3D scan of a camera trap
As Jeremy wrote in his excellent post on processing the 3D scan of a camera trap, a core component of our stereoscope is a shim that allows it to attach firmly over the lens of a camera trap:
Each different camera trap requires a custom shim that adapts the stereoscope to the camera. Creating the shim is not difficult, but it requires having access to the camera trap to take measurements. However, it is not practical to expect a designer to purchase each new camera trap to take measurements for shims, or for a researcher to ship a camera trap to a designer each time they need a new shim.
Indeed, our team is separated by the Atlantic Ocean, so shipping cameras, shims, or stereoscopes across is not economical. We are not alone, because lots of ecological research are international, so we thought it'd be good to figure out a way to reduce this physical transfer. One way is for me to produce a 3D scan of a camera trap, send it to Jeremy to design a shim, and for me to produce the shim locally, such as at a makerspace. This post will summarise my recent attempt at doing this, and click here for the full photo album.
Structured light 3D scanning
I'm lucky to be based at a university, and happen to know a couple librarians there who told me that they have a stash of 3D scanners. These scanners are often used to digitise items in the university museum collection. After some email correspondence, I scheduled an appointment with Catherine, the librarian in charge, for a 3D scanning orientation.
The scanner I was introduced to is the "Artec 3D Space Spider". This handheld device uses structured light as its mechanism to do 3D scanning. Basically, LEDs on the scanner project a light pattern onto the object you wish to scan. Then, a camera within the scanner records the distortion of that pattern as it moves over the contours of the object. This data is fed into dedicated software that computes a digital representation of what you are scanning.
Depending on the physical scale of what you'd like to scan, different 3D scanning techniques are used. For example, to scan things that are the size of a building, laser scanning is more appropriate. But in the case of something like a camera trap, I was told that structured light is good.
Anyway, I brought my camera trap to the orientation, meaning to give Catherine a sense of the size and physical complexity of what I'd like to scan. Since I had the camera with me, they kindly allowed me to stay and scan it after the orientation. Here's a photo showing the set up, and to give you a sense of scale:
The handheld 3D scanner is on the right, connected to its own power source and a laptop with a USB cable. On the laptop, you open need to start the Artec Studio program to receive real-time scan data from the scanner. This is a dedicated, powerful (and chunky!) laptop that the library designated to work with the Space Spider.
The camera trap sits on a rotating stand. This way, there's flexibility in how you perform the scan, whether by holding the scanner and moving it around the camera, or holding the scanner steady as you rotate the stand.
Performing a scan
The 3D scan was surprisingly straightforward and intuitive. Catherine started things off by activating Artec Studio on the laptop, and turning on the scanner. The LEDs started flashing, and as she moved the scanner around the camera trap, I could see an image of it starting to form on the laptop screen. This is best communicated with a couple clips I took.
The first shows Catherine holding the scanner, which was projecting its flashing LED light pattern onto the camera trap (full video clip):
At the same time, you can see the part being scanned show up on screen (full video clip):
It's important to hold the scanner at the right distance from its target. When this is the case, the part of your object being scanned will be highlighted in green in Artec Studio.
What about the base of the camera trap? Once we scanned everything we could in the first pass, I flipped the camera and repeated the scan. Artec Studio would then automatically align these scans to form a complete model of the camera trap.
If you watched the full video clip, you'll see that part of the rotating base was picked up in the scan. This is another problem solved with Artec Studio, which allows you to use a loupe tool to select the extraneous parts (e.g. the base), after which it will intelligently estimate what you intend to remove. I was impressed by how well this software automates this task.
While the whole process was not difficult, the scan took about 30 minutes to complete. I was also lucky, because the "automatic" scan mode and intelligent base removal by Artec Studio worked well for me. According to Catherine, this is not always the case.
Another pleasant surprise was that the Artec Space Spider is supposed to have trouble scanning reflective surfaces, such as the lens and flash on the camera trap. Normally, the solution is to either using masking tape to cover those parts, or use a special powder that you apply over them. However, somehow the scanner was able to pick up the entirety of the camera trap, lens, flash, and all, without a hitch!
Exporting the scan
Artec Studio has its own proprietary file format for storing each "project", but can also export scans to a variety of 3D model formats such as `ply` or `stl`. Note in the screenshot below that the model is exported separately from the texture, which in this case will be saved as a PNG image that "wraps" around the model.
I've uploaded the Artec Studio project files, and the exported models to the OSF (Open Science Framework), a platform for publishing open research:
https://doi.org/10.17605/OSF.IO/RXZAJ
It is these files that Jeremy worked with and described in his post. And check out the full album for more pics.
Final thoughts
The actual scanning was not hard at all, and it produced a usable 3D model that I shared with Jeremy for him to design a shim with. Where the difficulty lies is that an Artec Space Spider 3D scanner, the powerful computer it needs to work with, and the proprietary software license for Artec Studio are expensive. Going forward, Joshua suggested that we might try one of those mobile apps which can make a 3D scan using just a smartphone's camera. While those apps might still (sadly) be closed source, at least they are much cheaper than the set up I used at the library. The point is, we need to develop a workflow that works for as many ecological researchers as possible.
Stay tuned for more updates!
Acknowledgements
I want to thank Catherine Dack, the librarian in charge of 3D scanners at the University of Bristol library, and Zosia Beckles, the librarian who made the introduction. In my view, libraries and librarians are underrated, so I'd like to give them my thanks and priase.
1 comment