Over the past five years or so, the availability of 3D scanners on the market has increased dramatically. No longer is a 3D scanner only for those who can fork out tens-of-thousands of dollars or for a DIY fanatic who wants to hack an Xbox Kinect. There is now a slew of offerings that range from a free app store download for your smartphone to ultra-high-end dedicated metrology solutions. Regardless of your budget, there are some things you should know to make sure you choose the best scanner for your application.
First, we need to understand that 3D scanners are comprised of two subsystems – a capture system and a locating system. The capture system is responsible for registering the shape of the surface that is being scanned. This can also include the color and/or texture of the surface. The capture system plays a large role in the resolution of the final scan and contributes to the overall system accuracy. However, the lion’s share of an overall system’s accuracy is based on the locating system. The locating system is responsible for putting all that great surface data from the capture system in the correct location. Both systems must work together in concert to achieve a useable scan. Without a good capture system, you will have data that is in the correct location but may not have the required resolution to define critical features. Likewise, without a good locating system some surfaces may have great detail but are ultimately out of position.
Let’s have a look at the two subsystems of a typical laser scanner.
This is a Faro Quantum S arm with a blue laser scan head. The arm is the positioning system, it is mounted to the granite surface plate and that mounting location represents the origin of the XYZ coordinate system for the arm. As the operator moves the arm in space, angular encoders in each joint are used to report the exact XYZ location of the end of the arm. The capture system is the FAROBLU SD scan head. It uses a blue laser as the light source and a digital camera to register the shape of the laser line as it passes over the scanned surface. These two systems together create a dense point cloud that accurately captures a real-life object with high resolution.
Now let’s look at a typical structured light system.
This is the ZEISS COMET 5M blue light scanner. It uses a digital projector and a digital camera to project and capture a striped pattern that is cast onto the scanned surface. Because the projector is casting a 2D pattern on the part (as opposed to a one-dimensional laser line) the structured light system is able to capture the entire visible surface in one shot. But where is the locating system? In order to scan the entire part, we must reposition either the scanner or the part between shots. As long as we have overlapping surface data between each shot, the software running the scanner is able to “wiggle” the overlapping surfaces into position. Thus, the part itself (along with some fancy math in the software) becomes the positioning system. This results in a system that has a wide range of applications.
But what technology should I use for my project? We’re going to look at the raw point cloud data from some different scanned parts to help illustrate the strengths and weaknesses of each underlying technology. Please keep in mind that there are always exceptions and outliers and these examples should serve as general guidance and are based on our experience with many different scanning systems over the past decade.
This is a photo of an injection molded part with a clip on the end. The clip is a little less than ¼” wide. We scanned this with a laser scanner and a structured light scanner. Both scanners were able to scan the features of this part without any trouble and without surface preparation. Scan time was roughly equal although we did use an automated turntable with the structured light scanner so that session was relatively “hands-off”.
Here is that same part scanned with a laser scanner on the left and a structured light scanner on the right. With the laser scanner, you can make out the scan lines as they moved across the surface. You can also see where the operator sped up and slowed down as indicated by the change in density of the point cloud. The structured light scan looks nearly solid and complete. You can also clearly see the parting lines of the injection mold on the right.
All 3D scanners create some level of unwanted noise. This shows up as data points that are floating out in space and are not associated with the surface that is being scanned. Noise results in a reduction in accuracy and resolution. Many software programs are capable of filtering out noise, but this can be taxing on the computer system and should be used judiciously.
This is a photo of the feature block we used for this test. It is machined out of RenBoard and is approximately 7”x7”x2”.
Above is a scan of the feature block. We’re going to zoom into a top-down view of the front-left corner.
When we zoom in on the rounded corner of the block we can see the difference in the amount of noise produced by each scanner. While the difference may appear to be slight, we can see that the laser data on the left is fuzzier around the edge than the structured light data on the right. The data on the right has a nice crisp, clean edge with evenly spaced data points.
As with all light-based 3D scanners, light must exit the light source, land on the surface you wish to capture, then reflect that light back into the camera. If the surface absorbs or refracts that light (sending it off in all directions and not back into the camera) we end up with missing or inaccurate data. Lasers tend to have a much higher intensity of light and a much more focused spectrum of light than structured light systems. This gives laser systems the ability to more easily scan surfaces that are dark or reflective.
Here is a photo of a metal bracket that has been cast, then machined in some areas. The bracket is 6” tall.
We can see that the structured light system on the left had some trouble capturing the raw metal, especially in the machined areas. We were able to capture most of the surfaces, but it was at the cost of time. We had to set up the scanner to take much longer exposures in order to get this complete of a scan. On the right, the laser system produced a much more complete scan. We did have to fiddle with the settings a bit, but luckily there was a “Shiny Metal” setting that worked well. We were able to scan the part quickly and very little data cleanup was required.
All the examples we’ve used so far are smaller than a bread box. That doesn’t mean that either system is limited to the reach of the laser’s arm or what we can fit on the rotary table for the structured light scanner. Both systems are capable of scanning items much larger. Of course, there are trade-offs and extra considerations when scanning something as large as an entire vehicle, but it can be done.
Here we’ve scanned half of this car with a structured light system and half with a laser system. Both systems present their own challenges with a project of this size, but everything we’ve addressed already still holds true. So, the question is, what do you need from your scan data? Do you need to capture every single minute feature, or do you simply need your scan data to take up some space so you can design around it? Can you live with some noise that may compromise the accuracy of the scan or do you need the data to be as clean and accurate as possible?
We are fortunate to have several different 3D scanning systems that fit into a wide range of applications. If you’re wondering about what type of scanner you should purchase, or if you think we might be able to help with an upcoming project, let us know.