Behind Cloudy With a Chance of Meatballs' 3D Food FX (With Video!)

in #busy6 years ago (edited)

image.png

**When you're trying ** to create realistic-looking, computer-generated food, the first step is making a mess--at least that's what animators at Sony Imageworks found during their work on Cloudy With a Chance of Meatballs, out Sept. 18. "We've got a nice, three-story-tall building here, and lots of places where you can get hamburgers," says Rob Bredow, visual effects supervisor on Cloudy. So right off the bat "we were dropping hamburgers off the roof to see how it looked," Bredow says. "Basically, they'd just splat when they landed. It was pretty messy."

For the film version of Cloudy, based on the classic children's book, directors Chris Miller and Philip Lord created a backyard genius-like inventor, Flint Lockwood (voiced by SNL's Bill Hader), who creates a machine that turns water into food. Unlike Flint's other inventions--including a monkey-thought translator, spray-on shoes and ratbirds--which tend to cause more harm than good, the Flint Lockwood Dyatonic SuperMutating Dynamic Food Replicator (FLDSMDFR) could save the aspiring scientist's hometown, Chewandswallow, from a bleak future of sardine-only food supply. At first, the plan works; cheeseburgers, hot dogs, steak and ice cream fall from the sky like rain. But when the machine gets overloaded with requests, it begins to create bigger, more dangerous food weather that puts the town at risk.

"We wanted to approach it like this story was a Jerry Bruckheimer movie," Lord says. "If you're going to follow the disaster-movie paradigm, they always have a scientist, a reporter and a cop--so we tried to do silly versions of all those archetypes."

But when it came to creating the CG food, the directors were strictly hands-off. "The team at Imageworks are a bunch of supergeniuses," Miller says. "We let them know what we wanted, and they would find a way to make it happen. Invariably, it was better than we expected."

Animators were tasked with creating about 80 different kinds of food for the movie--and with variations, they ended up making about 150 models. But rather than go for a highly stylized look, like the buildings and people in the town, they chose to make the food look like the real thing. "Food is one of those things, kind of like a human face, where we're pretty good at recognizing what is good food, and what is bad food," Bredow says. "There are lots of things that make food not look appetizing. When we tried more stylized versions of the food, it just didn't feel like something you could eat. So the food is like 1950s advertising photographs of food--because that's what we all wanted to eat."

The filmmakers began shaping and coloring the food in the computer using reference photography, but sometimes that wasn't enough. To make the food believable, it also had to behave the way it would in the real world--so animators put on their aprons and hit the kitchen. One designer cooked bacon every way imaginable: microwaved, completely burnt, with a little grease and with a lot of grease, and everything in between; filmmakers then picked their favorite bacon. To see how Jell-O moved, animators filled a 5-gallon tub with Jell-O, trained high-speed cameras on it and tossed action figures and marbles at the stuff. And for the sequence where cheeseburgers fall out of the sky, the filmmakers dropped burgers from the top of Sony Imageworks' three-story Culver City offices to see how the food would behave in the air and when it hit the ground. The burgers, coming in at high velocity, all broke apart on impact. "It was fun to do," Bredow says, "but it wasn't what we wanted them to look like in the movie." VFX artists opted to create a stylized version of the cheeseburgers using a physics simulation with invisible springs that held some of the burgers together when they hit the ground. For the ones that broke apart, animators had to create digital versions of lettuce, pickles, cheese, buns and burgers that had all the characteristics of their real-life counterparts.
Advertisement - Continue Reading Below

**Video might not have ** been all that helpful for the burgers, but when it came time to animate the Jell-O, it was an important tool. "We could see there were multiple components to the way the Jell-O moved that made it look like Jell-O," Bredow says. "There was this high-frequency timpani drum wave that would fan out, in addition to the lower frequency sine wave. And once we got those two things moving at about the right speeds in our computer simulations, all of a sudden everyone was like, `oh, yeah, that's Jell-O!'"

To make the food look even more realistic, animators applied a lighting technique called subsurface scattering to the 3D food models. "Normally, we calculate the incoming light rays, the shape of the surface, and how much light reflects back to your eye," Bredow explains. "But for food, an important component is the light that goes into the body of the object and bleeds through. If you light a hamburger bun in real life, you don't get a hard line where the light falls off. You get this nice diffused bleed as the light goes into the hamburger bun and leaks back out. So if we didn't light that correctly, the food just didn't look as edible." Another renderer, called Global Illumination, gave animators the ability to calculate the bounce light between different objects, so "if you put a red object near a white wall, you got a subtle red bleed on the wall," Bredow says. "We've only once been able to use it before on a movie, and this was definitely the most ambitious use that we've ever done."

Once the food was modeled, animators began incorporating it into the movie, beginning with the cheeseburger rain. In one sequence, the residents of Chewandswallow are threatened by a massive spaghetti tornado that spews huge meatballs and garlic bread into the town. Animators created a simple version of the tornado that outlined how fast it would spin and where it would be positioned in the shot. The bottom of the tornado spun faster than the top, "so it was like somebody was twisting up spaghetti on the end of a fork," Bredow says. Next, the effects department outfitted the twister with about 3000 al dente noodles, then adjusted the speeds and added meatball projectiles--some of which fly out into the audience, thanks to Cloudy's immersive 3D.

One of the film's more ambitious set pieces was a giant food avalanche; the sequence was so complicated that it took about six months to complete. "There were 50,000 pieces of food, and each one has its own simulated properties--so the banana moves like a banana and the hamburger moves like a hamburger--and they all have to interact with each other," Bredow says. "And then when they hit the buildings, they have to break the buildings apart and send chunks of the building flying, and that all enters the same simulation. From a technical perspective, that's an enormous amount of data to deal with." Not to mention that filmmakers wanted to art-direct the speed and height of the avalanche, which required creating special tools--"invisible bulldozers," Bredow says, "to move the food into the positions we wanted, right before we rolled the camera, and then we let the simulation take over."

Animators knew they were done with a scene when their stomachs started growling. "It was clear the best-looking food was the food that made you want to eat it," Bredow says. "Anybody could look at the food when it wasn't working and say, "That's not edible. I wouldn't eat that.' But when we got it right, everyone was like, "Mmm, I'm hungry now. Let's go for lunch!'"
Source : https://www.popularmechanics.com/culture/movies/a4565/4331140/