Computer Graphics World Magazine Archived Article - November, 1997

Pioneers and Performers

 
Experts in the field offer their opinions about the state of the art and the technology of performance animation.

by Barbara Robertson

It`s been nearly a decade since the first 3D computer graphics (CG) characters animated with motion data rather than keyframe animation appeared onscreen: In 1988, the Kleiser-Walczak Construction Company introduced Dozo, a CG rock star who sang "Don`t Touch Me" as she danced using motion captured from a human dancer. That same year, deGraf/Wahrman brought "Mike the Talking Head," the first interactive 3D CG character, to SIGGRAPH.

How successful is this data-driven animation? Although 3D characters created with keyframe animation have been populating nearly every type of entertainment medium during the past few years, with some--Woody and Buzz, Draco, Casper, T Rex--even becoming film stars and household words, their data-driven cousins have rarely seen the limelight. Thus it seemed time to check in with some pioneers and practitioners in the field of performance animation to find out just what`s happening in this corner of the computer animation world.

We talked with Brad deGraf, CEO and founder of Protozoa (San Francisco); Michael Wahrman, pioneer and computer animation director (New York City); Scott Dyer, cofounder of Windlight Studios (Minneapolis), an animation studio; Larry Lamb, founder of Lamb & Company, an animation studio, and of Lambsoft, a software company (Minneapolis); Gregory Panos and Richard Cray, a technology developer and a stage performer and musician, who are cofounders of the Performance Animation Society (www.pasociety.com); Gary Jesch, president of Chops, a performance animation company (Incline Village, CA); David Sturman, director of R&D at Medialab, an animation studio in Hollywood and Paris; Chris Walker, founder of two animation studios, Mr. Film and Modern Cartoons (Venice, CA); Jeff Kleiser, founder of Kleiser-Walczak Construction Company, an animation studio in North Adams (MA), New York City, and Los Angeles; and Dave Verso, president of SimGraphics, a performance animation company (South Pasadena, CA).

CGW: Why isn`t performance animation taking off any faster--or do you think it`s evolving at a reasonable pace?

LAMB: I think putting body English into animation was a Holy Grail. It has been and still is a great idea. But it has been a very difficult thing to do.

VERSO: We`re maybe a third of the way toward becoming a mature technology. We started about five years ago in terms of having talking characters. It will probably take another 10 years for the processes and tools to evolve and for enough people to have experience. You can`t just buy a tool and get an animation; it`s not a magic button. It`s complicated. It`s a process.

WAHRMAN: Some people thought this technology would make animation easy. But it`s not about equipment, it`s about art. And nothing about art is easy. Jim Henson with a green sock on his hand was Kermit. Michael Wahrman with a green sock on his hand is Michael Wahrman with a green sock on his hand.

CRAY: When the prices come down and more talented performing artists can work with the technology at their own pace and on their own terms, it will become more creative.

STURMAN: The Jurassic Park of performance animation hasn`t arrived yet, but it will. The technology is mature enough to have something important happen.

CGW: Do you draw a distinction between performance animation and motion capture?

JESCH: Motion capture is a component of performance animation; one of the performance animation tools. It`s particularly good for lip sync. But we also use joysticks, midi sliders, and other devices. You could do a performance animation without a lick of motion capture.

DYER: The difference is in application. Motion capture is applying motion to a model. Performance animation is acting a character. At Windlight, many of the characters we animate are not humanoids, yet we use actors to drive the performances of these characters. In our system, the actor sees the character as he`s acting.

CRAY: Performance animation is really live theater.

DEGRAF: Think of the difference between performing Moxy live and capturing Michael Jackson dancing. You can capture Michael Jackson`s motion without seeing the character it will be applied to. But imagine trying to puppet a character without seeing the puppet. [Moxy, a real-time 3D character developed at Protozoa, appears on the Cartoon Network; data from a dancing Michael Jackson was captured at Digital Domain and used to help make a skeleton dance in a music video.]

STURMAN: For the most part, with motion capture you see the animation a day to a week later, but you can get high-quality images as in Andre Agassi [a TV commercial created at Digital Domain] and Michael Jackson, and the work being done at Windlight. At Protozoa, Medialab, Modern Cartoons, and SimGraphics, a performer sees what the character is doing. Most of our characters receive no post-treatment. It`s more like puppetry than acting. They`re driving a character, and everyone sees the result; the director sees what he gets.

WAHRMAN: Part of what makes performance animation so great is that it`s alive. It can be improvised.

WALKER: When a performer or director can see the character onscreen, rendered in real time, talking to them, the difference really hits home. That`s the distinction.

CGW: Why do you think some animators have such a negative opinion of motion capture and performance animation?

STURMAN: They`ve missed the point. We`re not trying to replace the animator. It`s not about that. It`s digital puppetry. Because it`s on a screen and there is image after image, it`s called animation, but in process it`s more like puppetry. Unfortunately, many of the performance characters are stiff, the shapes are blocky, the movements aren`t smooth, their facial expressions are rigid, and they don`t come off as alive. But that`s not a question of technology, it`s a question of experience and of designers.

VERSO: Some people have been burned. They`ve been promised things by people who didn`t understand the issues. Some have been trained another way and this is new to them. Sometimes it has been sold as a replacement. And some people compare this evolving form of animation to more mature techniques.

LAMB: For keyframe animators, it`s been an either/or situation that`s left them out in the cold. They couldn`t adjust the captured motion data without losing some or all of the subtleties. Animators just weren`t part of the party. Plus, it`s often used for low-resolution animation; not to capture acting but to do low-end animation. And, well, you get low-end animation. So unfortunately, there are many bad examples.

DYER: The more bad examples there are, the more people will think badly of it. There`s always a certain purity about traditional animation, and when motion capture is used, there`s a view that the animator`s art has been subverted. But this view is often held by people who haven`t experienced what they can do.

WALKER: We`re really creating another art form that`s a cross between cartoons and puppets. People haven`t accepted it. We`re still creating it. We have yet to really see it stretch its muscles.

CGW: Give us some examples of successful performance animation.

WAHRMAN: One of the most creative successes I`ve seen is Fabrique the villainous robot created by Medialab in a test for Starwatcher. I thought the Fabrique test was brilliant--ominous, mysterious, evil, genius. Another thing that blew me away was our second-take rehearsal for Mike the Talking Head. One of the most commercially successful real-time characters is without doubt Bert the Fish [created by Medialab for Nickelodeon (UK)].

WALKER: We just finished CyberLucy for CBS`s Wheel 2000 on Saturday mornings. The show is a children`s version of Wheel of Fortune and CyberLucy is that show`s Vanna White. She`s on a 9x12 screen in the set; the performer is off stage. The children go nuts when they see her. I think Digits `n Art (Montreal) has done some really good work. SimGraphics has really great technology.

DEGRAF: I think of Bert the Fish and Donkey Kong Country [both created by Medialab] as reaching a new level in terms of commercial awareness. In terms of return on investment, I think of our own Dev Null. I`m biased, but it`s been on the air for a year now. It`s a branded character...maybe not Ren and Stempy, but...

Also, to me, our Moxy was a big leap. We had Class A writers, a good performer, and state of the art software. Beyond that, we`ve all done characters all over the place.

LAMB: Brad`s Moxy. The demo of how Moxy worked is one of the best pieces of performance animation I`ve seen. The Boxer from Digits `N Art. It`s immodest, but I think our Huzzah showed a range of expression and emotion that was successful. We discovered that it`s really important to capture the entire performance. There is a continuity between facial twitches and finger twitches and when we tried to do faces separately, there was a discontinuity. We captured facial motion with an optical system, hand motion with a glove, and body motion with a magnetic system.

CRAY: Probably the silliest and simplest was a quick interaction between Sinbad and Flea on Protozoa`s demo tape. It was completely off the cuff and improvised. It wasn`t profound, but it made me think, "This is going to work."

VERSO: We just did a project in Malaysia. The client had just started up a facility and the people weren`t trained in a particular mindset. We were able to come in and they could quickly do an animation that was out of their own culture and their own aesthetic.

In terms of technical success, at SIGGRAPH we showed multiple characters in a scene. Even as crude as it was, the characters were performing stories.

STURMAN: I`m pretty interested in what Protozoa is doing on the Internet. The idea of an ongoing cartoon in VRML sounds like fun.

KLEISER: Digital Domain`s work on Andre Agassi. Our most successful work was on Judge Dredd. We captured the motion of people on motorcycles and applied it to digital stunt people that were flying [digital] motorcycles.

JESCH: I think one of the biggest commercial successes I`ve seen was Mario for Nintendo. They took a system and a performer on the road to WalMarts. [Mario was a real-time 3D character created by SimGraphics and performed by Charles Martinet.] I thought Suzy the Sex Therapist from Mr. Film was successful at suspending disbelief.

CGW: What is the primary market for performance animation or motion-capture animation, and how do you see that changing?

JESCH: I decided to design and perform my own character, so for me, it`s tradeshows and special events. I`m not a Hollywood studio, but I`m making money. It would never have worked, though, without the help of Motion Analysis [optical motion-capture technology], Quantum Works [Geppetto performance software], and 3Dfx [3D graphics chip]. It`s a thrill to go back to them and tell them this is a real business.

KLEISER: I think the technology is very handy for human-like motion. The most likely market is for a TV series with lots and lots of footage of characters walking around that`s on a tight schedule and has a tight budget.

STURMAN: Broadcast TV fits really well. It`s also very good for line tests. If you could block out a scene using live actors and use it as a reference, you could easily try 10 scenes with your characters, your snake, your rabbit, your lion, and use them as a reference for an animation. You won`t get fine-tuning, but you can get a rough idea in a day.

VERSO: Long-form animation--when time-to-market and cost are important and there are volumes of animation. We also have a lot of interest in location-based entertainment--not just trade-shows, but also putting interactive characters in places like Treasure Island in Las Vegas.

LAMB: Today, captured motion is just another tool along with keyframe animation for creating moving 3D objects. I believe motion capture opens the door for animation in corporate videos and lower-end venues.

DYER: Today, probably games, especially sports or action games. Secondarily, TV shows with performed characters. It`s working into features and commercials.

DEGRAF: Small broadcasters around the world who need differentiated content. A good application is a Muppet-style multiple-character show. I think there`s a type of show that could take the writing and show design of a puppet show and marry it with a 3D animation look that would be interesting and allow for great stories. Sort of Jim Henson meets Toy Story.

On the Internet the bang for the buck is really significant compared to any other way of livening up Web pages. We`ve done 65 episodes of Floops [for Silicon Graphics] and we`re doing 50 of Dilbert, sponsored by Intel. The medium is a good fit.

WALKER: For us, the primary market is television. We`ve been working on a new adult series, a political satire. We`re excited about it, but our instinct is that children`s animation is a better fit. We`re certainly moving into movies and want to do a full-on animated film.

WAHRMAN: My biggest disappointment has been that the Henson organization hasn`t used performance animation. They have the properties, the talent, the organization. They could do it.

CGW: What tools--hardware or software--need to be developed?

JESCH: I`d like to see face tracking under $5000. I`d also like to have spline-based tracking. Here`s what I mean: If you were to draw a line around your lips, you`d have a curve. When you talked, the curve would change shape.

WALKER: Faster computers. Faster render speeds at more affordable prices. As for the sensors--all the systems are good, and none is the best. Each has its advantages and problems. Optical systems have less spatial limitations than magnetic, but you don`t get real time with more than one character. The mechanical systems look beautiful, but they aren`t as accurate. We use a hybrid of all the technologies. Our focus is on production ergonomics, on making our performers feel comfortable.

DYER: The translation tools that can turn a doughboy into a dinosaur are not well developed or understood. We also need tools and interfaces for manipulating, filtering, and storing data.

DEGRAF: Bodysuits and face trackers are still too expensive. We still don`t have a good glove. They`re better, but quirky. The Holy Grail is to do unlimited characters in an untethered way in a large space with no occlusion, no noise.

STURMAN: Capture needs to be lighter, less intrusive, less susceptible to environmental problems. Hardware needs to be faster. We`re just starting to do multiple characters. We have an Infinite Reality with a gazillion processors and we can get two good-quality worms doing a lot together. And software needs to evolve. The software doesn`t support characters touching each other yet.

LAMB: The recording equipment is fairly well along. We`re finishing up two pieces of software that I think will address some fundamental problems. One is the ability to take motion from varying sizes and shapes of performers and apply it to varying sizes and shapes of characters. Another is making it easier for keyframe animators to adjust the captured motion.

CRAY: I have never believed that a performer should have to wear computerized clothing. I want control over motion without wearing something that`s pieced together like Frankenstein`s monster.

PANOS: The toolmakers have been working separately from the talent; performers haven`t been involved in the development of the tools. The sensors could be much less invasive. You could have body-tracking sensors that would be more like jewelry than clothing, and develop sophisticated gestural recognition software. It`s funny that people haven`t married virtual reality and performance animation, but they will.

CGW: Do you think this will ever make it into the mainstream or consumer market?

DYER: More quickly than you can imagine. You`ll put on a suit and you`ll get to kick and fight with a character or do a work-out tape or analyze a golf swing. We just need less expensive, less restrictive capture equipment. If you could get a bodysuit for $100, there would be an awful lot of applications.

Performance-Animation pioneer, Chris Walker of Modern Cartoons, turns game character Crash Bandicoot into an interactive character for Sony Computer Entertainment.
These Extreme Dinosaurs were created at Windlight Studios for a Foote, Cone, and Belding TV commercial advertising Mattel toys.

The animation was created by capturing data from performers who acted the part of dinosaurs.

 
Donkey Kong and his virtual pals have gone from Nintendo video game to TV stars, thanks to the Medialab Real-Time Performance Animation System used to animate these characters in real time as they co-host a new magazine show for kids on French TV.
Singer Richard Cray, shown here animating the "Sunny" character in an AOA optical facial expression capture helmet for
"A Musical Performance Animation" to introduce performing artists to the new medium via the special interest group: the Performance Animation Society, which he co-founded with VR guru Greg Panos.
 
At SIGGRAPH `97, real-time characters could be found dancing in the booths and walking the floor. SimGraphics demonstrated a multiple-character version of its real-time software with two dancers (insert).

Meanwhile, Protozoa`s Alive! software helped digital puppeteers control Reginold the roving bug reporter as he interviewed attendees (right).
Short Stack, the star of this TV spot created for the Louis London agency by Windlight Studios, is one of three delicious-looking characters animated using data captured from performers.
Meet Dr. Dexter Finnery Klaus, an evil geneticist who lives in an in-house project titled "Meat" at Protozoa.

Dr. Klaus is one of several real-time characters created at Protozoa for television (Moxy) and for the Internet (Floops and Dilbert).
Gary Jesch is a performer whose real-time animated character Chops makes regular appearances at trade shows and business meetings.

To animate Chops and other characters, Jesch uses a PC-based system with Geppetto software and Motion Analysis optical capture devices.
Computer Graphics World  -  November, 1997
Copyright © 2006: PennWell Corporation, Tulsa, OK; All Rights Reserved