How Pixar changed the way light works for ‘Monsters University’


From The Verge:

Pixar has a long history of building new software between projects (most recently overhauling its rendering software for Brave), but in the run-up to Monsters University, the changes cut particularly deep. As planning kicked in, director of photography Jean-Claude Kalache announced he wanted to change how light worked at Pixar.

The change has to do with ray-tracing — the technique of mapping out every ray of light in a given scene, as it bounces off walls and characters, casting shadows and producing reflections. It’s closer to the way light works in the actual world, where a single light source will bounce light into every corner of a room. It’s already common in partial-CGI environments like the Transformers and Iron Man movies, where Industrial Light & Magic used the technique to create metallic reflections. But for Pixar’s full CGI environments, mapping out the millions of beams of light was seen as too arduous to be practical. Instead, Pixar relied on manually placed shadows and a network of direct light sources, a setup that was becoming increasingly intricate as the models and setups became more advanced.

Global Illumination did away with all that, producing automatic reflections and shadows based on a simple placement of lights. If a wall is blue, the light coming off it will be exactly as bluish as it should be, at least to start with. Since Pixar is still, as Kalache puts it, “a company that makes cartoons,” the system leaves room for artists to blow out lights and shadows beyond what’s realistic, but the baseline is a naturalism that’s based on the physics of light. And instead of artists planning out soft lights and shadows, everything is moved into hardware. That leaves less work for the designers but more work for the render farm, which works at night while the designers sleep. “It’s all in the memory,” says Christophe Hery, Global Illumination’s core architect for Monsters University. Because of the intensely interconnected nature of the light, everything had to be processed at once, which resulted in incredible RAM demands. Describing one scene, Hery says, “We had the full campus. We had the trees, the grass, the people, the crowds, all held in memory at once. At some point, the rendering engine can try to be clever” about what doesn’t need to be rendered, “but you can only hide so much.”

Continue reading the rest of the story on The Verge