This post, by FSU Libraries’ Digital Scholarship Librarian Matthew Hunter, is a repost from the Immersive Scholarship blog at https://immersivescholarship.create.fsu.edu. For more information about the work the Immersive Scholarship team is performing, or to discuss project ideas, please visit their site. For more information about the Cosa project and other Classical archaeology projects, check out the FSU Classics Department site at https://classics.fsu.edu/
This past June, I traveled to Italy to contribute to the long-running archaeological excavations at the ancient Roman site of Cosa. Perched on a beautiful coastal promontory overlooking the Tyrrhenian Sea about halfway between Rome and Florence, the ruins of the ancient colony are surrounded by the sleepy modern town of Ansedonia and provide hints at both Republican and Imperial urban planning and building techniques later deployed across the Roman world. The current, FSU-led excavations at Cosa have been focused on a bath complex located within the ancient town, and add to a program of study that has explored the site since the late 1940s. The current multi-institutional focus on the bath complex, directed by FSU Classics professor Dr. Andrea De Giorgi, is meant to illuminate a particular iteration of one of the hallmarks of a “proper” civilized Roman colony town, the bathhouse, which were hubs for social life in the Roman period. The baths at Cosa are particularly interesting as a case study in large part due to the lack of a natural water source at the site. As there were no wells, springs, lakes, rivers, or streams to provide potable water, all water used at the site would have been rainwater collected into large cisterns built around the site. Strategically doling out this rainwater for necessities such as cooking and drinking, and for leisurely comforts such as fountains and the baths, was the work of a sustainability-minded town.
The current FSU excavations at Cosa began in 2013 when Dr. De Giorgi and a team of FSU Classical Archaeology graduate students set out to add to a long-running series of excavations carried out under the behest of the American Academy at Rome. Prior excavations have outlined the general city plan, uncovered a forum and details of the village, explored the houses of the residents, investigated the religious locus of the town in detailing several temples on the “Arx” (which is a Latin term for both the geographic summit of the site as well as its focal center), and investigated the long afterlife of the town’s settlement through the medieval period. The current exploration of the bath complex seeks to understand the extent of the town’s interaction with the larger Roman world as a center for maritime trade and commerce. Over the last decade, the FSU Cosa Excavations have produced a robust contribution to the understanding of Roman bath technology, Republican-era colonies, and more in various formats including edited volumes, articles, presentations, and dissertations – with the culmination of the project a slated entry into the series of major site reports under the University of Michigan Press imprint.
Under the direction of the current FSU team, recent excavations at the bath are invested in performing not only vital archaeological excavations to better understand the built environment, but to also incorporate cutting-edge archaeological methods and modern technologies as part of their investigations. Among these are the use of drone-based aerial mapping, LIDAR topology scanning, and (most importantly to my work) photogrammetric and 3D scanned digital recreations of excavation trenches, artifacts, and site features. The production of these digital records is intended to help future researchers access the scholarly material of the excavations in a format that is most suitable and available to them, increasing the generation of knowledge about the site and of Roman history in general. By embracing these technologies and the more openly-accessible opportunities they present for sharing their research widely, the team hopes to invite new audiences into accessing their work and increase the impact of their scholarship beyond readership of the site reports and scholarly publications.
As part of the university libraries’ partnership with the Cosa team, I have been assisting in these efforts by consulting on a digital database project and by working with the team to contribute to digitization efforts for the past several years. This has included, for the past two years, scanning artifacts with a high-resolution 3D scanner to create digital replicas for future study.
FSU Libraries + Classics
Though I have a background in Classics, my contribution at the site is not intended to be as a subject-matter expert. There are far more competent Classicists and archaeologists on staff supervising the excavation, and a fantastic team of student archaeologists each summer contributing to the hard and sweaty work of excavation. My job, rather, is to create a digital record of select finds deemed important enough to capture for future analysis in highly accurate 3D models. This is important for the team’s broader efforts of analysis and communication, as the archaeological site itself is only open for excavation during the month of June each year. Thoroughly recording all aspects of the process along the way helps analysis during the other 11 months of the year, and the addition of accurate 3D models can provide an additional aspect of data that the standard documentation photographs may not. This results in a flurry of activity to document, photograph, identify, describe, and scan as many finds and features as quickly as possible. This past season, for example, I was able to scan 65 models in just two weeks, which will allow for further analysis throughout the year, and bring the total of objects scanned to 109. These scanned models are in addition to the documentary “photomodels” of each excavation trench generated by the team using photogrammetry.
The process of scanning first requires the discovery of materials in the field by teams of students working under the direction of a team of trench supervisors with years of experience and graduate training in Classical Archaeology. When a potential artifact is discovered, the trench supervisor documents its location in the trench and its general characteristics, then sends it to the team working in the magazzino where it is classified and further recorded (including things like measurements and weight). Once the artifact is classified by the magazzino team, they work quickly to attempt an initial identification (i.e. type of pottery, type of coin, description of general sculptural features). They will then document this information and capture identification photos. Then, if the artifact is of particular noteworthiness or uniqueness, the team will set it aside for scanning, where it makes its way to my queue.
At this point, I use the library’s Artec Space Spider 3D to begin the process of scanning the object. This scanner uses a combination of structured light technology and real-time photogrammetry processing to capture surface topology and real-color images. Structured light scanning is a process by which a regular shape (usually a grid) of uniform dimension is projected onto a surface, then captured by a camera. Powerful software then analyzes the image to determine how that grid is altered by the surface of the object. When put together with enough other photographs of the deformed grid, the software can begin to triangulate features by comparing how the grid changes across the photo series.
Interested in Photogrammetry & 3D Capture?
Join the FSU Libraries Immersive Scholarship team this fall as we present our Photogrammetry Institute. For more details, check out the event schedule and prior event topics on our event page.
In addition to the structured light method, the 3D scanner I use also employs a trio of high-resolution digital cameras set at particular angles to capture an instantly-triangulated view of any object it is pointed at. By knowing the exact angle the three camera lenses are offset, the software that reconstructs the scan data can quickly patch together the surface and color information of a model in real-time. Paired with the higher-resolution structured-light approach, scans of objects ranging from coins all the way to statues can be captured in minutes with astounding detail.
Once each object is scanned, there are a few processing tasks I need to perform on the models to clean them up for presentation and eventual upload. The software paired with the Artec scanner automates most of this workflow, but still requires some decent processing power and time. This usually means that for every 10 minutes I spend scanning an object, I also have to spend 20 or so minutes performing hands-on cleaning and editing, with an additional 20-30 total minutes of hands-off processing time required for the computer to run through the various merging and texturing algorithms to produce the final model.
When things run smoothly, it usually takes anywhere between 20 minutes to an hour to finish the entire process of digitizing an artifact – from scanning to exporting a finished digital model. Not accounting for any errors in scanning or processing, the whole procedure is rather smooth and almost meditative. Where things get tricky are with objects on the extremes of size. One of the major limitations of the structured-light approach is the set size of the projected light-grid. While this allows for great accuracy when surface features deflect the grid, any features smaller than the grid are easily lost or left unrecorded. This is most commonly a factor on very thin objects (like coins) where the edge between two faces only deflects a small portion of the grid at any one time. This presents a problem in that the scanner can sometimes interpret this minor deflection as merely incorrect data capture – or “noise” – rather than actual surface information. And as good as the software usually is at tracking where the object benign scanned is at all times, thin edges like this are usually where the scanner begins to get confused, which often breaks the model.
However, once the model is scanned and processed, I then gather a set of important data about the object in a spreadsheet for upload into the Cosa team’s working database. The full database captures all aspects of the excavation and is a massive undertaking to organize and update throughout the year. My small portion of the database pulls the archaeological context for each digitized find, and adds data specific to the 3D modeling process. Eventually, the finalized information from this working database will be consolidated and uploaded to the FSU institutional repository, DigiNole for preservation and presentation.
But for the 3D objects I have scanned, however, the process of inclusion in DigiNole will happen much sooner. Thanks to the hard work of the Web Development team at the FSU Libraries, a new feature in DigiNole will allow for users to view and interact with 3D objects natively, much in the same way that the platform already supports PDF and audio/visual materials. This new feature will allow users the ability to engage with the 3D-scanned Cosa objects starting as early as this Fall, while research is actively ongoing. This is an exciting development for the project, as it allows for the fruits of FSU research endeavors to be held and cared for in an appropriate context within an academic environment. This is especially important, as for the past several years, the Cosa team has been uploading their 3D models of excavation trenches and objects to the public 3D repository Sketchfab. While Sketchfab allows for easy interaction with 3D models by the general public, not all of its content is scholarly (nor is all of it family-friendly, posing other problems). And though Sketchfab has done a great job supporting the work of cultural heritage institutions working to digitize and share their collections in 3D, their preservation and archiving obligations are very different than academic institutions’. We are proud to begin offering digital repository storage for 3D objects in the near future, and look forward to partnering with other researchers across campus to begin filling our repository with high-quality, high-impact 3D scholarship.
If you are interested in learning more about 3D scanning or photogrammetry techniques, be sure to check out our upcoming event series this fall, the Photogrammetry Institute, where we will focus on hands-on workshops with campus partners. Likewise, you can check out the recordings of the Spring 2023 events on our site here. For more information on services the FSU Libraries’ Immersive Scholarship team provides, feel free to explore our site at https://immersivescholarship.create.fsu.edu.
