The scanner has drawn a wall that isn't there — what do I do?
Reflections, mirrors and furniture close to a wall can trick a 3D scan into drawing surfaces that aren't really there. The fix is rarely to scan again — it's to remove the wrong surface from the drawing afterwards.
Short answer: yes, it happens, and it isn't unusual. A 3D scan builds geometry from whatever the sensor reads as a surface, and it can be wrong — about a mirror, a strongly reflective window, a sharp shadow edge from a piece of furniture. The fix is usually to remove the faulty surface in a later edit, not to rescan the whole room. The scan data behind the correction is never touched — a correction is laid on top of it.
Why does a scanner "see" something that isn't there?
A LiDAR sensor measures distance by sending out light and timing how long it takes to return. Most surfaces reflect diffusely — the light scatters in every direction, and the sensor gets a reliable number. A mirror, or a very glossy surface, does something else: it can send the light onward in one particular direction instead of scattering it, and the sensor can end up measuring the distance to whatever the mirror is reflecting, rather than the distance to the mirror itself. The result is a surface that appears to stand where there is really just air — or a wall that looks twice as deep as it actually is.
The most common causes of a misread surface
- Mirrors. The classic example. A bathroom mirror or a mirrored wardrobe door can be drawn as an entire new room behind the wall.
- Glossy, dark surfaces. A glossy black kitchen worktop, dark glass, certain screens — low reflected signal gives unreliable readings.
- Strong backlight through a window. Sunlight shining straight into the sensor can add noise to the readings near the window.
- Furniture close to a wall. A bookcase or a cupboard that almost touches the wall can, in some cases, be read as part of the wall or create a false edge.
- Very narrow recesses and projections. Sharp, thin geometry around door frames and corners is where automatic recognition most often struggles.
How do you spot the error?
Look through the drawing room by room after a scan, before you trust it. A typical sign is a wall that doesn't match the rest of the room's shape — a kink where there should otherwise be a straight line, or a small extra "box" somewhere that makes no structural sense. It helps to compare against where you yourself know mirrors and glossy surfaces stood in the room while you scanned. That's where to look first.
Do you fix a scan by rescanning, or by editing?
Both are possible, but they solve different problems. Rescanning the room is the right call when the error is large and systematic — for example if the whole room's shape is distorted because you walked the wrong route around it, or because the light was too strong throughout the pass. Editing afterwards is right when the error is local and well-defined — one wrong surface around a mirror, while the rest of the room is measured correctly. Rescanning an entire room to fix a single surface is rarely worth the time when a targeted correction solves the same thing in a minute.
Does a correction destroy the original scan?
No, not if the app is built the right way. A sound approach stores corrections as their own layer on top of the raw scan, rather than overwriting the scan data itself. That means you can always see what the scanner originally measured, and a removed surface is never permanently deleted from the system — it's marked as removed in the drawing you keep working on. If you're unsure how a particular app handles this, it's a fair question to ask before you trust corrections on an important project: "do you keep the original scan, or does the correction overwrite it?"
What about a door or opening that shouldn't be there?
The same principle applies to doors and openings, not just entire wall surfaces. The sensor can misread a deep shadow or an open sliding door as an opening in the wall, even where the wall is actually solid. Just as with a false wall surface, the fix is to remove the wrong opening from the drawing, not to rescan the whole room for a single detail.
How much does a single wrong surface matter for the overall measurements?
It depends on what the area measurement is used for. If the wall with the false surface belongs to a room where you'll later need a takeoff for paint or flooring, an error of even 20-30 cm in a wall's position can shift the area noticeably, because the error typically drags an entire line off with it — not just a small corner. If you're going to send the drawing to a contractor for a quote, it's therefore worth fixing that kind of error before you send the drawing on, rather than trusting the recipient to spot and correct for a surface that doesn't actually exist.
Can you prevent it next time you scan?
- Cover large mirrors with a towel or similar, if you can, before scanning the room.
- Avoid strong backlight. Don't scan straight towards a window with low sun coming through.
- Tidy up a little around furniture that almost touches the wall, if that's practical — it gives the sensor a cleaner edge to measure.
- Move calmly and steadily, especially close to corners and recesses, where the geometry is most complex.
None of these precautions guarantee an error-free scan. They just reduce how much you need to fix afterwards.
What HouseSense does
In HouseSense you can remove a misread wall, door or opening in edit mode — tap the surface and choose "Remove". The original scan data isn't touched; the correction sits as a layer on top, so you can always go back and see what was originally measured if you change your mind later or want to compare. The same goes if the scanner has mixed up a door and a window, or missed an opening entirely — both are fixed without rescanning the room.
How to approach it, step by step
- Go through the whole drawing before you trust it — ideally while you're still in the home and can check it against reality.
- Note where you had mirrors, glossy surfaces and strong backlight while scanning. That's usually where the errors sit.
- Remove local errors by editing. Save rescanning for the cases where an entire room's shape is wrong.
- Check that the app keeps the original scan, so a correction is never a guess with no way back.