Practical 3D Scanning: Scanner Metrics
Watch This Course
$179 USD
-OR-
Or
8 easy payments of only $22.38
Instant access. Easy checkout. No fees.
Scanner Metrics
09.40
| 00:00 | Before diving into more detail on 3D scanning technology, let's understand the metrics most commonly used to compare performance. |
| 00:08 | These metrics and others also vary between all the different scanners on the markets, so this will be helpful in not only comparing scanners that we might want to use or choosing a scanner to buy, but also in optimising the performance or efficiency of our scanner for each project. |
| 00:25 | The most commonly compared metrics will cover a resolution and accuracy, which to be clear are actually not dependent on one another. |
| 00:34 | The easiest to understand is accuracy, which is simply how correct the dimensions of the 3D digital model are compared to the real object that was scanned. |
| 00:44 | In other words, how much deviation is there between the points on the digital model compared to the real thing, with more accuracy resulting in less deviation or difference. |
| 00:55 | Of course, assuming that the scan we've captured is good and not distorted or damaged in any way. |
| 01:01 | What's important to understand is that the accuracy for scanners is usually advertised as "up to a certain number". |
| 01:07 | Up to 0.1mm for example. |
| 01:11 | So this is as good as we can expect, but that doesn't mean that this will always be the case. |
| 01:16 | The accuracy will depend on other factors like what we're scanning, how we're scanning it, and could be different between two scanners that both quote the same accuracy. |
| 01:27 | Also since the accuracy can depend on the size of the scan, this metric is often expressed as a percentage or in millimetres of error per metre. |
| 01:36 | So, the larger the scan gets, the more error we can expect, so the less accurate the scan will become. |
| 01:42 | The highest accuracy 3D scanners are achieving as low as 2 to 30 microns. |
| 01:47 | For reference, a human hair is around 70 microns thick, but these are generally designed for metrology rather than for mechanical design in motorsport, and as you'd expect, they're also extremely expensive. |
| 02:00 | Instead we'll usually be working with scanners with an accuracy of up to around 0.05 or 0.1mm. |
| 02:08 | That's around half of a thousandth of an inch in imperial. |
| 02:12 | This range is generally considered good enough for the majority of 3D scan use in the motorsport space, and any areas that need more accuracy can usually be checked with physical measurements. |
| 02:24 | To give you some reference, smartphones can have an accuracy as bad as 10mm, but most will be within a millimetre or so, which is still very usable depending on what we're working with. |
| 02:35 | It's generally agreed that for metrology purposes, there should be no less than 0.1mm of error over the entire scan. |
| 02:44 | When looking at buying a scanner, we need to consider how much accuracy is really needed, as more accuracy comes at more cost. |
| 02:52 | If we're designing parts where the precision is critical to its function, then we'll need higher accuracy. |
| 02:58 | But this won't always be the case. |
| 03:00 | In the likes of scanning an engine bay for clearance, or a body panel, it doesn't need to be perfectly accurate down to the tenth of a millimetre. |
| 03:09 | Moving on, the resolution is the distance between two points of data. |
| 03:13 | The space between each data point is an approximation, so closer data points means the space between them will be more accurate. |
| 03:21 | This is also often referred to as point distance. |
| 03:24 | A higher resolution, or smaller point distance, results in more points in the point cloud, or a finer mesh. |
| 03:32 | Either of these means a more detailed model, but as you'd expect, more data points also means a larger file size, which takes longer to create and also puts more load on our scanner or computer hardware. |
| 03:45 | Increasing resolution not only creates a bigger file size, but generally also increases the cost of the scanner. |
| 03:52 | It's common that the resolution will be adjustable within the software, but again, we need to consider how much resolution we really need, and this comes down to the level of detail of the objects we're scanning. |
| 04:04 | Simple objects require less resolution, and we won't gain much from having more resolution. |
| 04:11 | Whereas complex objects with fine details will require a higher resolution scanner, or risk not capturing the detail required. |
| 04:19 | Resolution values are typically in the same range of magnitude as accuracy, with most professional grade scanners coming in around 0.1 to 0.2 millimetres at the lowest resolution, and smartphones being around a few millimetres. |
| 04:34 | Again, these metrics aren't actually related, and it's entirely possible to have a scanner with low resolution, collecting fewer data points, but at a high accuracy, and vice versa. |
| 04:46 | While the accuracy and resolution of a scanner really dictate the quality of the 3D digital model, this doesn't paint the true picture of how efficiently the scanner will work for a particular job. |
| 04:58 | Before we dive into this, it's important to mention that the current state of this technology can result in quite different experience over the scanner options available. |
| 05:07 | In other words, while the object we're scanning and the environment that we're scanning in has a big impact here, some scanners just don't work as well as others, even at the same price point and specs, and the process can be either smooth and satisfying, or extremely frustrating. |
| 05:23 | I obviously can't comment on every scanner out there, as I haven't tried them all. |
| 05:28 | Online reviews will really be your best bet here. |
| 05:31 | However, the frame rate, field of view, and depth of field are all metrics that have a direct impact on the practical scanning process, or rather the efficiency of the scanner. |
| 05:42 | The frame rate expressed in frames per second, and the measuring rate in measurements per second, determine the rate at which data is collected. |
| 05:50 | A higher frame rate means a faster scan speed, but the computer's processing speed needs to be able to keep up, otherwise some scanners will automatically reduce the frame rate, and it won't move smoothly, and it can get very clunky. |
| 06:04 | The field of view and depth of field, or depth of focus, define the viewable range that the scanner can capture data from, and I've personally found that this has the biggest impact on the efficiency of the scanning process. |
| 06:17 | Jumping ahead a bit to what we'll cover in the next module, most scanners we'll be using work on the principle of triangulation. |
| 06:24 | Put simply, a scanner uses a central projector and two cameras separated at a certain distance. |
| 06:30 | The cameras view how the light from the projector lands on the surface of the target to understand its geometry. |
| 06:37 | The result is an imaginary pyramid shape that extends from the scanner and defines the visible area. |
| 06:43 | The depth of field, or depth of focus, defines the distance range in the shape that data can be captured from. |
| 06:50 | The field of view can be seen as a cross-sectional area of this, so the field of view at the nearest distance that the scanner can capture will be smaller than the field of view at the furthest distance. |
| 07:01 | As a side note, this is why scanners field of view is often advertised at the maximum distance, simply because it's a larger number. |
| 07:09 | In the end, the result is a volume and space that defines the range the scanner can capture data from, and generally speaking, data captured from the middle of this will be the most accurate. |
| 07:20 | The distance between the cameras, the focal length of the cameras, and various other attributes of the scanner all have an impact on this range. |
| 07:28 | A larger range allows for more flexibility in movement, and makes it easier for the scanner to understand its position as we move around the subject, as it has a wider window to pick up on references. |
| 07:41 | Because of this, it's common for scanners to have a recommended range for the size of objects that we're scanning. |
| 07:48 | For example, a scanner designed for parts around 50 to 500 millimetres in size has a field of view of around 100 by 150 millimetres, whereas a scanner for parts 100 millimetres to 3 metres has a field of view around 3 to 4 times the size at 380 millimetres squared. |
| 08:07 | Typically though, a large field of view is paired with a lower resolution, if we're comparing scanners at roughly the same cost. |
| 08:15 | Again, we need to consider what we're scanning. |
| 08:18 | For example, small detailed parts that benefit from high resolution can get away with a smaller field of view, but large, relatively simple scans, like exteriors of vehicles, chassis or engine bays, are a lot more efficient with a larger field of view. |
| 08:33 | It's also worth noting quickly that for some of the newer scanners, this range can also be adjusted to suit whatever we're working on, in case we're working in a tight space or can't get very close to the object. |
| 08:46 | The key point to understand from this module is that while it might seem that a more expensive scanner will usually be better, it depends on the job at hand. |
| 08:55 | A high end scanner with extremely high accuracy and resolution, with a small field of view, might actually be worse for scanning large areas, like vehicle exteriors, compared to a much cheaper and less accurate, lower resolution scanner with a larger field of view. |
| 09:11 | There are a range of other metrics to compare scanners that may or may not be important depending on our application, such as weight, size and the ability to work wirelessly. |
| 09:21 | Make sure you're clear about what's important to you and will work best for your application before committing to a purchase and keep in mind how scanner specifications can be shown by the suppliers, but this might not paint the full picture. |
