http://www.cgw.com/Publications/CGW/2012/Volume-35-Issue-4-June-July-2012/The-Royal-Treatment.aspx
The Royal Treatment
By: Barbara Robertson
Pixars
extraordinary run of successful films starring male characters took a
courageous turn in June with the release of Disney/Pixar’s 13th feature,
Brave, the studio’s first princess film. The conflict in this feature
centers on the relationship between Merida, a young “don’t wannabe a
princess,” and her mother, the elegant Queen Elinor. Merida inherited
her father’s fiery character along with his flaming red hair, rather
than her mother’s calm demeanor. She would rather be outdoors riding her
horse, rock climbing, and practicing archery like Fergus, her father,
than studying to be a princess and meeting her pre-ordained
destiny—marriage to the son of a rival clan leader. With a fairy-tale
setting in medieval Scotland amidst lush landscapes and kilt-wearing
clans, and a fairy-tale plot that includes a tricky witch, a spell that
must be undone, and plenty of action-adventure along the way, Brave
enters new territory for the studio.
Behind the scenes, Pixar’s toolmakers and artists entered new territory
as well, by developing and implementing new technology and methods for
the first time that affected everything from the landscape to the
costumes, from animation to Merida’s mop of curly hair.
Setting the Stage
During late summer of 2006 and again in October 2007, 12 members of the
Brave production team—including directors Mark Andrews and Brenda
Chapman, producer Katherine Sarafian, story lead Luis Gonzales,
production designer Steve Pilcher, and shading art director Tia
Kratter—traveled to Scotland to meet people, immerse themselves in the
landscape, and scout locations. In the Scottish Highlands, the team felt
the wind in their faces and pushed their hands down into the spongy
moss that softened the rocks and draped the earth. They saw lichen
dripping from trees shrouded in mist. They stretched out like snow
angels in fields of heather.
“We knew the world of Brave was going to be really, really rich,”
Kratter says. Merida would race through the world surrounding her family
castle, on Angus, her Clydesdale. She would run through it on foot,
following little electric-blue will-o-wisps. It was a landscape beyond
anything attempted before at Pixar.
“The thing about the landscape is what it gives you in spirit,” Pilcher
says. “To do something without that power and magnificence would be a
disservice. We built the ring of monolithic stones. The castle and the
black houses covered with thatch. The beautiful sky with sunlight moving
through mist. The forest, the highlands, the lochs. We had a full range
of emotions to work with visually.”
Although the castle and cottage were dense sets and artistic challenges,
the technical challenges had their roots in the vegetation. “We
prepared lots of set dressing, rocks, and trees, and put them into a
curving landscape,” says Pilcher. “But, the moss didn’t look right.” Nor
did the lichen, bracken (ferns), or grass. Modelers formed the terrain,
rocks, and tree trunks by hand, and set dressers placed the elements
into scenes. But, these landscape elements had no texture.
Pixar implemented a completely new
animation system to produce and perform Fergus, Elinor, Merida, and the
other characters in Brave. Dubbed “Presto,” the new system encompasses
character rigging, animation, layout, and simulation.
Painting with Code
“If you just render the geometry, it’s pretty, but it doesn’t look lush
and furry,” says supervising technical director Bill Wise. “We wanted
Spanish moss hanging from tree limbs, and clumps and hummocks of moss.
The Highlands of Scotland were like another character in the film, a
living backdrop for what was going on. We had never tackled as vast an
outdoor landscape, but we were able to generate it using insane
procedural geometry developed by Inigo Quilez. He’s a magician.”
Quilez describes the method he used to generate the 15 types of foliage,
rocks, and even small flies as “painting with code.” The code exists as
a Pixar RenderMan plug-in written in C++; that is, a DSO. When Quilez
started, he rendered with PRMan 15, then, as time passed, changed to
Version 16.
“It was quite fun using a compiler to produce assets,” Quilez says.
“Going from the flat world into a super-dense 3D world was all my work.
Moss with small clover leaves around it, bracken, hummocks, hanging
moss, all the leaves and pine needles, lichen, grass and flowers,
heather, birch trees, gorse, Scotch broom, the distant trees and rocks,
the small dots that were flies, all were specific pieces of code; all
the shapes, the colors, everything is in the code. We didn’t write a
tool that an artist would use; there’s no user interface. Usually we use
code to glue things together. In this case, we thought of code as
assets.”
At first, Quilez planned to hard-code only moss and grass, but the
result was so successful he ended up writing specific code for many more
types of vegetation. “I abstracted the code and found the parts they
all had in common, but in principle, each is different,” he says. “They
share the logic, of course.”
He treated the vegetation that grew on the rocks, trees, and up from the
ground differently from that without supporting geometry. For the
former: “We’d start with 3D models and go polygon by polygon in the
mesh,” Quilez explains. “For every quad, we would generate random
points, and from those points we would grow flowers, leaves, and
something else.”
For the latter: “When we didn’t have a mesh,” Quilez explains, “we’d
place cubes where we wanted things to grow. These weren’t polygons; they
were mathematical descriptions: This is the center, these are the
sides, end of story. The code would use that to generate detail inside.
We generated bushes out of nowhere.”
Quilez didn’t use typical plant-growing rules to produce the grass,
moss, and other vegetation. “The problem with L-systems and other
old-school techniques is that you have to encode the rules,” he says.
“If you want to change something, you have to change the rules, which
isn’t intuitive. When we wanted to change something, we’d go into the
code and make the change.”
The Scottish landscape provided a living
background for much of the action in the film. To coat the rocks and
earth with moss and clover (above), technical director Inigo Quilez
painted the landscape with code.
Master Gardener
To grow the moss, Quilez would use the orientation of the surface. If
the normals pointed down or horizontally, which suggested more water and
less sun, the moss would have more clover. If the normals pointed up or
vertically, which suggested more sun and less humidity, no clover would
grow. “So, the clover followed the shape of the tree or rock,” he says.
“We used these little tricks to make everything organic. If you just
drop flowers everywhere, it doesn’t look natural. You need driven
randomness.”
All the code executed at render time. There were no physical versions of
these plants, no files, no point clouds. As Quilez refined the look of
the vegetation, he’d render the elements to screen for review, and then
delete the render. He didn’t save images to disk. If a director or art
director asked for a different look—wilder grass, perhaps—he’d change a
number, then compile the code and render it again.
“Everything happened in the CPU when we rendered the frames for the
final shots,” Quilez says. “If we had stored all this detail on disk, we
would have had terabytes of data. We just had formulas and generated
the shapes at render time.”
That meant, though, that animators could not see the shapes until the
code rendered. “Luckily, it was quite quick,” Quilez says. “We could
have the ground plane with grass growing on it within 20 or 30 seconds
using 10 CPUs.”
Even so, about halfway through production, the crew developed a
real-time preview version in OpenGL for the animators and layout
artists. “We did a one-to-one match between RenderMan and OpenGL,”
Quilez says. “Of course, the quality wasn’t the same, but every blade of
grass was there in the real-time tool.”
Although lighting artists didn’t have feedback until after rendering,
that didn’t seem to be a problem, according to Quilez. “They didn’t have
to worry too much because it was quite predictable,” he says. “We
didn’t use raytracing or deep shadows for the grass. Instead, we faked
the occlusion and shadowing. Instead of casting rays, I took the
procedural signal, all the formulas and the fractals, and generated
occlusion and shadows myself. That way, all the vegetation responded to
lighting in a predictable way.”
For example, when the code would create a clump of grass, Quilez
darkened the blades in the center manually. Similarly, if a tree canopy
had a key light on the left, rather than lighting the left side and
darkening the right with shadow maps or deep shadows, he faked the
lighting with instructions in the code.
Magician at Work
“The process would be like this,” Quilez says. “A tree trunk gets
rendered. It calls the code that grows the leaves. The leaves grow with a
shader attached that generates the color, the same as for anything, and
passes it to the lighting. But, some of the coloring and lighting
decisions are in the code. The code doesn’t apply them; the decision
about what is dark and light just passes down the pipeline to the shader
and lighting tool.”
When vegetation needed to interact with characters, Quilez hand-coded
the animation, too. “Because we didn’t have data on disk, there was
nothing to simulate,” he says. “So instead, I told the code where the
interaction would happen by putting in spheres that defined a radius.
Then we had parameters we could tweak to move the blades of grass inside
the sphere.”
The parameters would specify when the blades bent, by how much, and how
long it would take to bend and bounce back. “It wasn’t that difficult to
hand-code the animation,” Quilez says. “I had three or four numbers to
change and everything happened automatically. Because things ran fast,
we could iterate fast.”
Quilez also paid attention to level of detail. To reduce the computing
load, grass and other vegetation far from camera would grow with fewer
curves and less detail. “In the end, we could render all the sets, which
are quite complex, in less than 2gb,” he says.
When Quilez showed his idea to other people at Pixar, some told him he
was, in effect, moving backward; that this was how people did things in
the old days before artist-friendly tools, back when programmers had to
write code to create images. Quilez disagrees with the notion that he’s
reverting to old-school days.
“You always think of CG as moving forward,” Quilez says. “But, it’s more
like a spiral. We have faster computers now. And this was super fun. It
was a dream. I love math, and to have a chance to use math to make
images and beautiful movies was super cool.”
For her part, Kratter, the shading art director, has her own opinion of
Quilez and his work. “When I first came back from Scotland and said I
wanted a dense layering of lichen, grass, and moss, I was told it would
be too expensive. But, Inigo [Quilez] spent six months working on his
code, maybe more, and it was like he gave us a secret present. We had
moss blanketing everything. We called him the wizard.”
Presto, Rig-o
This living backdrop, as Wise describes the landscape, is primarily
Merida’s world, a lush, green world. In the beginning of the story, we
see her as a child picnicking with her parents in the woods, but usually
when she’s outdoors, she’s alone. There are two exceptions—when she
visits the witch and when her mother changes. But otherwise, when she’s
with other characters—her parents, her wild little triplet brothers, the
three clan lords, the dozens and dozens of rowdy clan members—we see
them in and around the castle.
All these characters were rigged, animated, and simulated through an
entirely new proprietary system first implemented for Brave. Named
Presto, the system replaces Pixar’s previous character pipeline, which
was known internally as MenV (2x), and which included the studio’s
proprietary animation system, Geppetto.
Modelers at Pixar work primarily in Autodesk’s Maya. Presto takes over
from there and encompasses character rigging, animation, layout, and
character simulation. Cached scenes go from Presto to lighting and
effects.
Character supervisor Bill Sheffler began working on the new system in 2007 when
Ratatouille wrapped. “A code base over 20 years old or so can reach an
end of life,” he says. “You can’t make big changes and push yourself
further. We kept the best workflows, but this is a first-rate new code
base for us to expand on. The big benefit is that our IK systems are
better, sculpting and weighting tools are better, and we can do better
layering, referencing, and class structures within the rigs. And, it’s
more WYSIWYG.”
Some of the major changes affected the way in which the technical
directors rigged characters. “In the past, we would weight and sculpt by
hand,” Sheffler says. “Now we fill a new character full of a kind of
jelly, put an armature inside, pose the armature via simulation, and use
that to extract shapes for rigging. We can have a fully, nicely
simulated, rigged character in a couple days.”
The new system also automates the process of balancing blendshapes in
faces and makes the process of rigging multiple characters easier. “You
could think about a character as the sum of artistic opinions,” Sheffler
says, explaining that a character’s rig file contains those artistic
opinions, the decisions made as the rig evolves.
“It’s like a painting or drawing, except with a lot of ones and zeros,”
Sheffler says. “If you were to look at it over time, like a time-lapse,
you would see the artistic opinions in a particular syntax iterate and
change slowly. You want to have a nice way to put those opinions in a
global repository.”
Presto provides that global repository, which gives character riggers
flexibility and speed: They can move rigs in part or wholly from one
character to another. “Say you want to see the work on master characters
propagated into hundreds,” Sheffler says. “This is a class system we
couldn’t do in the past—pushing work from one to multiple places in a
dynamic fashion. Now, we can mix and match layers and define how one
inherits another—the shape of a face, maybe, or a control within a face.
We have a space in which the system can interpolate those artistic
opinions.”
To flex characters’ muscles and jiggle skin, Presto integrates
simulation—volumetric flesh and ballistic skin—more closely. This was
particularly important in Brave for Merida’s powerful horse Angus and
the svelte mother bear. In addition, the new system provides inputs to
external, proprietary cloth and hair simulation systems, each of which
incorporated new methods and technology developed for Brave, and the
various other simulation systems used at the studio, as well. All the
departments, including simulation, shading, lighting, and effects,
needed to redo workflows to connect with the new animation system. No
question, a major undertaking.
“The characters in Brave are the most complex we’ve made,” Sheffler
says. “We’re consistently upping the bar internally. And if that weren’t
hard enough, rebuilding the tools to do the film was an additional
challenge. It was a large tax on the technical side in many ways.”
At top, tailors modeled the clothes using
3D tools for the Brave characters, rather than creating costumes from 2D
patterns, as on previous films. At bottom, traditional articulation
with volume and skin simulations, less-defined musculature, and thick
fur helped give the bears mass and volume.
Bear-ly There
To move muscle and skin for those characters, Pixar used the PhysBam
system (developed at Stanford) for the second time. “We had started
working with muscle rigs for Bob in The Incredibles,” Wise says. “But
the first time we used the PhysBam volume simulator was to create big
and jiggly humans in Wall-E, the Captain in particular. We wanted him to
have believable mass, weight, and jiggliness. We also used it in the
big final sequence with the humans sliding down and piling on top of
each other. But, we hadn’t used it since, until Brave. So, using PhysBam
was not new, but we spent a lot of time experimenting with simulated
muscle rigs.”
To make Angus’s muscles look as if they expand and contract as he
gallops through the forest, the team ran volume simulations atop an
articulated, underlying musculature. “We’d build a 3D volume out of the
high-resolution skin mesh, fill it with tetrahedrals, and conform it to
the shape of the final skin mesh,” Wise says. “We ran the volume sim on
that 3D tetrahedral mesh and warped the skin mesh to the results of the
simulation.” As a result, the skin looks as if it stretched over the
muscles appropriately.
To jiggle the skin, the team used a second simulation. “We take the
surface mesh, after it has been warped with the volume simulation, and
then run a skin simulation,” Wise says. “That simulation slides in the
plane. It doesn’t break the silhouette; it slides along the surface. We
attach hair to the result of that simulation, and then run the hair
simulator.”
The bear was a different challenge. “The bear has looser skin and
flesh,” Wise says. “Having the bear feel alive, getting the feeling of
mass and weight, meant feeling that flesh move. So, we used more
traditional articulation with volume and skin simulations, less-defined
musculature, and thicker fur.”
Hair Brains
The studio estimates that 96 percent of the shots in Brave had some form
of simulation. To handle hair and cloth simulation, Claudia Chung led a
team of 16 simulation artists, five tailors and four groomers, and four
simulation leads. In addition to work specifically on shots, the team
handled changes to the pipeline caused by the switch to Presto.
“We have people with varied backgrounds,” Chung says. “Artists and
tailors, and one has a masters’ degree in physics. Some have backgrounds
in computer science and programming. We built our team to have a full
gamut of backgrounds and to rely on one another. The cool thing is that
as a simulation artist, you might understand how to program, to write a
simulation, and understand physics, but you also have to understand the
flow and dynamics of cloth and hair to make it realistic. That requires a
different eye. The artists are the people who know when something looks
wrong and can figure out how to make it right.”
Three senior scientists at Pixar, Andrew Witkin, David Baraff, and
Michael Kass, developed the studio’s FizT (for cartoon physics tool)
simulation system, originally developed for Monsters, Inc. to move the
cloth in little Boo’s T-shirt and the monster Sully’s blue hair. The
breakthrough came with unique methods they devised to trace and analyze
intersections and solve problems caused when hair and cloth intersect,
and especially when one part of a garment inadvertently finds itself
stuck inside another and can’t decide how to get back out. The three
scientists published their findings in a SIGGRAPH 2003 paper, and
received Scientific & Engineering Academy Awards in 2006 for the
cloth system they developed.
After Monsters, Inc., Pixar’s simulation artists continued using FizT to
move hair and cloth, with only a few evolutionary changes along the
way, for each film since. But, Merida’s hair was too much.
“Around the fall of 2008, Lena Petrovitch, our hair lead on Brave, and
Andy Witkin came together to work on a new system,” Chung says, “Lena on
the production side, Andy on the R&D side.”
Petrovitch, who had responsibility for hair modeling and simulation for
Brave, had identified a problem with FizT: To move the volume of hair in
Merida’s design without producing a tangled mess, every part of every
curly strand had to see every other strand, which FizT could do, but not
as fast as they would need. Merida’s thick hair was too long and curly.
“At that time, Andy was really into multiprocessing,” Chung says. “One
of the solutions he came up with was that at the initial start of the
simulation, the system could figure out which hairs cared more about
each other and, depending on algorithms, could send groups of hairs to
different processors. That way, we could run the simulations somewhat
independently while keeping track of whether the groups cared about each
other.” The team discovered that Merida’s hair ran best on eight
processors. Beyond that, the overhead caused by having the groups talk
to each other outweighed the advantages of multiprocessing.
Witkin also devised a solution to handle the volume of curls. “Before,
we ran the simulation on every point of the hair, however many points
were modeled,” Chung says. “With the new method, we filter the curl to a
core spring system, as if you had a core curve through the curl. That
provided dynamics in a more stable way.”
Chung suggests imagining a curly telephone cord to understand the
concept. If you swing the cord, it stays curly. “But, the motion is
almost as if a curve runs through the curl and that’s what moves the
curl,” she explains. “So, what we created was a way to simplify a curl
into a core curve, and simulated that. The motion was on this filtered
core curve. The fun was taking that physical model and figuring out how
to translate it into a point and spring system.”
The core curve solved another problem, as well. Again using the phone
cord example, if you were to fling a phone cord very fast, it would
uncurl and then bounce back into place. With the system Witkin devised,
the same thing happens with Merida’s curls: They lengthen, but at a
certain point, snap back into place.
“This was Andy’s model,” Chung says. “He was all about phone cords. His
way of explaining things was the key to understanding what the problems
were.”
Sadly, the new system, named Taz, was moving into full production when
Witkin died, September 12, 2010, in a scuba diving accident. “I remember
about that time doing weekly walk-throughs with Andy,” Chung says. “He
would visit every technical artist and help us figure out whether a
problem was a production problem or a core problem. He was very much
like a teacher. The very fact that I can explain this to you now is
because he was able to explain it to me.”
The tailors created the kilts by stitching
together 2D panels into an accordion shape, and then had the cloth
simulator relax the accordion into soft folds.
Hair She Is
To create Merida’s curly hair, Elinor’s long braids, Angus’s main and
tail, or all the other characters’ hair that Taz would move as the
characters delivered their performances, groomers started with a bald
model and then placed and defined key hairs. “Merida’s hair was
hardest,” Chung says. “Angus’s was the most complex.”
Indeed, Merida’s hair stylists spent months placing curves on her scalp
that represented the curls—1,500 curves that became 111,000 curves in
the final render. To create Angus the Clydesdale’s hair, groomers
hand-placed 111,000 curves that became 1.8 million in the final render.
Wise describes the process: “We lay down a curve in 3D space in Maya,
and click, click, click, put control points on the curve,” he explains.
“Then, we bend them. It’s very interactive. We define the curl and
sculpt how we want the curve to flow.” Shapes designed around each key
hair define a shell that fills with interpolated hair at render time.
“We spent months and months sculpting and tweaking to get the parameters
just right,” Wise says. “Merida’s hair was directed to within an inch
of its life. This forelock must go here. It must be this thick. It must
curl that way.”
The groomers attached the hairs to Merida’s scalp starting with the
innermost layers, working up her head to the outermost, positioning the
key hairs with the simulator in mind.
“Merida’s curly hair was so massive and interwoven together, the
challenge for Lena [Petrovitch] was in keeping it from becoming a rat’s
nest,” Chung says. “To have control over the artistic design, she
decided to groom Merida’s hair in an antigravity way, almost as if
Merida had touched a light socket.”
Once the groomers finished styling Merida’s curly mop, they sent all the
data to the simulator and turned on gravity. When the hair relaxed, the
artists evaluated the look. Did it sag too much, as if she had just
gotten out of the shower? Is it too stiff?
“We had a whole suite of calisthenic tests that we ran the characters
through to see how the hair behaved,” Wise says. “We turned their heads
left and right, animated them jumping up and down, twisted their torsos
with their arms out to the side. It’s an iterative cycle that goes on
until we have it just right. Later in production, we ran fast-motion and
slow-motion shots to be sure the simulator could handle those
situations and that the parameter settings were correct. People say,
‘[With simulation] you get the motion for free.’ That’s the funniest
thing. Yeah, it’s for free—after three years of development and months
of iterations.”
Once the crew untangled the challenges in creating Merida’s curly hair,
they moved on to Angus’s straight hair. “Those were our test pieces,”
Chung says. “Once we figured those out, our other characters fell into
place.”
The groomers grew Angus’s hair in exactly the way a horse’s hair would
grow, following the contour of his muscles. The challenge was in
creating appropriate motion for his long, silky mane and tail. “We
wanted it to move and hang luxuriously,” Wise says. “It was less of a
grooming challenge and more of a parameter-tweaking challenge. He has
some of the most extreme motion in the film.”
Angus might be swishing his tail in one frame, galloping through the
forest in another at speeds no horse could achieve, and coming to a stop
more quickly than would be possible in reality. However, the simulation
engine uses the laws of physics, those for velocity and gravity, to
move the digital hair, which sometimes resulted in hair moving in
non-artistic ways. So, the artists essentially tricked the simulator,
using various knobs, levers, and parameters to dampen the motion when
necessary and move the hair with “cartoon physics” instead.
That was true for the other characters, as well. For example, Merida
might turn her head quickly, causing her hair to cover an eye during an
important emotional moment. Or, worse. “We had to get rid of hairs from
Angus’s mane that ended up in Merida’s nose,” Wise says. “It isn’t like
every shot needed extra work. We put the simulation through such
rigorous testing before production that the intervention was about
artistic choices.”
An early goal, according to simulation groom and asset lead Emron
Grover, was to be able to send half the shots with Merida on through
without tweaking. “We got to the point where Merida’s hair wasn’t a big
deal, and that was huge,” he says. “She wasn’t easy. There were so many
shots. And, with all those vertices and hair layers colliding, the
simulation times were pretty high, but it was a huge help to use eight
cores simultaneously for her hair. Her hair was faster than her cloth
simulation.”
Dresses and Tresses
For cloth simulation, the team used the latest version of FizT, the
engine originally developed by Witkin, Baraff, and Kass for Monsters,
Inc. The big change was in how the tailors modeled the costumes that
FizT would move.
In the past, tailors created all the costumes for all the characters in
Pixar films from flat panels in the same way tailors in the real world
make clothes out of cloth cut from 2D patterns. The 2D planes fed the
simulator pristine UV coordinates. The system worked.
“You can get garments that look like real clothes because you’re
mimicking real life,” Baraff says. “The skill set needed to model
clothes is pretty specific; you have to understand real tailoring.”
There was only one problem: “The number of people who understand
clothing design is fairly few and far between,” he adds. “We have a few
of those, but we have a whole building full of 3D modelers.”
Thus, the R&D group developed a new system called C3D, for 3D cloth,
that made it possible to create the costumes for the hundreds of
characters in Brave by building the clothes on 3D models. They tested
the system on Toy Story 3 to help Ken with his costume changes, and put
it into full production on Brave.
Even so, for some costumes in Brave, the tailors adopted a hybrid
approach, using flat panels for ruffles and folds. “The 3D modeling
system was a good way to do fast iterating,” Chung says. “But sometimes,
when we modeled things on the organic bodies, the clothes didn’t look
right.” For example, the cloth in Merida’s father Fergus’s kilt looked
lumpy and sculpted when the simulator moved cloth modeled in 3D.
Instead, the tailors created flat planes, seamed them together in an
accordion shape, and let the simulator relax the pleated accordion into
soft folds.
“We started with Merida and Fergus,” Grover says. “Merida because she’s a
main character, and Fergus because we knew he’d be the most difficult.
We had two tailors working on those two. And then we got another tailor
and started on Elinor and the triplets.”
When characters wear more than one layer of clothing, the tailors would
model each layer in 3D (or the 2D/3D hybrid) and then send the heaviest
layers to the simulator first. Fergus had the most layers, usually 10 or
more including a shirt, a chain mail tunic, a leather tunic with metal
studs, the kilt, a leather strap, a belt around his waist, and a cloak
with a bear pelt on it. They started with his belt, worked inward, and
then put his cloak on top.
Although a cloth simulation could affect how the hair moved, the
simulation artists tried to avoid those collisions. “We didn’t want to
wait an hour for the cloth simulation and then another half hour to 45
minutes to see the hair,” Grover says. “So 95 percent of the time the
hair doesn’t collide with the clothes.” Adding an offset outside
Merida’s dress helped, and the trick was not obvious. Otherwise, some
shots would have been onerous.
In one sequence, for example, Merida rides Angus while holding a
tapestry on her lap. “She’s at full gallop,” Chung says. “We had to
simulate the horse’s hair, Merida’s hair, her dress, and the tapestry.
That was a moment. Fortunately, we prepared for hard shots like that. It
was the one-off’s that were harder.”
Chung singles out one shot in particular. “Merida takes off her hood in a
dramatic moment that lasted 600 frames,” she says. “Simulations are
progressive, one frame after another. If the shot were short, we could
have faked it. But, when it’s over 600 frames, everything has to be
stable.”
Internal Forces
The relationship between mother and daughter is at the core of Brave’s
story, and during the film, Merida and Elinor learn to appreciate each
other and discover how to work together to achieve a goal.
Metaphorically, at Pixar, departments that typically work separately
came together on this film and collaborated in new ways.
“We’ve gotten into a rhythm over the years, but sometimes we need to
change the process slightly,” Wise says. “For example, simulation and
character articulation [rigging] used to be separate. First of all,
because there wasn’t much character simulation, and second, because
simulation happened after a character was rigged and handed to
animation. On Brave, the departments needed to be more tightly coupled,
particularly in the case of the bear and the horse. The volumetric flesh
and skin simulations, and how they affected hair simulation, were part
of the rigging process. We had to have a lot of close collaboration and
coordination.”
The same was true of simulation and animation. Chung’s background is as a
simulation artist. She was the cloth lead for Up, and before that, a
simulation artist and tailor on Ratatouille, and a simulation artist on
the short film “Lifted.” “Some of my most fulfilling moments were when I
was working on something artistic and technically challenging with an
artistic partner,” she says. “I wanted to bring that experience to other
people. I made sure the simulation and animation departments worked
together.”
To facilitate that collaboration, the simulation group, knowing that the
animation department could not set up and run simulations, introduced
an animation preview. “It gave them a fairly low-stress, low-effort way
to run a fast representation of the hair and cloth simulation to judge
animation,” Chung says. “Animation really insisted that it be fast and
reliable; that it wouldn’t impact their creative cycle. If it blew up,
they wouldn’t use it, and simulation can be unstable. So it was a hefty
technical project.”
In addition, the animation supervisors and Chung agreed that they would
lockstep the two departments. “When animation kicked off, I kicked off a
technical artist, as well,” Chung says, “so they could work together.”
Thus, each animator had a geeky buddy; and each simulation artist had a
thespian pal. “I think both sides were apprehensive,” Chung says.
“Animators want to be creative and not stifled by something technical.
The technical artists don’t want to be directed. But in the end, I think
both sides realized each had something different to offer.”
There are shots, for example, in which Merida picks up her skirt and
sprints away while running her hand through her hair. Animators would
know how to create the performance. Technical artists would understand
the dynamics.
“An animator might say, ‘I want to hit this silhouette,’ ” Chung says.
“And the technical artist would say, ‘If you want to hit that
silhouette, you need to get her arm out of the way.’ It was pretty cool.
It’s what the animation supervisors and I wanted to happen. Both sides
realized that we are people who can work together.”
There’s another theme running through Brave, too, one embodied in the
title. “This movie is about being brave enough to look inside and see
who you really are,” says director Andrews. “About internal forces. All
the characters in the movie go through being brave.”
So, too, the studio, which was brave enough to tell a story about a
young girl who didn’t want to be a princess, yet would grow strong
enough within herself to command a queenly presence. The tomboyish
Merida learns to protect her family through collaboration and with soft
power—dominance without relying on physical force. And, she does this
without marrying a handsome prince.
Behind the scenes, making this story possible, was another form of
bravery: A crew courageous enough to create this film using new
simulation technology, an entirely new animation pipeline, and new forms
of inter-departmental cooperation. Merida’s hair, the animals’
musculature. Fergus’s clothes, moss growing everywhere. A show in which
96 percent of the shots had simulation.
“Brave was the most challenging film I’ve ever worked on,” says Wise,
who joined Pixar in 1994. “It was a hard, hard, hard film. The sheer
amount of new technology we had to implement made it hard. But, I like
to think the result is there on the screen to see. It’s probably the
most beautiful film I’ve ever worked on. I’m really proud.”
Barbara Robertson is an award-winning writer and a contributing editor for Computer Graphics World. She can be reached at BarbaraRR@comcast.net.