The Digital Michelangelo Project (original) (raw)

8/27/03 - Download ScanView: a program that lets you fly around our models of Michelangelo's statues - no license required.

About the project

Introduction

Recent improvements in laser rangefinder technology, together with algorithms developed at Stanford for combining multiple range and color images, allow us to reliably and accurately digitize the external shape and surface characteristics of many physical objects. Examples include machine parts, cultural artifacts, and design models for the manufacturing, moviemaking, and video game industries.

As an application of this technology, a team of 30 faculty, staff, and students from Stanford University and the University of Washington spent the 1998-99 academic year in Italy scanning the sculptures and architecture of Michelangelo. As a side project, we also scanned 1,163 fragments of the Forma Urbis Romae, a giant marble map of ancient Rome. We are currently back in the United States processing the data we acquired. Our goal is to produce a set of 3D computer models - one for each statue, architectural setting, and map fragment we scanned - and to make these models available to scholars worldwide.

The motivations behind this project are to advance the technology of 3D scanning, to place this technology in the service of the humanities, and to create a long-term digital archive of some important cultural artifacts. Our sponsors are Stanford University, Interval Research Corporation and the Paul G. Allen Foundation for the Arts. Our collaborators, a mix of computer scientists and art historians, include the Italian museums and institutions whose names are listed below.

Timeline

Since 1992 ProfessorMarc Levoy and hisstudents have been investigating methods for digitizing the shape of three-dimensional objects using laser scanners. Our early efforts in this area were optimistically called the"project to build a 3D fax machine". In 1996, we digitized a smallbuddha statuetteand emailed the computer model from Palo Alto to a fabrication facility in Los Angeles where an accurate replica was made, thus demonstrating at least the feasibility of building such a machine. Ours was not the only research group working on 3D digitization. By 1996 a group at theNational Research Council of Canada was regularly scanningmuseum and heritage artifacts. However, nobody had digitized a large statue with enough precision to serve as a primary resource for scientific work, and nobody had tried to digitize a large and coherent collection of statuary. Hence the impetus for the Digital Michelangelo Project.

Our project officially began in January of 1997 with atwo-year planning period. During this period we designed, built, and tested a laser rangefinder and mechanical gantry customized for scanning large statues. The requirements were demanding; we wanted to capture chisel marks smaller than a millimeter, we wanted to capture them from a safe distance, and we wanted to reach the top of Michelangelo's David, which stands 23 feet tall on its pedestal. The final system, which we call theStanford Large Statue Scanner, was fabricated for us byCyberware of Monterey, California. For those hard-to-reach places (we didn't move the statues), we used a second scanner - a jointed digitizing arm and small triangulation laser rangefinder made byFaro Technologies and3D Scanners Ltd. Finally, to enable us to scan the architectural settings of Michelangelo's statues (especially the new Sacristy of the Medici Chapels, which he designed), we used a prototype time-of-flight laser rangefinder made byCyra Technologies. All three systems also captured high-resolution color data.

The production phase of the project, from September of 1998 to June of 1999, consisted of scanning the artworks themselves. For this purpose, Professors Marc Levoy and Brian Curless, Dr. Kari Pulli, and 7 graduate and 14 undergraduate students and staff (listed below) relocated to Italy for periods ranging from a few weeks to a year. Professor levoy, who stayed the longest, managed to trade his condominium in Palo Alto for a memorable stay in avilla in Florence. The first 4 months (mid-September through mid-January of 1998) were spent converting the ground floor of apalazzo on the Arno Riverinto a temporarycomputer graphics laboratory, assembling and testing our scanners, and writing software. The next 5 months (mid-January through mid-June of 1999) were spent movingfrom museum to museum, digitizing artworks using our scanners, and sending the data back to our laboratory for post-processing. During this period of intense activity we scanned10 statues: the David, the four Unfinished Slaves, and St. Matthew, all located in theGalleria dell'Accademiain Florence, the four allegorical statues (Night, Day, Dawn, and Dusk) in theMedici Chapels, also in Florence, and the architectural settings of both museums. Our largest dataset was of the David - 2 billion polygons and 7,000 color images. Augmenting our full-time staff during the Winter academic quarter (January - March) were 13 additional undergraduates from theStanford Overseas Studies Program. For everybody, the project involved long hours -our scan of the David was done only at night - and unusual talents, like strength - the gantry weighed 1800 pounds, and it had to be erected each time we moved to a different museum. But the work was exciting, and the data we came back with - 250 gigabytes in all - was worth the effort. Of course, not everything was work. Here we are visiting theattic of Florence's Duomo, and here themarble quarries at Pietrasanta. Professor Levoy even tried his hand atcarving marble.

At the point this timeline was written (Autumn of 2000), we were working on building 3D models from our data. It's acomplicated task, made harder by the unusually large size of our datasets. For each statue and map fragment, we must first clean up the raw scans, thenalign themwith each other,merge them to form a seamless mesh, fill holes in this mesh, and finally map color photographic data onto it. The resulting meshes contain hundreds of millions of polygons, so even a simple task like displaying them on a computer is challenging. Thus, we're exploring new ways to compress,transmit, andrenderlarge polygon meshes efficiently. At present, we have built a full-resolution model of one statue -Michelangelo's St. Matthewand several medium-resolution models of a second - the head of David. Here are some newercolor renderings of the David. We have also built crude models of the other eight statues; here are computer renderingsof these models.

Some side projects

Although our primary goal while in Italy was to scan the works of Michelangelo, we became involved in several other projects during the year. One such project was the acquisition of anultraviolet fluorescence map of Michelangelo's David. Once mapped onto our 3D computer model, this data will show the location of waxes and other organic materials that have been applied to the statue over the centuries.Addendum, October, 2002: Conservators recently used our model to help plan amajor cleaning of the statue. Here's a recently published bookabout the scientific studies that preceded the cleaning, many of which used our 3D model of the statue. To coincide with the cleaning, we installed aninteractive kioskin the Galleria dell'Accademia. The kiosk allows museum visitors to rotate our 3D model of the David and play with its lighting. This allows them to examine aspects of the statue that are difficult to see from the ground.

Another side project was the acquisition of alight fieldof Michelangelo's statue of Night. Light fields, a technique from a new branch of computer graphics called image-based rendering, consist of dense arrays of images of an object, each image taken from a slightly different viewpoint. Once a light field has been created, perspective views from viewpoints not present in the original array can be constructed by combining pixels from the array's images. Sincerendering a light field involves only shuffling pixels, not rendering polygons, light fields can be viewed in real time on a PC having no hardware acceleration. Although light fields do not directly provide 3D shape information about an object, they might nevertheless be useful in some cultural heritage applications.

A final side project was the digitization of thefragments of the Forma Urbis Romae, a giant map of ancient Rome carved onto marble slabs circa 200 A.D. The map lies in fragments - 1,163 of them. Piecing this map together has been one of the great unsolved problems of archeology. Fortunately, the fragments are several inches thick, and the broken surfaces give us strong three-dimensional cues for fitting the pieces back together. Suppose one scanned these fragments; could a computer program be written that could "solve the jigsaw puzzle"? We are trying just that. To give us the data we need, we moved our computers and scanners to Rome, we had an additional desktop laser scanner brought in from Palo Alto to increase our throughput, and during a 3-week 24-hour-a-day scanning marathon in May and June of 1999, we digitized every fragment of the map. (In May of 2001, we scanned another 23 fragments that were unearthed after 1999, bringing the total to 1,186.) Under the banner of the"Digital Forma Urbis Romae Project", we are currently assembling these scans to create 3D models of each map fragment, and we are developing algorithms to help us solve the puzzle.

Creating a digital archive

Although the task of building 3D models from our raw data is only partially complete, we have received many requests for these models. To satisfy this demand, we have created a skeletalarchive of 3D models and placed its catalog online. The models in this archive are available to anyone, but for scientific use only, and users must first[obtain a license](/data/mich/#Obtaining the data) in writing from us. Although these licenses prohibit commercial use of the models, permission for such uses can be obtained by[applying to the Italian government](/data/mich/#Commercial uses). One obvious commercial use is the creation of accurate physical replicas of the statues. Here is a physical replica of the David we built as an experiment.

Although the methodologies needed to create and manage digital archives of two-dimensional artifacts have matured substantially in the last ten years, we are discovering that the jump from two to three dimensions poses new problems. These are problems of both scale and substance, and they touch on every aspect of digital archiving: storage, indexing, searching, distribution, viewing, and piracy protection. To address these problems, we have a pilot research project calledCreating digital archives of 3D artworks, sponsored by theNational Science Foundation(Digital Libraries Initiative). One outcome of this project is ScanView: a secure client / server rendering system that permits unlicensed users to examine our 3D models, but not extract the underlying data. If you want to fly around our models of Michelangelo's statues but you don't qualify for a license, try this viewer.

Current status of the project

Since the summer of 2004, we have done no further work on the 3D models of Michelangelo's statues - mainly for lack of funding and the time to seek additional funding. As a result, high-quality models exist only of the David at 1.0 mm resolution (56 million triangles) and St. Matthew at 0.25 mm (372 million triangles). (Update: in 2009 we completed a full-resolution (1/4mm)3D model of the David. The model contains about 1 billion polygons.) A model also exists for Atlas at 0.25 mm (approximately 500 million triangles), but it contains alignment errors. We welcome[requests for licenses](/data/mich/#Obtaining the data) from any research group or institution that wishes to help us finish this monumental labor (and solve some hard mesh alignmment and mesh completion problems in the process). The fragments of the Forma Urbis Romae have fared somewhat better; after 6 years of paid and volunteer student labor, models now exist for every one of the 1,186 fragments. These models, which total nearly 8 billion polygons, can be viewed without a license at thisweb site.

Photographic essays from the project

Click on a picture to view the corresponding essay. These essays are also linked into the narrative in the preceding section.
Two yearsof planning A palazzo onthe Arno River Our graphicslaboratory Our main scannerarrives at the lab Moving into theGalleria dell'Accademia
Our first statue -St. Matthew We beginscanning the David We finishscanning the David A 3D model ofDavid's head Some color renderingsof the David Other interestingbody parts
Scanning theMedici Chapel A light field ofMichelangelo's Night Scanning theForma Urbis Romae A kiosk forviewing David A replicaof David Exploring David:the book

Videos

From the director's private collection

Photographs of the team
1 2 3 4 5

Figure 1: Lunch at Il Fornaio, Palo Alto, May 29, 1998
From left to right: Marc Levoy, Szymon Rusinkiewicz, Sean Anderson, David Koller, Lucas Pereira, Alex Roetter, Jeremy Ginsberg, Matt Ginzton, Kari Pulli, John Gerth.

Figure 2: Packing up the equipment, Palo Alto, September 2, 1998
Clockwise from bottom-right: Sean Anderson, Lucas Pereira, Marc Levoy, David Koller, John Gerth, Kari Pulli, Szymon Rusinkiewicz. This is one of three shipments.

Figure 3: Preparing to scan the David, Galleria dell'Accademia, Florence, Italy, February 15, 1999.
Left to right: Kathryn Chinn, Jeremy Ginsberg, Sean Anderson, Lucas Pereira, Dan Perkel, Matt Ginzton, Alex Roetter, Wallace Huang, Maisie Tsui, Marc Levoy, David Koller, Unnur Gretarsdottir, Kari Pulli, Alana Chan, Dana Katter, Daniel Wood, Rahul Gupta, Szymon Rusinkiewicz, Brian Curless, Jelena Curless, and James Davis.

Figure 4: Down in Rome to scope out the Forma Urbis, Baths of Caracalla, Rome, Italy, April 13, 1999.
Left to right: Jonathan Shade, David Koller, Kari Pulli, Sean Anderson, Marc Levoy Lucas Pereira, Szymon Rusinkiewicz, James Davis, and Maisie Tsui.

Figure 5: Valedictory dinner in New Orleans, July 28, 2000, two days after presenting apaper on the projectat Siggraph.
Left to right: Sean Anderson, David Koller, Steve Marschner, his wife Heidi, Marc Levoy, Szymon Rusinkiewicz, Kari Pulli, Matt Ginzton, Jeremy Ginsberg, James Davis.

Some of these links are woven into the narrative in the beginning of this web page.

Our team

Faculty and staff (in alphabetical order)

Graduate students

Undergraduates

Our Italian collaborators

In Florence

In Rome

In Pisa

Our sponsors

Data and software

**Note:**We receive many requests from people who would like us to scan objects for them. Since our primary mission at Stanford is research and teaching, we really can't operate as a scanning service bureau. Simple3D runs a greatweb site about 3D scanning, with links to scanning manufacturers, software, books, etc. Raindrop Geomagic maintains another good web page of 3D scanning resources.

The press box

Our project in numbers

Number of objects scanned 10 statues + 2 buildings + 1,163 map fragments
Smallest and largest objects scanned 1 inch (map fragment) and 23 feet (the David)
Spatial resolution of data 0.29mm for geometry, 0.125mm for color
Complexity of largest dataset 2 billion polygons + 7,000 images (the David)
Size of largest dataset 32 gigabytes (the David)
Total amount of data captured 250 gigabytes
Size of largest scanner 24 feet tall, 1800 pounds
Total weight of equipment shipped to Italy 4 tons
Number of people involved 32 (not including subcontractors and collaborators)
Average time to scan a statue 1 week (except the David, which took 1 month)
Total time spent scanning 5,000 man-hours
Total time spent processing data 4,000 man-hours (so far)
Cost of project $2,000,000

Publicity about the project

**Note:**For a good overview of the project, look first at the article written for the Stanford Campus Report, then at this file of frequently asked questions. You can also check out a posterthat visually describes the project, and images of everything we scanned.

Readers should be aware that other than the Stanford Campus Report, the articles listed above have not been selected by us or edited for accuracy. Some of them, and others that have appeared in the lay press, contain misquotes or factual errors. They should be read with caution.

Regarding specific "news events", if you're interested in our discovery that the David is 3 feet taller than history books say he is, we talk about ithereand[here](publicity/faq.html#height of the David). If you're interested in our observation that David's eyes diverge, listen to thisinterviewby Noah Adams on National Public Radio's All Things Considered (June 13, 2000).

**Notice:**The images of Michelangelo's statues that appear on this web page are the property of the Digital Michelangelo Project and the Soprintendenza ai beni artistici e storici per le province di Firenze, Pistoia, e Prato. They may not be copied, downloaded and stored, forwarded, or reproduced in any form, including electronic forms such as email or the web, by any persons, regardless of purpose, without express written permission from the project director Marc Levoy. Any commercial use also requires written permission from the Soprintendenza.

**Clarifications:**The prohibition in the preceding paragraph against copying does not apply, of course, to the ephemeral downloading and copying associated with browsing these web pages. Also, permission is hereby given to reuse the project's logo image in its entirety, even to incorporate it into web pages, but only as a clearly labeled representation of the Digital Michelangelo Project. All other use requires the express permission of the project director.

Copyright © 1997 - 2003 Marc Levoy
Last update: