How a failed Photoshop session made me realize there's a real market for AI photo merging
A few months ago I had a great group photo from a trip — except one friend was missing. She'd left early that day.
I opened Photoshop to try adding her in from another photo. Twenty minutes of fighting with selection masks later, I gave up. The edges looked terrible and I didn't have the patience to learn the right workflow just for one casual photo.
What bothered me wasn't the result — it was that this felt like a completely solvable problem that nobody had made easy yet.
So I started googling
I typed in things like "how to add someone to a group photo" and "merge two photos together." What I found wasn't a gap in tools — it was just how many people had the exact same frustration.
Reddit threads, Quora questions, Facebook groups. People asking how to add a family member who missed the reunion, how to fix a group shot where someone blinked, how to put their kid into a photo with a character from a trip they couldn't afford.
The demand was clearly there. I figured I'd build something and see if people would actually use it.
Who actually needs this?
Once I started paying attention, the use cases multiplied:
- Family reunion where grandma couldn't make it
- Product shots without a studio budget
- Fitting clothing on your body before buying
- Group photos ruined by bad timing
Most of these people are not designers. They're not going to learn Photoshop. But they'd happily spend 2 minutes typing a prompt if it meant getting a usable result.
Building it
I built aiimagecombiner.app to test the idea. The core interaction is simple: upload 2–5 photos, add an optional prompt describing what you want, pick an aspect ratio, generate. No masking, no layers.
Getting the AI blending to look natural was the hard part. Straight image stitching looks fake immediately — the lighting never matches. The prompt-guided generation approach helped a lot with this.
Stack ended up being Next.js + Cloudflare Workers + Claude Code for moving fast.
What I learned about the niche
The positioning matters more than the feature list. People searching for this aren't thinking "AI image compositing." They're thinking "how do I put two photos together." The language gap between how builders describe tools and how users search for them is real.
Still early days. Curious if anyone else has explored this space or has thoughts on the demand side.
The part I'd be careful with is that you may have discovered multiple businesses at once.
"Add a missing family member to a photo" is a very different buying moment from product photography, virtual try-ons, or fixing group shots.
The risk isn't lack of demand. The risk is proving demand across several use cases and ending up with positioning that feels broad to everyone and urgent to nobody.
I wouldn't rush to solve that in a comment, but I do think there's a real first-buyer decision hiding underneath the search-intent insight.
That’s a really good point — I hadn’t framed it that way before. I’ll need to think more about the positioning. Thanks for the thoughtful feedback.
I think that's the right thing to think about.
The dangerous outcome isn't picking the wrong use case. It's trying to validate several at once and ending up with signals that are hard to interpret.
Curious where you land after you've sat with it for a bit.