DNEG 谈《Coyote vs. Acme》如何让卡通逻辑在实拍世界中成立
DNEG on making cartoon logic work in Coyote vs. Acme
DNEG 视效总监 Alexander Seaman 与动画总监 Keith Roberts 在 fxpodcast 中讲述《Coyote vs. Acme》的制作,团队把长片动画与视效技术结合,在实拍画面里保留 Looney Tunes 的喜剧感。
In this fxpodcast, we speak with Visual Effects Supervisor Alexander Seaman and Animation Director Keith Roberts about DNEG’s extensive work on Coyote vs Acme. Working with director Dave Green (Teenage Mutant Ninja Turtles: Out of the Shadows) and Production VFX Supervisor George Murphy (Maleficent, King Kong, The Matrix Revolutions, The Matrix Reloaded), their teams combined feature animation and visual effects techniques to preserve the particular comedy of Looney Tunes within a photographed world.
Both came to the project with considerable affection for the material. Seaman recalls being dressed as Roger Rabbit for a childhood birthday and immediately volunteering when the opportunity arose. Roberts was involved from the outset, drawn to a production that could use both sides of DNEG’s capabilities. That enthusiasm mattered, because faithfully translating these characters into 3D required a great many deliberate departures from conventional CG practice.
The falling piano is a particularly good example. Its impact has the abrupt, violent timing of a Warner Bros. cartoon, while the car underneath it needs enough recognisable weight and behaviour to remain believable. Even the decision to use a cartoon piano involved experimentation. The team tested different approaches, including whether a realistic piano performing the action would be as funny.
Their working method was to establish the piano’s action and Wile E.’s performance first, then work out how the surrounding world could accommodate them. Seaman explains that the car destruction incorporated a practical special effects stunt, which the team augmented and retimed. The apparent collision between two kinds of physics was carefully constructed around the animated gag.
That flexibility extended to the characters’ frame timing. Early tests placed animation on ones and twos against live action plates. On twos, a pose is held for two frames, an essential part of the familiar cartoon vocabulary. But Roberts found that Wile E. walking beside a live action person needed animation on ones, with movement updated every frame. The film required a mixture, chosen according to the performance and its relationship with the photography.
Roberts was equally careful about how much movement the animation contained. Who Framed Roger Rabbit was an obvious reference for bringing drawn characters into live action, but the target here was specifically Chuck Jones. The team needed his economy of posing and precise timing. The flowing curves and overlapping movement that CG can produce so readily could easily pull the performances away from that style.
“Their comedy was all about clean timing and about the spare quality of the animation,” Roberts explains.
A major part of preserving that quality was the sketchvis process. Almost every shot passed through artists drawing directly over the live action plates. The team reached around ten people at one point, although Roberts recalls much of the film being handled by about four. They were encouraged to explore the performance freely, without first having to solve the photographed lens or the limitations of a 3D rig.
Those drawings gave Green a rapid way to evaluate alternatives. The artists could produce several versions of a shot or sequence in a week, allowing the director to choose moments and develop the comedy before committing to the more involved 3D work. The animation team then had to find ways to reproduce the strongest drawings, sometimes through substantial adjustments to the rigs and cameras.

This helps explain some of the film’s striking views of Wile E. There are close, wide angle compositions that give him a dimensional presence while retaining the familiar graphic arrangement of his face. Seaman points out how narrow the margin is between a recognisable character and a version that simply looks wrong. A lens that works beautifully for the actors may distort the cartoon beyond that point.
The character’s rendering camera therefore often used a different lens from the live action photography. The team could adjust his perspective to preserve the required expression and silhouette. Where photoreal elements interacted with the cartoon rendering, these separate camera treatments created further work, but they gave the artists essential control over the character’s appearance.
Much of the performance had already begun on set. Puppeteer Dave Barclay provided the actors with a physical Wile E., giving them someone to look at and, crucially, someone to react to. A small head movement could acknowledge a line or prompt a response. For a character communicating so much through expression and gesture, those exchanges helped establish the rhythm of the scene.
The intention was often to rehearse with the puppet and then photograph the actors with a simpler eyeline reference. In practice, the interaction with Barclay could produce the preferred take. Seaman’s team would then remove both puppet and puppeteer, retaining the timing that made the scene work. Some shots became more complicated still when performances from different takes had to be combined.
The puppet itself could be simplified to suit the shot. Its ears, nose and arms were removable, and sometimes Barclay could perform with little more than a pair of eyes. This reduced the material obscuring the background while preserving something for the actor to respond to.
The rendering process required comparable attention to the characters’ graphic identity. After conventional plate preparation, the work moved through the animation pipeline using rigs designed to support much more extreme manipulation than a typical photoreal character. DNEG developed toon shaders in Clarisse, initially rendering relatively flat images to check that the animation read correctly.
The outlines were a substantial technical undertaking. Generated in Houdini from the outlines of the shading network, they were adjusted according to factors including movement, their position on the character and distance from camera. Artists worked specifically on line treatment, because its effect on the finished character was so significant.
A line around a paw could not simply grow thicker in proportion to the paw’s approach to camera. Characters at comparable depths also needed consistent line weights. These decisions helped maintain the impression of a drawing through changes in scale and perspective, with the results checked against the reference model sheets.
Lighting demanded restraint. Seaman describes having to steer artists towards simpler key lighting and flatter results. Much of the lighting treatment was handled in Nuke using roto shapes and simple light treatments, allowing carefully placed accents without modelling every surface in light and shade. The team could spend time finding the right highlight along Wile E.’s nose while keeping the overall image faithful to the cartoon.
The tunnel sequence tested these relationships in a different setting, placing photographed actors into a cartoon environment. On-set lighting anticipated elements such as the moving tunnel lights. The VFX team then used body tracks to support selective relighting and to cast cartoon shadows onto the ground.
The desert presented another substantial challenge. It needed the warmth and familiarity of the original cartoons, expressed through a photorealistic environment.
“It needed to be realistic, but you could never actually go to that place,” Seaman says.
Unreal Engine helped establish ideas, blocking, and camera moves for Green. That work passed into animation, where Roberts’ team developed the gags and the cameras needed to present them. The environment had to remain flexible around those choices, leaving much of its final visual development until surprisingly late. For Seaman, allowing such a large virtual setting to follow the evolving animation was one of the more challenging aspects of the production.
Listen to the full fxpodcast to hear Seaman and Roberts fully discuss how much detailed work supports moments that feel wonderfully spontaneous on screen.
来源:fxguide · fxguide.com

