A panorama of the area behind the Temple of Castor (Post Aedem Castoris)

(Click on the image above to view it at full resolution. Or view it as a QuickTime VR panorama: small version (500KB), big version (1.7MB). Don't forget to zoom in using (Shift) and (Ctrl).
Or save it to your hard drive, launch it in a QuickTime Window, and stretch the window.)

A week on an archaeological dig in the Roman Forum

Photographs and text by Marc Levoy
(with help from Jen Trimble and Danielle Steen)
June 30 - July 4, 2003

My involvement over the past five years on the Digital Forma Urbis Romae Project has sparked my interest in exploring how computer graphics and computer imaging might be used during an archaeological excavation - to help record the site. So when my collaborator on that project, Professor Jennifer Trimble of Stanford's Classics Department, together with Darius Arya (Institute for Roman Culture) and Andrew Wilson (Oxford University), got permission to begin a new excavation in the Roman Forum, I was on bended knee asking her if I could join the dig for a week. To my delight, she said yes (after consulting with her co-directors). Their permission was to begin five exploratory trenches in the area behind the Temple of Castor (Post Aedem Castoris). Here are some pictures I took during that torrid, terrific week in Rome.

Unless otherwise noted, all pictures were taken using a Canon EOS-D30 digital camera, edited using Photoshop 7, and downsized for web viewing. The panorama above was acquired using a handheld camera and stitched together using PanaVue Image Assembler 2.02a.

At left is a map of the Roman Forum, taken from Samuel Ball Platner's The Topography and Monuments of Ancient Rome (1904), courtesy of Prof. Felix Just, Loyola Marymount University. At right is an annotated version. The asterisk shows where I stood to take the panorama that appears at the top of this web page. The other annotations are explained below.

At left is a view (numbered 1 on the annotated map above) of the north wall of the massive Domitianic structure connected to the palace complex on the Palatine Hill above. The true function of this building is still debated. In the midground is trench A (blue rectangle on the map), roped off with striped tape. At right is another view of the trench, taken later in the week.

At left is a view (numbered 2 on the annotated map) of the inside of the Domitianic structure. Its walls are 25 meters tall and over 3 meters thick. Turning 90 degrees to the right gives us a view down the ancient Vicus Tuscus. Roughly paralleling this road, we see (at right) the area of trench E (green rectangle on the map).

The team on this dig consisted of 12 graduate and undergraduate students from Oxford, and a matching number from Stanford. At left, part of the team takes a break in the shade. At right, co-directors Andrew Wilson (extreme left) and Dar Arya (extreme right) pore over the plan for trench E, assisted by Stanford PhD student Lide de Jong (left center) and site assistant Dora Cirone (right center).

Although my main reason for joining the dig was to learn the methods of field archaeology, I also expected to work. At left I help roll a broken column out of the way of trench A. These things are heavy! How the Romans quarried, transported, and erected the gigantic columns of the Pantheon is beyond me. At right, I clear the topsoil from a strip in the same trench. (The picture at left was taken by Jen using her digital camera.)

Here are my trenchmates. At left are Zoe Cox and Matt Symonds, both from Oxford. Matt supervised the digging while I was there. At right are Marden Nichols (from Stanford) and Matt. The other supervisor of this trench is Danielle Steen (pictured later). I strived mightily to pick up a British accent from Zoe, but I wasn't there long enough.

The other supervisor of trench A was Danielle Steen, one of Jen Trimble's PhD students at Stanford. In these two shots, Danielle and Jen try to reconstruct the convoluted history of the wall that borders the trench (visible behind Zoe and Matt above).

Meanwhile, elsewhere on the site Lide (at left) struggles to learn the intricacies of the Leica total station. As I quickly discovered, you don't get far with this machine without reading the manual. At right, Andrew tutors Stanford undergraduate Leila Ben-Youssef on the operation of the slightly less intricate dumpy level.

The problem with joining a dig for its first week is that not much happens during that week. Well a lot happens, but not much in the way of finding antiquities. So, in a fit of what Jen called "dig envy", I snapped these pictures of an Italian dig - evidently further along - elsewhere in the Forum.

So did we find anything during that first week? Sure we did! First, Matt found a marble block just below the topsoil, possibly in situ, although tilted slightly, which made us suspicious that it had been dumped there during recent excavations. Matt was reluctant to pose beside this questionable find, but I insisted.

With only 30 minutes to go before quitting time on my last day, while I was clearing a large area in trench A, one of the swings of my pickaxe was rewarded with the resounding ring of steel on marble. Switching to a trowel and working under Matt's careful supervision, I uncovered one corner of a large marble slab, apparently in situ. Boy was I excited!

I must have taken a hundred pictures of that silly slab. I'll restrain myself and include just one. Danielle and I measured its location relative to a nearby wall, and it matched an ancient drain slab found during a previous excavation (yellow arrow on plan). However, Jen tells me that after two more weeks of excavation, it was decided that my slab wasn't in situ, and probably isn't a drain cover. Harrumph, who needs the truth anyway?

Finally, here are solo portraits of (two of our three) fearless leaders, Jen Trimble and Andrew Wilson. Don't you think Andrew would look natural carrying an Indiana Jones bullwhip?

P.S. As I told anybody who would listen, our dig was in exactly that corner of the Forum that has most fascinated me since childhood - the complex, partially excavated ramparts of the Palatine Hill. As proof, I offer this crude (and slightly romanticized) pencil sketch I made of our site 18 years ago. At right is a photograph from the same viewpoint, taken during my week on the dig.
P.P.S. There is a direct connection between my week on this dig and my involvement in Stanford's Digital Forma Urbis Romae (FUR) Project. The FUR, also known as the Severan Marble Plan, is a giant marble map of ancient Rome, which comes down to us as a jigsaw puzzle with 1,186 pieces. Here is fragment #018a. It clearly shows a portion of the Temple of Castor ([Cas]toris). However, its depiction of the surrounding structures disagrees with what we find on the ground. Morever, the features incised on this fragment were never present on the ground at a single point in time. This raises interesting questions about the authority and temporal consistency of the Marble Plan - questions that Prof. Trimble is addressing in her research.

More games with digital imaging: high dynamic range photography

This photograph shows the ancient ramp leading from the antechamber near our tool shed up to the Palatine Hill. This is a challenging photographic problem, because the ramp is unilluminated, except for sunlight streaming through two openings at the far end. No single image can capture the full range of intensities in this scene. My solution was to use high dynamic range (HDR) photography. The first seven images were taken one f/stop apart by a digital camera, which I mounted on a tripod to prevent motion between shots. These images were fed into Paul Debevec's HDR Shop software, which combined them to create a single photograph in which each pixel is represented using a float-pointing number, instead of the usual 8-bit number (between 0 and 255). Such an image can be stored using a variety of HDR-enhanced file formats. However, since current-day web browsers cannot read these formats, and everyday computer screens cannot display them, I reconverted my HDR photograph using a non-linear "tone mapping operator" to create a normal (8-bit per pixel) JPEG file, shown at right above. By using a non-linear mapping, one can do a reasonable job of preserving both darks and highlights, although subtle gradations are inevitably lost. The operator I used was from R. Fattal et al., Gradient Domain High Dynamic Range Compression, Proc. Siggraph 2002.

So what did I learn from my week as an archaeologist? First, I learned that real archaeology is not an Indiana Jones movie; it's a slow, methodical, scientific process. Of course, science has exciting moments, as the pictures above attest, but these may be separated by days of hot, dusty, fruitless labor. Moreover, a lot of time is spent measuring and recording, rather than digging. As Jen and Andrew instructed us, archaeologists destroy as they dig; what separates them from mere looters is that they record what they find. During this particular dig, the team employed the recording method of the Museum of London Archaeological Service (MoLAS).

I also learned that, in addition to being labor-intensive, the recording process is only moderately accurate, and in some cases inadequate for the crucial task of planning where to dig next. Fortunately, the last ten years have seen a revolution in digital imaging technologies, some of which might be useful. Examples include imaging rangefinders (a.k.a. scanning total stations), location-aware digital cameras (i.e. cameras equipped with differential GPS sensors), and computational imaging - in which multiple images are combined computationally to create a single synthetic image of very high spatial resolution, dynamic range, depth of field, and so on. Multiview imaging might also play a role. The panorama above is a simple example, but how about a Matrix-style "Bullet Time" flyaround of the trench? One could be made simply by walking around the edge of the trench with a handheld, location-aware video camcorder.

Needless to say, new technology must be introduced thoughtfully; otherwise, it will bury the archaeologist in a mountain of raw data whose interpretation takes longer than the manual process it is replacing. But the possibilities are enticing, and computer scientists are jumping into the fray. Indeed, several annual computer graphics and computer vision conferences now devote a special session to cultural heritage applications. The coming decades will be exciting ones in the development of archaeological practice.

levoy@cs.stanford.edu

	At left is a map of the Roman Forum, taken from Samuel Ball Platner's The Topography and Monuments of Ancient Rome (1904), courtesy of Prof. Felix Just, Loyola Marymount University. At right is an annotated version. The asterisk shows where I stood to take the panorama that appears at the top of this web page. The other annotations are explained below.
	At left is a view (numbered 1 on the annotated map above) of the north wall of the massive Domitianic structure connected to the palace complex on the Palatine Hill above. The true function of this building is still debated. In the midground is trench A (blue rectangle on the map), roped off with striped tape. At right is another view of the trench, taken later in the week.
	At left is a view (numbered 2 on the annotated map) of the inside of the Domitianic structure. Its walls are 25 meters tall and over 3 meters thick. Turning 90 degrees to the right gives us a view down the ancient Vicus Tuscus. Roughly paralleling this road, we see (at right) the area of trench E (green rectangle on the map).
	The team on this dig consisted of 12 graduate and undergraduate students from Oxford, and a matching number from Stanford. At left, part of the team takes a break in the shade. At right, co-directors Andrew Wilson (extreme left) and Dar Arya (extreme right) pore over the plan for trench E, assisted by Stanford PhD student Lide de Jong (left center) and site assistant Dora Cirone (right center).
	Although my main reason for joining the dig was to learn the methods of field archaeology, I also expected to work. At left I help roll a broken column out of the way of trench A. These things are heavy! How the Romans quarried, transported, and erected the gigantic columns of the Pantheon is beyond me. At right, I clear the topsoil from a strip in the same trench. (The picture at left was taken by Jen using her digital camera.)
	Here are my trenchmates. At left are Zoe Cox and Matt Symonds, both from Oxford. Matt supervised the digging while I was there. At right are Marden Nichols (from Stanford) and Matt. The other supervisor of this trench is Danielle Steen (pictured later). I strived mightily to pick up a British accent from Zoe, but I wasn't there long enough.
	The other supervisor of trench A was Danielle Steen, one of Jen Trimble's PhD students at Stanford. In these two shots, Danielle and Jen try to reconstruct the convoluted history of the wall that borders the trench (visible behind Zoe and Matt above).
	Meanwhile, elsewhere on the site Lide (at left) struggles to learn the intricacies of the Leica total station. As I quickly discovered, you don't get far with this machine without reading the manual. At right, Andrew tutors Stanford undergraduate Leila Ben-Youssef on the operation of the slightly less intricate dumpy level.
	The problem with joining a dig for its first week is that not much happens during that week. Well a lot happens, but not much in the way of finding antiquities. So, in a fit of what Jen called "dig envy", I snapped these pictures of an Italian dig - evidently further along - elsewhere in the Forum.
	So did we find anything during that first week? Sure we did! First, Matt found a marble block just below the topsoil, possibly in situ, although tilted slightly, which made us suspicious that it had been dumped there during recent excavations. Matt was reluctant to pose beside this questionable find, but I insisted.
	With only 30 minutes to go before quitting time on my last day, while I was clearing a large area in trench A, one of the swings of my pickaxe was rewarded with the resounding ring of steel on marble. Switching to a trowel and working under Matt's careful supervision, I uncovered one corner of a large marble slab, apparently in situ. Boy was I excited!
	I must have taken a hundred pictures of that silly slab. I'll restrain myself and include just one. Danielle and I measured its location relative to a nearby wall, and it matched an ancient drain slab found during a previous excavation (yellow arrow on plan). However, Jen tells me that after two more weeks of excavation, it was decided that my slab wasn't in situ, and probably isn't a drain cover. Harrumph, who needs the truth anyway?
	Finally, here are solo portraits of (two of our three) fearless leaders, Jen Trimble and Andrew Wilson. Don't you think Andrew would look natural carrying an Indiana Jones bullwhip?
	P.S. As I told anybody who would listen, our dig was in exactly that corner of the Forum that has most fascinated me since childhood - the complex, partially excavated ramparts of the Palatine Hill. As proof, I offer this crude (and slightly romanticized) pencil sketch I made of our site 18 years ago. At right is a photograph from the same viewpoint, taken during my week on the dig.
	P.P.S. There is a direct connection between my week on this dig and my involvement in Stanford's Digital Forma Urbis Romae (FUR) Project. The FUR, also known as the Severan Marble Plan, is a giant marble map of ancient Rome, which comes down to us as a jigsaw puzzle with 1,186 pieces. Here is fragment #018a. It clearly shows a portion of the Temple of Castor ([Cas]toris). However, its depiction of the surrounding structures disagrees with what we find on the ground. Morever, the features incised on this fragment were never present on the ground at a single point in time. This raises interesting questions about the authority and temporal consistency of the Marble Plan - questions that Prof. Trimble is addressing in her research.