3D Scanners are set to be a growth industry, that according to Resarch and Markets, is set to reach around $4.08 billion in 2018, with a compound annual growth rate of 14.6% from 2013 to 2018. Source.
The reason for all this growth lies in the wonderous technical capabilities of 3D scanners; with the soon to be advent of personal 3D scanning, as shown in Google’s latest smartphone product, or in the Microsoft Kinect, we could see a future where we all have a 3D camera in our pockets, and with personal 3D printers, we could soon be duplicating any part of our physical world in plastic, at-whim.
For those who are looking to purchase a 3D scanner for home or business use, or are just curious about the different options available to modern professionals for digitizing physical data, we have compiled this informational list along with some purchasing considerations. Many of these scanners can also be built at home for much cheaper than an off the shelf system, so if you’re handy, you could be scanning at home on the cheap. So grab a cup of coffee, sit back, and enjoy!
Structured Light Scanners
These use striped patterns of light to detect objects, with two major methods of creating these patterns being laser interference and projection. The former uses two planar laser beams to create an interference pattern of regular, equidistant line patterns. This method is very useful for creating laser patterns with unlimited depth of field, meaning objects can be as far away or as close to the lasers as the power of the lasers will allow. However, the high precision required for this setup leads to very expensive systems, meaning they are out of reach for the majority of the consumer market.
The less expensive and arguably more common (in the consumer space) SL scanner operates off the projection principle, whereby what is essentially a standard video projector beams a set of vertically alternating black/white lines onto an object. One or two cameras observe the displacement of any of the stripes, or any variation in their width, to derive 3D point data.
- Better used in low-light environments
- Can be built at home using kits
- Very fast – some systems can even scan objects in motion, and are extremely accurate compared to laser scanning
These are very common in the consumer market, and come in several flavors, ranging from large, tripod mounted units for gathering data over large distances, to small hand held ones that can measure the texture on a person’s face. They operate on different principles (acousto-optic deflectors, vertical-cavity surface-emitting laser, etc), but the two most common ones that you will actually encounter primarily operate on time of flight and triangulation, which are relatively simple (the vertical-cavity-whatever is used by NASA to align docking ports on the space shuttle).
While time of flight can be used for short-range scanning, it is typically used when scanning long distances, such as the inside of the Colosseum or the outside of a mansion. It works similar to a laser rangefinder; a laser pulse is fired from the scanner, and the time it takes to make its round-trip is computed to find the distance of the point in 3D space. Taken all together, these points create a 3 dimensional point cloud that can be used to reconstruct a large object. The NextEngine pictured above uses triangulation.
Triangulation is used in handheld scanners, and operates by firing a laser at an object and recording it’s position using it’s location in the field of view of a camera. Since trig is not my specialty, I present to you Wikipedia’s explanation of the triangulation;
“The length of one side of the triangle, the distance between the camera and the laser emitter is known. The angle of the laser emitter corner is also known. The angle of the camera corner can be determined by looking at the location of the laser dot in the camera’s field of view. These three pieces of information fully determine the shape and size of the triangle and gives the location of the laser dot corner of the triangle. “
To speed up the process, a planar laser (aka a laser stripe) is used to acquire multiple points at a time.
- Can be used in virtually any indoor lighting condition
- Laser light can be damaging to eyes, do not stare into the beam
- Can scan parts of any material
- Great depth resolution (divots, cracks, etc)
This one is pretty cool; multiple photos are stitched together using feature recognition algorithms to create a 3D model. This method is potentially the fastest scanner out there, depending on the setup. Companies like Infinite Realities use photogrammetry setups with multiple (50-120) cameras to take instant snapshots of humans to produce extremely accurate 3D models. Other companies, like Google, use photogrammetry to stitch together aerial photos and create topographical 3D models of the Earth. However, if you only have one camera, it can be pretty slow, as you have to walk around the object and take many pictures to get a good result. But, it’s pretty cheap – all you need is some software and a camera and you’re good to go. Depending on the quality of both, you can get great or mediocre results.
Photogrammetry uses feature-based algorithms to identify unique features on a part and stitch a model together based on how those features change from photo to photo. This means that if you’re trying to make a 3D model of a flat surface, it’s better to have more clutter on it, as this gives those algorithms more features (corners, rounds, protrusions, etc) to pick up on.
- Either one camera with many pictures, or many cameras with one picture.
- Easiest to start with – grab a camera and some free software and go.
- Make sure whatever you’re scanning has identifiable features. If you’re having trouble getting good results (say, scanning a toy on a flat table), add more objects to increase the amount of available features.
- Works great with aerial photos!
Contact probes use an arm or other sort of moveable mechanical carriage with a probe on the end. This probe travels along the surface of the object, using the placement of the mechanical linkages in the arm to feed coordinate data back to the computer. Some of these machines can be automated, others are operated by hand; users hold a pen-like tip on the end of a mechanical arm, and trace lines across a surface to build coordinate data.
Since this type of scanning method is extremely accurate, it (along with others), is often used for inspecting manufactured parts for accuracy, to compare them to the CAD model. These devices are called CMMs, or Coordinate Measuring Machines. Contact scanners also trace lines as opposed to scanning in surfaces, so many of these lines can be brought directly into a CAD package as editable curves – and have functions for creating reference planes and finding hole centers via geometric inferences.
The biggest drawbacks to these scanners are their limited scanning ability; as they can only digitize where the probe touches, they are not suitable for scanning surfaces of objects – those surfaces must be reconstructed using the scanned curves. Additionally, another major drawback is of course the necessity for the scanner to actually touch the surface of the object. If this is not possible or contact would damage the object, these scanners are useless – or worse. Lastly, these scanners are very slow – dragging the point across the surface can be a very slow and painstaking process.
- Very accurate, capable of creating editable curves and geometrically inferred details in CAD packages.
- Some older systems can be relatively cheap.
- Very accurate
- Must contact the work surface
So if you’re looking at purchasing a scanner to start your own scanning business or just play around, there a couple factors you should take into account;
Portability - Can to bring this scanner to the objects, or will you have to bring the objects to the scanner? If it’s the former, a handheld laser scanner would be a great idea. If it’s the latter, then a full-on photogrammetry setup would be ideal.
Accuracy - This is directly related with cost, but how accurate do you really need the scanner to be? .1mm would be fine for most projects – if you’re just reverse engineering parts, you may need less accuracy than if you’re doing manufacturing inspection.*
Speed - What will you be scanning? Is it something that can sit still for awhile (like a bicycle), or not (like an organic object)? Also, factor in how much time it might take you to stitch together the scan’s results – it might be worth it to pick up a live-stitching scanner, like the Artec Eva, over something that takes multiple separate scans that you then have to stitch together after the fact, like the Next Engine.
Cost - Unfortunately there’s no two ways around this – while scanner prices are set to come down, the more you pay, typically, the better results you’ll get. Still, you can weigh the different options against each other and choose a solution that best fits your budgetary needs as well as your operational ones.
*A note on accuracy: When you read a scanner’s spec sheet, the two biggest factors when it comes to accuracy are going to be the “Resolution” and the “Point Accuracy”. “Resolution” is a measure of how small an object it can read. For instance; .5mm means that any object smaller than .5mm will not register. “Point Accuracy” is a measure of how accurate the registered scan data points are; so a measurement of .1mm means that any point that was picked up could be out of place by .1mm up or down. This will usually mean your scans have a “waviness” or “roughness” to the surface that is not present in the actual part.