A rug photographed flat tells you its colour. A rug photographed in a room tells you whether you want it.
Nobody has ever bought a rug because of a photograph of a rug.
I mean that fairly literally. A rug shot flat, on a white background, from above, communicates a colour and a pattern and nothing else. It does not tell you how it sits under a sofa, what it does to the light in a room, whether the scale is right, or how it reads next to painted walls. Those are the things that decide the purchase, and none of them survive a product shot on white.
This is the problem with selling anything for a home. The object is not the point. The room the object makes is the point. And a room costs money to photograph.
The arithmetic that keeps small brands stuck
A styled room shoot means a location or a built set, a stylist, a photographer, furniture and props that are not yours, and a day. That is manageable for a hero piece. It stops being manageable the moment you have a catalogue.
Say you have forty pieces across three colourways. Photographing each one properly styled is a hundred and twenty setups. Then somebody asks whether the same rug would work in a modern apartment as well as a panelled dining room, and now it is two hundred and forty, because a room is a set and you cannot repaint it between shots.
So what actually happens is that small brands pick one aesthetic, shoot their five best pieces in it, and put everything else on white. Which means the majority of the catalogue is being sold with the one kind of image that does not sell it.
I have been on both sides of this. It is the single most expensive gap between a small studio and a large one, and it is not about talent.
What changed
You can now take a single existing product photograph and generate it inside a styled room, and the result holds up.
That is a narrower claim than it sounds, and the narrowness is the point. This is not about imagining products that do not exist yet, which is a different and genuinely interesting use I have written about separately. This is the unglamorous version: you already have the piece, you already have a decent flat photograph of it, and you need it to appear in eleven different rooms by Thursday.
An AI Image Generator built for this takes the source image and places it into a generated setting with the lighting, the scale and the surrounding furniture that make it read as a real space. The same rug can appear in a bright coastal room, a dark panelled study and a modern apartment, from one photograph, in an afternoon.
We use this at Kevin Francis Design. Our pieces are made to order, which means a great many of them have never physically existed in a styled room and never will, and generated room settings are how we show them at all. We have been public about the environmental cost of that choice and about what we do to offset it, which is written up on our GreenAI Initiative page rather than being something I will summarise here.
What actually matters when you try it
Most of the marketing around these tools talks about the wrong things. Having used them on real product for a while, three things decide whether output is usable.
Does the product survive the process
This is the whole ballgame. A generated room is worthless if the piece in it is no longer the piece. Colour has to hold, texture has to hold, and scale relative to the furniture around it has to be right.
Wool is a genuinely hard test. A hand-knotted pile has a depth and a slight irregularity that models love to smooth into carpet, and once that is gone you are photographing a different product. Check the fibre at full resolution before you trust anything else. A customer who receives a rug that does not match its image is a return, and worse than a return, a review.
Does it look like your brand or like the internet
Left alone, these models produce a consistent, pleasant, slightly anonymous aesthetic. Warm neutrals, mid-century legs, a fiddle leaf fig. It is fine, and it looks like everybody else's.
That is death for a brand whose entire proposition is a point of view. The New Regency thing I do is specific: classical proportion, saturated colour, layered pattern, a bit of gilt, a bit of levity. If a generated room comes back beige and tasteful, it has failed even though it is technically correct.
So the useful capability is control. Feeding in your own palette, holding a consistent styling direction across generations, and being willing to reject output that is competent but generic. That last part is a discipline rather than a feature, and it is where most brands give up.
Does it stay consistent across a whole catalogue
Forty products that each look like a different photographer shot them is worse than forty products on white, because now the inconsistency reads as carelessness rather than as a budget constraint. Consistency across the range matters more than any individual image being spectacular.
Where I still book a photographer
I want to be straight about this, because the vendor pitch never is.
Real photography still wins for anything that carries the brand rather than the product. Campaign imagery, editorial, a real installed room in a real client's house. There is a quality to a genuine space that has been lived in, with genuinely imperfect light and objects that arrived over years rather than in one generation, and I can tell the difference. So can anyone who looks properly.
What changed is not that photoshoots stopped being worth it. It is which images need one. The hero shots do. The long tail of the catalogue, the colourway variants, the seasonal refreshes, the third and fourth room setting for a piece that already sells, those never justified a shoot and mostly never got one. Now they can exist.
Practically, that means the photography budget goes further, because it is spent on the ten images that carry the brand instead of being spread thin across a hundred that merely have to exist.
If you are going to try it
Start with one product that already sells. You want a baseline, and a piece with no sales history tells you nothing about whether the imagery changed anything.
Generate two or three room settings, put them up against the current listing image, and watch what actually happens to engagement and conversion. Then test across materials, because results vary enormously by surface. A matte painted finish is easy. Glass, polished brass and deep-pile wool are not, and whatever settings worked for one will not transfer.
And look at every image at full size before it goes live. The failure mode is not a bad picture, which you would catch. It is a beautiful picture in which the product is subtly wrong, which you will not catch on a phone.
For anyone curious about how we think about this more broadly, including working with brands on it directly, there is more on our AI-directed campaign work.
The part worth remembering
The tool is not the interesting bit. The interesting bit is that showing a product in context stopped being a budget question.
For a category where nothing sells on its own merits, where a lamp on white tells a shopper nothing and a chair needs a room around it before anyone can want it, that is a genuine shift. It does not make a mediocre product look good. It just stops a good one from being invisible.
