What is “3D digitising”?
3D digitisation or 3D reconstruction is a collection of processes to read and recreate the depth of an object so that they can be displayed in a digital space. Displayed above is an example of a 3D model that can be found in one of UNE’s Pedestal3D galleries. Pedestal3D is UNE’s online platform for delivering 3D media, it is a comprehensive software that boasts in-built tools such as the ability to measure, create cross-sections, adjust lighting and add annotations, and it can all be accessed anywhere on the web. Through this platform, these 3D objects can be placed directly into any website from this very blog through to Moodle’s units and beyond.
The value of these models as teaching resources is understated, given their capacity to be an equaliser between campus-based students and the vast population of online students. Bridging the gap by bringing a facsimile of object-based learning into the homes of every student with a basic internet-capable computer or tablet; these resources broach even beyond the capabilities of access to original progenitor artefacts/specimens. For example taking measurements through online tools leave no damage to the original artefact/specimen. A virtual cross-sections for visualisation becomes possible. Even the lighting tools allow for more freedom and agency than any controlled lab environment can provide. Comparative studies have never been easier, given the ease of creating virtual grab bags of resources, with each new resource joining an ever-growing collection of tools for an educator to access whenever required. With units building assessments out of these resources in Archaeology, Paleozoology, Historical Inquiry and Museum Studies already – the time is now to jump on board and bring the 3D dimension into your teaching.
There are numerous methods for making these types of 3D models – from designing them from scratch, extracting segments from CT scan data, constructing the model through photogrammetry to mapping out the geometry using various types of light-based scanning. The subject of this article will be a cross-examination of the latter two. For context, we will break down both in a concise manner.
.
What is photogrammetry?
Starting with modern photogrammetry, we find a term used to define a broad slice of computer algorithms designed to utilise the overlapping features found in photos taken circling the subject. By turning each pixel into a point of data the photogrammetric processes can then compute matches and create geometry from triangulating this data. Thus the workflow becomes simple enough – take your photos, input them into your photogrammetry software and allow it to reconstruct the information. There are numerous pros and cons associated with this that will be quickly touched on:
- Low cost start up – an affordable DSLR camera and a computer are the bare minimum to beginning this method.
- Colour Realistic – by nature of the data points coming from a modern camera, your model typically comes out with close to real-life colour.
- Easy for newcomers – the complicated portions of the work are hidden behind the software in most cases, allowing volunteers to begin working with minimal training.
- Scale – the model’s scale is not inherently generated in a photogrammetry process, though this can be supplemented in various software using manual scale bars.
- Light dependent – photogrammetry can be very dependent on the correct light conditions to create the best possible result, this can and often will rule out its use outside of fully controlled environments.
- Time consuming – the algorithms used to construct data from every pixel in a DLSR camera shot are not quick, even with an average computer; often a specialist computer built for this purpose is recommended.
- Divorce of data and process – Because a dataset must be captured in stable light conditions this means processes occur after the entire dataset has been planned out and captured. Thus any errors made in this portion of the process won’t be recognised until a significant time has been spent on the process; making backtracking a chore.
Not quite cons, but definitely considerations:
- Upgrade costs are less friendly – upgrading from a basic setup and moving to more specialised equipment can often become an expensive venture.
- Knowledge of photography is recommended – whilst not mandatory, a basic knowledge of balancing out photography settings will help immensely in refining a better product.
.
What is light-based scanning?
The newer method on the proverbial street1 is light-based scanning, a more specialised form of reconstruction that has become a mainstay in the fields of reverse engineering and surveying. Varying across numerous companies and their ranges, the units designed using this technology range from hand-held, turntable dependant to larger-scale pieces focusing on landscapes. The in-built flashing lights, lasers and cameras all work simultaneously to highlight, record and convert data into depth maps of the subject, oftentimes able to display this process in real-time on an accompanying monitor. The pros and cons are as follows:
Pros:
- Speedy capture – an equivalent sized object will take approx. 1/10th of the time it takes to make a photogrammetry model with an equivalent machine.
- In-built scale – As scanners use multiple cameras and receive information from light pinging of the subject, they can inherently capture scale with sub-millimetre accuracy.
- Real-time reconstruction – visually seeing the data collection can go a long way to helping a technician adapt on the fly and capture a better data set.
- Less light-dependent – whilst still optimal, a full controlled lighting environment is not required, in part due to most scanners’ own lighting.
- Professional suites – Due to the higher buying point and more specialised nature of Light Scanners, they come with attached software to run said hardware. These suites are tailored to their equipment and are given ample documentation and support.
Cons:
- Cost – there is no avoiding this, most 3D scanners will come at a non-consumer friendly purchase point.
- Colour – depending on your investment in the technology, a lot of lower end scanners will sacrifice the quality of the cameras that specialise in colour. This can sabotage the visual appeal of models.
- Specialist equipment – by nature of the highly specialised equipment involved, any repairs or maintenance will often require sending the equipment away leading to downtime.
With these summarised comparisons, the takeaway will highly depend on the needs of the project being worked on. Yet in the Learning Media’s recent experience, we’ve found the Light Scanners a more robust solution to 3D digitisation and reconstruction. The advantages manifest in multiple portions of any given project – the literal speed of machine processing, the lack of manual scaling, and the “on-the-fly” adjustments minimising reshooting have made significant improvements to the operational time a project can take. Using a recent project as a case study we have the perfect example to demonstrate this.
1 This is an entirely relevant comment as whilst the basis of Photogrammetry could have roots as far back as the Ancient Greeks, modern Photogrammetry arguably only really beats out modern Light Scanning by 20-30 years.
.
Case Study: Boof the Horse
A project involving the 3D digitisation and reconstruction of 89 horse bones began near the end of Tri 2 2021, aiming to have the whole set complete and fill in the missing components for the ARPA course in time for Tri 1 2022. It had begun initially using photogrammetry as it was the best fit at the current time. Working through COVID related circumstances, the method had produced 30 models by the close of business on December the 24th, a period of three months or so required for this output. Realising the end of the project was near impossible to reach in time for Tri 1, we opted to switch to using the newly acquired Artec Space Spider Scanner. The expectations had been that this would be ultimately faster than the photogrammetry but the results spoke far more than any assumption could. Using a light tent to simulate the best lighting conditions possible, we managed to capture the remaining 59 bones in 19 days!
Photogrammetry was producing approx ~0.5 models per day in comparison to the ~3.9 models per day using Light Scanning. That’s 7.8 times more efficient. The normal consideration may be, so what did this speed cost? The answer is next to naught. The geometry was on average, superior by a small amount – owing itself to the multiple cameras working in tandem on our Artec Scanner. The colour was on average, equal to photogrammetry, the one area we expected it to fall short. The scale was on average, vastly superior given to the measurements being recorded with sub-millimetre accuracy. The biggest loser in all of this project was the technician’s wrist which suffered some mild irritation from holding the scanner for long periods of time. A success overall I would argue.
The end result of this project was able to deliver the scans ready for T1, and prepare for the future applications of 3D scanning as mentioned here by Melanie Fillios:
“These scans provide a digital record that can be printed to create an accurate replica of each specimen. What a gamechanger! In both the teaching and research realms, Jackson’s expertise has drastically augmented what we’ve been able to achieve.”
– Associate Professor Melanie Fillios (Director Place Based Education and Research)
If you’d like to further explore this cutting edge technology and incorporate it into your teaching and learning – make sure you contact the Learning Media team! We are happy to organise a prototype or demonstration or chat about your next project.
Contact the Learning Media team on mediarequests@une.edu.au.