In 1988, Pixar introduced RenderMan, built on a system called Reyes (standing for Renders Everything You Ever Saw, and named after Point Reyes, CA). At the time, raytracing was too expensive for the hardware limitations of the day, so physically accurate lighting, reflections, etc, were simply not possible in feature film. Ingeniously, the Reyes algorithm allowed RenderMan to render photorealistic images by faking these effects … without raytracing.

This system allowed Pixar to make the first fully computer-animated feature film, Toy Story (1995). It also powered classics like The Abyss (1989) and Jurassic Park (1993). But as the industry grew, so did the expectations for realism. With the exponential advancements in compute it became possible to use new techniques like ray tracing for production frames to achieve better reflections, shadows, and lighting. Reyes alone wasn’t enough anymore, and so RenderMan evolved.

As hardware became more powerful, Pixar could now use more sophisticated techniques to deliver more realistic lighting, reflections and photorealism. The answer was hybrid rendering, combining Reyes for efficiency with ray tracing for certain effects like ambient occlusion and global illumination.

Films like Ratatouille (2007) and Up (2009) used new technologies, from ray tracing to point based methods, to calculate new levels of visual fidelity to great success. But the demand for realism kept growing, and a new question emerged: Why not just ray trace everything? The answer was simple: computers were not fast enough … yet, and then it became a question of not “if” but “when” would it be possible.

With RenderMan 19, Pixar introduced a completely new architecture, RIS (RenderMan Integrator System), which was a full path-tracing renderer. This system abandoned Reyes entirely and used a method called Monte Carlo ray tracing to simulate real world lighting. Now, instead of faking effects, RenderMan could calculate every bounce of light, every shadow, and every reflection with real physics. This was a huge leap forward. Films like Finding Dory (2016) and Coco (2017) took full advantage of these improvements.

RIS opened up a whole new way of working with RenderMan interactively, as artists didn't have to wait for final images, or buckets, before they made creative choices, but instead worked on an image which was interactively changing as they made creative choices.

However, RIS had a weakness: it was slow to render final images. Ray tracing requires huge amounts of computing power, and RIS only used CPUs. To the rescue came state-of-the-art denoising which made it possible to render production images fast and efficiently using full path-tracing combined with a denoising pass.

Meanwhile, GPUs were getting faster but had memory limitations. Limited memory meant GPUs couldn’t be used on the complex assets for feature films … although it would take years, it was clear that eventually those memory limitations would go away, and GPUs would be ready for feature film production. So the planning was made for the renderer to follow RIS, RenderMan XPU.

With RenderMan 27, Pixar introduces XPU, a new architecture designed to utilize both CPU and GPU power … together, simultaneously. Rather than forcing artists to choose between one or the other, XPU dynamically distributes rendering tasks to whichever processor handles them best. It’s not just a performance upgrade, it’s a philosophical shift.

For decades, rendering was an exercise in patience. Artists would submit frames, wait overnight, and hope their lighting or shading choices landed. But with XPU, that feedback loop is drastically shortened.

Artists no longer need to wait for full renders to see meaningful results, during interactive rendering sessions artists can make adjustments, try new ideas, and work interactively with a scene without missing a beat. This new immediacy empowers artists to explore more creative paths, while the director gives feedback. Creative decisions can be informed by what is seen, not what is imagined.

Multi-GPU support pushes this further, allowing teams to tap into multiple GPUs for heavier scenes. By balancing CPU and GPU workloads, XPU makes the most of the strengths and limitations of existing hardware, just as Reyes and RIS did before it. In short, XPU changes what it feels like to render. It’s no longer about what the machine can handle; it’s about what the artist wants to try. RenderMan becomes part of the creative conversation, not a delay.

Pixar’s rendering innovations have always mirrored and planned for the exponential growth of computational power. The efficiency and innovations of Reyes ushered in a new era of digital visual effects. Hybrid ray tracing pushed realism. RIS perfected light simulation. XPU now unlocks interactive, multi-GPU rendering. Each innovation has been driven by the needs of artists, proving again that “The Art Challenges the Technology, and the Technology Inspires the Art.” 

As we look to the future, RenderMan will continue evolving, pushing the boundaries of what’s possible in rendering for feature animation, visual effects, and beyond.

Are you interested in a deep dive into the history of RenderMan and its historic contributions to feature animation and visual effects? Then check out this great article highlighting the IEEE Milestone Award!