It started with someone else's release.
OpenAI shipped a new image model and within hours my timeline was full of generations. Anime portraits, ghibli stills, product mockups, the same handful of genres the internet always converges on when a new model drops. I joined in and made my share of all of them.
Then I started wondering what people weren't using it for yet.
The interesting space at any new model launch is the unclaimed corner. Everyone makes the obvious thing. The opportunity sits in the second click weirder than the obvious one. So I went looking.
The first prompt
I uploaded a photo of myself, then a photo of an outfit I liked but didn't own. The prompt was crude, something close to "wear me this outfit", and I clicked generate without expecting much.
The output came back wearing it.
Not "wearing it" in the funhouse mirror, six fingered, plastic skinned way most early try-on demos did. Wearing it like the photo had been taken that way. The body was mine. The fit was right. The lighting was consistent with the outfit's source photo. It was, technically speaking, a finished image.
I tried another outfit. Same result. A third. Still good.
The friend share
I sent the outputs to a few friends. Not as "look what I built", but as "look what I prompted." They came back asking for the prompt. I sent it. They tried it on themselves. Then they sent me theirs.
That is the moment, looking back, that decides whether something becomes a project or stays a screenshot. Friends were excited. Not polite excited. "Send me the prompt, can I do it with my girlfriend's outfit" excited.
But they were excited about a prompt. Most of them were never going to copy paste model parameters into ChatGPT every time they wanted to see themselves in something. The friction killed it before they used it twice.
So the actual idea was small and obvious. Take the prompt, hide it behind a button, give people the button.
That was StudioPic v0. A button on top of a prompt that worked.
The early framing
I shipped it framed as a virtual try-on. The pitch on the homepage read "Wear any thing." with the last word in a vermillion italic accent. The supporting copy talked about uploading a photo of yourself, dropping in any outfit, getting the picture you wished existed.
This was honest copy. It described exactly what the product did mechanically. Photo in, outfit in, image out. Try-on.
I shared it with my contacts. Some were friends. Some were tailors. Some were people who had seen my generations online and wanted to play. I waited to see what they would do with it.
What they actually said
Within a few days the messages back to me had a shape I didn't expect.
"Photographers are in trouble 🤣", a friend wrote, unprompted, no clarification, sent after generating two photos.
"Yuppp surprised how it pulled this off from a random selfie", from another, the same week.
"Sweet. Works perfectly fine."
"This be cool guy, Nice Boss... Maintained body size But if you zoom in a bit it didn't fully maintain my facial features maybe because the picture was far shaaa but this is very legit Nice Nice one."
Nobody was telling me they were trying clothes on. They were telling me their phone had stopped being a phone and started being a studio. The word "photographer" kept appearing, never about a person, always about a category. As in, the category of person you used to need.
And then, from a tailor I had shared it with:
"So I go dey use am test cloths before I sew them."
(Pidgin, roughly: I'm going to use this to test clothes before I sew them.)
Try-on existed in the data. That was in fact exactly that user's use case. But it was the second place finisher. First place was a different request entirely. Make me a picture I didn't have to book a photographer for.
What I had actually built
In Lagos, a studio photoshoot runs ₦40,000 to ₦80,000 for an hour. Add the makeup, the wardrobe coordination, the half day off work, the second outfit if you want variety, the prints. It is a Saturday and a month of budget decision.
For most people, most of the time, the answer is no. Not because they don't want a nice photo. Because the friction is enormous and the trigger ("I'd really like an updated portrait of myself") is small. The result is that most people don't have a single decent picture of themselves taken in the last three years, except by accident at a wedding.
What I had built, by accident, while testing a new model in the unclaimed corner, replaced every line in that quote except the smile. Backdrop, handled. Lighting, handled. The photographer's eye for composition, encoded into the model. The outfit, pulled from any catalog photo. The cost, under five dollars a month for thirty of them. The turnaround, about thirty seconds.
I had not realized I was building this. I was looking at the wrong dashboard.
The pivot
The day I rewrote the homepage took an afternoon. The day I admitted to myself what I had been building took longer.
"Wear any thing." came down. The new headline went up:
Two words. A period. The whole product strategy follows from that period. The portfolio became a portfolio of portraits, not outfit experiments. The pricing page stopped competing with try-on apps and started competing with a Saturday at a studio. The voice across every screen rewrote itself overnight to point at the outcome (a finished portrait) instead of the mechanism (outfit transfer).
The product code didn't change. The frame did.
The honest part
I am not going to claim AI photography has replaced photographers. It has not, and it will not. A great photographer reads your face, finds the angle that flatters your jawline, makes you laugh until you forget the camera is there. The output of an hour with someone who is good at this is qualitatively different from anything you can summon out of a model. If you can afford one and the moment matters, you should hire one.
What AI replaces is the median photoshoot. The one most people don't book because they cannot justify the cost or the day. Anniversary portraits people meant to schedule and never did. Holiday cards that became selfies. The headshot that is still from 2019 because updating it is a project. The tailor's mockup before the cloth gets cut.
That is not a small market. That is almost everyone.
The photographer near you was never your competition. Avoidance was. The thousands of decisions every year where someone considered a portrait, looked at the price, and quietly decided to live another year without one.
What StudioPic is now
Solo or family. Pick a look. Send a photo of yourself. We make the studio shot in about thirty seconds, in your browser. The free tier gives you one watermarked photo a day so you can prove to yourself the thing actually works on your face. The Pro tier (₦5,000 or $9 a month, depending on where you are paying from) lifts the watermark, raises the cap to thirty photos a month, and unlocks 4K downloads. Print quality. Framed on the wall quality.
That is the entire product. Two flows, two tiers, one promise.
What I learned
Pivots usually look obvious looking back. From inside, they look like noise. A few users using the thing wrong. A quote that sounds like a joke. A tailor mentioning he is going to use it for a thing you didn't build it for. Paying attention to the noise long enough to realize it is the signal is the actual job.
I had been telling people I was building a try-on tool.
I had been replacing my photographer.
I just didn't know it yet.
If you have been meaning to update your portrait, your headshot, your family card, try it. studiopic.art. First photo is free. Or browse what other people have made first.
Top comments (0)