3D Models as Analytical Tools

This is the fourteenth in a series of posts exploring 3D modeling in Mediterranean and European archaeology. For more in this series click here. We hope these papers will start a discussion either in the comments of the blog or on Twitter using the #3DMedArch hashtag.

Ethan Gruber, American Numismatic Society


Several years ago, when I was still a graduate student at the University of Virginia, I took part in a sort of Digital Humanities speed dating event hosted by the Scholars’ Lab, for whom I was employed as a web developer at the time. This event was geared toward creating connections between technologists and humanists at the university. I remember a brief encounter with two archaeologists in the Anthropology department (which was separate from my own Classical Archaeology program, which was part of Art History department) who saw only dubious benefit of 3D within the discipline. Models were seen by my colleagues as more of a sexy technological endeavor with little or no scholarly application. I discussed with them some of my own work in lighting simulation of 3D reconstructions, similar to the work by recently undertaken by Bernie Frischer and John Fillwalk on the solar interaction between the Horologium Augusti and the Ara Pacis, and so I hoped they developed a greater respect for the technology after enhancing their understanding of it.

We have seen in previous blog posts the value of 3D models in archaeological documentation and pedagogy, and so I wanted to discuss the value of architectural models as scholarly tools, enabling us to visualize the built environment and test hypotheses in ways that were previously impossible. My aim in this post is not necessarily to detail the results of these simulations (since you can find open access papers about most of these projects), but rather the provide a synopsis of the thought processes which go into creating the reconstructions and the evolution of my own sunlight simulation methodology, which has grown more sophisticated over the years.

The House of the Faun

My first foray into lighting simulation was purely accidental. In 2008, I used Autodesk Maya to create a 3D model of Pompeii’s House of the Faun, one of the largest and best known houses in the city. The model was created for a graduate seminar taught by John Dobbins to test hypothetical sight lines in a typical Italian atrium house. Sight lines are often presumed to be an integral part of Roman architecture. Architectural historians quite often depend on two dimensional plans to formulate their hypotheses. In theory, the open doors of the atrium house would invite passersby to gaze inside, revealing an architectural and social hierarchy of sorts: one may see the gardens beyond the tablinum, but may only enter this space if he or she is of higher social status, invited by the paterfamilias into this more intimate environment. The House of the Faun is a more special case, for just beyond the peristyle is the exedra containing the famous mosaic of Alexander the Great, and beyond this is yet another, larger peristyle. As it turns out, the sight line analysis of the 3D model of the house reveals it is basically impossible to see the exedra from the street, when basing the model upon the standard reconstructions found in printed illustrations.

FIG1 faun85Figure 1
View from the fauces to the Alexander exedra in the House of the Faun

Near the end of the semester, I happened upon a Maya script written by a lecturer of architecture at TU Delft, Thijs Welman, which allows a user to input the latitude and longitude of the structure, as well as the precise minute in time (back to 2000 B.C.), in order to accurately set the model’s sunlight angle. Once the angle is set, the model can be rendered with physically accurate light and shadows. I corresponded with Thijs briefly several years ago, and he was surprised that his script had found its way into the hands of an archaeologist–it had merely been intended to accurately render sunlight in the models of architecture students. He even extended the script, by my request, to animate the sunlight entity in Maya, enabling the rendering of timelapse videos.

I proceeded to render the exedra and the mosaic contained within on the equinoxes and solstices of 100 B.C., the approximate date of the mosaic’s installation. In the summertime, the angle of the sun was high enough that the mosaic was completely enshrouded in shadow. In the winter, however, the low sun projected six long shadows across the mosaic, certainly not an ideal viewing condition. 

FIG2 faun87Figure 2
About noon, December 21, 100 B.C.: commonly accepted reconstruction. Click image for walkthrough and timelapse animation.

When John Dobbins and I went to Pompeii the following summer (2009), we went into the House of the Faun and began to take note of peculiar architectural features which ultimately led us to conclude that when the new Roman overlords colonized Pompeii following the Social Wars, the new Roman patron of the house planned serious renovations of the peristyle, installed the exedra, the Alexander mosaic, and removed several columns and part of the portico to enhance the experience of encountering the space. For further details, please see our paper, presented at CAA in 2010.

FIG3 faun88Figure 3
About noon, December 21, 100 B.C.: alternate reconstruction.

Timelapse videos are available for both the summer and winter solstices.

The House of the Drinking Contest

Also in 2009, I began working on a reconstruction of the House of the Drinking Contest, a late Severan period house in Seleucia Pieria, the port city of Antioch (also published with Dobbins in the proceedings of CAA 2010). There were two purposes for this model: 1) test John’s sight line hypothesis published in Antioch: A Lost Ancient City (2000, ed., Christine Kondoleon) and 2) import images of the house’s elaborate mosaics into the model and apply/improve the lighting simulation methodology first tested on the House of the Faun to recontextualize the artworks–to experience them in the same environment that the ancient inhabitants of the house once (theoretically) experienced them. It was during this project that I began delving more deeply in the technological and theoretical aspects of the simulations and found that these methodologies had been applied first in 1994 by Simon Ellis (published in the fourth TRAC proceedings) and in later publications. Ellis’ methodology has been tremendously influential on my own scholarship, and I consider him to be one of the true pioneers in the adaptation of 3D models to archaeological research. The conclusions that he reached in his houses were quite similar to my own in the House of the Drinking Contest: that the setting sun in the spring and summer months shone through large doors of the triclinium, illuminating a grand central mosaic. It is at approximately this moment that guests would be gathered by the patron to take part in the symposium, with the most brilliant work of art illuminated with its glass and polished stone tesserae visible to all in a way that would otherwise never be visible earlier in the day or other times of year.

FIG4 render112Figure 4
Triclinium: June 21, A.D. 230, 6pm.

FIG5 planFigure 5
Plan of the House of the Drinking Contest

With respect to sight lines, Dobbins hypothesized that irregular intercolumniations within the colonnade bordering on the north side of the courtyard allowed for direct views from rooms on the north side of the house to the south. But what lay south? The archival photographs from the 1930s provide some glimpse, but by some stroke of luck, I was able to find a georeferenced photograph in Panoramio (discovered through Google Earth) of the same viewpoint taken from within one or two hundred meters from the site. Incorporating this photo into the model, I was able to simulate this hypothetical view through picture windows, glancing toward the shoreline of the Mediterranean and Mount Casius beyond. I have subsequently found a plethora of literary and archaeological examples of views to mountains and seas. But in archaeology, as we know, context is everything. How exactly do we know that there wasn’t a structure directly south of this house blocking this view? We don’t know for certain, but we can be reasonably sure this was not the case, based on the topography. The House of the Drinking Contest and those those houses built to the east and west were built on a terrace–the land sloped downward to the south, toward the sea. So while there may have been more houses to the south, they would have been lower to sea level, and therefore unlikely to have obstructed the view from this house. I believe this is an important point to drive home to anyone modeling reconstructions for scholarly purposes: artifacts must be placed back into context within the architecture, but one needs to take into consideration the larger topography surrounding the architecture. How do the walls and roofs affect the illumination of the mosaics? How does the natural or urban environment interact with the house itself?

FIG6 render121Figure 6
View toward the sea and Mount Casius to the south.

The Temple of Artemis at Ephesus

The final test case for 3D as a platform for scholarly analysis I would like to discuss (briefly) is a reconstruction of the Temple of Artemis at Ephesus. I modeled this in a graduate seminar on Anatolian archaeology to test a hypothesis postulated by Anton Bammer that one could view the epiphany of Artemis from within the temple: a sight line from the central pedimental window, toward and above the rear wall of the altar, to the cult statue below.

FIG7 fig17 altar view linesFigure 7
Witnessing the epiphany of Artemis from the pedimental windows.

While this may have been possible in the Archaic temple built by Croesus in 550 B.C., it would not have been possible in the 4th century temple, based on archaeological measurements. I applied the sunlight simulation methodology to observe the temple at different times of day throughout the year, but nothing particularly notable stood out. Unlike in the House of the Faun, where only the walls and roofs of the house affect the lighting within it, a structure of the size of this temple is largely unaffected by the built environment surrounding it. There was, however, potential for the physical topography to affect the lighting of the temple. The Austrian Archaeological Institute was gracious enough to provide me with their 10 meter resolution Digital Elevation Model data for Ephesus. To the southwest of the temple (between the temple and the Hellenistic/Roman city) lay the mountain, Panayirdağ. It was necessary to observe the interaction between this mountain and the sunlight, and in the waning hours on the winter solstice, shadows cast by the mountain fell upon the temple. It is not apparent that the architects of the temple took into deliberate consideration the sun’s effect on the structure, unlike those who designed the Augustan monuments in the Campus Martius in Frischer’s and Fillwalk’s project.

FIG8 sight lineFigure 8
Hypothesis applied to the reconstruction


Three dimensional models and simulations have enormous potential in furthering archaeological scholarship. They enable us to test hypotheses that could otherwise never have been examined, to observe the environmental conditions and context in which artworks and artifacts were placed in their own time, and, furthermore, formulate ideas that would never have been imagined by our predecessors before the digital age. John’s and my work on the House of the Faun is an example of this, I think. Simulations within the 3D reconstruction altered our perception of the ruins, and the next time we visited them, and we developed a different interpretation of the evidence.

The final issue that must be noted is that everything I have discussed is somewhat theoretical. While sunlight angles themselves are grounded in mathematical fact, reconstructions of ancient architecture are inherently theoretical. We create models which are plausible, based on data we have available, but these data are often incomplete or confusing. Thus, the simulations are not absolute truth, but rather a potential explanation of the past. The flexibility of 3D models allows us to address alternative scenarios based on new evidence or interpretations.


Ellis, Simon. 1994. “Lighting in Late Roman Houses.” In TRAC 94: Proceedings of the Fourth Annual Theoretical Roman Archaeology Conference, held in Durham, 65-72.

——–. 2000. Roman Housing. London: Duckworth.

——–. 2007. “Shedding Light on Late Roman Housing.” In Housing in Late Antiquity, edited by
Luke Lavan, Lale Özgenel, and Alexander Sarantis, 283-302. Leiden, Boston: Brill.

Gruber, Ethan and John Dobbins. “Illuminating Historical Architecture: The House of the Drinking Contest at Antioch.” In the proceedings for Computer Applications and Quantitative Methods in Archaeology 2010 held in Granada, Spain 6-9 April, 2010.

——–. “Modeling Hypotheses in Pompeian Archaeology: The House of the Faun.” In the proceedings for Computer Applications and Quantitative Methods in Archaeology 2010 held in Granada, Spain 6-9 April, 2010.

Lippolis, Isabella Baldini. 2007. “Private Space in Late Antique Cities.” In Housing in Late Antiquity, edited by Luke Lavan, Lale Özgenel, and Alexander Sarantis, 197-237. Leiden, Boston: Brill.


One Comment

  1. Wow !


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s