As much as we like to tout the amazing abilities of 3D scanning, just like everything else in life it has its weaknesses and problems. Though it is used in many areas of life like self driving cars, real estate and land mapping this article is going to focus strictly on where the issues lie in the manufacturing realm. Many of the issues are going to be very situational so we’ll try to cover them the best we can and at the very least hope to get your brain turning to anticipate any issues (no one likes learning the wrong kind of “something new” everyday in the middle of a time sensitive project).
3D Scanning Occlusion
The English language is amazing, isn’t it? We have 10 words to describe the same thing but often hear “there’s no word for it in the English language” when someone on a TV show is trying to explain the meaning of a word in a foreign language. “Occlusions” are just some very smart and very obnoxious PhD’s word for shadowing or the shadow effect that has somehow made it to be industry standard lingo (sorry if you’re an obnoxious PhD reading this, but you kind of bring it on yourself). If the idea is still unclear, imagine this (or draw from experience): you’re flying at 40,000 feet over a mountain range. As you’re looking out of the window, you can see the full front of the mountain near the plane but as you look out further you begin to only see other mountains top halves, eventually only seeing just their tips. You know that those mountains have a back side and valleys running in between them, you just can’t see them. The valleys and hidden mountain faces that you can’t see are what machine vision people consider to be occlusions, which will produce patches of missing data in the scan. Or think back to how Frodo hid behind rocks near Mount Doom to avoid the eye seeing him, your call.
A lot of times the missing data is of no importance for the end goal of the acquisition (measurement, counting, etc). There are some solutions to this problem-if it is a problem. The first and easiest solution would be to really analyze the position of the scanner and see if there is another position that will get all you need.
The next easiest solution may just be to add another scanner mounted at an opposite and opposing angle (or-up to a whole ring or row of scanners if the brand of scanner you’re looking at allows it). This would produce separate scans and depending on your end goal processing the scans separately might be enough. However, your end goal may require stitching together all of the separate scans which exponentiates the complexity. This can quickly rack up a bill and slow down the project.
Another solution might be to scan the object and then rotate it and scan again. This again can bring in even more complexity. If you’ve been saying “yes, this is (probably) going to be an issue for me” up to this point, we can only highly recommend contacting an experienced professional — unless you have a large budget, a huge project timeframe, love to regret choices and wake up every night from nightmares about work, then by all means play ball.
Material
The material(s) of the object being scanned can cause issues. One of the biggest culprits are shiny surfaces, as you have probably already educationally guessed. If you’ve ever looked in the mirror and shined a light into it, and angled back at your face as a young child trying to understand reflection (as the author did) you clearly understand the repercussions. The intensity of the light coming back to the receptor is much higher than it should be and can cause the data to be scattered and showing high points that don’t exist. There are a few levers to be pulled to possibly fix this but unfortunately it’s all situation and scanner/technology dependent beyond that.
Another material issue that can arise is material opacity and refraction. 3D scanning through glass will distort an image so there are some corrections to be made in that instance. A sub requirement would be that the scanner needs to be parallel with the glass plane. Attempting to scan anything with a clear-ish non-flat surface, while not impossible, is going to be expensive. You have been forewarned. In this same vein, some plastics can also be penetrated by laser even if they aren’t see-through by the naked eye. You’ll just have to ask the scanner manufacturer beforehand if the scanner you’re looking at will have this issue. And to answer you’re question before you ask, no this isn’t the same as a 1998 Sony HandyCam (IYKYK).
Speed
If this 3D scanning process is in line (production) or time dependent somehow, “you’re gonna have a bad time.” Just kidding, if you’re not a South Park fan disregard that last line (or go google the clip for a 2-minute break), time dependance is not a deal breaker.
There are 2 different “speeds” that I’m referring to though. The first is scan speed. It’s dependent on the scanning technology you’re looking at, but scan speeds are not instantaneous. Take for example, a line profiler. That can be integrated into a continuously moving line, but the capture mechanism can only be triggered so fast to build up a 3D scan. On the other hand, a structured light camera/scanner needs to have a still object so it can only be integrated in a spot where there’s a dwell time.
The second is the processing speed. Guess what? It’s also situationally dependent. It’s going to depend on:
- How many points create a full scan.
- How much transformation has to happen to the scan to anchor it, so it is in the same 3D scanning space every time (see the general rules in the next section)
- How much the scan must be pre-processed: smoothed, trimmed etc. before measurements can be attempted (if it’s needed)
- How many measurements and the type of measurement (some measurements require heftier math than others)
- Whatever extra post processing needs to be done (yeah you could just feed the values into a PLC for decision making, but I also know that your boss wants all the data, all the time, and for it to output into some file format to go into another system so everyone has access to it but never reviews it. No, it isn’t just a problem in your company and yes you will go blue in the face repeating to person after person (or the same person) that they can actually see it and where it’s at, in meeting after meeting)
We have had scan results range from milliseconds to minutes for processing. Don’t be immediately scared by this, it usually is in the seconds range. Plan for pizza, aim for french fry (seriously go watch the clip).
Repeatability/Accuracy
The end all-be all of what you’re looking for just might be too much for the scanner to handle. The scanner specs as well as the acquisition setup are going to be of the utmost importance on this. Without going down the 100 different rabbit holes, here’s the general rules: get the object as close to scanner as possible while still enveloping the whole area of interest and scan the product in the same position every time (I know robotic arms seems like a great solution, but test the position repeatability (x,y,z) of the very end of it through all the axes, it might surprise you).
Need More about 3D Scanning?
This article is only a slight general grazing of issues you may run across. We tried to keep it in the general 3D scanning realm and technology agnostic. We are also trying not to repeat the same information in multiple articles. Most problems that you may come across can be cured or worked around. If you have more questions, please feel free to contact us. A 30 minute phone call is definitely worth avoiding a 3 day hangup in the project, ask us how we know!
Learn more here or Contact Us today!