The idea that food can think is so bizarre, it's a relief that the first inkling appeared in a children's book, Judi and Ron Barrett's Cloudy with a Chance of Meatballs, and in Sony Pictures Animation's (SPA's) subsequent animated feature. Now, in SPA's Cloudy 2 that notion has come to full fruition.
Cody Cameron and Kris Pearn direct the pun-filled comedy starring returning voice actor Bill Hader as Flint Lockwood. Both directors were story artists on Cloudy with a Chance of Meatballs.
In that film, Flint Lockwood's Diatonic Super Mutating Dynamic Food Replicator (DSMDFR) turned water into food with sometimes disastrous results.
"The idea of sentient food came about two years into the first film," Pearn says.
"We have gummy bears attack a plane in that one," Cameron adds.
In Cloudy 2, the foodie characters become even saltier. Flint discovers that the DSMDFR is still operating, but now rather than spaghetti tornadoes, it's making food-animal hybrids. The first film was a disaster film with food attacking from the sky. Cloudy 2 is a monster movie.
FOOD "PEOPLE," like Pickle and Barry (the strawberry), eat up the screen in this “fruit-loose” comedy.
"I basically posed pickles and strawberries in action scenes in the backyard," Cameron says. "Then, we started talking about food creature combinations. [Character Designer] Craig Kellman came up with maybe 80 hybrid combinations, and then other members of the crew started in. We had food tribesmen, food hybrids, and food-animal monsters."
"Foodimals," if you will, like the Watermelophant, Flamango, Cantalope, Fruit Cocatiel, Hippotatomus, Shrimpanzee, Tacodile, and the especially dangerous Apple Pie-thons.
"Lewis Carroll used the portmanteau in which the sound and meaning of two words combine into one," Pearn says. "A lot of our food creatures are portmanteaus."
The foodimals and other food creatures - a strawberry named Barry, pickles, wild scallions, and so forth are visual as well as verbal puns that serve a tantalizing purpose beyond making us laugh.
"The food creatures represent Flint's creativity," Pearn says. "The hook of this film is that there's a meta thing going on. Flint is chasing himself through the film. By accepting the creatures, he accepts himself."
"He goes through an artist's journey," Cameron says. "He learns to be creative and to trust himself."
Cameron and Pearn began their journey on the film at SPA in July 2010, but the team at Sony Pictures Imageworks didn't get into the meat of production until March 2012.
"It was definitely the shortest production SPA has scheduled, and then we ended up shaving a couple months off the end," says Pete Travers, visual effects supervisor. "It was originally scheduled for February 2014, then December 2013, and then September 2013. The last five months were harrowing."
The nutty schedule meant that Animation Supervisor Pete Nash and the five supervising animators organized a crew of 80 to 100 animators more organically than usual.
"Typically, a lead animator has a team with the same artists," says Supervising Animator Joshua Beveridge. "This time, the schedule was so fast, we worked together. Sometimes we had 50 animators, sometimes 10. It was hard, but we made it our responsibility to exchange information." Most animators worked out of Imageworks' Vancouver studio, but some animators and supervisors cooked up the mouth-watering performances in LA.
"Using our own software, we could sync the computers and control each other's machines," Beveridge says. "At every meeting, we were talking face to face using our version of Skype and drawing over each other's work. I didn't expect it to work as well as it did."
Because the film is a sequel, the animators didn't have to work out an underlying concept. "In the Cloudy world, we cue from a UPA-style of animation, which is an extremely limited animation," Beveridge says, referring to the studio that produced such cartoon characters as Mr. Magoo and Gerald McBoing-Boing.
"But, UPA limitations didn't determine how everything moved," Beveridge continues. "The concept does. Some of the food-character designs are funnier when we animate them like believable visual effects creatures with strong weight. Others are funnier if they don't move because the design itself is so funny. Every character was different. We had only one overall rule: Any change has to be funnier. We'd go with whatever made the directors laugh."
From all accounts, the most difficult character was Barry the strawberry. "It was tricky to keep Barry on model and have him look good with all the seeds," Pearn says. "We had to subtract seeds around his mouth."
Cameron provided Barry's voice. "Barry had to emote," Cameron says. "He became one of our key leads in the film. We had a big conversation happening between design and animation."
ANIMATORS gave the carnivorous Double Bacon Cheespider, one of the most dangerous foodimals, the same weight and type of performance as a visual effects monster.
Shape and Bake
All told, there are approximately 50 food-based characters, 10 human stars, and thousands of extras, including the Livecorp nerds where Flint works and the citizens of Sanfranjose. Flint and his friends all appear throughout the film.
"It's a buddy film, so most shots have the entire group," Beveridge says. "We overhauled how the faces work to give the animators more intuitive controls, but the bodies are almost identical under the hood to the characters in the first film. One advantage of a sequel, though, is that we could perfect and harness in our design rules. We were more brutal about the eye shapes, which we felt were a little mooshy in the last film. We made them sharper and clearer."
To resurrect the character models and rigs used for Cloudy with a Chance of Meatballs, the team created what Visual Effects Supervisor Pete Travers calls a "back-in-time snapshot" of that previous film.
"We brought back the same versions of software, the same rigs, even the same hardware," Travers says. "We even put the turntables of the characters into the old version of QuickTime. It was an interesting experiment. Once we had done that, the R&D department brought everything we wanted from the past into the Cloudy 2 production environment, and we went from there. It was remarkably successful."
Because the crew needed to serve up the population of Swallow Falls again in this film, they also revived the crowd system used on the first film. However, when the action moved from the tiny Swallow Falls Island into the more complex world of Livecorp and the Greater Sanfranjose area, the directors asked for more diversity.
Kellman again provided the artwork. "He must have done 50 sketches of nerd-like employees," Travers says, and added, laughing, "He probably had plenty of reference here at our company." Model Supervisor Marvin Kim and his crew then had to scramble those shapes into a palatable and manageable crowd system.
"In most crowd systems, you have a mondo character from which you create lots of variety," Travers says. "You build 'guy A' and try to build 'guy B' out of that topology, and if that doesn't work, you modify the topology."
Eventually, the modelers sculpted a topology from which they could create a variety of body types and mix those with multiple hairstyles. "People often think of rigging as bones and joints, but our mondo rig would access model shapes, like blendshapes," Travers says. The result was akin to applying a facial animation system to an entire body.
"You could move a slider, and a head would morph into another shape," Cameron says. The trick was in rolling out a system in which the animation snippets would work with all the different body types.
"Short nerds walk differently than tall nerds," Travers says. "So, we had a character-creation engine that would semi-bake out the character, and animators would create specific cycles for those characters. That required a little more brute force on the back end, but it was far more effective."
Travers estimates the maximum number of individual characters they created with this system at 4,000. "That was for an auditorium sequence," he says. "The directors wanted an overall feeling that Flint entered a real world, not a simplistic world. All the characters mattered. And, even though they were generic, some had speaking parts. Beyond 4,000 it would be hell, but this is how to create crowds."
Beveridge led the team responsible for animating the sequence in which Flint brings his friends back together, and their introduction to the foodimal creatures in the jungle. "There's a reveal in Sardine Circle," he says. "When we had hundreds of characters, we split the shots into sections and had five or more animators on one shot, with one animator responsible for the characters closest to camera. We tried to avoid that, but sometimes we just had to. By the end of the movie, we started sharing sequences with two supervisors per sequence. It might not have been the most efficient time-wise, but we got higher-quality work. Each of us brought different skills to the table, and we leaned on each other's strengths."
After a few scenes at Livecorp and in Sanfranjose, the action moves to the jungle. "In the first film, Flint's lab represented a mood ring for him," Pearn says. "It was always a graphic
response to where he was emotionally. So, when he first enters the jungle, we play it more like a garden with the camera low. Later, after he makes poor decisions, it's overgrown and feels more like Angkor Wat."
Adds Cameron: "We have some food-looking jungle plants -a take on onion shapes, and a pancake bog that looks like something from a John Ford Western - but it's more like a giant garden."
Unlike the first film in which many zany effects happen within one environment, Cloudy 2 's action has Flint and his friends encounter foodimals in many environments. "We reveal the foodimals in the Sardine Circle, and then Flint and his friends travel down the coconut river and there's a new environment in every shot," Travers says. "We built all those environments from scratch."
The process began with Travers and Production Designer Justin Thompson looking at concept art. "It was Impressionistic with more of a storybook look than the typical precise CG environments," Travers says. "I loved the style. Justin turned to me and asked, 'Can we do this?' I said 'Yeah, but it isn't one person working with Photoshop making a tree from a brushstroke. We need to quantify and translate it into the CG world.'"
Fortunately, Travers had experience with Impressionistic computer graphics. As a CG supervisor at Mass Illusions, the seasoned veteran had supervised the first painterly sequence for the film What Dreams May Come, which won an Oscar for best visual effects in 1999, one of 20 films for which he has served as a CG, digital effects, or visual effects supervisor. Travers received a BAFTA nomination in 2005 for his work on The Aviator.
"What Dreams May Come was a real eye-opener for me," Travers says. "I saw how imprecise you could be, and that within the imprecision you can create beauty and capture a style."
The artists had a limitation with Cloudy 2, though: They had to relate it to the first film. The answer was to keep the similarity with the first film but create a more painterly environment in the background. Whisking the two together would require some magic.
"We called it 'depth styling,'" Travers says.
A TOOTHY Tacodile Supreme hovers behind Tim Lockwood, Flint’s father, as Watermelophants, Wildabeets, Hippotatoes, Meatbalruses, Cantalopes, and various pickles and berries look on.
In the Soup
Often, artists will create establishing shots by using a technique called "level of detail" in which they reduce the complexity of the geometry as it recedes into the distance until, in the far distance, the 3D geometry becomes a matte painting. The crew on Cloudy whipped up a new technique.
"One of my pet peeves is to watch a movie with pristine, precise lighting on objects and then a matte painting takes over and there's a stylistic gap," Travers says. "We did a blend based on Z depth. We render a depth map and strip out lighting as the jungle plants get farther away."
The environments in these establishing shots were 3D geometry that extended back into infinity. "The skies were matte paintings, but the rest is 3D," Travers says. "The background looks like 2D because it's stylized, but this is a stereo film and you can truly see the depth of everything. The shots are beautiful. I wish they were longer."
To create the look, the artists preserved the geometry and original textures, but flattened the lighting and saturated the colors. For tools, they used a pipeline consisting of Autodesk's Maya and The Foundry's Nuke and Katana. "Our environments relied on the artistry of every color and lighting artist on every shot," Travers says.
Rather than composing shots with a photographer's eye, the team saw the shots as a painter might, looking at shapes and colors to balance the composition. "We'd pick plants and make them super-bright orange or turquoise," Travers says. "Usually when you put shots into rendering and start adding lights, the texture becomes muted. When you flatten the lighting, you enrich the color."
Thus, a scene might have an eggplant-colored night scene with tangerine-colored plants. "It's a super-rich environment," Travers says. "The artists would ask, 'Like this orange?' and we'd say, 'More orange.' I would tell people 'Good job' for making a shot un-integrated. It isn't what we usually do."
Although the images in Cloudy with a Chance of Meatballs were highly color saturated, Cloudy 2 pushes saturation even beyond the first film.
"If someone tracked the record for the most saturated movie, I'm sure we broke it," Travers says. "It was unbelievable how much we cranked the color. Digital projectors can capture it, but you will not see the whole richness and color on film. Film can't handle it."
The depth styling also created problems for Arnold, the raytracing-based renderer that Imageworks used. "We were breaking Arnold," Travers says. "Arnold is not exactly rigid, but it is a disciplined, physics-based renderer, and the renders we got out of the box didn't look like what we wanted. The characters were all Arnold-centric. But, for these environments, we had to break Arnold into multiple passes and then rely heavily on compositing to adjust the mattes for anything in the background."
Travers points to other areas in which Arnold excelled. "We used global illumination a ton in this movie," he says. "There's a shot of Chester, the bad guy, wearing a vest that glows, especially in the dark. We used the vest as a light source so, when he walks down a tunnel, we see the light from his vest casting on the walls. And, the cavern sequence is awesome. The interior of a mountain has rock candy and pools everywhere with caustics. People asked if we were really going to raytrace this. I said, 'This is what Arnold is for. Why have a raytracer if not to do something like this?' So we did full-blown raytracing, and it looks so pretty."
To maintain control over the composition, the artists would often dress the shots to the camera and sometimes even asked modelers to add geometry after animation. "It was a bit of a nightmare to do that, but it was worth it," Travers says. "The whole point of the movie was to create a world we haven't seen before. It isn't photoreal. We could do whatever we wanted. There are frames in this movie that look like artwork. They couldn't be better balanced from a composition standpoint."
Animators used a cartoony, “UPA animation” style for main characters.
Inside this zingy world, effects animators needed to peel back and pare their physics-based simulations in ways that would fit the tangy actions. For simulation, the crew used Side Effects Software's Houdini and proprietary software.
"[Animation Supervisor] Pete Nash did an amazing job bringing characters to life in a whimsical world," Travers says. "But the jumpy, jerky motion was difficult for the effects animators. My analogy is to Road Runner cartoons where you see a character pause and then drop quicker than gravity. Simulation software gets confused with that style of animation in a movie."
For shots with the characters shooting down coconut milk rapids, the effects artists applied physical simulation properties and then worked with scaling the simulation and the characters until the fluid had the consistency they wanted. The artists also sent the gooey maple syrup in a breakfast bog rocketing off to infinity with cartoon acceleration.
integrating Impressionistic, painterly backgrounds with cartoony foregrounds was not, ahem, a piece of cake.
"I look at some of the shots and marvel at them," Travers says. "Ten years ago this wasn't even possible. We didn't have this capability of integrating the physically based simulations into a cartoony animation feature."
Sometimes shots had to go back to animators to adjust, but more often, the simulation artists had to fix flakey details. "We have talented people who know where to take shortcuts, where to re-map a simulation, scale up, scale down, trim out undesirable effects," Travers says. "They'd fire off a bunch of simulations and then use 80 percent of this and 20 percent of that."
The directors wanted Flint's coat to move in a certain way, for example, to create a satisfying silhouette. "When his arms become twice as long and the coat still has to behave, it took brute force," Travers says. "The simulations on this film required talent and artistry. Some of the most creative people we have are programmers."
In Cloudy 2, Flint, the main character, travels what the director calls an "artist's journey," in which he learns to be creative and to trust himself. So, too, the crew on this film. These artists learned to work with shapes, color, lighting, and simulation in juicy new ways to produce the lip-smacking feature the directors wanted.
"Our rule was the food had to look delicious," Pearn says. "The cheeseburger might want to eat you, but subconsciously, you want to eat it."
Barbara Robertson is an award-winning writer and a contributing editor for CGW. She can be reached at BarbaraRR@comcast.net.