What used to be a resource-intensive prototyping process for engineers is now happening on computers in virtual worlds, with new products tested and perfected long before any materials hit the factory floor or cars zip down the track.
“I could show you things we’ve made in Omniverse and you would have a hard time deciphering if that was a photograph of something real in the physical world, or whether it was a simulation,” said Jeremy McCaslin, who directs fluid dynamics product management at Ansys, part of Synopsys.
Ansys — a Canonsburg-based simulation software firm that was acquired by California’s chip-design company Synopsys for $35 billion in July — recently announced that it will integrate Nvidia’s Omniverse technology into its current simulation products. Nvidia is the global leader in computer chips, and its Omniverse platform allows users to access a 3D virtual reality.
Like a lab wind tunnel in a physical environment to test out the aerodynamics of a new car, Omniverse is a virtual environment in which all of Synopsys’ simulations can be tested, Mr. McCaslin said.
3D rendering of hydrogen burner in Nvidia’s Omniverse
“We have lots of tools where you can visualize the virtual assets that Ansys-Synopsys creates,” he said, “but none of them have the level of photorealism and real-world interactivity that Omniverse has.”
Omniverse isn’t new. Nvidia officially released the platform in January 2022, and Ansys has used the system for projects before. But now, the virtual-reality juggernaut is getting woven into the everyday offerings of Ansys-Synopsys’ simulation.
Read More
Mr. McCaslin recalled a client that sells hydrogen burners to companies such as asphalt manufacturers that use the burners to heat up bulk materials.
By animating the hydrogen burner inside of Omniverse and seeing where it would be in the real world, “actually sitting there out in this desert with the hot sun baking on it,” Mr. McCaslin said he was able to visualize how “hostile” the actual working environment is.
“There’s dust everywhere, because you got trucks driving around, and maybe dust is being ingested,” he said, “or it’s sitting in the hot sun, and so it’s being heated by radiation from the outside, and that is going to conduct through the material.”
He started to think about all the new ways the burner might fail. Being able to move around virtually in the product’s future location in Omniverse gave him a “deeper, richer insight into the virtual asset when you see it operating in its working environment before you’ve actually cut metal.”
Many people associate Nvidia with gaming, as the company became the computer-chip tycoon it is now — the first company in history to be valued at $4 trillion. It was mostly known for its pioneering engineering in graphics processing units, or GPUs, that could handle the intense visual rendering that video games demand.
Those video games, Mr. McCaslin said, are getting more and more realistic — sometimes so real that you wonder “was that a video game or reality?”
But what the video games never had quite right, he said, was the physics.
“Imagine you’re playing some game and there’s an explosion in the background, or you see like a tornado in the video game,” Mr. McCaslin said. “It looks real enough that you recognize it as a tornado or an explosion. But the actual physics of that event ... isn’t correct.”
Omniverse, on the other hand, allows engineers to engage in that physics-based assessment of the world alongside the visual illusion of reality.
Digital twins
Omniverse is like the digital twin of the real world: An accurate replica of reality where all the same rules apply, he said.
“A digital twin is basically a virtual asset that is communicating with the physical asset in its working environment,” Mr. McCaslin said. That twin often reflects changes or developments to the physical thing in real time.
Think of a Tesla self-driving car — its digital twin renders all of the car’s surroundings in real time as it drives, updating with pedestrians and other vehicles present, and preparing to make an emergency response if the input changes.
Being able to make a digital twin of a satellite in space is especially helpful, given that it’s essentially impossible to build and test in space. So simulation in the real world has to happen in extreme environments like the desert, said Anthony Matarazzo, a sales director with Synopsys. Now, the physics of space is accessible through Omniverse, and engineers can play around with what the satellite will look like in orbit, interacting with space debris or the dust on the Moon’s surface, he said.
Mr. McCaslin said the goal is to make the transition as seamless as possible for clients who have troves of data already on Synopsys systems. With “the push of a button,” data will be converted from existing Synopsys projects into Omniverse thanks to Universal Scene Description, or OpenUSD, which was created by Pixar, the animation studio, more than a decade ago to make a kind of universal language or file format for the intensive data demand of its 3D worlds.
Nvidia’s website describes OpenUSD as “an open and extensible ecosystem for describing, composing, simulating, and collaborating within 3D worlds,” and essential to its Omniverse platform. Ansys-Synopsys is part of an alliance specifically for OpenUSD technology, alongside Apple, IKEA, Microsoft, Coca-Cola and Epic Games who are also members.
“We’re enabling engineers to step inside their designs,” Mr. Matarazzo said.
“For the first time ever, you have engineers, analysts, manufacturing operators — they all work together in the same virtual environment to simulate and validate the entire [system] collaboratively in real time. So everybody’s looking at the same source of truth. No silos, no delays, no guesswork.”
In the future, he said, that could look like “collaboration at scale,” where engineers in Austin, Texas work with researchers at Carnegie Mellon University and manufacturers in Taiwan, all in one virtual universe.
That kind of thing could “collapse” current development cycles, Mr. Matrazzo said, and that’s the goal. Get everything faster and cheaper but most of all, “more innovation per engineer per hour.”
And that need for speed is something that is fueling fire under all industries touched by AI, and it’s a need verbalized by President Donald Trump and countless other industry leaders emphasizing the urgency associated with a global AI “war” or “race.” More specifically, though, it’s framed as a race against China and its cutting-edge AI innovation.
“We never get time back. We have to go faster,” Mr. Matarazzo told the Post-Gazette after the $35 billion Synopsys acquisition. “From autonomous vehicles to robotics to data centers to energy — the common denominator is time and intelligence. That’s it.”
Simulation, then, is the force multiplier for the expansion of AI infrastructure, he said.
“We’re using Nvidia's hardware to accelerate our software, and using AI so we can accelerate the simulation data exploration,” Mr. Matarazzo said. “So instead of waiting a day or a week to run 100 designs of experiments, you can run 1,000 in a day. Now you have a greater data lake to then train those models and make better decisions.”
Omniverse is helping to generate “more data faster,” he said, and that in turn feeds into the AI training models that further hone the Omniverse platform.
Mr. McCaslin said the same: “AI is really a data science management problem,” and Omniverse being open-source to the public, is “readily connected to those libraries” of data.
The embedded video is a rendering of Synopsys client Astec Industries's hydrogen burner in the Omniverse. In the 3D rendering, everything from the real location of the burners to its thermodynamics are on display. (Courtesy of Astec Industries and Synopsys)





Comments
Welcome to Post-Gazette subscriber-only commenting community. Please review our community guidelines.