Each new Pixar film employs newer and better technology, but Finding Dory introduces an unprecedented amount of new software to their production pipeline. The company’s chief technology officer Steve May, who worked on Finding Nemo as the supervisor of the shark sequence, says that the process of how they make films has changed a lot since then, but “mainly computers are way faster and algorithms are way better.” Finding Dory introduces three completely new technologies and major improvements in one of their older pieces of software.
While their animation software RenderMan has been around since the late 1980s (used in all of their films as well as many films that have won Oscars in the past 15 years), an entirely new version of the software was employed on Dory. There are basically two kinds of light: direct light and indirect light (which bounces off something else). In Finding Nemo, Pixar could only afford direct light and the lighting artists would create invisible light sources around the scene to mimic what indirect light might look like. That process is intensive as it’s all created and designed by hand.
Finding Dory is the first Pixar film to use the RenderMan RIS architecture software, which is able to create both the direct and indirect light. This allows the lighting team to spend less time trying to mimic reality and more time making creative decisions. Finding Dory has a lot of water, and the software is also able to deal with reflected and refracted light in water, which impacts the color of the water. With something more complicated like a splash, the software is able to reflect and refract light on every single drop of water. Each shot in Finding Dory has billions of individual light rays per frame, with probably ten reflections and refractions in each ray.
On Finding Nemo, creating the look of water meant a lot of work by hand. Now all of that can be created automatically in the software. The software is also able to create foam, aeration inside the water, which adds another 100 reflections and refractions — all of which was not possible before this film. Fish tanks are very complicated, and in Finding Nemo Pixar had three or four people work for six months to add in the reflections on the tank and water surface. This was done by hand for every shot of the tank that you see in that film. In Finding Dory, it’s all accomplished automatically.
One of the biggest challenges with lighting a computer-created scene is that it takes a lot of time to get feedback to see what a render will actually look like. Sometimes this takes minutes, but often times it takes hours for a single frame render. For Finding Dory, Pixar employed a Katana, a DCC app which creates more reactive renders and which works with RenderMan through a plugin. As they move a light source in a scene, the artist see a noisy-looking live render of the results which gradually improves in resolution.
Pixar’s proprietary animation system Presto is not new: they’ve been using since Brave. But they’ve added new abilities to the software for this film. Sully in Monsters Inc. originally had tentacles, but Pixar was forced to get rid of them because they were too hard to animate. They have added a new ability to allow artists to simply draw tentacles rather than program tons of different points along each tentacle. The result is almost like bringing computer animation back to the hand-drawn days.
On Finding Nemo, they built blueprints to help them animate how each fish swims. On Finding Dory, they have created an animation recipe to help make the animation of the swimming more automatic, with controls to allow animators to change and alter how Dory swims. While it doesn’t create the swim animation automatically, it helps take a lot of the tedious work out of it so animators can concentrate on the creative side of it.