VFX Soup: Tintin VFX Supe Joe Letteri Talks 3D and Mocap
COW Library : Autodesk Maya : Debra Kaufman : VFX Soup: Tintin VFX Supe Joe Letteri Talks 3D and Mocap
The Adventures of Tintin: The Secret of the Unicorn. Courtesy Paramount Pictures/Columbia Pictures
On December 13, Autodesk sponsored a screening of director Steven Spielberg's The Adventures of Tintin for the visual effects community and featured the movie's visual effects supervisor Joe Letteri discussing how the movie was made with moderator David Morin, a consultant to Autodesk.
Joe Letteri (right) with Autodesk consultant and moderator, David Morin at the LA screening of The Adventures of Tintin. Please click on image above for larger view.
Letteri, who has worked at Weta Digital since 2001, has won four Academy Awards in Visual Effects as well as countless other honors for his work on two films in the The Lord of the Rings trilogy (The Two Towers and The Return of the King) as well as Rise of the Planet of the Apes, Avatar, King Kong, and many others. He's currently working on The Hobbit.
After I finished LOTR: Return of the King, [producer] Kathy Kennedy called me to say that Steven [Spielberg] wanted to talk to me about The Adventures of Tintin. This was in 2003/2004. Steven's first exposure to Tintin was French press coverage of The Raiders of the Lost Ark when reviews kept referring to Tintin. I hadn't been exposed to Tintin so when I knew Steven wanted to discuss this with me, I got the books and read them, and really enjoyed them.
The trick with casting Tintin was how to fit the actors into the world of the author, Hergé. There were concerns such as how to handle the fact that any child actor playing Tintin would age over the three or four-year period of making the movie and that it wasn't clear how to deal with the twins -- the two Inspector Thompsons -- in live action. Around then, Jim Cameron invited Steven and Peter on the set of Avatar to shoot a five-minute test. They shut down Avatar for that day and shot that test, and Steven loved working with way with the virtual cameras. So the decision was made to go with performance capture and the virtual cameras. We began prepping for Tintin while Avatar was shooting, and as soon as we finished with that, we started with Tintin.
Pre-production took about three years. Hergé worked like visual effects artists; he looked at a lot of reference material. His estate was very helpful in letting us look at a lot of the material he used, including the model he used for the Unicorn, the ship that plays a central role in the story. Our art direction took as much as we could from consulting with all the resources we could find. But Hergé was much more cartoony with the characters, especially Tintin, who was basically a fat bowling ball with dots for eyes. We did thousands of iterations to come up with him. Haddock was more straightforward than Tintin.
We started with lots of concept art and then got into 3D modeling and testing, picking lenses…everything to get the characters to work in performance. The movie also had lots of sets, and creating them was a two-step process. The previsualization team blocked out the action and then the art direction department came on. We made all the costumes. Everything was the same as a live action film up to the point of pounding in nails or cutting cloth. Most of the costumes were created in Maya Cloth. But previz stayed at the previz level; we didn't build all the details until we had a cut. We didn't want to expend effort on scenes that might get cut.
Director/Producer Steven Spielberg (right) with Producer Peter Jackson (center) behind-the-scenes - and below:
We started without a script. We would do long takes and then Steven would pick the moments he wanted. Sometimes Peter would block out the action he wanted, but we didn't have dialog, so we played it broadly, which helped us to understand how the characters developed and related to each other.
Director/Producer Steven Spielberg (center) behind-the-scenes. Photo credit - this and above: Andrew Cooper. Please click image above for larger view.
Performance capture was shot for 30 days on a big stage at Giant Studios in Los Angeles, and then there were a couple weeks of pick-ups in New Zealand. We used Autodesk MotionBuilder on the stage for real-time playback. The virtual camera was based on the one used on Avatar. Although it was a bit lighter, there were really no changes from the Avatar camera. The capture technology was also the same thing we developed for Avatar. Shooting a film this way is harder for the actors. There's no time out for set changes, so they are really working a full day. Performance capture strips acting down to its basics: You're in a minimal costume, with other actors and that's it. On set, Snowy was a prop. Snowy needed to steal whatever scene he was in. He needed to be key-frame animated to do that. Hands were also key-frame animated. We used Maya for modeling and animation and wrote various Maya plug-ins for other features we needed such as hair.
Captain Haddock (Andy Serkis), Tintin (Jamie Bell), and Snowy. Please click image above for larger view.
Capturing facial expressions was another issue. When we did Lord of the Rings, we had no way to capture facial animation. We captured Andy's facial expressions for King Kong by using 150 markers on his face. James Cameron found that too restrictive and didn't want to use it for Avatar. Instead, Weta Digital developed head rigs as part of a facial-capture system, and the actors playing Tintin characters used the same rigs.
Weta Digital developed head rigs as part of a facial-capture system to create realistic facial animation and expressions. Captain Haddock (Andy Serkis) and Tintin (Jamie Bell). Please click image above for larger view.
Tintin (Jamie Bell) and Captain Haddock (Andy Serkis). Please click image above for larger view.
Steven cut the film in six weeks post-shoot, and turned it over to us. If the scenes survived intact, they stayed as they were. But many scenes took bits and pieces from several performances. We did do animation on top of the performance capture material. Because Tintin is so plain looking, it required a lot of massaging to bring out what [actor] Jamie Bell was doing. There is no literal translation of the mocap data. We had to go over Tintin again and again. Haddock and Sakharine also had a lot of detail, although their performances were easier to drop in.
Steve Spielberg does get a lighting consultant credit. We never got around to getting a lighting consultant, so it ended up that way. Hergé has very little use of shadow; his world is very bright. But that didn't work for us -- it made the world too flat and featureless. Steven said let's go film noir, so we went very dark. This is one area in which we turned 180 degrees from the books.
Steven chooses a film noir feel. Tintin (Jamie Bell), and Snowy. Please click image above for larger view.
Tintin (Jamie Bell). Please click image above for larger view.
The 3D process was really pretty straightforward also. We rendered one eye to the Avid for editing. Steven did a flat 2D cut. For 3D, we showed him several stylistic choices with convergence, where to place the characters in the scene and so on. In post production, we did hardware renders with stereo cameras, and got most of it refined that way. Once we constructed the 3D rules we were going to play by, it worked itself out.
There were also some visual effects elements, such as the stormy seas, that required new simulation techniques. Other elements included rendering Snowy's white fur, mist, God's rays…everything we would have cheated in 2D, you can't cheat in 3D.
Shots like this of stormy seas and those with god rays, fur, etc., required new simulation techniques. Rackham and Sir Francis Haddock's ships do battle. Please click image above for larger view.
We never really conquered "uncanny valley," in which the closer the character gets to looking really human, it begins to get scary or creepy. The only way to overcome that is to have a real human play the role. How you can try to go beyond uncanny valley is to understand human communications well enough so the animation doesn't look like a dead person talking. There is still a lot to learn regarding facial animation, skin and so on. Most importantly, there's a certain amount of artistry in making a scene look right…and not letting it out the door if it doesn't.
Left to right: Captain Haddock (Andy Serkis) and Tintin (Jamie Bell). Please click image above for larger view.
Left to right: Thompson (Simon Pegg), Tintin (Jamie Bell), Thomson (Nick Frost), Snowy, and Captain Haddock (Andy Serkis). Please click image above for larger view.
Everyone talks about motion capture and characters, but it doesn't automatically give you what's on the screen. When I see Haddock, I see Andy, but most people don't know Andy. We look at the actor/character frame by frame to see if we feel the same way in each frame. And if we don't, we do it again. We're still trying to preserve the guts of the performance.
The Adventures of Tintin was the first fully animated feature done at Weta Digital. The Academy considers Avatar as live action. To us, there is no difference between live action and animation. Will these two techniques converge? I think that's just what you saw with The Adventures of Tintin.
Official "The Adventures of Tintin" Featurette with Steven Spielberg and Peter Jackson - In Theaters December 21
All images from THE ADVENTURES OF TINTIN, from Paramount Pictures and Columbia Pictures in association with Hemisphere Media Capital. Unless otherwise noted, photo credits: WETA Digital Ltd.
The Academy® is Copyright ©2011 by Academy of Motion Picture Arts and Sciences. All rights reserved.
Cinesite Creates Effects for Disney's JOHN CARTER
Great Effects for Good Omens: Collaboration with Milk VFX
|Related Articles / Tutorials:|
Art of the Edit
TV & Movie Appreciation