At the recent GPU Developer Conference (GTC), NVIDIA announced a range of new hardware, software and cloud solutions and partnerships. One of the most consequential and exciting areas of news is a series of announcements related to Omniverse.
On a related topic: What is Generative AI?
Omniverse Brings Industrial Metaverse to Life
NVIDIA’s Omniverse is a computing platform based on Universal Scene Description (USD), an open ecosystem for developing 3D worlds in support of what NVIDIA refers to as the “industrial metaverse.”
NVIDIA sees the metaverse as the next generation of the internet – a 3D-connected internet – where users will go from 3D world to 3D world to work, collaborate and play. Developed by Pixar, USD has been adopted by Apple, Autodesk and many other companies, with NVIDIA calling it the “building block of the metaverse.”
One of the most important aspects of Omniverse is that it is designed to be as realistic as possible. This means all of the laws of physics applies, objects cast shadows, light reflects causing distortion and colors look different in the shade. This allows developers to create a digital twin of almost any environment, which can accelerate the deployment of new services. For example, a retailer could use Omniverse to simulate traffic flows in different store designs and test hundreds of options before doing any construction.
For more information, also see: Top AI Software
NVIDIA Announces Enhancements to Omniverse
At the conference, several of NVIDIA’s announcements were related to the use of 3D worlds in design and visualization and the management of warehouses and factories, and in the automotive industry.
Below is a summary of the main Omniverse announcements at GTC23.
Growing the Ecosystem with New Omniverse Connectors
NVIDIA announced several new USD-based connectors to Omniverse, including with Siemens Teamcenter NX Process Simulate, Rockwell Automation Emulate 3D, FlexSim, Blackshark.ai, and Unity.
NVIDIA also announced the general availability of Bentley Systems LumenRT for NVIDIA Omniverse, which was announced last year. LumenRT for NVIDIA Omniverse allows companies to harness Omniverse to deliver 4D visualizations of digital twins.
NVIDIA now has hundreds of connectors to Omniverse – from software to sensors to cameras – creating a broad development ecosystem that enables companies to create more powerful and flexible solutions based on visualizing a solution before actually developing it.
For more information, also see: Top AI Startups
New Features in Omniverse
New features in the Omniverse platform itself include:
- Real-time ray-traced subsurface scattering shader – This is the process of accurately capturing what happens when light hits an object and is refracted or split or shattered through different types of surfaces. NVIDIA was the first company to do real-time photorealistic ray tracing a few years ago, and the addition of real-time ray-traced subsurface scattering in Omniverse is impressive.
- Performance improvements – Improvements around runtime data transfer and scene optimizers will support operating more complex digital twins, including for a factory, warehouse, city block and more.
- More SimReady assets – SimReady is a standard for creating photo-realistic 3D assets that include standard metadata that allows anyone on the Omniverse platform to use them. NVIDIA has now made hundreds of free SimReady assets available to customers.
- Generative AI – New generative AI capabilities include Text-to-Material, Text-to-Code, and Audio2Face.
The company also introduced NVIDIA Picasso, a new cloud service that developers can use to build and deploy generative AI-powered text-to-image, text-to-video and text-to-3D applications.
NVIDIA is partnering with Getty Images, Adobe and Shutterstock to provide developers with USD-based content they can easily incorporate into their workflows.
For more information, also see: Cloud and AI Combined: Revolutionizing Tech
NVIDIA Omniverse Cloud and Microsoft
One of the more intriguing announcements at GTC was that Omniverse Cloud will be available on Microsoft Azure.
The availability of a subscription-based, platform-as-a-service Omniverse offering via the Azure Marketplace will make it easier for enterprises to adopt and begin experimenting with the solution. NVIDIA Omniverse Cloud on Microsoft Azure will include onboarding, direct access to NVIDIA Omniverse experts and 24×7 enterprise support and training.
NVIDIA also announced that it is collaborating with Microsoft on advancing the “industrial metaverse” by providing access to Omniverse for hundreds of millions of Microsoft Enterprise users. In addition to Omniverse Cloud on Azure, the two companies are integrating Omniverse into Microsoft 365, including Teams, OneDrive and SharePoint. At a basic level, this opens the door to remote collaboration on Omniverse objects and environments. Omniverse is also being connected to Azure IoT to enable real-world inputs from Azure Digital Twins to Omniverse models.
This collaboration with Microsoft on the industrial metaverse is already being played out extensively in the automotive industry. Omniverse has the ability to connect diverse teams using a variety of software packages across concept and styling, design engineering, car electronics, smart factory, retail experience, and autonomous driving.
Connecting teams this way has the potential to make the manufacture of automobiles far more efficient. This includes the use of digital twins that will enable teams across each phase of manufacturing to test out designs in the virtual world before committing dollars in the physical world. BMW, Mercedes Benz, Volvo, Toyota, General Motors, Lotus Cars, and Lucid Motors all are using Omniverse in parts of their workflows.
BMW Opens the World’s First Entirely Virtual Factory
By far the most interesting announcement related to manufacturing at NVIDIA GTC was that the BMW Group is starting a global rollout of Omniverse across all of its factories, and is creating the first entirely virtual factory in Debrecen, Hungary.
Due to come online this year – approximately two years before the completion of the physical factory in 2025 – the virtual factory will allow workers to get a head start on designing the factory layout, training of robots, and even sales and marketing, which will save money and increase safety.
On a related topic: The AI Market: An Overview