Intel Noise Canceller brings the animated film Adams 2 to life

The latest film, Adams 2, is a collaboration between visual effects, animation and digital entertainment company Cianceite and Intel Corporation to take visual and animated effects to new heights of beauty and realism. This animated feature is based on Intel Deep Learning Acceleration technology built into intel ® xeon ® scalable processors and Cinemat rendering farms (distributed parallel cluster computing systems), and is optimized through the Intel oneAPI tool, Intel ® Open Image Denoise. Intel ® Open Image Denoise is an AI-accelerated denoise reducer that achieves superior visuals as part of the Intel ® OneAPI rendering toolkit.

“

When we started making Adams 2, we wanted it to be bigger, better, and more interesting than the first one. To achieve this, we need advanced tools and techniques to help our artists create beautiful, groundbreaking visuals that add color to the story. We're lucky enough to work with Intel to give our artists everything they need to achieve the effect we want in our films.

——Laura Brousseau

Co-director of Cinesite's "Adams Family 2"

”

In Adams Family 2, Cinesite used a complementary combination of Intel oneAPI software development tools running on Intel platforms to improve rendering efficiency by up to 25%(1), saving thousands of hours of render production time, while still completing production tasks ahead of schedule despite a significant increase in both production scale and rendering needs.

Hank Driskill, head of CG for animated films at Cinesite, said: "Eliminating noise from renderings often creates a lot of work that requires visual effects artists and lighting engineers who are already shouldering the heavy burden. ”

Image noise gives animated vision a bad look, including speckles, grain, or deviations in light, color, and shadows, greatly diminishing the visual appeal of a movie. Image noise can also affect production time, quality, and budget.

Driskill said: "In the production process, any method that can optimize the process helps to improve the quality of the film. If noise can be removed early and the render footprint cut off, the rendering time can be shortened, giving artists more time to iterate over a shorter duration to improve image quality. ”

Intel Open Image Denoise uses machine learning algorithms and efficient deep learning noise reduction filters to remove noise from images that are trained to process large samples on a pixel-by-pixel basis. It can distinguish the parts of an image that are really needed, such as geometry, texture detail, and pattern, from rendering and pixel noise in order to clear the latter.

Rendering tools used to make Adams 2 include:

Gaffer, an open source, node-based visual effects production software for "look" development, lighting, and rendering.

Autodesk Arnold, which recently integrated intel Open Image Denoise and took on the heavy lifting of ray tracing and rendering to present a beautiful 3D world.

● Intel Open Image Denoise noise-cancels every frame of a movie.

By using the Intel oneAPI tool, Cinesite was able to significantly reduce the rendering time of each frame of a film, creating more high-quality scenes during production. Instead of spending more time troubleshooting samples, adjusting rendering quality, or re-rendering, artists are able to focus on more creative aspects of filmmaking, such as developing more elaborate, engaging visuals.

Swipe up to read

(1) Test Date: Results are from tests conducted by Cinemat between 2020 and 2021. 10% to 25% increase in rendering efficiency / Saving thousands of hours in rendering time / Reducing 15 hours per frame per shot to 12-13 hours. Cinesite Configuration: Render farm with 18 core Intel ® Xeon ® Scalable processors (W-2295), workstations with 2nd generation Intel Xeon processors (W-2135 and W-2195). Rendering tools: Gaffer, Arnold, and Intel® Open Image Denoise optimizations.

Actual performance varies by use, configuration, and other factors. See www.intel.com/PerformanceIndex for more information.

Specific costs and outcomes may vary.