Not a member yet? Register for full benefits!

A 3D Scanner for which No Object is Too Big or Too Small

3D scanners are wonderful devices. Whereas the 3D printer may be considered the normal way to get a 3D construct out of a virtual environment and into the physical world, a 3D scanner is the way to get a pre-existing physical object in.

Another facet such scanners have in common with their 3D printing brethren is how primitive they are; how early in the development of the technology we still are. Typically your 3D scanner must be of roughly the same dimensions as the object you are scanning, with the 3D scanner being the larger object – able to fit your scanning target completely inside.

There are exceptions to this rule, but since a 3D scanner has to literally scan every surface and side of a given object, and from every direction, most demand that the object is inserted inside them. Most of those that are the exception tend to resort to greater sophistication in error correction as they'll be carried round the object by a human in most cases, again so they can scan all sides of the object. The remaining exceptions simply don't scan all sides of the object at all: they simply take a photograph from their perspective and extrapolate the 3D shape from lighting and shadow in the resulting image. The non-visible sides of the object having to be put in by hand when it is scanned into the virtual environment.

Further, each type of 3D scanner generally has an intended purpose. There is a target market it was built for, and that target market demands a certain resolution. It would be silly to say the least, to try to scan a car for example, with a 3D scanner designed to scan microbes. Not only does the increased resolution demand an increased sensor density that jacks up the price, but it would scan the car in absurdly small increments and produce a massive file as a result.

Conversely trying to scan a microbe with a 3D scanner designed for small hand held objects is not going to be helpful. The scanner lacks the fidelity to even see the microbes in the first place, and will simply image the slide.

What would be ideal would be a 3D scanner that doesn't have these limitations. Something modular that could be expanded and shrunk as necessary, and something with a variable sensor fidelity. Able to scan in greater detail when it is required, and scan things not visible with the naked eye in some areas of the object, then scan other areas with a much lower resolution, decreasing the scanning time and filesize of the resultant 3D model in areas where such high detail simply is not required.

Such a scanner would be wonderful, and would truly be a game changer for every field where a physical object such as a body, vehicle, even a building needs to be scanned and then worked on. Whether to look for injuries, defects, signs of disease or stress fractures, or simply to take any random physical object and quickly turn it into a model for use in a 3D world.

Of course such a scanner still doesn't exist. Not completely. But, this article would not of course exist, if somebody had not at least given it a go.

Enter the Fraunhofer Institute for Integrated Circuits in Erlangen, Germany. They approached this problem from a fairly novel set of circumstances that in itself exemplified the problem with current 3D scanners, and whilst their problem was specific, there are countless other issues with 3D scanners that their solution would benefit.

One aspect of vehicle safety testing for any design is to crash-test it. Send it at high-speed into a wall or another vehicle in a testing facility. Once this is done, whether the collision is head on, side on, or at an angle, the resulting wreck must be investigated. Visual inspection alone is not enough; every component's position both before and after the crash must be investigated. What moved? Why did it move? If we look at where it moved to, we can determine how much of the impact force it absorbed directly. All these are useful elements of assessing how safe a given vehicle design actually is.

Problem is it is not really possible to take the vehicle apart manually; in the process of taking a badly damaged vehicle apart, the normal connection points are likely to be buckled, and/or torn. So you have to force it apart, and in doing so, run a real risk of moving the components you are trying to get to from the positions the crash put them in.

So a method of externally scanning the car to get the placement of all the little bits inside it would be absolutely ideal.

The first and most obvious method is to X-ray the car. But, there's a problem. X-ray scans are inherently 2D. They can tell that a piece has moved from a side-on perspective, but they cannot tell the relative positions front-to-back of the pieces they are scanning. Worse if you try to scan deeper – from front to back of a car wreck say, then some pieces of the wreck will be dense enough to completely occlude others, and may even prevent the x-rays from penetrating completely.

CT scans offer a far better way of doing things. The Computed Tomography scanner completely surrounds the car, takes pictures continually and builds them up into a 3D composite you can view from any angle. Only one problem: A car is far too big to fit into a CT scanner.

Even if one was built specially for the purpose, what happens when they need to crash-test a bus or a truck? Build another, larger one for that? What about a train?

It becomes impractical to build a special-sized CT scanner for every type of vehicle being crash-tested. The only reason we get away with the single-sized ones in general usage is because people tend to be pretty much the same size, and so a one-size-fits-all paradigm was possible.

What would be far more practical would be if the CT scanner could change size depending on what is being scanned. That is what the researchers have done; created a prototype scanner that can be used to scan a wrecked car, a wrecked bus, even a wrecked passenger airliner should the situation arise.

It is in essence the first 3D scanner that changes size according to the target to be scanned.

This image was created by the XXL CT scanner; the working name of the prototype CT scanner which can change its size depending on what you wish to scan.
Image copyright Fraunhofer IIS


This is of course a long way from a home scanner that could sit on your desktop when you wish to scan a bottle, or be driven over when you wish to scan a car, but the principle is the same. The core element is a turntable exactly like that found in every other 3D scanner, or even in a microwave oven. The turntable is the biggest part of this scanner, able to accept an 18 wheeler truck. If extra size is required, additional pieces can be joined onto it, and additional motors added. In theory it would work the same in reverse; just pop out concentric rings of turntable until you reach the desired size, stepping the motor down as you do so.

The turntable doesn't have to move very fast; it just has to keep the target turning at a smooth rate and steady speed. A the object turns, an X-ray source is positioned on one side of it, and a detector on the other. These can be affixed to any ring of the turntable, which is what allows it to change size as required. The detector is four meters wide, although narrow in height. Both it and the source are constantly moving vertically. Up and down in a slight parabola, always directly opposing one another. Each time the source moves down, a slice of the vehicle is scanned. Each time it moves back up, another slice is taken, as by then the turntable will have moved the object just round enough that half of it is not present on the original slice.

From that point it works as normal; the data is streamed to a computer system to reconstruct the object – innards included – as a fully de-constructable 3D object.

Currently the system is capable of a 0.8mm resolution, meaning it can pick up objects as small as the head of a pin on the scanned object but would have trouble differentiating smaller objects such as individual strands of human hair. The research team is working on this, and attempting to up the resolution still further by further refinement of the X-ray source. Their goal is a dynamic resolution of up to 0.4mm, which would be more than enough to detect any crack or ripple in the metal.

Other than improving vehicle safety, the institute hopes to offer versions of the scanner to security firms wishing to examine a suspicious container without opening it, detecting explosives or drugs no matter how well hidden they are within the container – by scanning the inside just as thoroughly as the outside, with the x-ray source. However, to do this on a realistic scale, they will also have to work on upping the scan speed. If you are working security, an hour per container at a busy shipping port would be far, far too long.

Whilst it is clear it is just a prototype at this stage, the researchers' resizeable scanner works, which is the litmus test of all such devices. Further, the simple engineering principles the basic scanner setup uses, will be applicable to smaller, more mainstream scanners, offering a potential way for them too to become dynamically resizing, and maybe even capable of a dynamic resolution at scan time, the same way as the Franhofer's prototype does.


Taking a close look, whatever the scale

Staff Comments


Untitled Document .