Blackshark.ai raised $20 million to develop its replica-Earth Tech
Blackshark.ai, the Austrian startup behind a digital globe flying in Microsoft Flight Simulator, raised $20 million in Round A to develop and expand copies of Earth technology.
Blackshark.ai, the Austrian startup behind a digital globe flying in Microsoft Flight Simulator, raised $20 million in Round A to develop and expand copies of Earth technology. The potential applications for planetary digital twins are many and varied, and the company has an edge even in mapping giants like Google. Last year the world saw a fully walkable and surprisingly (though not 100%) accurate globe in Flight Simulator. We called it a "technical miracle" and later detailed how and by whom it was made.
Blackshark.ai was spun off from game studio Bongfish with the intention of leveraging world-building technologies beyond gaming environments, as founder and CEO Michael Putz told me. At the heart of their technology is turning popular 2D images into accurate 3D representations using machine learning, a little smart guessing, and a lot of computing power. The details are here, but in fact the Blackshark.ai system has a good understanding of how other buildings look from above, even with optimal lighting and imperfect images.
The machine learning system they built can extrapolate from imperfect contours by taking into account area (residential or commercial), roof type (sloped or flat), and other factors such as air conditioning. Building using all of this. Of course, the most difficult part is not how to do it once, but how to do it regularly billions of times to create a modern 3D representation of every building on the planet. As Putz explained, "Even if you could afford to buy all the processing power, it would be difficult to build a backend that would provide it! It was a real problem we had to face."
A solution often needed for AI-powered services was optimization. Putz said that the process of calculating a 3D model for each building on Earth, which initially took about a month to calculate, can now be completed in about three days, with an acceleration of about 300 times.
“The ability to regularly update based on new satellite imagery is very important to our business proposal,” Putz explains. Much 3D map data, such as those found in Google and Apple Maps, is based on photogrammetry, aerial photography combining multiple aerial photographs, and comparing parallax data to determine size and depth.
This gives excellent data when taking pictures. If you want a 3D map of the Chicago area last week instead of two years ago, and you want this level of freshness to be known to the world as much as possible, satellite imagery is your only option today. But this also requires the 2Dto3D method mentioned earlier. Putz noted that while the Blackshark.ai 3D Maps and Google and Apple Maps are superficially similar, they are not really competitors. All of them create a realistic "canvas", but they are very different in design.
“Google Maps is the canvas for local businesses,” he said. What matters to both the company and the user is location, reviews, directions, and more. "For example, for flooding, climate change use cases, we can provide 3D data for Seattle, and others specializing in water physics and fluid modeling can use the real world as a canvas. Our goal is to have a searchable planet surface. will become."
What is the total flat roof area for this San Diego area? Which regional airport has an open area of 4,000 square meters? How do wildfire risk areas match the updated wind model? It's not hard to figure out how this could be useful. "This is one of those ideas that the more you think about it, the more use cases you get," Putz said.
“Obviously there are government applications, disaster relief, smart cities, autonomous industries (driving and flying). All of these industries require a synthetic environment. "Hey, we want to do this" was necessary as well. And these 2D3D models are the only way to solve this huge problem."
The $20 million round was led by M12 (Microsoft Venture Capital Fund) and Point72 Ventures. Putz was delighted to be aboard several familiar advisors, including Google Earth co-founder Brian McClendon, former Airbus CEO Dirk Hawk, and former Y Combinator COO and current Applied Intuition CEO Casar Younis. (These people are giving advice rather than joining the board, as misplaced earlier in this paragraph.)
Scaling is about bringing a product to market rather than building it. Of course, more engineers and researchers will be hired, but the company has to move quickly from "smart startups" to "global suppliers of 3D synthetic earths." Lunch. Thus, a sales and support team will be created alongside "the rest of the mega companies," Putz said.
In addition to the more obvious use cases he lists, there are possibilities for metaverse applications. However, it's not very hot in this case, but if AR/VR/etc applications are interesting to you, from games to tour guides, we wanted to base our virtual experiences on our newly created Earth version. , can do.
Not only that, but worlds outside of us can be created in the same way, so if you want to create new contours by mixing planet contours (who can blame you?), lord. Can't you hear me? Expect to see something similar in Blackshark.ai or a new generation of increasingly detailed simulations of complex markets and processes on the Earth's surface as new funds are spent.