Issue: Volume: 24 Issue: 8 (August 2001)

All Mime

By Karen Moltenbrey

When you think of legendary dancers, a few people may come to mind-Fred Astaire, Gene Kelly, Mikhail Baryshnikov. The same holds true for musical talent. But when it comes to pantomime, there is only one universally acclaimed genius-Marcel Marceau. "He is one of the few undisputed masters of an art form," says Jeanine Thompson, a theater professor at Ohio State University and one of the artist's few protégés. "When you think of who it is that will be handing down the codification of the art form, it will be Marcel Marceau."

To ensure that Marceau's legacy is preserved for generations to come, Ohio State's Advanced Computing Center for the Arts & Design (ACCAD) captured the intricacies and precision of the mime's trademark movements using the College of the Arts' new motion-capture system. The daylong mocap session was held during the French artist's recent visit to the university, where he taught classes and performed as part of a residency program. The digital data collected from the shoot will become part of the Marcel Marceau archives, established by the university in collaboration with the Marcel Marceau Mime Foundation in New York.
Through motion capture, the students and faculty at Ohio State University have preserved the movements of renowned mime Marcel Marceau, so they can be enjoyed by future generations. (©1998 Roger Pic.)

The idea to augment the school's existing Marceau archives with 3D data followed on the heels of the decision to purchase the motion-capture equipment. But persuading the 78-year-old mime master to participate took some doing. According to Thompson, Marceau had been introduced to the technology years ago, when it was not nearly as developed as it is today. "He was shown [an early] magnetic system that was very cumbersome, and the data that was acquired was not that good," she says. "It left him with a definite opinion about never being involved with motion capture."

However, Thompson cajoled the performer into examining the college's Vicon 8 (Vicon Motion Systems; Oxford, UK) optical motion-capture system during his residency at Ohio State. She and others from ACCAD worked through the school's spring break period to devise ways of placing the system's reflective markers-which track a person's movements-on the body so that they could capture the essence of Marceau's work. "When he arrived, we showed him a demonstration of me performing some of his signature movements, and he saw that we indeed captured everything, even his breath initiation [before] each movement," explains Thompson. "The result didn't look like computer lines and dots on a screen. It showed a living, breathing form."

Because of the importance and delicacy of the Marceau shoot, and the fact that it was the school's first real project using the system, the department asked alumnus Jeff Light, motion- capture supervisor at Industrial Light & Magic, to oversee the project. Light, along with ILM mocap engineer Doug Griffin, also used the session as a teaching opportunity to show the students some "tricks" they had learned over the years while working on projects such as Star Wars: The Phantom Menace and The Mummy.
Initially wary of motion-capture technology, Marceau quickly adapted to performing for the mocap cameras, which at times required him to slightly alter his body positioning so as not to occlude the reflective markers. (Photos courtesy The Lawrence and Lee

What differentiated the Marceau shoot from most motion-capture sessions is that the technicians did not know what kind of future applications the data would be used for. Therefore, their main concern was gathering the greatest amount of data possible, which meant using as many cameras, reference points, and reflective markers as they could.

The group began the session by setting up the college's 14-camera MCam Vicon 8 system around the perimeter of a capture volume "stage" that was 14 feet wide, 10 feet long, and 9 feet high. This space was large enough for Marceau to move without his performance being encumbered, but small enough to capture the data at high resolution.

Next, the team placed 50 small reflective markers on the performer's body, making sure his arms, legs, chest, head, feet, and face were covered. During Thompson's demonstration prior to the shoot, the group discovered that at times her movements occluded the markers, resulting in a loss of data. But when she adjusted her limbs slightly, there would be enough space for the cameras to "see" the markers. "To rectify that, all I had to do was extend my arm maybe a few inches," Thompson says. "I wasn't sure if asking Marcel to do that would be pushing him too much, but he was fine with it."

A number of markers were also placed on the mime's abdomen and upper chest to capture his breathing. "His breathing is a very subtle motion but something that was extremely important to him. And after watching his performance, I know why," says Light. "He brings the character to life with his first breath. And that differentiates a living being from a dead doll that's moving around, which is often the effect you can get with computer graphics-a look that is less than fully alive."

Another important aspect of Marceau's performance was the movement and positioning of his feet, which are as important to a mime as they are to a dancer. "We wanted to make sure we marked his feet sufficiently to capture the detail of what they were doing and how they made contact with the floor," says Light. And like a dancer, Marceau felt more comfortable performing while wearing specific footwear. "I was hoping that we could buy a pair of shoes in his size that we could safely destroy, because the adhesive used to attach the markers is very strong and could damage his shoes. But he insisted on using his own shoes." The group fixed the markers to an older pair of Marceau's shoes, and when they removed the reflectors, it indeed ripped holes in them. "It made my heart sink, but he was okay with it," Light adds.
Studying 3D representations of Marceau's movements enables theater students to examine the data from all angles, giving them a more complete view of the movements than would be offered merely by watching a videotaped version of the mime's performa

The technicians also set up three additional stationary motion-capture cameras operating from different angles around the perimeter of the capture space so they could capture Marceau's facial expressions at the same time they captured his body movements. "But even with state-of-the-art technology, you really can't get good facial, hand, and body data all at the same time and integrate it automatically as a complete whole," Light says. Ideally, Light would've preferred having a crew of people use handheld mocap cameras to zoom in close to the mime's face during his performance. But he admits that would have resulted in a much larger production and could have endangered the shoot by obtruding on the artist's performance.

"His performances require a great deal of concentration, and we had to be respectful of his space once he began," says Light.

The technicians also used regular video cameras to record the performances, so the film footage could be used in the future to complement the digital data. "They could use the video as a backdrop that's synchronized to the motion-capture data and make sure it matches up to the video," suggests Light.

Some of the mime's routines were 10 minutes in length-much longer than the 30-second performances Light typically captures. This required the group to have enough disk space to store the several gigabytes of information. It took a Pentium III computer all night to process the 2D data from the cameras into 3D models of the points moving around in space. The shorter pieces took far less time to process, and the group was able to reconstruct that data as point clouds on stick figures within the Vicon software, so Marceau could view it after his performance, to make sure he was satisfied with the results.

"We showed the data as it was applied to the stick figures, so we could concentrate on the pure motion, and not the problems that sometimes arise when bringing a particular 3D character into the equation," says Light. For fun, the group also applied the data to a generic character within Kaydara's (Montreal) Filmbox environment, so Marceau could also see how the movement looked when applied to a 3D model.

Another factor that made this project different from others Light has worked on is that there was no definitive target to which the data was to be applied. "The performance is normally less important than how it will look on the final character or creature. So I can cheat by having the person perform the action a few times before I know it's right," notes Light. "In this case, we were making an archival record, so the only thing that mattered was the performance. It was up to Marceau whether he thought the performance was acceptable or whether he wanted to do it again."

That often meant performing more than one take. "He's a perfectionist," adds Light. "He would rehearse some of the longer, emotional pieces three or four times before doing an actual take. Then, if he was unhappy with the way his arm moved, he would do it again." Light was especially impressed with Marceau's physical endurance, noting that he performed his physically demanding routines for five hours straight.

Marceau owns the digital data, as he does the rest of the archival ma terial. He has a greed to let the university use it for teaching purposes, albeit on an individual-case basis. "None of us understands all the possibilities for using this data," Thompson says. One option is to compare Marceau's motion-capture data alongside that of students, to show them exactly how their movements differ from those of the master they are trying to imitate.

"It's only when you look at Marcel Marceau's movements in pure motion form [on the computer] that you start to realize what an amazing talent he is. He works in a realm that most actors don't even know exists," says Light. "It's so impressive the way he communicates an idea and captivates the audience through his body language alone. I don't think anyone in latter part of the twentieth century is identified with pure motion as he is."

Karen Moltenbrey is a senior associate editor for Computer Graphics World.