Cinnafilm and Nvidia improve video upconversion

The video optimization specialist Cinnafilm develops alongside Nvidia a breakthrough in Artificial Intelligence (AI) image processing for high-definition and ultra-high-definition upconversions.

This technology, which will be commercially available later this year within Cinnafilm's flagship conversion platform, PixelStrings, can be seen for the first time at NAB 2025 in Las Vegas, April 6-9, 2025.

By integrating Cinnafilm's image processing engine with image processing Nvidia RTX AI, this collaboration will offer an interesting scaling software solution for professional video. This technology combines Cinnafilm's decades of video engineering and GPU-accelerated signal processing expertise with Nvidia's next-generation AI tools, setting a new benchmark for speed, scalability and visual quality impeccable in the media industry.

This collaboration leverages Cinnafilm's image processing engine, known for its Tachyon standards and frame rate conversion, noise reduction, resynchronization and resolution scaling, with new Nvidia RTX AI image processing, offering AI-powered real-time super-resolution, based on RTX Video Super Resolution but customized for Cinnafilm. The result is a single-pass production pipeline that transforms low-resolution content into ultra-high-definition masterpieces from any source with unprecedented detail and clarity.

A breakthrough in conversion

Cinnafilm estimates that the improvements in quality and detail are 25-30% higher to those of its current resolution upscaling solution, which is a significant improvement. While other upscaling tools rely on generic algorithms or generative AI that creates hallucination errors, this joint technology offering will leverage the latest Nvidia Tensor Cores and Cinnafilm's expertise in motion compensation and noise management to minimize artifacts, focus intelligently, and preserve cinematic image quality at scale.

“Our AI collaboration with Nvidia represents a breakthrough for video upconversion,” he highlights. Lance Maurer, CEO and founder of Cinnafilm. "It was paramount to ensure our PixelStrings customers had access to safe and effective AI tools without introducing the headaches of artifact distortion, or adding the potential usage risk currently associated with generative AI. Video engineering is a delicate business, and partnering with Nvidia ensures our AI strategy is based on a stable and refined process. We see a bright future with AI across our product suites, and our first foray into AI will be a game-changer for our customers and the industry. in general.”

Cinnafilm has been a pioneer in the GPU-based image processing since 2004, using the Nvdia CUDA architecture to power its award-winning tools. By incorporating the RTX AI Image Processing SDK, the company now leverages advanced machine learning models optimized for Nvidia RTX GPUs. Whether scaling and normalizing content from multiple camera sources at a large event or enhancing legacy file library content for modern distribution channels, this solution delivers unparalleled automated performance with uncompromising quality for the viewer.

Richard Kerris, VP of Media and Entertainment at Nvidia, says, "Advanced speed and scalability are key to accelerating AI image processing. By collaborating with Cinnafilm, the new solution pushes the boundaries of what's possible by leveraging advanced signal processing and AI enhancement, delivering impressive results, computational power, and workflow efficiencies that will benefit creators and businesses alike."

This joint innovation will be available both on-premises, as SaaS or in hybrid environments, making it completely scalable and automatable to satisfy any level of demand.