It’s easy to forget that Pixar is a technology company.
Mostly because the invention that they do is in service of a narrative.
When done successfully, your attention is held by that narrative, not
but the technical achievements that made it possible.
With Inside Out,
its newest feature due later this year, Pixar had its own unique set of
technical challenges to overcome. A bigger vision led to scaling
problems, the duality of the film’s narrative meant creating not one,
but two worlds and visual languages — not to mention a main character
made entirely of light.
Inside Out is the story of a young girl named Riley who
moves to San Francisco from the midwest with her parents. To some
extent, the story is about how she adjusts and how her relationship to
her family changes along with their locale.
But, in parallel, there is an internal story being told about her emotions — Joy, Sadness, Fear, Anger and Disgust.
Emotions which influence how Riley feels, acts and reacts to events in
her life — as reflected by some very clever construct building that I
won’t spoil here. This dual story led to some sticky issues when it came
to representing and differentiating them in ways that contributed to
the story.
I spent some time at Pixar recently, watching just under
an hour of Inside Out and talking to some of the creative people behind
the movie. Though there were dozens of anecdotes, a couple of them stood
out to me as unique examples of Pixar’s ability to solve technical and
creative problems through tool building.
A More Human Camera
One of the major technical hurdles to overcome was how to
shoot the movie in a way that communicates the tumultuous, expressive
world of emotions — yet can also transmit the subtleties and nuance of
our ‘outside’ human world.
Typically, when you want to direct a camera in a virtual
world for an animated film, you do it point-by-point. If the desired
effect is a mechanical, tracking, dolly or even handheld shot, each of
those is programmed in by a camera operator to mimic the real-world
equivalent.
For Inside Out, there were a couple of unique issues
because the worlds inside and outside Riley’s head had to be
significantly different. They looked and felt different from art
direction and design viewpoints, of course. But they also had to feel different to the viewer. So different virtual camera techniques were used to film the two worlds.
In addition, as the story progresses, the camera
techniques move from a swooping, 30’s-style mechanical camera into a
much more modern hand operated camera style.
“We’re just like a live action camera crew,” Director of
Photography Patrick Lin said. “Our big tool, our main tool is a camera —
only, our camera is virtual. But although it’s virtual, it’s
mathematically true. It has lenses, focal length, focus, F-stop,
distortion, depth of view and everything. We mimic real camera movement
as if it’s on track, dolly, crane, steadicam or handheld.”
The more mechanical, on-track mimics of standard filming
techniques were used in filming the world outside of Riley’s head. This
lent it a more realistic feel. Inside the head, however, the world of
emotion deserved a more handheld, stedicam-type vibe. Looser and free
form techniques, in general, were used to highlight its much more
malleable and surreal landscape.
In order to do this, they created rough physical cameras
and attached sensors to them. They projected those cameras into the
virtual world and had human operators walk around a physical space,
allowing all of the subtle details of a camera operator’s ‘performance’
to inform the scene.
Things like an ever-so-slightly ‘missed’ focus that makes a
scene feel more energetic as a character moves quickly. Or the very
human flex of a knee stepping down off of a curb to follow the action.
Some of the designs look like black boxes with mechanical
camera wheels attached to them in a similar configuration to actual
camera cranes. Others look like shoulder-mounted blocks of wood with
wands attached — one rig was even a cobbled-together affair with a cheap
plastic tripod dangling from the bottom that gave it mass like a
steadicam.
Lin had previously used the technique while making The
Blue Umbrella, a short that ran before Pixar’s Monsters University. The
version they used then was significantly improved for use on Inside Out.
“On Blue Umbrella it was really in its infancy. The
gearbox that we have, that is actually built by one of our lead layer
artists. He also built a focus ring, too, that can actually do live
focusing, so that we can get that perfect focus more naturally.
Everything we do has to be deliberate, and nothing is accidental.”
For Inside Out, Lin and the photography department used
extensive camera capture techniques. The scenes are set up virtually,
blocked in by a process called Layout — also in their wheelhouse — and
then animated. The camera’s movements through the scene are then shot
just like a live action film, following the ‘actors’ as they move
through the scene.
Both handheld and ‘gear box’ rigs were used to get the
camera movements just right, depending on the act of the movie and
whether we are inside or outside Riley’s head. It’s a solution that’s
both technology heavy — and very aware of the value of a human touch.
RenderMan Saves The Day
One of the main characters in the movie is Joy, as voiced
by Amy Pohler and featured prominently in the trailers and other
marketing. As you can see here, Joy glows. Not only does she glow, but
she’s actually a full on light source.
Having a light bulb walking around in your scenes presented some difficulties to Pixar’s lighting staff.
Character lighting lead Angelique Reisch says that Joy’s glowing nature was one of the tougher technical challenges to overcome.
“We needed really believable light cast from her, so when
she was animated, we wanted to feel that light moving around on the
set,” she said. “Additionally, we wanted it to be pretty detailed. If
she picked up, for example, a frog, we want to see the light between her
fingers. We don’t want to just put a spotlight on there and make the
whole thing bright.”
Joy is a main character, not a single-scene issue to be dealt with, so she was going to be in a lot
of the shots of the movie. In addition to the other lights (some scenes
have as much as 175) in the shot, they had this mobile light source
walking through, causing havoc.
Pixar CEO Ed Catmull started work on what would become RenderMan
with his research at the University of Utah in the 1970’s. RenderMan
became an internal tool used by ILM and Lucasarts to create
photorealistic imagery for their use in movies — it’s been used in
Titanic, Star Wars and Lord of the Rings, among many others. In 1989, it
was offered to the public for the first time. Now over 25 years old, RenderMan is a legendary Pixar product, and one of the things that initiated its purchase by Steve Jobs in 1986. Recently, RenderMan was even made free for budding filmmakers with a non commercial license.
When Reisch’s lighting department needed a way to
represent the light coming off of Joy in a way that would be accurate to
her body’s exact structure, and not just a ‘big light’ that glowed
around her, they were stumped. Until they saw what RenderMan was working
on.
“RenderMan was working on what they call geometric area
lights, or a geometry light,” says Reisch. “What this light does is it
allows you to select a model and then…turn that model into a light
source. This was music to our ears.”
RenderMan’s geolight wasn’t scheduled to be ready for
Inside Out’s production in time, but after some internal ‘red flags’
being raised, Pixar’s global technology and lighting people got together
with RenderMan and were able to run it out in time.
“Around December 2013, I was telling everybody, “
All I want for Christmas is Joy’s geometry light working
”,” says Reisch. “We deployed it with very little testing in March of 2014 into production, but thankfully, it ended up working out and we used it in almost every…actually, as far as I know, every shot in the film.”
Of course, audiences won’t notice. That’s how Pixar knows
it has done its job correctly. When problems are solved or tools are
built there, they’re in service of the only thing that matters: the
story. The temptation to flaunt a technical filmmaking feat on screen is
set aside, and the lighting effects on Joy — or the camera work inside
Riley’s head — are in the end incredibly subtle. Most people will never
know the work that went into them or even notice her innate glow. But
they will feel it, and that’s what matters.
I’ll have more on the movie from my time with it and
interviews with creative staff like director Pete Docter, producer Jonas
Rivera and supervising animators Shawn Krause and Victor Navone in the
weeks to come.