This company wants to be the Google for the 3D world
Physna is launching a search engine for 3D models. It's a first step toward digitally indexing the physical world.
Even a once-in-a-generation pandemic couldn't keep Dennis DeMeyere from jumping at a life-changing opportunity.
DeMeyere was a technical director in the Google Cloud CTO's office, one of Google's burgeoning businesses. The work was good. "I had a lovely time there," he said. But DeMeyere, who is originally from Michigan, kept in contact with venture capitalists in the Midwest, and had the chance to meet Paul Powers, the founder and CEO of Physna.
Powers showed DeMeyere software he had been working on that takes a geometrical approach to understanding the volume of any object. The software maps the relationships of small fractal shapes — tiny triangles, as DeMeyere puts it — to create a millimeter-perfect virtual representation of a physical object. What Powers had built was more accurate than anything DeMeyere had ever seen, he said, and it got him thinking about the possibilities. "Then he shows me the customer list … a who's who of Fortune 50," DeMeyere said. He was so interested in joining the company that he dropped everything to start.
"When I moved, I jumped on a plane with two checked bags and moved into a hotel, until I could rent a fully furnished home," DeMeyere said. "I have been so busy and excited, I haven't bothered to have anything else shipped out to Ohio."
"It's so rare that something comes out that's so genuinely new," DeMeyere said. "I had the best team and the best job — I saw something that was technically unique."
What DeMeyere saw in the company just a few months ago is now ready for anyone to take advantage of. On Tuesday, Physna is launching a public search engine called Thangs (a name that DeMeyere was quick to point out he inherited rather than chose). The team is thinking of the product as a cross between Google and GitHub for anything with volume. It could well end up being the stepping stone that sets Physna, or someone like it, up to do with three-dimensional objects what Google did to words and images on the internet.
On its surface, Thangs looks rather like the Thingiverse, Yeggi or other sites that 3D-printing enthusiasts use to upload and search for models of things to print out. And Thangs definitely has that functionality — a search for "Pikachu" will show you several models of the adorable Pokémon that users have uploaded — as well as searching every publicly available 3D model database. What it also shows you is a graphical, collaborative history of each model, kind of like going back in the edit history on Google Docs or version controlling on GitHub. And if a model is actually part of another model, it can have its own page with its own edit history, allowing you to change each requisite part without having to reupload the entire thing.
You can set up a profile and comment on or collaborate on any public model, as well as store as many models as you would like. Say you're working with some friends on pulling together a massive 3D-printed Iron Man costume for the next time you can go to a Comic-Con; each person could take a different piece of the armor and work on it separately then have all their work ladder up into one project folder, without anyone having to act like a project manager and keep track of each individual piece.
"Never thought I'd be commenting on a 3D Pikachu model, but here I am," one early user wrote.
An individual model page on Thangs.Image: Physna
Thangs is just the latest effort by Physna, which will help raise awareness of its product with the general public. But the company already has a growing enterprise business. Physna offers what DeMeyere calls an enterprise SaaS product that's supposed to alleviate some of the opacity in complex corporate supply chains.
Many companies end up ordering accidentally the same part — or identical parts from several vendors — because keeping track of every item needed to make things like smartphones, cars and planes, let alone every iteration in every country you've ever produced in, is an extremely difficult task. To date, companies have tended to rely on product lifecycle management, or PLM, software to be able to keep tabs on everything. But for anything that relies on user input, especially when it might involve people in several countries speaking multiple languages, it's easy for things to fall through the cracks. "Geometry is a universal language," DeMeyere said.
Physna's search product allows companies to search by 3D models, rather than by text. If a company orders the same wheel but has logged it three different ways from two different suppliers, Physna's software will be able to tell if it's the same product. It can even tell if something else in a company's product catalogue is a close enough match, as well as if there are slight differences that might be imperceptible to the naked eye. DeMeyere said that one company that makes french-fry baskets was inadvertently buying the same parts 100 different ways for different products before Physna showed them.
"It's solving the fact that people are imperfect," DeMeyere said.
DeMeyere said that Physna's models have 10,000 times more data per model than those built with traditional scanning software because of "geometric decomposition." Physna can take any CAD files and turn them into volumetric files to be used in its engine. It's a similar concept to how video game designers create objects, laying them out in a series of flattened, connected polygons (along with other data about the object's topography). Physna's algorithm then pattern-matches that object with everything else uploaded to the engine to figure out if it's a match. DeMeyere tells companies, "Give us the objects, we'll break them apart with math."
The long-term goal is to provide a similar offering to Thangs for enterprise customers. It's a lot easier to cut down on duplicative part ordering when you have full knowledge of what's already available and how exactly it might fit into the product you're building. DeMeyere said some clients have already signed contracts expecting a product like this in the future.
That being said, DeMeyere said the team is concentrating on what it now has on the market. The team is already running at three times the board's customer-acquisition targets, DeMeyere said. "What's more important is making them work well first," he said.
Physna first launched in 2015, and has raised just shy of $9 million to date, according to PitchBook. In July, DeMeyere joined the team, along with two new board members: Mark Kvamme, a partner at Drive Capital (previously of Sequoia), and Jason Warner, the CTO of GitHub. Between Warner and DeMeyere, it's easy to see how the "Google and GitHub for 3D" concept has crystallized within the company.
Thangs is just the start of what DeMeyere envisions for the volumetric data on Physna. "I want to build a geometric cloud," he said. It's a tough sell right now, given that so much of this is still conceptual. "The problem is, nobody has anything to compare this to," DeMeyere said.
But if you accept that the way we interact with the internet in the future is going to evolve, perhaps it's not that crazy an idea. The internet to date has primarily been experienced on computer screens of varying sizes, which is fine for text and flat images. But it's possible that the future will have far more depth — whether that's through augmented-reality headsets the likes of which Microsoft and others are developing, or VR that Facebook is building, or even volumetric displays — and we'll need companies built to understand that 3D data.
DeMeyere said he wants to build software that would allow any smartphone with the requisite sensors (think newer iPhones, or the lidar scanner Apple put on the latest iPad Pro) to be able to scan any object in the real world and create a model that could run through Physna's engine. For businesses, this could be game-changing for anyone who has a broken part in front of them and wants to know if they have any more in stock or if they can just buy a new one. For everyone else, it could be the three-dimensional version of what adding a camera to the smartphone opened up.
But that's assuming the world goes 3D. If not, Physna still has its enterprise business, and myriad design and DIY enthusiasts who could find value in Thangs. But if it can figure out how to translate the value in knowing the intricate details of three dimensions in ways that the search engine defined the modern internet, there might be some depth there.
Correction: This post was updated to correct where DeMeyere is from.
Mike Murphy ( @mcwm) is the director of special projects at Protocol, focusing on the industries being rapidly upended by technology and the companies disrupting incumbents. Previously, Mike was the technology editor at Quartz, where he frequently wrote on robotics, artificial intelligence, and consumer electronics.