Every lab scanner on the market publishes an accuracy figure in microns. Almost every one of them is measured the same way: a certified calibration artefact, a controlled scan path, ideal surface, ideal light. The number is real. It is also the easiest scan that machine will ever perform.
What the figure actually measures
Quoted accuracy is usually trueness on a single, geometrically simple body. It tells you the sensor and the calibration are sound. It does not tell you how the scanner performs when it has to stitch dozens of individual captures into one arch, which is where error accumulates.
A scanner with excellent single die trueness can still produce a full arch that is a few hundred microns out at the terminal molar, because stitching error compounds along the arch. That is the error a lab actually feels, and it is not on the datasheet.
The cases that separate scanners
Three case types expose the difference. Deep undercuts, where the geometry occludes the camera and the software has to interpolate. Highly polished or wet surfaces, where reflection defeats structured light. And full arch with an articulator, where the machine has to relate two models in space rather than capture one accurately.
A scanner that handles all three is a different class of machine to one that scores well on a die.
What to ask for instead of microns
Ask whether it does dynamic articulator scanning, not just static. Static tells you where the models sit at one position. Dynamic captures the movement, which is what you need for anything occlusally demanding.
Ask about camera count and resolution — dual 5 megapixel cameras capture undercuts a single camera geometry cannot reach. Ask whether the manufacturer validates the machine on dentures and full arch specifically, or only on crown and bridge.
Then ask for a scan of your own worst model. Any serious supplier will do this.
Trueness and precision are not the same thing
Two words get used interchangeably in scanner marketing and they mean different things. Trueness is how close a measurement is to the real dimension. Precision is how repeatable it is.
A scanner can be highly precise and consistently wrong, which is the worst combination because nothing in your workflow will flag it. Every scan agrees with every other scan, so the operator has no reason to suspect anything, and the error appears at seating on every case equally.
When a supplier quotes a single micron figure, ask which one it is. If they cannot answer immediately, the number is marketing.
A test you can run in an afternoon
Take one model you know fits. Scan it 5 times, removing and replacing it on the stage between each scan. Superimpose the 5 files.
Tight agreement means good precision. Now compare the average against a physical measurement of a known dimension on that model — a caliper reading across two defined points is enough. Agreement there means trueness.
Then do the same with your worst full-arch model. The gap between the two results is the number the datasheet never gives you, and it is the one that predicts remakes.
Software is half the machine
Scan quality is not only optics. The stitching algorithm decides how individual captures are aligned, and how aggressively it smooths where data is thin.
Aggressive smoothing produces beautiful-looking scans with invented surface. Conservative handling produces scans with visible holes, which is uglier and far more honest, because you can see where to rescan.
Ask which behaviour the software defaults to, and whether you can turn smoothing down. A scanner that hides its gaps is harder to work with than one that shows them.
Go deeper
Scanner comparison and specifications: intraoral scanners compared for 2026.
The UP3D UP1000 runs dual 5 megapixel cameras at up to 5 micron precision, with both dynamic and static articulator scanning, and is validated for dentures and full arch rather than crown and bridge alone.
Pricing
Request a Quote — 954-874-6325
The UP3D UP1000 desktop lab scanner is quoted per configuration — voltage, software, extraction and freight all change the number.