At the time of publication: Series A | Total funding raised: ~20mn USD
****
Alphabet/Google has proved via Search and Maps that ‘organizing the world’s information’ is immensely valuable to the world. Google Maps has seen tremendous improvement in the past decade and has been a feat of engineering and is updated almost real time. Maps enable businesses like Uber, Lyft, Doordash and increase productivity. Google Maps has around a billion monthly users.
Satellite imagery is getting better and better and we are getting very highly realistic maps. Below is an image captured by Planet Labs’ Skysat satellite.
Maps are generally rendered in 2D - for example, you figure out the depth of the buildings accurately.
Blackshark.ai is a startup that has created the first (almost) accurate 3D map of the entire planet and is poised to cater to multiple applications.
The 3D map is like a ‘digital twin’ of the planet earth and can be used as a ‘synthetic environment’ for simulations and visualization among other things. Blackshark is the first to use satellite imagery to generate 3D maps at scale.
For example, check the ‘before and after’ pictures of landscapes in different cities in the 3D Maps of Blackshark.
Blackshark explains that over 99% of the buildings are not in 3D.
The startup’s origin is interesting. Blackshark.ai was a key development partner for the critically acclaimed Microsoft Flight Simulator 2020 video game. (Blackshark itself was a division of another gaming company that was spun off to focus on creating automated scalable 3D mapping environments). Blackshark developed the solution using “Bing maps, Azure and AI and smart guesswork” to create a 3D version of the world.
As the CEO Michael Putz explains in an interaction with Techcrunch:
“It’s easy to see why reconstructing a 3D building from a 2D map would be hard. Even figuring out a building’s exact outline isn’t easy. What we do basically in Flight Simulator is we look at areas, 2D areas and then finding out footprints of buildings, which is actually a computer vision task,” said Putz. “But if a building is obstructed by a shadow of a tree, we actually need machine learning because then it’s not clear anymore what is part of the building and what is not because of the overlap of the shadow — but then machine learning completes the remaining part of the building. That’s a super simple example.”
While Blackshark was able to rely on some other data, too, including photos, sensor data and existing map data, it has to make a determination about the height of the building and some of its characteristics based on very little information.
The obvious next problem is figuring out the height of a building. If there is existing GIS data, then that problem is easy to solve, but for most areas of the world, that data simply doesn’t exist or isn’t readily available. For those areas, the team takes the 2D image and looks for hints in the image, like shadows. To determine the height of a building based on a shadow, you need the time of day, though, and the Bing Maps images aren’t actually timestamped. For other use cases the company is working on, Blackshark has that and that makes things a lot easier. And that’s where machine learning comes in again.
“Machine learning takes a slightly different road,” noted Putz. “It also looks at the shadow, we think — because it’s a black box, we don’t really know what it’s doing. But also, if you look at a flat rooftop, like a skyscraper versus a shopping mall. Both have mostly flat rooftops, but the rooftop furniture is different on a skyscraper than on a shopping mall. This helps the AI to learn when you label it the right way.”
“And then, if the system knows that the average height of a shopping mall in a given area is usually three floors, it can work with that.”
The result is a remarkable and realistic rendering of the real world.
Below is how a 2D picture of a building is rendered into 3D with AI and some guesswork.
Blackshark now has a constantly updated 3D map of the entire planet with 1.5 Billion buildings and 30 million square kilometers of vegetation.
And the CEO explains these maps are updated constantly:
“Our platform can process petabytes of satellite data in near real-time and accurately extract semantic information such as building footprints/heights, land use, bodies of water or infrastructure assets such as streets or rail tracks.”
That is an enormous head start. Now it wants to focus on non-gaming applications and I think there is huge market potential and a number of applications for Blackshark’s platform.
Brian McLendon (co-founder of Google Earth) who sits on the board of Blackshark says,
“These 3D environments are not just beautiful to look at, but given the availability of semantic information they can be queried, searched, updated rapidly and changed for simulation purposes in an instant. This is immensely powerful and really sets the product apart from anything else I have seen”
Some potential applications:
Simulation environment for Autonomous Vehicles
Simulation environment for (Smart) City planning
Simulation and Visualization for Construction and Property Development
Collaborative 3D Environment where ‘crowds’ can collaborate, build and visualize new construction and infrastructure. Imagine many collaborating online and creating properties and buildings together.
Simulation environment which Governments can use to plan infrastructure development
Simulation and Visualization environment for Drones
Alternative data to track indicators for economic activity, growth and change in specific assets - can be useful for both Governments and Hedge Funds
Environment monitoring like deforestation and simulating impact of disasters etc.
Property value predictions based on location, density, nearby assets etc.
The startup, based in Austria, is already focusing on a few of the above and have announced partnerships with Epic Games (Unreal Engine) and Nvidia.
Blackshark can continue to improve its mapping accuracy and increase the assets it can label accurately which can enhance its attractiveness for the applications mentioned above. The mapping accuracy and automated labeling capabilities will also help it be the go-to 3D mapping API for ‘metaverse’ applications.
Blackshark down the road may also want to acquire a few adjacent capabilities inorganically to double down on certain verticals like construction.
Blackshark.ai is a high potential startup. It can be the go-to API or provider for a number of applications in a large, growing market. It has a good head start and has proven its capabilities in the Microsoft Flight Simulator game. If a Metaverse future becomes a reality, then Blackshark stands to benefit immensely.