Suppose you've just excavated a large building. You call it a palace. There's a good bit of it remaining, but not really enough to understand the function of all the spaces or how the entire building might have been used. But you have some ideas. You're a decent artist, so you put your concept of one of the main spaces on paper (see the results of Ann R. K. Yologist's reconstruction at the left; hover over to enlarge). The image endures for decades as the epitome of ancient royal reception rooms. The power of pretty pictures of the past perpetuates and persuades. I'll return to this image later.

Simulating the construction of an ancient structure on a computer is not the same as drawing a reconstruction on paper. The computer demands greater precision, more detailed information, and vastly more difficult decisions about how buildings were put together. Because we need to ask different questions in order to re-create an ancient built environment precisely, the process necessarily leads to a deeper understanding of the ancient world than possible with traditional visualization methods. Thus, using advanced graphics software (see a listing of some of the software solutions at the left; hover over each image to enlarge) can lead to new insight about ancient societies. The virtual worlds derived from these computer models can then be used, for example, to test hypotheses about the use of space, about the impact of different lighting methods, and ultimately about behavioral issues such as privacy and territoriality. By exploring detailed and complex simulations of ancient spaces in real time, we can really begin to understand how people of the past lived.

I'll mention some of the computer techniques we use to re-create ancient architecture, and how the process of producing a digital reconstruction differs from the one used to draw ancient buildings by hand. I'll discuss some of the conversations we have with archaeologists as we nudge them toward thinking about the implications of their finds in three dimensions, rather than only in two. Then I'll review some of the insight gleaned from our 3D computer modeling work on buildings from around the ancient Mediterranean world. I'll conclude with a look at some of the really cool stuff currently under development, and how the new digital archaeology will dramatically change how we will learn about the ancient world.

First, a little background. Ever since historians began chronicling buildings of the past through images, at least since the early 13th-century work of the French architect, Villard de Honnecourt (see one of his drawings at the left; hover over to enlarge), the primary means of visualizing architecture have been via plans, sections, and elevations (first published as a set by Albrecht Durer in 1547). The main reason for this centuries-old dependence on a few basic drawing types is the historically close relationship between the architectural profession and the archaeologists who, in the 18th and 19th centuries, supplied details of ancient monuments to architects for their new buildings, being designed in the Neoclassical style.

These building details were supplied in the format needed most by architects, namely plans, sections, and elevations (see the drawings by Stuart & Revett and how their details show up in the St. Pancras church, London, soon after; hover over each image to enlarge). It was precisely at that time that our profession developed and matured, with excavations often led or co-directed by architects. Now, in the midst of the Information Age, the Digital Revolution, and the new millennium, we still rely too heavily on those same static, two-dimensional representations of what is really a dynamic, behavior-influencing four-dimensional environment.

The idea, that representing architecture in arbitrary frozen slices through its spaces can somehow provide meaningful insight into the actual use of the spaces, is simplistic and biased. Simplistic because in two dimensions alone, we can not in the least begin to understand the complexities of construction, the visual impact of decorative detailing, the nuances of lighting, or the behavioral issues of privacy, territoriality, and viewsheds (see the rendering of the courtyard of the Hellenistic House of Many Colors, Olynthus, Greece, extracted from the Learning Sites virtual-reality educational package; hover over to enlarge). And biased because the traditional approach to documenting and visualizing space relegates the study of architecture to the same level as the study of individual artifacts isolated from their context and societal function. This is not to diminish the study of pottery or stone tools, since they too could benefit from interactive 3D modeling, but the built environment is a fundamental modifier of human behavior as well as an encapsulation and reflection of that behavior.

Thus, in order to truly understand how buildings were erected, why they were erected, and how they were used by real people, we must re-create them and experience them in simulations of their original complexity, which means in three dimensions (actually four dimensions, if you add the capability of walking through the spaces in virtual reality).

[ed. note 10/2019: At this point in the original presentation, Sanders showed the interactive VRML-based virtual reality re-creation of house AM600 from al-Meragh, ancient Nubia, Sudan, dated to the 1st c. CE. However, at the present, VRML is no longer supported by browsers, so what you see below is the same house reprogrammed in WebGL, the current standard for displaying interactive worlds online. Use your mouse or touchpad to walk through the world, updated from the original 2001 version (head for the silhouetted person). The red triangle in the upper right, with the 'i' is a fly-out menu to manipulate the sun's position and turn on and off a mini-map for orientation. Use the 'c' key on the keyboard to freeze the world so you can access the fly-out. Also, certain features in the world are highlighted in green, which signifies a hotspot for additional background information; again, use the 'c' key to freeze the world while you access the interactive supplements.]

The kinds of information needed to create a credible 3D digital re-creation of an ancient built environment include: drawings, photographs, actual materials, descriptions and measurements of the remains, and detailed discussions with the excavators (see the screen grab at the left from the virtual reality re-creation of a trench at the excavations at Poggio Colla, Tuscany, Italy; hover over to enlarge; ed. note 10/2019: in the original presentation, this was a virtual reality model). In order to replicate field conditions in the computer accurately and precisely, we need as much information as possible about the excavated remains. Each piece of evidence is used to corroborate others. But for a full reconstruction, the analysis of the evidence becomes much more thorough and the archaeologists become tested in their ability to envision how the building they are studying may actually have worked--and in doing so become much more aware than they ever would otherwise of how the ancient peoples thought about their spaces.

In order to achieve that level of investigation about the ancient built environment there needs to be a give-and-take between the model-builders and the excavation staff, and the questioning must be accomplished in language both understand, that is, in archaeological and architectural terms, not in computer jargon. Both sides need to be able to make the kinds of critical evaluations about the evidence that go well beyond what would be needed for traditional paper-based research or publication. And a reply such as, "we don't have enough evidence," is not good enough, especially if it's a euphemism for "I never thought of that."

For example, we ask: How high was that doorway? How were doors built in this period (what wood was available, what metal was used)? How was that door attached to the wall? How did it pivot? What kind of lintel was used? How did the lintel join the frame or wall? Which way did the door swing open? How was it secured? How did the door move against the floor? And that's just for one door. Simply drawing a door swing in plan does not force an archaeologist toward this level of understanding (see the screen grab showing the interior of the dining hall at Palike, Sicily, taken from the Learning Sites virtual world re-creating the building; hover over to enlarge). When we simulate an architectural element in the computer, we actually need to build it piece by piece and have it work just as it would have in real life. In VR, the materials do have to join correctly, be of the correct size structurally, and function as they did originally. The same detailed questions need to be asked about roof framing, the way rafters meet the top of a wall, how did the roofing sit on the rafters, how much light by what kind of lighting sources would it have taken to find your way through the building, would wall decoration ever have been seen? And more.

From analyzing all of these little construction details for an entire building, it is possible to learn about ancient behavior. All these little questions we ask and decisions that must be made replicate the subconscious decisions that the ancient builders made. The builders' decisions were all based on localized cultural conventions regarding how a feature is supposed to be built, to look, and to function. Through careful analyses, we can begin to understand those cultural conventions by using theories and observations from environmental psychology, semiotics, and ethnoarchaeology (but that is another paper).

Let me quickly run through how the evidence from the field becomes a fully interactive 3D model (see at the left some of the evidence we have collected to build a 3D model of an ancient Assyrian palace; hover over to enlarge). We begin by studying drawings, photos, measurements, ethnographic evidence, and trench descriptions so that we understand the remains and the site.

We then add relevant information into a CAD program--one that models directly in 3D as the data is being entered (it is not simply an extrusion of the plan; see a sample screen grab at the left showing a wireframe diagram of an Assyrian palace in 3D modeling software; hover over to enlarge). Care is taken with layer names and layer creation to isolate all important aspects of the building elements--each surface face, each detail. The model is constructed to our internal specs with an eye toward its later importation into virtual reality viewing software.

Then come the detailed questions. As the model develops, we are in constant contact with the field team or other experts regarding the materials that remain and were likely used during the building's occupation. We build up an inventory of materials and colors (see the beginnings of texture mapping at the left, also from our Assyrian palace model; hover over to enlarge). Color is real issue, since no two computers deal with color the same way; nor will any two monitors display the same color equally. This situation gives us no end of headaches.

Farther along, we add the appropriate lighting, and then cameras to the scene so that we can generate renderings for evaluation (see a sample rendering of our Assyrian palace model at the left; hover over to enlarge.

Once the modeled spaces are approved, we move into production of furnishings and other artifacts, either from the excavated evidence itself or from comparable buildings elsewhere (see a nearly completed rendering of the palace model at the left; hover over to enlarge).

Spatial tests then begin, either with static renderings to focus on specific lighting or viewpoint characteristics or in real-time via virtual reality. We can then get a sense of what the people who built the buildings actually saw as they moved through the spaces (see the sample screen grab of the throne room of the Assyrian palace from our virtual world; hover over to enlarge).

Let me give you some examples--it would be best to show these in VR, but due to lack of time, I will fall back to simple renders--even though they're 2D. If anyone is interested in a live VR demo, see me later. The advantages of a 3D model include the ability to modify it easily and to generate renders from any viewpoint. Too often hand drawings become solidified despite their errors or conjectures because it is just too time-consuming to redraw them.

A. The Northwest Palace at Nimrud (northern Iraq). Here are two views of a throne room in an ancient Assyrian palace (hover over the images at the left to enlarge). The Learning Sites rendering, taken from our virtual reality model, allows us to study lighting, structure, the relation of sculpture to processions, the exact viewpoint of people in the space, and the relationship of the king to the wall relief program--in real time to see the effects. Note in the images, how each render treats the ceiling, colors, and scale.

B. Tsoungiza, Ancient Nemea (Greece). Here we can see the difference between traditional 2D representations ( the cient's floor plan) and Learning Sites interactive reconstructions (the exterior rendering; hover over the image to enlarge). Our virtual reality model allows researchers to study how floor levels influence the use of space, how space influences architectural decisions about window placement and light floow, how the roofline and door at rear might work, and how light, smoke, ceiling heights influence interior functions.

C. Nemrud Dagi, Turkey. For this remote mountaintop sanctuary, our interactive model allowed scholars to envision the size of dexiosis reliefs and their location in relation to the colossal statues, examine what could be seen and what was blocked by the reliefs, how much time it takes to walk around site, what colossal sculptural details could be seen from the terraces. None of these could be studied previously, even if one could visit the site.

D. The hestiaterion, Rocchicella-Palike, Sicily. For this unusual building, a soldier's dining hall, our 3D model envisioned and tested door swings, coursing patterns, use of a specific foot measure, floor-level changes, and the impact on spatial use by windows and light, the structure of rooms, door details, jambs, lintels, and fascia boards in much more detail than could be possible by traditional means.

Those are examples of some of our current work. We also have an R&D department that continues to generate new means whereby our experience of the past can be enriching and enlightening (such as the java-based time slider, seen at the left; hover over to enlarge).

Where will these kinds of visualizations lead in the future (hover over the image at the left to enlarge)? Simple extrapolations of current technologies and current applications show that virtual worlds will continue to become more detailed and complex with each leap in the capability of graphics cards. Today's worlds are already inhabited both by avatars (3D characters that become your surrogate in the world) and by freely roving or guided independent agents that respond either to your movement or your queries or move to their own preprogrammed actions. We can watch simulations of ancient behavior, and see how they play out in the virtual environment. We can become time-traveling anthropologists collecting information with no impact on the simulated cultural activities or directly interact with the virtual agents for guidance in understanding the ancient built environment. Virtual ancient worlds will become gloriously cluttered with the artifacts of the culture, from furniture, to pots, to tools, to weapons, to foodstuffs. We'll have sounds and smells (the technologies for both already exist). Each artifact in the virtual world will link to complex semantic networks offering text, 2D images, voice, and video background data about the selected object or architectural feature.

We'll have dynamically changeable virtual worlds in which hypotheses can be tested in real time right before our inquisitive eyes. Walls can be adjusted in size or material, doorways changed in height, the artifact distribution changed, and we'll watch as mudbrick decays. When we've configured the virtual ancient world to our satisfaction, turn the timedial ahead 1000 years or 5000 years and see if the virtual results matches the excavated record. And there's more. We'll be able to test hypotheses at two scales--large, using full-scale projected or wall mounted holograms or small, via PDAs attached to augmented reality glasses mounted into a wireless headset connected to the Web or a local server providing virtual reconstructions as we wander around actual archaeological remains. These too have already been demonstrated and a version of them already in use in for archaeology. When we add that the data behind these complex worlds will be coming from digital and noninvasive data gathering techniques, then we can truly appreciate the profound affect these burgeoning changes will have on our profession.

I've tried to demonstrate that the process and the results of creating 3D computer models of ancient architecture can lead to new insight about how ancient cultures used their built environment. The process also creates a more critical evaluation of architectural evidence than traditionally accomplished when architecture treated as merely another artifact type to be quickly drawn, minimally published, and simply studied as a container for other, supposedly more significant, artifacts. The behavioral clues encoded in the built environment are not easily deciphered, but much more easily understood when the spatial system is re-created so that it's analysis comes close to a first-person experience. All this can come to pass as long as we realize that our study and visualizations of the past need not be constrained by the methods or technologies of the past, and that our questions about the past can only benefit from these new means of visualizing the answers.