Background and First Steps in the Field
I never thought in a million years I’d be able to create faithful 3D copies of anything. How I got to the point where I am today is all pretty much luck and happenstance, but I wouldn’t change anything, even for all the skulls in the world.
During my first year of undergrad at McGill University in Montreal, I discovered my passion for archaeology. I had always loved history, learning about past peoples, events, and societies, but I never really understood what anthropology or archaeology even really was. All it took was one class, Intro to Archaeology, and a couple of books on evolution and cultural anthropology, and not only did I find the topic interesting as heck, I also found myself to be pretty good at interpreting data and supporting my opinions with archaeological or ethnological evidence.
Fast forward 8 years, and, with the exception of a couple breaks of a year or two to pursue other passions (culinary arts, travel, languages and literature), I arrived at the point where I had a master’s degree in bioarchaeology from a rather well-known school, a few good connections, and the hope that I could work pretty much wherever I wanted with my background and CV to back it up.
Turns out, I was still just one of the many recent grads without a specialized, useful, and above all, modern skillset to set me apart from the rest of my colleagues. So did I do what a rational person would do? Pivot? Use the skills I already had to make the best of a subpar situation? Nah, I decided to keep studying.
And so, once again, I found myself in another master’s program, this time rooted in anthropology, osteology, and forensics, in the oldest still-functioning university in the world, the University of Bologna. My program was the only one of its kind in the whole country, with a specific focus on studying human skeletal remains, and guided by some of the biggest names in Italian anthropological studies today. The program was headed by Maria Giovanna Belcastro, one of the best-known bioarchaeologists and physical anthropologists working in Italy. She would later become my supervisor during my research internship. In addition, the other students and I would study under Cristina Cattaneo at the Forensic Anthropology and Odontology Laboratory (LabAnOf), who is most well known in the field of forensic science for her work identifying the remains of Mediterranean migrants lost in shipwrecks, and Gino Fornaciari from the University of Pisa, the leading figure in modern Italian paleopathology.
Although each of these scientists would contribute much to our understanding of the human body and how to interpret and study human remains, it was in Milan at the LabAnOf where we were presented with the theories behind facial scanning and reconstruction. Sculpture had always interested me as a younger man with more free time on my hands, and now, as an anthropologist, I saw a way to combine one of my favourite art forms with my academic interests. From that point on I dedicated much of my studies to learning how to reproduce skulls in order to perform a “manual” reconstruction of ancient and modern faces using the Manchester Protocol, a theory that combines mathematical regression formulas with standard tissue thicknesses to reproduce as faithfully as possible the visual aspect of an individual’s face in life using only the skull.
In order to properly use this method, one must combine digital and plastic methods of reconstruction. A model of the skull is used as a base to apply various levels of plasticine or clay to reproduce the soft tissues of the head and face. Obviously, in an archaeological context the original skull cannot be used as it could be damaged or destroyed. In the past, creating a facsimile of the skull was achieved by making a standard negative cast of the skull, using various methods, in order to create a positive reproduction in plaster or resin. As this can be time-consuming and can create distortions based on the personal skill of the anthropologist or artist involved, in recent years more modern methods have been proposed. One of these methods involves using a 3D printed model based off of a textured or non-textured mesh of the skull, scanned using laser, structured light, or photogrammetry.
About two months ago we finished all of our classes and exams in the program, and in August we spent two weeks excavating human remains from the high medieval cemetery of Badia Pozzeveri near Lucca, Tuscany, Italy. Then we passed on to the practical portion of our studies, each one of us choosing a different internship at European institutions (mostly Italian). I chose, or rather was chosen, to assist in an ongoing research project with Dr. Belcastro to study the human remains excavated from a VII century CE necropolis currently located in the modern Italian region of the Molise, once home to the powerful pre-Roman tribes of the Samnites and one of the key settlement areas in Southern Italy of the Germanic Longobards in the Early-Medieval period. In the VII century CE, this area was under the control of the Duke of Benevento and within his domain there was a rather unique blend of Germanic, Italic, Byzantine, and Steppe-Nomad cultural traditions. This area would conserve its independence, even after the conquests of Charlemagne in Northern Italy, until the arrival of the Normans in the XI century.
Applications and Methods of Photogrammetry in Academics
I had been introduced to photogrammetry during our session in Milan, but finally I saw my opportunity to apply it in a practical manner for this research project in Bologna. I was to study the remains of the so-called “Horselords of Campochiaro” (individuals buried in the same grave as their horses) in order to determine sex and age at death, as there was some doubt that all of the individuals were males, and that at least one of them could have been a “symbolic” burial of a female. Since there were so many well-preserved crania from this site, I thought this might also be the way to perfect my photogrammetry skills and to apply the techniques of facial construction that had so interested me during my studies.
At this moment, I am currently waiting for a response from one of the 3D printing companies here in Bologna so that I can send my meshes off to be printed into an actual physical object. In the meantime, I have found quite a few traumatic lesions and an as-yet undescribed paleopathology about which I will write a nice, tidy scientific article for publication before my internship period ends. Although the facial reconstructions will have to wait, and won’t be an “official” part of my master’s degree, I will continue working here at the University of Bologna into the next year, which hopefully will allow me the opportunity to finally look into the face of some of our ancient ancestors.
Everything I have learned thus far regarding photogrammetry and 3D modeling I’ve more or less taught myself in the last three months. Through trial and error, I was able to learn which techniques work better than others, and I finally found a program that was able to make models that more or less satisfied my immediate needs for the project.
My first foray into photogrammetry was meant to be more or less a one-off. I had hoped to do a facial reconstruction on a particularly intriguing skull, and for that, I needed a 3D printed model to work off of. I hadn’t yet thought of the idea of creating a database in order to create a permanent online home for some of our University’s many, many skeletal collections— that would come later after I saw how successful my first model attempt had been.
Before I discuss how I went about creating my model, I should preface by saying that our department does not have a professional, or even nearly professional photo studio. We have a small, poorly lit room with a vertical tripod and supporting incandescent lights that we use to photograph small, fragmentary pieces of bone from a static perspective. In addition, we also have a pair of “scientific quality” LED lights that were probably purchased about 20 years ago. In order to somewhat rectify this, all of my models have been captured on film by setting up a stool in the middle of our lab, using only the overhead fluorescent lighting supplemented by our ancient LED lights and the flashlight from my cell phone. It’s quite sad, but at least it works (to an extent). This has often led to uneven colours and/or shadows in my models, but there’s not much I can do about it for the time being, unfortunately.
So, my first model was of the skull of an adolescent male with a particularly traumatic lesion, seemingly lethal, running obliquely along the cranial vault, most likely caused by a sharp, thin blade (such as this one). As I was quite inexperienced and really had no idea what I was doing, I had not yet devised a method to be able to fully capture the skull, including the base. The skull was carefully placed onto a black fabric background, and I proceeded to take about 50-55 photos, mostly from parallel to the floor to perpendicular to the floor, without ever really shooting below the model. I took about 4 photos from each angle (level, inclined, more inclined, and finally from the azimuth). Once I had all my photos captured, I transferred them to my laptop where I used a free, open-source program called Meshroom to create my point cloud. Meshroom works rather well, though it can be a bit difficult to start with, as there are many parameters that can be changed in your workflow, and it can be daunting for someone who has never used a similar program before. Through mostly trial and error (and a few online video guides), I was able to complete my first model to a satisfactory degree.
After the creation of my textured mesh for my third model, I wasn’t very happy with the result and decided to try out another program to make my models. I was familiar with AutoCAD and was interested in trying out one of their other programs, ReCap Photo, which is specifically designed for photogrammetry. Luckily I was able to get an education license as the full version is quite pricey.
Using ReCap Photo is a pretty simple process: all you have to do is take the best pictures possible, then upload them into the software. From there your photos will be sent to Autodesk’s main servers (somewhere in the US I believe…) where they will then wait in a cue to be processed; this step will take your photos, create a point cloud, transfer that point cloud into a mesh, then apply textures. Once this is complete, your model will be available for download on the ReCap Photo application. After downloading, you can use the built-in tools provided by Autodesk to cut and trim your model (getting rid of any supports or backgrounds captured in your photos), fill holes, change the model orientation, sculpt, smooth, and remove any particles. There is even a feature that allows the software to detect any obvious issues, regardless of how small, allowing you to peruse and fix critical problems while skipping others.
Because most of the modeling is done on external servers, you are rather limited in the number of variables and options you can tweak while making your model. However, since no actual structural algorithms are being applied on your personal computer, your system won’t grind to a complete halt and begin to overheat, and artists and researchers with less powerful hardware can also make use of this software. Another huge advantage of ReCap Photo is the inclusion of post-processing tools included with the software, something which Meshroom doesn’t provide, making touch-ups or modifications to your models nearly impossible. Whether you use Meshroom or ReCap, it’s pretty necessary that you have a PC that runs Windows, as Macs aren’t supported. I really have no idea about Linux support since I’m not a sociopath (kidding, kidding, we like to have fun here). In order to publish to Sketchfab, most models will need to have their meshes decimated (basically reducing the resolution in terms of vertices) to reduce the size of your uploadable file. Both Meshroom and ReCap Photo have options for exporting decimated models.
Failure and Success: The Road to Innovation and Progress
Looking at my models published on Sketchfab, one can easily see how I was able to improve over the last couple of months. The first models all include a scale marker, black cloth background, and provenience of the object. As I explored other models on Sketchfab, I saw that many of the published skulls included the entire skeletal district, base included. This perplexed me a bit; how would I ever be able to capture the entire skull, with the mandible in anatomical position, without building some sort of complex harness system? I tried lifting the skulls with black cushions or sponges, but these didn’t help much. You can see that attempt in these models:
These models also display a much “cleaner” appearance compared to my first model, as it was around this time that I was learning how to properly use the post-processing tools in Sketchfab.
I tried experimenting with various support systems for my skulls, including a carved wax candle, until I finally invented my “patented photogrammetric skull pedestal,” seen here:
This pedestal allowed me to raise the skull without otherwise obstructing the view of the cranial base, allowing me to create full “closed” models for the first time. One can see a major difference between this model:
which was created from a skull supported by a series of sponges and plasticine scaffolding, and this model:
which was the first skull to be photographed using the pedestal system. While the former model is a complete skull with mandible, an attentive observer will quickly notice that most of the inferior aspect of the object, including the cranial base and internal mandible, was not modeled and instead is composed of a simple smooth fill texture, while the latter model has nearly every aspect of the skull replicated in three dimensions. This was really a turning point for my photographic technique, and it has more or less remained unchanged until today.
Using this setup, I usually take about 60 to 80 photos, depending on the context and how complex the object is. I usually take a series of five or six photos in vertical sequence, starting from as low an angle as possible, to nearly vertical, with one photo exactly at level with the object. This process is repeated from various overlapping views of the object at different angles.
I firmly believe that my latest model is definitely one of my best, having a rather high texture resolution as well as having little to no artefacts, distortions, or weird shadows created in the photographic process.
This skull, while not of one of the so-called knights, is still one of the most complete and serves as a kind of counterbalance to the other modelled skulls, as it is the only one that comes from an individual that has been tentatively described as female— at least until we begin genetic and chemical testing.
Do Try This at Home
If this article has at all piqued your interest in photogrammetry, or you’re a researcher who could see this technology helping your studies, my advice would be: just do it. Unlike other forms of scanning, such as CAT, MRI, laser, or structured light, photogrammetry is extremely accessible to just about anyone who owns a camera. There are even smartphone apps (such as SCANN3D) that allow you to make small 3D objects in under 20 minutes. Although this is still a relatively new technology, the accessibility of this tech to all levels of society has created a wealth of tutorial videos and articles on the internet that allows even the most technologically illiterate of us to become nearly professional-level 3D artists.
Most of your early work will be trial and error, figuring out how best to photograph your subjects, what types of light to use, and which software works best for your needs. Since the initial start-up cost of using photogrammetry is so low, and it is completely non-destructive in nature, meaning that any mistakes or problems can usually be resolved with a little bit of time and elbow grease. If worst comes to worst, you always have the undo button.
The most difficult aspect of photogrammetry for 3D modeling neophytes will most likely be the photographic aspect itself, as understanding how to properly photograph your subjects will take time and practice unless you are already a professional photographer. With the wealth of resources we now have at our disposal, there is no time like the present to start your new photogrammetric journey!
The fields of cultural heritage, history, museology, archaeology, and anthropology are rapidly becoming more in touch with new forms of technology, both popular and scientific, and I can only see this technological integration increasing in the future as younger generations spread into these fields and the cost of specific hardware falls sharply. The most obvious application for photogrammetry and 3D models in archaeology and anthropology is the ability to disseminate information, copies of artefacts, and the material and biological record to an extremely large, worldwide audience of both private individuals and researchers who might otherwise never have the opportunity to study or view said artefacts because of geographic or financial limitations.
Although archaeological databases have existed since the start of the new millennium, it has only been recently that many universities and private institutions have begun to share these databases with the public. I see this type of academic approach, and public access in general in academia, becoming much more frequent in the upcoming years as the applications of simple photogrammetry software become more known and more used in the various fields of cultural heritage. I have high hopes that one day a young academic from an impoverished corner of the world will be able to study such incredible works as the Sistine Chapel, the Mona Lisa, or even the mummies of Ӧtzi the iceman and Tutankhamun, down to the most minute detail and all to scale, without ever leaving the comfort of their home or university.
Another application for this technology, and other types of scanning and rendering tech, would be in museology, best exemplified in the ability to explore certain artefacts or human remains from various angles and distances, perhaps even with integrated annotations, all without ever moving or touching the object itself. Many museums have begun to use this technology, but as photogrammetry improves and perhaps even matches other types of scanning in the future, I’m sure this will become much more common, as even relatively untrained museum staffs would be able to create high definition 3D models without having to rely on outside agencies.
Even my colleagues, who professed themselves to be rather technologically illiterate, have begun to try out photogrammetry after I explained to them the basic steps and how easy it was to start, even without specialized equipment. Although they are still just beginners, you can find their profiles on Sketchfab (Martina Demaria and Teresa Nicolosi)
As you can all see, I’m a big fan of photogrammetry, so much so that I’ve infected many others in my faculty here in Bologna with my passion for easily made, beautiful 3D models. I hope to continue my journey in photogrammetry, whether continuing in academics, or for more artistic purposes after this program. If you’d like to contact me regarding photogrammetry, or are interested in learning more about anthropology, osteology, or the horsemen of Campochiaro, drop me a line.