Inside Out came out in 2015, produced by Pixar Animation Studios for Walt Disney Pictures and directed by Pete Docter. It grossed $858.8 million worldwide and won an Oscar for Best Animated Feature at the 88th Academy Awards. Inside Out 2 was released on June 14, 2024, and is already a smash hit. Kelsey Mann directed the new film in his feature debut. According to Variety, Inside Out 2 scored $155 million in its first weekend of release, considerably overtaking Dune: Part Two ($82.5 million) and Godzilla x Kong: The New Empire ($80 million) as the biggest opening of 2024. In the international box office, it opened with $140 million, which is enough to beat Frozen 2 ($135 million) as the biggest overseas animated opening of all time.
The original film Inside Out featured a tremendous cast to convey the inner workings of a young girl’s brain. In the first film, in the mind of a young girl named Riley, we are introduced to a series of personified basic emotions that influence her actions: Joy, Sadness, Fear, Disgust, and Anger. Joy is voiced by Amy Poehler; Phyllis Smith voices sadness; and Anger by Lewis Black. For the sequel, Bill Hader and Mindy Kaling, who had voiced Fear and Disgust in the first movie, have been replaced by Tony Hale and Liza, Lapira. Additionally, there is a whole new set of emotions, most notably Envy, voiced by Ayo Edebiri, and Anxiety, voiced by Maya Hawke.
The first film was rendered with Renderman REYES and the new film used RIS. Marsha Ellsworth was the character shading lead on Inside Out 2. “We had to do a lot of tricks. We had to learn how to work in this physically based world when we’re trying to do something magical and animation, and we’re stylized for aesthetics.”
In Pixar’s Inside Out (2015), the emotion characters were composed of multiple elements to create an ethereal look, including a glowing core, hovering particles, edge volumes, and strands of dots. For Inside Out 2, Pixar faced the challenge of recreating these characters while updating them with newer technology and introducing new emotions to Riley’s adolescent mind. The team aimed to preserve the original look using RenderMan’s REYES mode but had to adapt to RenderMan’s newer RIS path tracing framework. They re-implemented procedural primitives in Houdini, enhancing inspection, debugging, and dynamic behaviour. The original look was developed using RenderMan’s REYES mode, which made non-physical behaviours easy to implement and control. However, the team encountered a number of challenges re-implementing these behaviours in RIS, RenderMan’s newer framework. RIS is an energy conservation model, so in some respects it was harder to generate the look in RIS than the earlier REYES. ” With REYES 80% of your job was just getting it to look right and not look broken, but then the last 20% was easy, as you already had all these controls set up,” explains Jacob Kuenzel, who was the character shading principal artist on Inside Out 2. “With RIS, it’s more like 80% of the work is done upfront for you by the renderer. Things look really good right out of the box, but then if you want to really tweak something or break physicality (as this film required), that last 20% of the image is much harder.”
Lighting
Given the non-physical design of the emotion characters, extra close collaboration was needed between the Shading and Lighting departments to achieve their final bright and appealing look. A separate lighting rig controlled her appearance in various lighting scenarios, ensuring consistency and efficiency in rendering.
One particularly interesting aspect was allowing Joy or any of the other characters to seem ‘in the scene’ in terms of lighting when they themselves are glowing. Here, the team had the concept of glow-darkening. Joy, for example, is not unlike a light bulb. Imagine looking at a light bulb with all this energy illuminating the room. So how to you shape the lighting so she feels that she’s in an environment and there’s is actually light from the key light(s) in the scene? How do you shape her? “We introduced this thing ‘glow darkening, and actually, it was introduced in the first film. So from noting the direction of a key light, you can actually darken the glow, so you can feel the energy from light with its tint shaping her.” explains Ellsworth. “Then Joy is not just absolutely flat with glow, but we are able to shape in the light and shade in these very unique ways, and it looked good.”
Things get really much more complicated when characters interact or they hug, “which happens in this film a lot; -there is a lot of hugging,” jokes Ellsworth. “How do you deal with all of the characters casting glows? Well, Joy casts a glow on everything, and there are complex rules of Joy and her comping. For example, if her glowing body is behind Sadness, but her arm is wrapped around them, it gets very complex, but there are controls for that to direct the glows to allow the compositor to make it what is wanted.” The compositing was done in Nuke, and there was a lot of very complex comp work required for this film, more than a usual Pixar film, oftentimes to help solve the nonphysically plausible effects. Many of the looks are also distance-dependent, with a LOD setting and optimization so the characters’ ethereal look reads correctly from any distance.
Another interesting technical challenge came from the particles. In the first film, the original team used particles that were geometrical spheres. On the new film, the team ended up using camera-facing discs. “We used the RiPoint primitive in both shows, but the RiPoint primitive in RIS is a little different than the RiPoint primitive in REYES,” Kuenzel explains. “When I tried to use spheres, there were some issues with our path trace subsurface, where I think there was a bias issue or something between these tiny little spheres. At any rate, I got slightly better behaviour from the camera-facing discs.” The problem with using discs is that on the silhouette, the disc is facing the camera, but for a light wrap, backlighting and other subtle effects, one wants it to behave as though the surface is pointing at an angle to the camera. This meant that the team needed to tell the “physically correct RIS” renderer to geometrically face the camera but, for illumination, have a normal in another direction. “And we hadn’t tickled the RIS into that before. So somewhere in the light sampling, it effectively said, ‘oh, if my geometric normal faces this way, I don’t care – as the lights that are over there, and we were kind of breaking reality.” 
Amusingly, one of the great innovations from the Renderman team has been the powerful denoiser. But especially in this film, the team has to carefully manage all the various components of the Renderman pipeline, as many of the emotions visual effects, particles, glints and sparkles would otherwise be seen as noise and removed during the AI denoising.
Ana Lacaze was the Character’s Shade and Groom supervisor for Inside Out 2. While the first step for the shading team was to match the look of the first film, quickly the team also had to address the render times for the new RIS pipeline. “We’re definitely keeping an eye on the expense,” she explains. “The hair was a perfect example of something that ballooned and that we had to bring it back down to a reasonable level.” Because of the way that the hair was implemented, it was expensive, with a very high render time initially. “It was actually mostly related to the shadowing.”
In the first film, the team used Shadow Maps, “In RIS, we can’t do that anymore; we don’t use shadow maps. …We needed to create spheres around the discs, which could then actually create the shadows,” Lacaze explains. “Also using the opacity was pretty expensive, so we had to manage to change that a little bit to figure out how to make it a little bit less expensive.”
Additionally, once the shading solution was resolved for Joy, each of the other emotional characters did not use exactly the same recipe or script. They were all similar, but for example, only Joy has trailing particles. “There was also a second type of particle that we called soft particles, which is seen on Disgust or Sadness,” Lacaze outlines. In these characters their particles kind of meld with their skin a little more than Joy. “And then we had what we call ‘sugar crystals’, which was more what Anger and Fear used, where the particles are a lot more visible and more responding with specular instead of having a transmissive quality.” Of the new set of characters, they all fell in between these three shader worlds. For example, Anxiety is very ‘sugar crystal’ and is closer to Fear in look than Joy.
All these advancements allowed Pixar to maintain the beloved look of the original characters while taking advantage of new technologies for Inside Out 2.
In the Inside Out films, a person’s emotions are portrayed by characters with wigs mimicking their respective human’s own personal style. For example, if they wear glasses, etc. One of the most important of these is matching the hairstyles of the original character back to their set of emotions. In the first feature released in 2015, these emotion grooms were adapted from the original human wigs through an in-house Maya plugin that required the artist to place a list of landmarks defining correspondences between the source ‘person’ and target ’emotion’s’ scalp surfaces. These correspondences allowed both scalp meshes into a common UV space, and subsequently, every hair curve was warped by the scalp normals at the corresponding hair root locations. This technique produced a rough approximation of the new wig refit which could then be refined using typical Maya tools such as lattice deformers.
Since the first film, the grooming pipeline at Pixar has been re-engineered into a core component of Pixar’s Presto animation system. In the new Presto workflow, hair grooms are composed by layers of guide curves at the scalp. Full-fidelity rendered hairs are then generated procedurally by processing the groom layers through a series of non-destructive shaping operators. The Presto hair grooms are structured similarly to a deformation rig while also sharing direct access to the underlying character’s animation. The new wig refitting enables the transferring of hair grooms across multiple characters with minimal artist input. In Inside Out 2, this allowed for a diverse range of styles and wigs for various human models to their multiple emotion characters, in a total of 45 assets.
There were also extensive rigging improvements, including hands and mouths. In Inside Out 2, most characters have traditional hands with four fingers and a thumb. Pixar aimed to avoid rigging each character with unique topology from scratch, as they had previously done. They utilized new articulation tools, curvenet and Profile Mover, to deform subdivision surfaces independently of shape or topology. For the first time at Pixar, they created a fully articulated hand curvenet as a shareable rig. This allowed for fully rigged hands for all characters without additional rigging work on the geometry.
Another challenge was crafting Anxiety’s expressive mouth. Anxiety is a unique character akin to a cloth puppet. To animate her, a specialized mouth rig setup that could provide flexible and precise control was required. Anxiety’s design required a rigging solution that could accommodate her unique features.
The rigging design for the character Anxiety involved a three-layer mouth corner rig to manipulate the outer silhouette and mouth corners and refine the outer silhouette without affecting tooth placement. Additional lip controls, divided into nine sections, allowed for wavy shapes. A curvenet setup on the lips and head enabled detailed control over the facial shape. For the teeth, curves wrapped around the lips and individual tooth controls allowed for precise positioning. These solutions provided animators with flexibility and precision to create dynamic facial expressions for Anxiety.
copyright © fxguide, LLC
