AI Interior · space reconstruction from a single photo · GIPTER
Every photo saved to your Pinterest board is someone else's home. AInterior takes one photo of the space you are in right now and shows you how it could look, with your window positions and structure left untouched. Whether you change only the décor, the furniture as well, or redraw the whole room is something you choose at the start.
AFTER · Full redesign
Why we built it
You have been thinking about redoing the living room for years, but when it comes time to start there is no obvious way in. You have saved hundreds of photos, yet none of them tell you whether that style suits your home, and when you ask for a quote you cannot explain what you want changed or how far. AInterior was built to close exactly that gap.
The photos on interior design forums and Pinterest are lovely, but they will not tell you whether that style works with your window positions, your sofa and your worn-out wallpaper.
Most services that transform a space from a photo hand back a result with walls torn out and windows moved. It looks good, but almost no one can actually build it.
Someone on a jeonse lease (a long-term deposit lease) who cannot do construction, someone with a budget of KRW 200,000–300,000, someone who just wants to swap the sofa — none of them even know where to ask, so they stay stuck.
If you have not decided what to change and how far, you cannot even pick up the phone. Decisions start with a result you can see.
Change scope · what makes it different
This is the most important choice in AInterior. Anything outside the scope you pick is preserved. Choose “furniture too” and the walls, floors, windows and doors, and fixed lighting are left alone, so you see a result where only the movable furniture has changed. Change scope is applied ahead of style, tone and color.
Swaps out fabrics and décor items: cushions, rugs, lamp shades, frames, plants. This is the range where a budget in the hundred-thousand-won range can still change the mood.
Swaps movable furniture such as sofas, tables and cabinets. It needs no construction, which makes it the scope renters choose most often.
Touches the walls, floors, lighting and finishes. The structure and window positions stay, so this is a result that can turn into a real construction quote.
Rethinks the composition and layout outright. Choose this when a move or a major remodel is coming up and you want to see the possibilities broadly.
All four images above were generated from the same single living room photo, changing nothing but the scope. The narrower the scope, the more of the original space remains.
One photo, three directions
These results were generated with the change scope fixed at “furniture too” and only the design style varied. The view through the window, the windows and doors, the walls and floors and the proportions of the room are all still there. Styles are chosen from Modern, Minimal, Natural, Classic, Industrial, Mid-century modern, Scandinavian, Japandi, Bohemian and Art Deco.
Six space types
Six space types are supported: living room, bedroom, kitchen, study, kids' room and office. Once you pick a space, the questions that follow are about what actually causes trouble there — storage, circulation, daylight, child safety — and your answers are reflected in the result.
Coming next
Once the office results hold up, we extend to commercial spaces such as cafés and shops, and to rental housing management. The office samples were our first check that the change scope idea proven in homes still works outside them.
How it works · 5 steps
There is no separate survey. The questions sit inside the flow you are already using, so by the time you have finished choosing, your requirements are laid out. Below are screens from the working prototype (PoC v7.7).
We show you a Before / After first. You see what you are going to get, then begin.
Take one yourself or pick from your gallery. If you have no photo, you can try it first with six example spaces.
Décor only, furniture too, finishes too, full redesign. This choice shapes the result more than any other.
Ten styles plus a regional feel, tone, wood tone and color, and any extra conditions. In a hurry? Skip ahead with “recommend one for me”.
Compare it side by side with the Before and say how much you like it. Watch an ad and you can generate another result.
What you can choose
You do not have to fill in everything. Set the change scope and the style and you already get a result; adjust the rest when you want something more precise. The options below are the ones actually implemented in the prototype.
Which space it is, and what you mainly use it for, changes the furniture and the way people move through it.
It is applied ahead of style, tone and color. Anything outside the scope is preserved.
We ask both “how would you like it to look” and “what bothers you right now”, and both feed into the result.
Ten styles combine with a regional feel. The same “Natural” gives a different result depending on whether it leans Korean or Japanese.
Pick a recommended combination and it is all set at once; pick “adjust it myself” and you can tune each item in detail.
Put in the realities of how you live. The free-text answers collected here are what decide the next features.
Partial revision
There is no need to throw the whole result away and start over. Name only the element you want changed and everything else is kept while it is redrawn. What people say at this moment is, in itself, the most accurate demand data there is.
What makes it different
It is not that the technology is special. The order of the questions is different. Instead of asking “what style would you like” first, we ask “how far can you actually change things” first.
Anything beyond the scope you chose is left untouched. Instead of a result with walls torn out and windows moved, you see a result you can actually carry out from where you are now.
Whether you are renting, what your budget is, whether construction is possible — these come in first. That opens up demand the existing market has never answered.
It does not have to be right the first time. Requests like “just brighten the lighting” or “keep the sofa” bring it closer to the picture you have in mind.
GIPTER is a company that has been reconstructing spaces in 3D from video and sensor data and measuring their dimensions. Transforming a photo is where that technology starts.
The change scope idea proven in living rooms applies unchanged to offices and commercial spaces. The space changes; the way the user decides does not.
A prototype implementing all 13 steps of the user flow is already up and running. Nineteen result samples confirmed that option combinations reproduce reliably.
Pricing
AInterior's first goal is not revenue. It is to find out what people find most uncomfortable about their own space. So basic generation is free and ads cover the cost. Paid features are added only where free users' behavior has shown a real need.
Daily generation limits and subscription prices will be fixed during the beta by comparing cost per user against ad revenue. We do not build paid features first and then push people toward them; only what free users ask for repeatedly becomes a paid feature.
Roadmap
Today it handles a single photo, but the goal is an AI spatial platform that understands real dimensions and redesigns accordingly. The 3D space reconstruction technology GIPTER has built up in its core business comes in at the final stage.
Who made it
GIPTER has spent years measuring spaces and facilities from video and sensor data and reconstructing them as 3D data. It runs national R&D projects that deal with real spaces, such as drone-based facility safety inspection, and AInterior is an attempt to bring that technology into ordinary people's spaces.
What we're really collecting
Building the prototype, we found something more important. The things people say when they see a result — “just change the sofa”, “we're renting, so we can't do construction”, “I'll only spend KRW 200,000” — are already the business.
We collect behavior data that no survey produces: what people actually click, how many times they regenerate. That data decides what we build next. Everything collected is anonymized and used only for analysis.
Get early access
The Before / After images on this page are samples generated in-house for the AInterior prototype (PoC v7.7) and used for validation; no external images were used. Results in the live service will vary with the photo you upload and the conditions you choose. Operating terms such as subscription prices and generation limits will be fixed after the beta.
We use Google Analytics cookies to understand how our website is used. With your consent, they help us see which pages are read, as anonymous statistics. They are not used for advertising. Privacy Policy
