Tips for accurate phone measurements

The checklist

If you read nothing else, run through this before a measurement that matters:

1. Lights on, curtains open.
2. Wipe the lens.
3. Calibrate slowly for ten seconds over a textured floor.
4. Measure something of known size first.
5. Aim at real edges and corners, never at glass, mirrors or blank wall.
6. Stay within a couple of metres of what you are measuring.
7. Hold the phone with two hands and move slowly.
8. Measure everything twice; keep the pair that agree.
9. Split long spans into overlapping sections.
10. Confirm anything you will cut, order or build with a tape or laser.

The rest of this guide explains why each of those works, so you can adapt when a room does not cooperate.

Light

The camera is doing the tracking, and a camera in dim light produces a noisy, blurred image with few usable features. Tracking then drifts more and calibrates worse. Turn on every light in the room and open the curtains. If the only light is a window behind you, your own shadow will fall across what you are measuring; move so the light comes from the side. Avoid low direct sun into the lens, which overexposes part of the frame and hides features in the glare. How AR measurement works explains what the camera is looking for.

Surfaces to avoid

Camera-based tracking needs visual detail, and a few surfaces provide none or, worse, provide false detail:

• Glass. The camera sees through it and a depth sensor may see through it too. A tap on a pane lands on whatever is behind. Tap the frame.
• Mirrors. A mirror shows the app a second room that is not there. Keep mirrors out of frame during calibration, and never place a point on one.
• Plain white walls. No features, no tracked points, no plane. Keep the skirting line, a socket or a picture in view so the app has something to hold on to, or measure along the floor edge instead of across the wall.
• Glossy and reflective floors. A polished tile reflects the ceiling lights as moving highlights, which the tracker can mistake for features. A rug on the floor fixes it.
• Repeating patterns. Identical tiles or a regular grid can fool the tracker into matching the wrong tile. Include something irregular in the frame.

Phones with a LiDAR or time-of-flight sensor are far more tolerant of plain walls and dim light, but glass and mirrors are still a problem for them. Which iPhones and iPads have LiDAR covers what the sensor does and does not change.

Hold steady, and stay close

Hold the phone in both hands with your elbows in. Move it slowly and smoothly; fast movement blurs the image and drops tracked features, and a sudden swing is the most reliable way to lose tracking altogether. When you need to turn to see the other end of a wall, turn your whole body rather than flicking the phone.

Distance matters twice. A point placed far from the phone carries more error, because a small angular error at the camera becomes a bigger positional error at the target. And a long measurement means more movement between the two taps, which means more drift. Stay within a couple of metres of what you are measuring where you can, and for a long wall, measure it in two or three overlapping sections from a fixed feature such as a door frame and add them up.

Start and end on real edges

The app places a point where your tap meets a detected surface. A tap in the middle of a wall relies entirely on the app’s model of where the wall is; a tap on a corner, a frame edge or the line where the wall meets the floor lands on something the tracker can actually see. Corners and edges are also where you want the measurement to start and end anyway. Aim for the bottom corner rather than the top one, because the light is usually better and the skirting line gives the tracker a strong feature.

Take your time over each tap. Bring the reticle onto the edge, hold the phone still for a moment so the app can settle, and then tap. A tap taken while the phone is still moving lands wherever the model was at that instant.

Measure twice, and use a reference

A reference object also tells you which kind of error you have. If the app reads the two-metre tape as 2.06 m, the session has a scale problem from a rushed calibration, and every measurement in it will be proportionally out. Restart and calibrate slowly. If it reads 2.00 m and a wall still comes out differently on two attempts, the wall measurement is drifting; measure it in shorter sections.

The phone itself

• Clean the lens. A fingerprint on the lens softens every frame, and a soft frame has fewer features. This is the first thing to check when a phone that used to measure well suddenly does not. Wipe it with a cloth, not a sleeve.
• Mind the case. A case with a thick lip around the camera can vignette the frame, and a cheap lens protector can add haze. If measurements are consistently poor, try without.
• Battery and heat. AR runs the camera, the sensors and the processor hard. A phone that is hot, or in a low-power mode, may reduce its frame rate and processing, and tracking quality drops with it. Do not measure a whole house on a phone at 5%, and if the phone is warm to the touch, give it a minute in the shade before an important measurement.
• Close other apps. Anything else using the camera or heavy processing in the background competes with the tracker.

When precision needs a laser

A well-taken phone measurement is typically within a centimetre or two over a short distance in good conditions. That is the right tool for a floor plan, a furniture check, a listing, a first quote or a rough quantity. It is the wrong tool when the number goes straight into a cut or an order.

For those, use a tape or a laser distance meter: a laser measures a single point directly and is precise over spans a phone can only estimate. Use a physical tool for worktops and fitted furniture, made-to-measure blinds and curtains, flooring and tiles cut to fit, anything structural or load-bearing, anything to do with gas, electrics or fire safety, and any purchase where being a centimetre out means sending it back. The phone gets you to the plan quickly; a physical measurement is what you sign off on. TritonMeasure keeps both together, with room to note which measurements have been confirmed by tape, so the record you export says what was checked. How accurate is AR measurement goes deeper into where the line falls.

Frequently asked questions

Why are my phone measurements inconsistent?

Usually one of four things: a rushed calibration that got the scale wrong, poor light or plain surfaces that starve the tracker of features, fast movement that lost tracking, or a dirty lens. Fix the light and the lens, calibrate slowly over a textured floor, measure a known object first, and measure everything twice.

How far away should I stand to measure with my phone?

Close enough to see both ends of the measurement clearly without walking, and ideally within a couple of metres of the target. Error grows with distance from the phone to the point and with the amount of movement between taps, so split anything longer than four or five metres into overlapping sections.

Does a phone case affect AR measurements?

It can. A thick lip around the camera can cut into the frame, and a hazy lens protector softens the image, both of which reduce the features the tracker has to work with. If measurements are consistently poor, try the same measurement without the case.

Is a laser measure more accurate than a phone?

For a single distance, yes. A laser distance meter measures the point directly rather than estimating it from a tracked model, and it stays precise over spans where a phone accumulates drift. Use a phone for planning and layout, and a laser or tape for anything you will cut, order or build to.

Estimates, not certificates: every phone-based AR measurement is an estimate, TritonMeasure included. Verify anything safety-critical, structural or expensive with a physical tape or laser before you cut, order or sign.