Choose a 3D Scanner for Your Mapping Project

The right way to choose a 3D scanner comes down to one question: what am I projecting onto, and how close does the audience stand. Brand, range, points per second, all of it follows from that. A 3D scanner captures a surface as a point cloud, you turn that into a model, and you prepare your mapping on the model. The right scanner is the one whose accuracy matches your real need. Rarely the one topping this month's comparison chart.

I have been surveying surfaces for projection for 15 years. Facades, museum interiors, statues, stage sets. The "which 3D scanner" question comes back on every slightly awkward project, and the answer changes every time. Here is the decision tree I run on site, before I even look at a spec sheet.

Useful accuracy first, maximum accuracy never

Most guides rank scanners by accuracy, finest to most expensive. Wrong end of the stick. A pixel projected from 30 metres away easily measures 1 to 2 cm on the wall. Surveying that wall to a tenth of a millimetre buys you nothing: the projector will never resolve that detail. You would pay for precision your image cannot use.

My field rule: scan accuracy should stay under the size of one pixel on the surface. For architectural mapping seen from the street, a centimetre-level survey is plenty. For a statue viewed from a metre away, where every fold of drapery counts, you need to drop below the millimetre. Between the two, the projection distance decides.

This ties straight into up-front sizing: the target pixel size is worked out from the viewing distance, exactly like the projector count for an installation. Lock that pixel size first. It tells you which scanner accuracy to aim for, and therefore which family of machine.

Three families of scanner, three jobs

There are broadly three ways to capture a surface in 3D. They do not play in the same league.

The fixed laser scanner, for spaces and facades. A tripod that sweeps the space around it and returns a dense point cloud in minutes. The Leica BLK360 reaches about 60 metres and gives roughly 6 to 8 mm accuracy at 10-20 metres (Leica Geosystems spec sheet, checked 2026-07-09). The FARO Focus Premium pushes to 200-350 metres of range for a stated accuracy around 1 to 2 mm at 10 metres (FARO, same date). This is the tool for building, room and monument surveys. You set it up at several stations, register the clouds together, and get the full geometry of a place. To scan an interior or a facade before a mapping job, it is the default answer.

Structured light, for objects and sculptures. Here the scanner projects a light pattern onto the object and reads its deformation to rebuild the relief. Artec scanners comfortably drop below a tenth of a millimetre (Artec3D, how to choose a 3D scanner, 2026-07-09). It is the right pick when you map a statue, a bust, a carved set piece, anything you can handle. Short range, huge accuracy, colour capture. Forget it for surveying a whole square, though: you would be there all night.

Photogrammetry, when the hardware budget is zero. No scanner at all: a DSLR, or a drone for the high stuff, and software that rebuilds the 3D from dozens of photos. Low hardware cost, medium accuracy that swings with light and image overlap (Formlabs calls it the "least expensive and least accurate method", Formlabs guide, 2026-07-09). It is often the first reflex for a facade when no laser is on hand. The real work happens in the software, and I compared the options in the dedicated photogrammetry software guide.

A quick mental grid to sort it fast:

• place, room, facade, range beyond 10 m: fixed laser scanner
• object, statue, set piece under 2-3 m, fine detail: structured light
• zero hardware budget, medium accuracy acceptable, plenty of post-processing time: photogrammetry

Rent, buy, or subcontract

Picking the technology does not yet tell you how to get your hands on it. A pro laser scanner runs into tens of thousands. Three routes, and I rank them by how often you scan.

Subcontract. For a one-off, this is almost always the smart call. A surveyor or a scanning service shows up, surveys, and hands you a clean, registered point cloud. You pay for the survey, not the machine or the learning curve. On a one-shot mapping, you do not want to be learning a BLK360 during build week.

Rent. If you string two or three surveys across the year and you can already drive the thing, renting a fixed laser makes sense. You keep control of the site work without tying up the capital. Just check the registration software is included and that you can use it before the day.

Buy. Only worth it if 3D surveying becomes a real line of your business, several times a month. Below that, the machine sleeps in a flight case and depreciates. An entry-level structured-light scanner stays more affordable and can pay back faster if you regularly map objects.

When you need no scanner at all

The "I'll scan it" reflex costs time and money, and it is often pointless. Cases where I tell a client to put the scanner away:

• A flat, regular surface. A straight wall, a screen, a stretched sheet: a laser distance meter and a tape measure do it. Three readings, a rectangle, done. Scanning a plane to get a plane is wasted energy.
• Usable plans already exist. Plenty of recent buildings have DWG drawings or a BIM model. Starting from those is more accurate and faster than a scan, especially for simple volumes. Ask before you unpack the gear.
• The geometry is easy to model by hand. A cube, a cylinder, an arch: ten minutes in a 3D app beats a million-vertex point cloud you then have to clean. Scanning shines on the irregular, not the geometric.
• You are after an approximate projection. If the content is abstract and forgives a slight offset, a rough survey passes. Save precision scanning for mappings that hug the relief fold by fold.

The scanner is for the organic, the weathered heritage piece, the carved set, the tortured facade. Not for what a tape measure reads in two minutes.

Once the scan is done

The raw point cloud is not a working model yet. You often need to lighten and retopologise it so a media server can digest it, then load it into your prep chain. The clean model imports into Lumeo to place projectors in a 3D scene, check coverage and overlaps, and validate the look before a single machine is rented. The survey is the first brick. On-site projector calibration is the last, and it hinges directly on how good your scan was. For the full chain, from sensor choice to mapping, start from the 3D scanning for projection guide. And if you are stuck between three machines with a quote due, send me the context: the projection distance and the detail you need usually settle it in five minutes.