avoided the question.
Use AI when you need something acceptable this week, your current photo is genuinely old, you are testing how a different look reads before committing, or you need variations for different channels and cannot rebook a session for each.
Hire a human when this photo is going on signs and vehicles for the next three years, when you work in a market where clients will be looking closely, or when you have tried the AI route and nothing came back that looked like you. Three hundred dollars amortised over three years is eight dollars a month, which is less than the assistant subscription in your monthly tool budget.
The middle path most working agents land on: a real session every few years for the primary photo, AI in between for social profiles, seasonal variations and the headshot on the newsletter.
The disclosure question
A headshot is not listing media, so the image disclosure rules covered in the piece on MLS and state requirements do not apply to it in the same way. California’s AB 723 governs images of the property, not of you.
The standard that does apply is simpler and older: your marketing must not be misleading. A generated photo that looks like you, dressed how you dress, is your photo. A generated photo of a noticeably younger, thinner or different-looking person is a misrepresentation of the professional the client thinks they are hiring, and no rulebook is needed to see the problem with that.
Some brokerages have their own policy on generated imagery in agent marketing. Worth a two-line email before you update every profile.
If your existing photo is fine apart from what is behind you, a dedicated generator is overkill: Canva’s background replacement does that job for nothing.
A photograph is half of a profile. The other half is the paragraph next to it, and there is a method for that in the piece on writing an agent bio with AI.
Common questions
How many photos do I need to upload?
Most tools ask for somewhere between ten and twenty. Give them the upper end, and make them genuinely different from each other. Twenty varied photos beat forty similar ones every time.
Can I use the free Canva tier?
To find out whether this approach works for your face, yes, and it is the sensible first move. The realism ceiling is lower than the dedicated tools, so treat it as a test rather than a source of your final image. Two credits every 24 hours is enough to learn something.
Will it work if I wear glasses?
Usually, with more failures. Frames confuse generation and you will see warped arms, missing hinges and reflections that make no sense. Upload plenty of photos in the glasses you actually wear, and budget for a lower usable ratio.
How often should I redo it?
Whenever the photo stops matching the person who walks into the listing appointment. In practice that is every two or three years, sooner after a significant change in how you look. The cheapness of AI makes it tempting to redo it constantly, but recognisability is worth something too.
Sources
- Published 2026 comparisons of AI headshot generators, covering pricing, output volume and turnaround.
- Vendor documentation on input photo requirements.
Prices in US dollars, compiled from 2026 comparison data and checked in August 2026. Verify on the vendor’s own page before purchasing, and check whether you are buying once or subscribing.