Friday, March 29, 2024
Subscribe Now

Voice Of The Crew - Since 2002

Los Angeles, California

Beowulf and 3-D

-

By Henry Turner
Beowulf in 3D is a unique experience, raising not just questions about future of cinema, but also posing unique problems that the film’s hefty postproduction process had to solve to bring the ancient tale to life.
Buzz Hays, who produced the 3D version of Beowulf at Sony Imageworks, says the production team knew from the start that Beowulf would be in 3D. “Our pipeline for production is designed for both 2D and 3D, but we have separate crews in terms of the slightly different disciplines involved in creating a 3D version. We’re dealing with creating a virtual world. On the simplest level we’re adding a second stereo camera to that virtual world, although it gets a lot more complicated fairly quickly. But in essence that’s what we do — we build the other eye for the movie.”
Shot in motion capture and rendered digitally, Beowulf uses the techniques made familiar through films such as The Polar Express (also directed by Robert Zemeckis), and well-known characters such as King Kong in Peter Jackson’s 2005 remake and Gollum from The Lord of the Rings trilogy. Actors wearing special Spandex suits set with hundreds of reflective markers act out scenes on a vast soundstage, where their motions are recorded by scores of sensors placed all throughout the stage. The data achieved captures the motion of the reflective dots on their bodies, which after being processed digitally and artistically rendered, re-creates the actors as animated characters, which are then integrated into virtual environments. With more and more added detail the effect gains such realism that it is not a cartoon or live action, but a highly stylized hybrid with a look entirely its own.
“Beowulf being an entirely CG film, it all exists in the computer,” Hays says. “What we do is add a second virtual camera to those scenes as soon as they’re finalized in 2D. A 2D version is rendered, and from that a second slightly different angle is created to achieve the 3D effect. It’s much like if you were on a live action shoot and shooting one camera, and you got to go back to that same shoot and add a second camera and reshoot everything. But a virtual world is a perfect digital world, so we can re-create things exactly.”
Render times were extensive because of atmospheric elements such as hair and costumes. The added data to create the difference of angle came from the same 2D virtual model. “We approximate the distance between the eyes, and vary that distance for different effects,” Hays says. “Everything was done 3D, the characters, backgrounds, everything. The detail is incredible all the way into the background.”
Though the film is shot without cameramen, a DP does lend a hand in post. “Once we’ve captured all this and started applying it to the models, the DP helps direct how the film is lit, but we’re lighting it on the computer.” Of greatest importance is deciding camera placement and motion — something made unlimited by motion capture and CGI. “The DP has control of that, and fine tunes the moves.” An example of how spectacular these moves can be is best seen in a shot that shows a rat on a rafter grabbed by an owl that flies out a window, over a valley and into cave.
Hays points out that because the details are added in post, the shoot is uncommonly short. In the case of Beowulf, the soundstage production wrapped in a mere thirty days, leading to post production processes that consumed a considerably longer period of time.
In terms of how 3D was used, Hays says that Zemeckis asked the 3D team to go to the limit. Yet greater emphasis was placed on adding depth to scenes, rather than poking audience members with virtual projectiles. “If you tell a story in 3D and start throwing stuff at the audience all the time, it really distracts, and is more theme park than filmic.” He also points out that unless care is used, 3D can become fatiguing for an audience. “When you go shot to shot with hard cuts, you don’t make the object of attention jump around in 3D space because that’s too hard on the eye and brain to track.”
For Hays, the art is in the details, not the novelty. “For us it’s the subtlety of the detail that really sells the 3D than a more overt use of depth.”
Sho Hasegawa, senior effects technical director at Imageworks, oversaw the creation of all the costumes and hair styles in Beowulf. Hasegawa knew that hundreds of costumes and hairstyles would be needed for the film — everything from the unique look of the villagers and guards to the drapery of the nobles and armor of the warriors. Because of the huge scope of the project, he looked to live-action production as a model for how the virtual characters could all be costumed and coiffed with both artistry and economy. A modular approach was chosen, through which variations of the same costume could be used to clothe a number of characters.
To achieve this, costume designer Gabriella Pescucci was brought over from Italy. “Gabriella came over in 2005 and was here for two months doing photo shoots,” Hasegawa says. “She’d designed on paper all these looks and we worked with her incredibly closely, asking her how to do it in real life if you needed, say, ten costumes, and had to do it with five. She had answers. Changing tops, bottoms, switching stuff around, alterations of lapels and so forth, she was invaluable in terms of explaining how she’d have done it practically. She brought boxes and crates of costumes from Italy, and we shot photos of every piece — leather armor and chest plates, chain mail skirts, dresses, tunics, jewelry, everything — with body doubles on turntables and walking video footage to get a full range of motion.”
Hasegawa also points to the skill of the digital artists who further varied the appearance of the costumes. “The shader guys did a brilliant job taking the few geometric variations we had provided them and using color and so forth create more variations beyond that.”
Textures were taken from high res photos of actual costumes, but also rendered by artists. “Geometrically, the amount of detail modeled was the most I’ve ever dealt with,” Hasegawa says.
Hairstyles were also done modularly. “The hair was sourced from reality as much as possible, it was sort of like a collage of information.” Photos of real hair as well as artists’ renderings built the hairstyles strand by strand, long and flowing or greasy and short depending on the character.
“The facial hair and beards were built in a modular way as well,” Hasegawa says. “The beards were built so you can turn off pieces of it, and have long hair and a full beard on one character, while the guy beside him has the same long hair but a goatee, etc., etc., to create variations. Then the color variations take it further.”
Interestingly enough, the actual science of perception plays a role in designing a CGI motion capture film, demonstrating the ever-expanding skill set required by digital filmmakers. “Research has shown that we determine difference in a crowd by how much skin is showing and by the silhouette outline of the individuals — if that silhouette is similar, we’ll think they all look the same.”

Written by Henry Turner

- Advertisment -

Popular

Brad Allan

Over the Weekend 8/9/21: Night Court‘s Markie Post Dies, The Suicide...

0
Unfortunately, this past weekend was one full of sadness as a number of prominent and beloved people from across our industry passed away. First up, Markie...