John Kahrs recalls working on the Walt Disney Animation Studios feature Tangled and seeing a myriad of 2D concept art and drawings that helped bring the story to life. Embarking on his own film called Paperman, Kahrs felt that “those drawings were something I didn’t want to leave behind when we went into the final product of the computer animation. I thought, ‘Isn’t there a way we could bring drawings back into the final frames of animation?'”
And that’s precisely what transpired with Paperman, a now Oscar-nominated film in the Best Animated Short category that merges CG and hand-drawn animation in telling the story of lonely man in New York City who meets a beautiful woman on his way to work. fxguide spoke to Kahrs about the origins of Paperman and how its unique painterly style was realized by a small team of artists at Walt Disney Animation Studios.
– Above: watch the entire film.
How Paperman began
The story of Paperman, says Kahrs, begins “way back to when I was kind of like the guy in the film, living in New York City and was not a very happy guy. You’re in your early 20s and you kind of think, life should be great, but New York’s an intimidating town with that feeling of being surrounded by people but being totally alone and trying to find a connection that meant something.”
“Every morning on my way to work I would go through Grand Central Station,” Kahrs continues, “and sometimes you’d meet eye to eye with people, just strangers, like a pretty girl or something, and you’d think is there a connection? You feel that connection for a split second and wonder who that person was. That’s the core idea of it – what if two people were really perfect for each other, and they had that chance meeting? And what if they were separated – how would those two people get back together again? And how could a little bit of magic and fate intervene to bring them back together?”
Kahrs worked as an animator at Pixar, and then at Walt Disney Animation Studios, before proposing his Paperman idea to Chief Creative Officer John Lasseter. But in realizing a proof of concept approach for the film, the director wanted to find a way to embrace the 2D style of animation he thought was still something audiences loved and so produce a hybrid with the 3D animation he’d had experience in. “I started thinking about ways of merging the hand-drawn lines and get the best of both worlds,” recalls Kahrs, “trying to find the expressiveness of the line, and keeping that, and then hanging onto the stability and dimensionality and the refinement and motion that you find in computer animation.”
The tech solution
Searching for a tangible way to merge 2D and 3D, Kahrs happened upon the work of Brian Whited, who had been developing a 2D animation tool. “He was chugging away on this program called Meander and as soon as I saw what he was doing, we hijacked him and his software to work in the service of Paperman.”
Meander is a hybrid vector/raster-based drawing and animation system that gives artists an interactive way to craft the film, not just toon-shaded renders. “We didn’t want to do something where you send it off and the computer does something to it and the next morning it comes back and it’s done – we wanted something that spoke back to the artists,” notes Kahrs.
The small team at Disney embarked on some early tests, including close-ups of the girl with her face filling most of the frame. “I knew that if we could have a character be very close to camera filling the field of view you’d have the audience believe that character was fully alive, you’d feel that emotion, and forget about the technology. When we finally showed it to John Lasseter, he knew it wasn’t a cheap trick – it was a quality technique that looked good, had depth and it didn’t look like a gimmick. That was the moment we got the go ahead.”
Step-by-step: the animation process
So how was the final animation completed? Here’s a step-by-step look at the process, as presented at SIGGRAPH LA 2012.
CG animation – the scenes and characters are modeled, rigged and animated in the same way a typical CG animation would be completed (although with simpler models since the CG renders do not represent the final look).
Motion fields – motion fields are created on the CG animation from which the final lines will be advected. These are rendered per-frame per-element with the result being each pixel in the image having a 2D offset. The effect of this is to describe where that point on the CG model will be in the next or previous frame.
Silhouette ribbons – to preserve the ‘line’ look from traditional animation, silhouette ribbons were created for the characters. Here, a character was divided into “topologically-cylindrical components by offsetting the silhouettes perpendicular to the camera direction.” This helped ensure that lines near the character’s silhouette remained on the silhouette.
Motion paper – a paper texture advected by the motion vectors gave a temporal coherent texture to the line and paint.
Final lines and paint – using Meander, final line artists drew key drawings on top of the CG renders, relying on two techniques to then compute the non-key drawings but preserve the temporal coherence of the final-line stage.
Motion pasting – the first technique was called motion pasting, in which “a group of strokes in one frame are selected by the artist and then ‘pasted’ into another frame, such that through all intermediate frames, the strokes are advected by the motion fields defined by the CG animation.” The ultimate effect is that the intermediate frames are automatically drawn.
Motion betweening – this was the second technique. Here, strokes are drawn on an initial key frame and the corresponding strokes drawn on the next keyframe. The inbetweens are then created by automatically computing correspondences, advecting strokes via motion vectors, and geometrically blending the corresponding strokes.
A new approach
Kahrs says that even though the approach to animating Paperman was novel and relied on custom technology, the end result was one that didn’t necessarily draw attention to itself, and always retained the 2D qualities he had been looking for. “I actually had some pretty silly ideas initially about how it would happen. First I thought, what if we were rotoscoping line work over existing animation – how would you project those lines onto the animation? But it turns out the end result is we are kind of dragging the lines around in that flat screen space. It doesn’t just look like a texture map that’s just plastered on there. There’s a flatness about it that I really like.”
We've been a free service since 1999 and now rely on the generous contributions of readers like you. If you'd like to help support our work, please join the hundreds of others and become an fxinsider member.