Skip to content

Feature

Last updated

See the colour on your wall, not on a card

A paint chip lies. It is two inches square, held under shop lighting, next to nothing it will actually sit beside. Put the colour on the real wall instead.

Change a wall colour

One surface

walls change, everything else stays

Your light

rendered in the room's actual lighting

30 seconds

per colour, so compare several

Choosing paint from a card is guesswork, because a colour is not a fixed thing. It shifts with the light in the room, the direction the window faces, the floor it sits above, and the furniture in front of it. The same grey reads cool and institutional in a north-facing room and warm and soft in a south-facing one. Changing the wall colour from a photo of your own room removes the guess: the colour is rendered on your wall, under your light, next to your things.

What this mode does

Input
A room photo with the wall clearly visible.
What changes
Wall paint, wallpaper, paneling, plaster, tile, brick, or another surface finish.
What stays
Furniture, flooring, lighting, architecture, and camera position.
Use another mode when
Use Restyle when the whole room feels unresolved, and a tester patch for the final paint choice.

Why a paint card misleads

A colour has no fixed appearance. What you perceive is the light the surface reflects, which means the same paint genuinely looks different depending on what is illuminating it and what surrounds it. A north-facing room in the United Kingdom receives cool, indirect light for most of the day, and greys mixed with blue read colder and flatter there than the card suggested. The same shade in a south-facing room with warm afternoon sun can look almost beige.

Surrounding surfaces compound this. A wall above a warm oak floor picks up reflected warmth from it. A wall opposite a large green garden picks up green. Colour theorists call this simultaneous contrast, and it is why a shade that looked perfect in the shop looks wrong at home even though nothing about the paint changed. A two-inch card held in your hand shows you none of these interactions, because it is surrounded by your hand.

Rendering the colour onto a photograph of the actual wall puts all of it back: the room's light direction and temperature, the floor reflecting up, the furniture sitting against it, and the scale of the surface. Scale matters more than people expect. A colour that reads as a pleasant accent at card size can read as overwhelming across an entire wall, and there is no way to anticipate that from a sample.

Wall colour, or a full restyle?

Use the focused mode when the walls are the question. Restyling a room gives the model licence to change the furniture, the floor, the finishes, and the lighting all at once, and if you were only trying to choose a paint colour you now have six variables moving and no way to isolate the one you cared about. The narrower instruction is not merely more polite, it is more useful.

It is also more reliable. A single-surface instruction has fewer ways to go wrong, and checking the result takes seconds: you look at the wall. Auditing a full restyle for whether it also quietly changed the flooring takes longer and is easier to get wrong. When a client has approved everything except the colour, regenerating the entire room to change one surface risks losing the parts they signed off on.

The case for a full restyle is when the colour is a symptom rather than the problem. If the room feels dated overall, changing the wall shade will not fix it, and you will work through six colours discovering that. Restyle first to find the direction, then use a focused pass to settle the exact shade within it.

What you can and cannot ask for

Anything that is a treatment of the wall surface is fair game. Paint in any colour, wallpaper, limewash, microcement, tongue-and-groove or shaker panelling, exposed brick, tile, and timber cladding are all changes to the finish rather than to the building. So is a feature wall: changing one wall while leaving the others is a normal request and generally works well.

What is declined is anything structural. Adding a window, enlarging one, removing a wall, or opening the space up are requests to alter the building, and they are rejected rather than half-attempted. That constraint is deliberate and it is what makes the output trustworthy: if the tool would move a window on request, you could not trust that it had not moved one when you did not ask.

Two practical limits are worth knowing. Very dark shades on a wall the photograph has underexposed give the model little to work with, so shoot with the lights on. And a wall largely hidden behind a tall bookcase will only change where it is visible, because the model renders the surface it can see rather than the surface it infers. If you want to judge a colour properly, photograph the wall with as much of it in view as you can manage.

Taking the colour from the render to the tin

The render settles a direction, not a product. Brand names are blocked in the prompt deliberately, because naming a product would imply a match the model cannot promise: it renders a colour you described, not a mixed and catalogued paint. So describe the shade and let the image answer the question you actually had, which is whether a deep blue-green belongs on that wall in that light. Which tin delivers it is the next question, and it is answered in a shop rather than here.

Take the render with you and hold cards against it. That inverts the usual order, where you pick a card first and discover the room later, and it narrows a wall of two hundred nearly identical greys to the three worth carrying home. What no render will tell you is sheen, because a flat image cannot show you how much light a surface throws back. Matt swallows light and hides a bad wall. Eggshell and satin return it, which lifts the colour a shade and shows every dent in the plaster. The same colour in two finishes reads as two colours, and hallways and kitchens usually need the washable one whatever the render suggested.

The tester patch still earns its place, and the render makes it cheaper rather than unnecessary. Paint the shortlist onto lining paper rather than straight onto the wall, so you can move each one around the room and look at it against the floor, beside the window, and on the darkest corner. Check it in the morning and again under the lamps at night. The render takes you from a hundred candidates to three; the patch takes you from three to one. Doing it in that order is what saves the money, not skipping the second step.

Why change wall color works

The room stays the room

A focused mode touches the walls and leaves the furniture, floor, and lighting alone, so you are judging the colour rather than a whole new scheme.

Colour in context

A shade next to your floor and your sofa is a different proposition from the same shade on a card. This is the comparison that actually decides it.

Cheaper than tester pots

Four testers, four patches on the wall, and a fortnight of looking at them costs more and tells you less than four generated versions side by side.

Settle a disagreement quickly

Two people arguing about a colour are usually imagining different rooms. Seeing both on the actual wall ends it.

Finishes as well as colour

Panelling, plaster, limewash, wallpaper, and exposed brick are all wall treatments, not just paint.

Feature walls

Change one wall rather than all of them, which is the usual question when someone is considering a darker shade.

How it works

  1. 1

    Upload the room

    Shoot the wall you are thinking about, in the light it normally gets. Daylight and lights on.

  2. 2

    Name the colour or finish

    Describe it plainly: deep olive green, warm off-white, tongue-and-groove panelling painted charcoal.

  3. 3

    Compare a few

    Run three shades and put them side by side. This is the comparison a paint card cannot give you.

Related tools

Go deeper

Frequently asked questions

Ready to try it on a real room?

Change a wall colour