Cinesite has long struggled with the challenges that come from noise in their rendered imagery. which took enormous efforts. When we looked around at what was available, there were only two high-quality, machine learning-based denoisers: Intel and Nvidia.
The obvious choice was Intel® Open Image Denoise. Intel has always been a great partner and this time was no different. They jumped in with their experts and, within days, enabled the tool in our pipeline. The visual quality with Open Image Denoise is outstanding—far beyond what we’ve seen with Nvidia. –Hank Driskill, head of CG – Animation, Cinesite
Since 1994, Cinesite—a renowned VFX, animation, and digital entertainment studio—has conceived, crafted, and innovated digital eye candy. Its singular focus: Combine amazing artistry, engineering, and technology to breathe life into otherworldly visual experiences that span superheroes, fantastic beasties, … and Uncle Fester?
Yes, they’re back.
The Addams Family 2 is Cinesite’s latest feature. Produced by Metro-Goldwyn-Mayer (MGM), the sequel was released in October 2021 with a new twist: its vivid, photorealistic visuals were made possible in large part by Intel® Open Image Denoise, an AI-accelerated denoiser for superior visual quality and part of the Intel® oneAPI Rendering Toolkit.
Meeting the challenge of noise reduction
Noise obscures the desired high-fidelity image, introducing things like specks, granularity, or variations in light, color, and shadows.
Even worse, noise impacts time, quality, and budget. According to Hank Driskill, Cinesite’s head of CG, “Battling noise in renders often creates an extreme amount of work that lands on the already burdened shoulders of effects artists and lighters.”
That scenario was untenable for The Addams Family 2.
“The new film is visually much grander in scope,” says Kenny Chang, head of Lighting at Cinesite. “The first film confined the family to the mansion’s interior and had a dark, brooding, gothic feel. The sequel takes them on a cross-country road trip that includes many exterior environments, monuments, and touristy destinations. The lighting was much more ambitious, requiring a lot of volumetrics to integrate atmosphere—things like distance haze, fog, cityscapes, and landscapes.”
To achieve that, “Cinesite needed advanced software tools and technology that would allow our artists to create ground-breaking, beautiful visuals to accompany our story,” says Laura Brousseau, the movie’s co-director.
So the studio turned to Intel, using Intel® Xeon® Scalable processors as the powerful foundation of its render farm and Intel Open Image Denoise for sophisticated ray tracing and denoising capabilities.
The result?
If you watch the first The Addams Family and then the sequel, you’ll clearly see the visual quality—across the board—has taken a large step up. And Intel Open Image Denoise is an absolute factor in that.
Accelerating movie development with advanced ray tracing
“In production, anything you can do to optimize the process is a win for movie quality,” says Driskill. “So if you can do denoising and cut off the render footprint earlier, that means the renderers take less time, which gives the artists more time to iterate on improving image quality in fewer calendar days—which is always your enemy.”
To stay ahead of an ever-present deadline in the face of increased scope and scale and immense rendering demands, Cinesite used a complementary combination of software development tools running on Intel-based platforms.
The software
- Gaffer* open source, node-based VFX software was used for “look” development, lighting, and rendering.
- Autodesk* Arnold managed the heavy lifting of ray tracing and rendering to deliver the sequel’s elegantly styled 3D world. Particularly for exterior elements—like sunlight shining through leaves and foliage—Arnold’s tracing capabilities delivered highly accurate shadow casting, reflections, and refractions. This freed up artists to focus on the creative side of their work. (Intel Open Image Denoise was integrated into Arnold’s 7.0 release, further expanding the rendering tool’s power and efficiency.)
- Intel Open Image Denoise was used to remove noise from the imagery with its collection of machine-learning algorithms and efficient deep-learning-based denoising filters, which are trained to handle a wide range of samples per pixel. It distinguished an image’s “good stuff (e.g., geometric shapes, textural detail, patterns) from render and pixel noise and cleared away the latter. Not only did denoising create more brilliant scenes, it also significantly reduced the hours-long rendering time on every shot—from around 15 hours per frame per shot, to 12-13 hours, according to Cinesite. To put this in context, Cinesite rendered approximately 175 shots per week, running them in parallel and 24/7 on the render farm. Using Intel Open Image Denoise, machine-power time was reduced by over 500 hours per week.
Before Denoising
After Denoising
The hardware
- 2nd Generation Intel® Xeon Scalable processors (W-2295) were the 18-core workhorses of Cinesite’s render farm, operating 24/7 to distill in-process images into beautiful, photorealistic phantasmagorias.
- Intel® Xeon® processor-based workstations (W-2135 and -2195) were utilized by approximately 300 artists who designed, iterated, and tweaked thousands of images using local versions of Arnold and Intel Open Image Denoise for fast feedback before sending them to the server farm for further optimization.
The development of this movie continues to explore and embrace advancements in rendering technologies to deliver a rich, stylized, CG world; what we call ‘Dilapidated Elegance’.
Before Denoising
After Denoising
It’s about the storytelling – rendering fast with Intel® technologies
High-fidelity, photorealistic experiences are part of our daily interactions—from scientific visualizations and architectural design to gaming and animation. Intel Open Image Denoise is an established part of the equation, improving image quality via powerful machine learning algorithms that selectively filter noise in 3D rendered images.
Want more?
Explore the Intel Open Image Denoise Library, part of the Intel® oneAPI Rendering Toolkit.
Learn about oneAPI, a simplified, unified, cross-architecture programming model.
Why Denoising Matters
Denoising doesn’t change an image in any way. Rather, it reveals the image by clearing away render noise, refining and smoothing it so an artist can better determine if or where it needs to be tweaked.
This is a big deal because rendering complex images can take tens—even scores—of hours, which significantly reduces the number of times an artist can iterate and refine it.
Intel® Open Image Denoise takes the rendering process to the finish line a lot faster. For The Addams Family 2, it cut as much as 50% off the core hours according to Hank Driskill.
Notes and Disclaimers
Testing Date: Results are based on data conducted by Cinesite 2020-21.
10% to up to 25% rendering efficiency/thousands of hours saved in rendering production time/15 hrs per frame per shot to 12-13 hrs.
Cinesite Configuration: 18-core Intel® Xeon® Scalable processors (W-2295) used in render farm, 2nd gen Intel Xeon processor-based workstations (W-2135 and -2195) used. Rendering tools: Gaffer, Arnold, along with optimizations by Intel® Open Image Denoise.
Performance varies by use, configuration, and other factors. Learn more at www.Intel.com/PerformanceIndex.
Your costs and results may vary.
Intel technologies may require enabled hardware, software, or service activation.
Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.
*Other names and brands may be claimed as the property of others.
Related Content
How Tangent Studios Creates Production-Quality Rendering
Watch
Add High-End Photorealistic Rendering to Any Application
Watch
Save Rendering Time & Increase Image Quality with a Free Open-Source Library
Watch
Fidelity-first Ray Tracing in the Heterogeneous Compute Era: See Image Quality Comparisons between Intel® Open Image Denoise & Nvidia Optix
Read
Advanced Ray Tracing APIs Proposed for the oneAPI Specification
Read
Intel® oneAPI Rendering Toolkit
Accelerate high-fidelity rendering and visualization applications with a set of rendering and ray-tracing libraries optimized for XPU performance at any scale.