A photo of a lived-in interior loses its boxes, laundry and dishes, then catches up on the light. The flaws of the property stay visible. The full workflow, no account needed.
Real workflow, real results. Pan and zoom freely.
Use this workflowThe real workflow: the lived-in living room is emptied of what can be tidied away, the light is corrected, and the crack and the damp stain are still there.
At a glance
Nobody photographs a tidy home. There are boxes against the wall, laundry on the sofa, dishes on the coffee table, cables on the floor. This workflow removes all of it in one pass: the photo enters the Media node, an Image node empties it of its movable objects, item by item as named in the instruction.
The second pass handles the light. A badly exposed interior, with dark corners and the yellow cast of a ceiling bulb, is corrected the way a photographer would in post-production: shadows lifted, white balance neutralised, daylight recovered at the window. Nothing is added or moved, only the exposure changes.
What separates this workflow from a beautifier is what it refuses to do. The crack, the damp stain and the flaking wall stay visible: left free, the model repaints and fills, and the viewing turns into a disappointment. Seedream 5 Pro is chosen here precisely because it keeps the flaws, where a more obliging model erases them along the way. A third render shows the same room after the works, and that one is a projection, not a photo of the property. All three for under a quarter of a dollar.
Photograph the room as it is
Without tidying, at chest height, in daylight. The mess is not a problem, it is the raw material of the workflow; what matters is that the framing shows the volume and the window.
Name what has to go
The instruction lists the objects one by one: boxes, laundry, dishes, magazines, shoes, cables, posters. An object that is not named stays put. The second half locks the flaws of the property, do not delete it.
Correct the light
The second node no longer touches the scene, only the exposure and the white balance. That is what takes a listing photo from amateur to acceptable, without inventing anything.
Check what stayed
Compare the render with the original right next to it: the crack, the stain and the worn floorboards must still be there. If they are gone, the instruction is too weak and needs hardening, not publishing.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The direct use case: an occupied home that has to be photographed before the move. The boxes go, the laundry too, and the buyer sees the volume instead of the seller's belongings. The flaws stay, and that is what keeps the listing standing at the viewing.
Rental platforms penalise cluttered photos without saying so: they read badly as thumbnails. Tidying the photo is often enough to change the click rate, without touching the property itself.
A piece of furniture photographed in a lived-in room shows badly. The instruction is rewritten around the object: remove everything except the chest of drawers. Same gesture, applied to one subject rather than a whole room.
The second pass can be used alone, removing nothing. A badly exposed interior, blocked corners and a yellow cast, is corrected as in post-production, without anything moving in the scene. Useful on a photo that is already tidy.
Decluttering is the step before virtual staging. Empty first, furnish second: a model asked to add on top of existing belongings returns a room twice as cluttered.
Nothing ties this workflow to homes. A retail unit full of stock, a desk covered in files, a workshop: the principle does not change, you name what can be tidied and lock what needs repairing.
That is the part listing the flaws to keep, one by one, and it reads like padding. Without it the model repaints the wall, fills the crack and erases the damp stain, which turns a tidy photo into a lying one.
The model only removes what you name. A power strip, a pinned poster or a pair of shoes left off the list will stay in the render, and stands out all the more because everything else is clean.
The light correction comes after the tidying. Fixing the exposure first, then emptying the room, drops the shadows in the wrong place: the model recalculates the light on a scene that has since changed.
The third render repairs and furnishes: it is a projection of the works, not a photo of the property. It belongs in a listing when labelled as such, next to the real photo, never in its place.
Tidying a photo is, hiding a flaw is not, and that is exactly the line this workflow draws. It removes what can be tidied away, never what needs repairing: a box leaves with the move, a crack stays after the sale.
Because on an interior, Flux Kontext beautifies on its own: it empties the room cleanly but repaints the wall along the way and erases the damp stain. Measured twice on the demo living room. It stays the right model for the third render, the one allowed to repair.
It will if you name it, but furniture is not an object you tidy away: removing it changes what the photo shows. For a room you want to show furnished differently, the furnish an empty room workflow starts from a bare room and says so.
Under a quarter of a dollar for all three renders. The price shows on each node before you click, and the first two passes alone, the ones that tidy, cost around a dime.
Yes, the canvas works on mobile and the photo is taken straight into the Media node, so a room can be handled on the spot. The creating on mobile guide covers the canvas gestures by finger.
No. Your media stays in your workspace, it is not handed to any training run. The question matters all the more here, since this is the inside of your home.
The sale listing photo handles an object for sale, a car or a piece of furniture, and produces a cover image. This one handles a whole room and its light. Both share the same rule: you tidy, you do not hide.
That is the promise of the workflow and the canvas lets you check it at a glance, with the original next to the render. In the demo, the crack on the left wall, the damp stain on the ceiling and the flaking paint by the switch are present in the first two images.
Declutter a room with AI
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.
Seedream 5 Pro
Flux Kontext
Nano Banana