DisneyResearch|Studios technology receives Sci-Tech Oscar
DisneyResearch|Studios, a close partner of ETH Zurich and led by Computer Science Professor Markus Gross, has been selected to receive a Scientific and Technical Award from the Academy of Motion Pictures Arts and Sciences for its Machine Learning Denoiser technology. The Academy recognises technologies that have proven to add significant value to the process of making motion pictures.

Rendering is a crucial process in movie production and one of the most computationally expensive steps. It is the process of tracing light through multiple interactions of millions of light rays with virtual scene geometry and materials. Due to its stochastic nature, rendering generates visible noise, slowly reducing with more computation. Denoising is a quick post-process on rendered content that removes rendering noise and alleviates the need for prohibitively expensive rendering until convergence.
Before this point, denoisers could only reduce some of the rendered noise before introducing objectionable artifacts. The invention of the Machine Learning Denoiser provided substantial noise reduction, transforming otherwise unintelligible noisy render images into clean ones, shaving off the majority of the rendering cost compared to other denoisers. The Machine Learning Denoiser was the first to exceed ILM’s high-quality bar for photorealistic rendering and a cornerstone for making modern movie production possible.
“The Machine Learning Denoiser is one of our earliest and most impactful technologies,” said Markus Gross, Chief Scientist, Research & Innovation, Studio Technology, and Professor of Computer Science at ETH Zurich. It is trained on terabytes of noisy and clean image pairs from older movies rendered with Disney Animation Studios’ external page Hyperion and Pixar’s external page Renderman. During training, the multi-scale kernel-predicting convolutional neural network (CNN) learns to reconstruct the clean rendered image in a short post-process.

“The Machine Learning Denoiser is one of our earliest and most impactful technologies.”Markus Gross, Chief Scientist, Research & Innovation, Studio Technology, and Professor of Computer Science at ETH Zurich![]()
Rendering is one of the final steps in the production of a film. All the elements that make a movie, including video, audio, visual effects, images, and text must be integrated together. Essentially, the source material is converted into what will be displayed as the final footage.
Since its integration at Disney Animation, Pixar, and Industrial Light & Magic in 2018, the denoiser has been used on over 100 movies and shows, starting from “Toy Story 4,” “Ralph Breaks the Internet,” and “Avengers: Endgame.” The project was a team effort across the studios with DisneyResearch|Studios’ Thijs Vogels, a former computer science student from ETH Zurich, Fabrice Rousselle, and Gerhard Röthlin joined by Disney Animation’s David Adler and Pixar’s Mark Meyer as the award recipients. More than 20 collaborators have contributed to this technology over the years, such as the authors of the SIGGRAPH papers that detail the technology and the ILM rendering engineering teams for pipeline optimisation, which enabled its use at a large scale and paved the way for the commercial release in Renderman. “Finally, I would like to mention Marios Papas, the lead of our rendering team, who has been leading the cross-studio denoising project since 2018,” said Markus.
DisneyResearch|Studios drives cutting-edge research and technology innovations that enhance the efficiency of the media production pipelines while empowering storytellers with new tools to realise their creative visions. The world-class researchers at DisneyResearch|Studios operate in close partnership with ETH Zurich and work at the intersection of machine learning and visual computing to revolutionise how we produce movies and create media content.
Academy Scientific and Technical Award
The Academy of Motion Picture Arts and Science recognises and celebrates all aspects of the arts and sciences of moviemaking through renowned awards for cinematic achievement, including the Oscars®. The Scientific and Technical Awards are three different Honorary Awards that are given by the Academy during the annual Academy Awards season. The awards recognise original developments resulting in significant improvements in motion picture production and exhibition.
The Scientific and Engineering Award is given for scientific achievements that produce a definite influence on the advancement of the motion picture industry.
The award ceremony will take place on 29 April 2025.
More information
- external page DisneyResearch|Studios
- external page Press release (Academy of Motion Picture Arts and Science)
- external page Disney's Hyperion Renderer
- external page Pixar's RenderMan
- Markus Gross
- An Oscar for Medusa (ETH article, 11.02.2019)
- external page Oscar-worthy smoke signals (ETH article, 09.01.2013)