Pixar Animation Technology: How Pixar Makes Its Films

Pixar uses a proprietary software suite called RenderMan for rendering, along with Presto for animation and USD (Universal Scene Description) for scene data. These tools, combined with a highly parallelized render farm of thousands of CPUs, allow artists to create the studio’s signature 3D animated films. This article explains the core technologies, how they work together, and the common questions people have about Pixar’s pipeline.

Background: From RenderMan to Presto

Pixar’s history is deeply tied to its software. The company began as the Graphics Group of Lucasfilm in 1979, and its early work on Star Trek II: The Wrath of Khan (1982) featured the first fully computer-generated sequence in a feature film. That sequence was rendered with an early version of RenderMan, which has been the studio’s core rendering engine ever since. In 2012, Pixar introduced Presto, a proprietary animation system that replaced its earlier Marionette tool. Presto was designed to handle the increasing complexity of models and the need for more intuitive, real-time manipulation. Today, Pixar’s pipeline is built around three pillars: Presto for animation, RenderMan for rendering, and USD for data interchange.

Core Technology: Presto and RenderMan

Presto is a node-based animation tool that runs on Linux workstations. It allows animators to create and edit keyframe animation with a real-time viewport that shows the final look of the scene (using a technology called “final quality” preview). This is a major advantage over older systems where the animator worked with simplified proxy models. Presto supports complex rigs, muscle simulations, and cloth dynamics, and it is used for every character and object in a Pixar film.

RenderMan is the renderer that turns the 3D scene into the final 2D image. It uses a technique called REYES (Render Everything You Ever Saw) for rasterization, but since version 21 (2015) it has also included a path tracer (XPU) that simulates the physical behavior of light. RenderMan handles global illumination, subsurface scattering, and motion blur. It is also used by other studios, including ILM and Weta, and has won multiple Academy Awards for technical achievement.

How the Pieces Fit: The Production Pipeline

The making of a Pixar film is a highly organized workflow that moves through several stages:

1. Story and Layout: The story is broken into shots, and a rough 3D layout is created with camera angles and character blocking.
2. Modeling and Rigging: Artists create 3D models of characters and sets (using tools like Maya, but often re-imported into Presto). Riggers add controls for facial expressions and body movement.
3. Animation: Animators use Presto to set keyframes and refine motion. They can see the final lighting and shading in real time, which helps them make better creative decisions.
4. Effects and Simulation: Hair, cloth, water, fire, and other dynamics are simulated using custom solvers (e.g., for hair, Pixar uses a tool called Fizt).
5. Lighting and Rendering: Lighting artists set up virtual lights, then RenderMan renders each frame. A typical frame can take 10-30 hours to render on a single core, so the studio uses a render farm with thousands of cores.
6. Compositing and Final: The rendered images are composited with any 2D elements, color-corrected, and output to film or digital formats.

Render Farm and Computational Power

Pixar’s render farm is one of the largest in the world. For Toy Story 4 (2019), the farm had over 100,000 CPU cores. Rendering a single frame of a Pixar film can take anywhere from 1 to 30 hours, depending on the complexity of the scene. To speed up the process, RenderMan uses parallel rendering: the frame is divided into tiles, and each tile is rendered on a different core. The farm runs 24/7 during production, and a single film can require over 100 million CPU hours.

Universal Scene Description (USD)

USD is an open-source framework developed by Pixar for describing, composing, and sharing 3D scenes. It was first released as open source in 2016. USD allows different departments (modeling, animation, lighting) to work on the same scene simultaneously without conflicts. It uses a system of layers and references, so a character model can be reused across multiple shots, and changes to the model automatically propagate to all shots. USD has become an industry standard, adopted by companies like Apple, Nvidia, and even the film industry’s “Alliance for OpenUSD” (founded in 2023).

Common Misconceptions and Limitations

One common misconception is that Pixar uses a single “magic” software. In reality, they use a combination of proprietary and third-party tools. For example, they use Maya for some modeling and Houdini for certain effects. Another misconception is that the rendering is done on a GPU. While RenderMan supports GPU rendering (XPU), most of the heavy lifting is still done on CPUs because of the massive memory and precision required. A limitation of the pipeline is that it is extremely complex to learn; Presto is not available to the public, so artists typically learn similar tools like Maya or Blender and then adapt.

Alternatives and Comparisons

Other major animation studios use different approaches:

• DreamWorks uses its own Premo animation software and MoonRay renderer, which is also open-sourced as OpenMoonRay.
• Blue Sky Studios (now closed) used a proprietary tool called CGI Studio.
• Independent artists often use Blender (free and open-source) with the Cycles renderer, or Autodesk Maya with Arnold.

Pixar’s RenderMan is also available as a commercial product, so even independent artists can use the same renderer that Pixar uses, though the full pipeline (Presto, USD) is not freely available.

Frequently Asked Questions

Is Pixar’s software available to the public?

No. Presto and RenderMan’s internal tools are not publicly available. However, RenderMan is sold as a commercial product with a free non-commercial license. USD is open-source and can be downloaded from the official repository. Presto is not available, but artists can use similar tools like Blender or Maya.

How long does it take to render a single frame?

It varies widely. A simple scene might take 1-2 hours, while a complex scene with many characters, effects, and lighting can take 20-30 hours on a single core. Because the farm has thousands of cores, the actual wall-clock time is often less than an hour per frame.

Does Pixar use AI to make its films?

Pixar uses machine learning for some tasks, such as denoising rendered images and automating certain animation clean-up. However, the core creative work—modeling, animation, lighting—is done by human artists. AI is a tool, not a replacement.

What is the difference between RenderMan and Presto?

RenderMan is the rendering engine that calculates the final image. Presto is the animation software that lets artists pose and move characters. They are separate tools that work together: Presto sends the scene to RenderMan for final rendering.

How does Pixar handle hair and cloth simulation?

Pixar uses custom simulation solvers. For hair, they use a tool called Fizt (which stands for “Fizt is a zillion times faster” but is actually a physics-based hair solver). Cloth is simulated with a custom cloth solver. These are integrated into Presto, so animators can see the results in real time.

Can I use USD in my own projects?

Yes, USD is open-source and free to use. It is widely supported in tools like Blender, Maya, and Houdini. You can download it from the official USD repository and use it to manage complex 3D scenes.

Related

• Open Source Problems Nobody Admits
• Create New Service Units in systemd?
• Obsidian Plugin Deletes Work: Fixes & Data Recovery