Late June, two of the most prominent international conferences in computer graphics were held in Copenhagen: the Eurographics Symposium on Rendering (EGSR 2025) and High-Performance Graphics (HPG 2025).
With a record average of 180 participants per conference – and many attending both – the week-long events brought together researchers and industry professionals from across the globe to explore the latest advances in rendering techniques, real-time graphics, and interdisciplinary applications used in video games, product visualization, commercials, computer animations, and animated films. The conferences also served as an informal meeting point for discussing future challenges and opportunities in high-performance graphics.
One of the most pressing questions was how AI will transform rendering, says Jeppe Revall Frisvad, Associate Professor in Computer Graphics at DTU Compute, who served as Chair of EGSR 2025 and Local Chair of HPG 2025:
“A central theme – also among the keynote speakers – was the interplay between artificial intelligence and the traditional methods of simulating light, which we have relied on in computer rendering for many years. Will there be peaceful coexistence between AI and classical, physically based 3D graphics built on mathematics, geometry, and physical formulas? Or will AI replace it all with data-driven approaches?”
This is a timely discussion, especially as new AI tools are emerging that enable users to input a storyline and generate a photorealistic film sequence. Could the same soon apply to animation?
What is rendering?
Rendering is the process of digitally simulating light, shadows, materials, and surfaces in 3D graphics to make them appear realistic. All physical objects in a 3D scene are described using mathematical models (algorithms) that define the physical properties of surfaces – such as how they reflect light.
A light source is then placed in the 3D space, and the path of light is traced through the scene, pixel by pixel, ultimately determining the colour that appears on screen. The mathematical model ensures that light is reflected differently depending on the materials it interacts with, mimicking how objects would behave under real-world lighting. This classical method is known as path tracing.
In animated films, it is the director who sets the scene – describing to the technical team what the rendering should achieve, what the narrative is, and how the environment should be designed. The technical team then translates this vision into a 3D universe.
Pixar and the physical world
One of the speakers at HPG 2025 was Per Christensen, Principal Scientist at Pixar Animation Studios (Pixar RenderMan). He holds a Master’s degree from DTU and a PhD in computer science from the University of Washington.
In 2010, he and his colleagues were honoured with a Scientific and Engineering Award (Academy Plaque) at the Oscars ceremony for their pioneering work on point-based rendering for indirect illumination and ambient occlusion. This computer graphics technique, significantly faster than traditional ray-traced methods, has made it possible to achieve colour bleeding effects and realistic shadows in complex scenes for motion pictures - transforming the visual realism of animated and live-action films alike.
Per Christensen and his colleagues do not currently use artificial intelligence in their rendering workflows:
“AI is out there, and it’s incredibly exciting. A lot is happening very quickly. But it’s not something that will change our work in the immediate future – for those of us calculating pixels,” he says.
“Our world is very physically based. It’s light simulation. We have light sources, objects, cars, environments – it’s a simulation of how the physical world behaves. That’s what our software calculates. If you do it with AI, it’s a completely different approach. If you’ve trained a model on enough films, maybe it can be done that way. But that’s not happening next year.”
However, the situation may be different for those who work with idea generation:
“Let’s say you imagine a film about cars with different personalities that meet and interact. If one person can sit down in an afternoon, write a prompt, and have AI generate an entire film, that’s incredibly impressive – but also a bit unsettling,” says Per Christensen.
He also points out that Pixar has a distinctive visual style that is consistent across films – and protected by licensing rights.
