Obsessed with technology?
Subscribe to the latest tech news as well as exciting promotions from us and our partners!
By subscribing, you indicate that you have read & understood the SPH's Privacy Policy and PDPA Statement.
News
News Categories

NVIDIA Maxine platform reduces video call bandwidth consumption by a factor of 10

By Vijay Anand - on 5 Oct 2020, 9:01pm

NVIDIA Maxine is billed to solve video conferencing woes and increase quality

NVIDIA Maxine cloud-AI video streaming platform to connect the millions working and studying remotely.

Designed to solve the internet's top source of traffic, video streaming content, NVIDIA Maxine could be just the magic that video conferencing platforms need. 

NVIDIA Maxine is a cloud-native streaming video AI platform that allows service providers to bring new AI-powered capabilities to the more than 30 million web meetings estimated to take place every day. Video conference service providers running the platform on NVIDIA GPUs in the cloud can offer users new AI effects -- including face alignment (where you appear to be facing the camera),  gaze correction (to simulate eye contact even if your camera isn't aligned to the screen), super-resolution, noise cancellation, face relighting and more. Other possibilities include animated avatars with realistic animation driven by your voice and emotional tone in real-time, as well as auto framing that allows the video feed to follow the speaker even if one shifts away from the screen.

Data is processed in the cloud rather than on local system, so all the new added functions can be enjoyed without any new requirements on the client systems.

But beyond new features, first and foremost, NVIDIA Maxine promises massive bandwidth savings by using AI to analyze key facial points of each person on the call, and re-animates the face in the video to the viewer on the other end. This allows video conferencing to stream far less data across the internet as opposed to streaming the entire screen of pixels for every frame. In conjunction with a new video compression algorithm run on NVIDIA GPUs, the bandwidth consumption is cut to one-tenth of the usual H.264 streaming video compression standard. This ultimately means a drastic reduction in costs for service providers and smoother video conferencing experience for those on the receiving end.

The Maxine platform integrates technology from several NVIDIA AI SDKs and APIs. In addition to NVIDIA Jarvis, the Maxine platform also leverages the NVIDIA DeepStream high-throughput audio and video streaming SDK and the NVIDIA TensorRT SDK for high-performance deep learning inference. The AI audio, video and natural language capabilities of the Maxine platform were developed through hundreds of thousands of training hours on NVIDIA DGX systems.

The NVIDIA Maxine platform is now available for early access to computer vision AI developers, software partners, startups and computer manufacturers creating audio and video apps and services. At the moment, cloud services that have lent their endorsement are Google Cloud, Amazon AWS, Tencent Cloud, Oracle and Microsoft Azure.

Also on the topic of collaboration, for 3D design content collaborators and digital artists, NVIDIA's Omniverse that we reported last year is now available in open beta.