The reason that nobody’s attempted a sitcom in 3D animation is simple.
It’s because writers and producers rework dialog until the last minute,
a reality that would reduce any whiz kid character rigger to
budget-busting basket cases. But that didn’t stop Toronto-based DKP
Effects from collaborating with the producers of Game Over to try it
anyway. While UPN cancelled the replacement series half-way through its
run, the production strategies developed by the Toronto company point
the way to making 3D viable in prime time.
In order to get the greenlight, a 3D show can’t cost more than the
average live action half-hour. The brainchild of David Sacks (The
Simpsons, The Tick), Game Over was about a family of 3D videogame
characters and so couldn’t be shot live action, and it wouldn’t work as
2D animation. Also raising doubt was the feeling that existing
pipelines for 3D were too rigid to support a comedy that would be
tweaked constantly. Fortunately Sacks met DKP brass who believed that
they could produce a series with 288 characters for a budget comparable
to a live action show.
"The pilot was a hairy time," said John Morch, DKP’s VP, business
development. "That was a hundred times worse than doing six shows in
eight months. At that point no one knew the direction of the show. We
were reinventing the way that animation was produced and building the
pipeline at the same time they were making the pilot." The DKP team
knew that if they were going to pull it off, the process was going to
have to mimic that of building a video game.
Fortunately DKP had been evolving production management software called
the Grid. Jeff Bell, DKP’s VP of R&D explains that the Grid is
custom code that pulls together many off-the-shelf programs — Maya V5.0,
ParticleIllusion V3, Shake V2.47, Final Cut Pro V4 and Jaleo V3.5 — so
that they can be accessed out of a central server from a node on each
animator’s desk. "We’re trying to eliminate all the technical bits and
make it more like doing live action," he said.
"Last-minute editorial changes can be made," said Bell. "They flow
through the pipeline from what the animator sees when he loads up a
shot in Maya to what a compositor or editor sees. If comedic timing
didn’t work, our producer would add that to the shot and the next time
the animator opens the shot, that asset is in place and ready to do
layout on."
Optimizing for speed was critical throughout the process and that meant
opting for lightweight geometry in modeling. Animators used polygonal
surfaces rather than NURB patches or Subdivsion surfaces. The team used
identical base skeletal set-ups for each model. to allow automation of
the setup of animation control rigs. Animators could then move
comfortably from one character to another. Rendering was accelerated by
avoiding procedural textures. File textures were made into
memory-mapped textures, instead of file-based texture, which requires
more memory consumption by the renderer.
The game may be over on the Carsey Werner Mandebach series, but this
experiment showed that animated sitcoms are ready to move to the next
level.
Post your comments below
