4/17/01

John Sanford, News Service (650) 736-2151; e-mail: jsanford@stanford.edu

Computer technology enlisted in effort to assemble pieces of 1,500-year-old jigsaw puzzle

Thirty newly discovered fragments of a giant, third-century map of Rome will travel from Italy to Stanford in late May.

"It's unprecedented," said Associate Professor Marc Levoy of the Computer Science and Electrical Engineering departments. "Antiquities rarely leave Italy unless they're on a major tour of, say, the Metropolitan Museum of Art in New York."

Laura Ferrea, an Italian archaeologist and museum curator who works for the Capitoline Museums, will accompany the fragments to Stanford.

Levoy is the reason the fragments are coming here. Using laser rangefinder technology, he and his team of student researchers will spend a day scanning the 30 pieces into a computer. The data then will be used to create digital, three-dimensional models of the fragments, which will be included in a database containing 3-D models of the other 1,163 pieces of the map known to exist.

"[The Italians] are very interested in getting a complete database of these map fragments, and they trust us," Levoy said.

The map is called the Forma Urbis Romae, or the Severan Marble Plan of Rome. It once was attached to a wall of the Templum Pacis in Rome and measured about 60 feet across and 42 feet high, detailing every street, every building, every ground-floor room and every staircase in the city ­ a feat of map-making that never has been matched in modern times.

But as the Roman Empire crumbled in the fifth century, people began stripping the map for the marble, which they used in construction or to make lime for concrete. Eventually, the plan fell off the wall, and most of the building collapsed.

The 30 fragments coming to Stanford were discovered in 1999 during restoration work on a nearby building. It was the first time fragments had been found since 1956.

"In Rome, repair work and archaeology go hand in hand," Levoy said. "You can't dig in the city without expecting to encounter remnants of its ancient past."

As an undergraduate, Levoy majored in architecture, earning a bachelor's degree and a master's from Cornell in 1976 and 1978, respectively, and he has always maintained a keen interest in art.

"One semester, I was bored with the courses in my major, and I kind of blew them all off and wrote a 50-page paper on Michelangelo for a history course I was taking," he said.

Now, in his career as a computer scientist -- he earned his doctorate in computer science from the University of North Carolina at Chapel Hill in 1989 ­ Levoy is working to enlist technology in the service of the arts and humanities.

"If you think about the teaching of art history, it's been revolutionized in the last 50 years because of the development of high-quality color photography," he said. "What color photography has done for the graphic arts, I'm hoping that 3-D models can do for the plastic arts. So if you want to study a sculpture, you can rotate a model of it around the computer screen, re-light it and so on."

The story of the Digital Forma Urbis Romae Project began in early 1997, when Levoy went to Florence to give a talk about three-dimensional scanning and find some advocates for a sabbatical project he wanted to do: Scan the works of Michelangelo. (For more information about this project, visit http://graphics.stanford.edu/projects/mich.)

In Florence, he met with a group of government officials, including Susanna Le Pera of the Archaeological Superintendency of the City of Rome. Le Pera had studied the Forma Urbis.

"She said, 'Are you looking for something else to scan? How about a giant map of Rome?'" Levoy recalled.

In 1998, Levoy; Brian Curless, a faculty member at the University of Washington; and 30 researchers, including students and staff from Stanford and the University of Washington, traveled to Italy to scan, in three dimensions, sculptures and architecture of Michelangelo, as well as the pieces of the map.

Funding for the initial scanning came from the Interval Research Corp. and the Paul Allen Foundation. Current work on the project is made possible largely by a grant from the Stanford President's Fund.

Levoy and members of the team spent the entire academic year in Florence. In May and June, they commuted to Rome to scan fragments of the Forma Urbis.

It took six people, including Levoy, 25 days of around-the-clock work to finish the job. At one point, the ceiling of the museum basement they were working in began to collapse.

"I have felt like Indiana Jones on more than one occasion on this project," Levoy said, sitting in his third-floor office of the Gates Computer Science Building.

One of driving forces behind the Forma Urbis project is to create a database of the map fragments and distribute it to archaeologists. A majority of the fragments are sealed in crates in the basement of the Museum of Roman Civilization in Italy and, thus, inaccessible to most scholars.

At present, Levoy's team has digitally reconstructed about 70 percent of the individual fragments. These three-dimensional images have been entered into a database. Levoy said he expects the remaining 30 percent of the fragments to be digitally reconstructed by summer.

On a computer screen, the fragments look astonishingly real, like pieces of cold marble floating in the vacuum of space. They can be effortlessly rotated and inspected from various angles. Depending on the size of the fragment, each 3-D model consists of between one to 50 million tiny triangles. Each triangle measures 0.25 millimeter on a side.

A computer program could help researchers determine how pieces of the puzzle fit together, Levoy said. Recently, two of his doctoral students, David Koller and Natasha Gelfand, have started to develop algorithms to do the job.

Only about 15 percent of the map is known to exist, but these shards are believed to represent some of the most important and famous parts of the city: Researchers already have identified fragments representing the Colosseum and Circus Maximus. In all, roughly 200 pieces have been identified and, in some cases, fit together. About 500 pieces have not been identified, and about 400 have no surface incisions -- scholars have theorized that they might represent parts of the Tiber River, the center of plazas or, perhaps, the border of the map. Nobody knows for sure.

However, Levoy notes that it will be impossible to solve the entire Forma Urbis puzzle.

"But if we can fit together even a few pieces, we might be able to identify the location of monuments ... that have been lost since antiquity," he said.

However, regardless of whether a computer succeeds in piecing together fragments, the database still will be valuable for archaeologists, Levoy said.

"Putting archaeological artifacts online is important, but it's uniquely important in this case because there are so many pieces," Levoy said.

Jennifer Trimble, an acting assistant professor of classics and project collaborator, agrees.

Trimble advises Levoy and his crew about what information archaeologists likely would be looking for from such a database. When efforts to assemble the map electronically are undertaken, she also will be able to help determine whether a match is right or wrong.

"The map doesn't just tell you about all the glamorous sites, like the Colosseum," she said. "It tells you about housing and commercial space and how people lived in the city and worked -- all at the micro level. All this information would be hard to get without tearing up large swaths of modern Rome."

She described Rome as the world's first "mega-city." The Forma Urbis can help answer questions about how people were kept healthy, how they were fed and how authorities enforced the peace in this large metropolis, among many other things, Trimble said.

"One of the striking things is that, for the most part, ancient Rome was not zoned in the way that modern cities are," she said. "Monumental space, commercial space, living space and religious space tended to be mixed together. People working now on how to solve urban sprawl want to mix it up in the same way. See? Ancient history is deeply relevant."

-30-

By John Sanford