The unique challenge of the AR cloud is creating virtual worlds that can be streamed to mobile.
The project is a partnership between the City of Helsinki and Umbra, a platform that optimizes large visual datasets for streaming over mobile, a challenge that's going to be increasingly vital with the anticipated rise of Augmented Reality applications.
The Helsinki model was created from aerial photographs that date from 2015. The resulting texture-mapped 3D mesh accounts for more than 30 square miles of the city.
Why go to the trouble of making a virtual map of a city when a satellite can easily snap photos?
It's a step toward what's called the AR Cloud, which is most easily understood as a digital copy of the real world.
Right now, GPS does a decent job of permitting mobile applications to pull up relevant information about the world (restaurant reviews, for example) based on general location. In the future, AR apps will access a digital overlay of the real world whenever you train a camera on a given landmark.
Pointing your device at a statue, for instance, may prompt an AR app to pull up relevant visual or textual information. Through AR, the visual world will become a giant repository of information, just as the internet is today.
But whoever controls the AR Cloud -- that virtual overlay -- will have an outsized influence over the information AR apps access. There's a push right now to keep the AR Cloud open, which may be the only way to ensure it doesn't fall under the influence of profit-motivated companies.
Helsinki has decided to grab the bull by the horns and create its own virtual overlay, which anyone can access and, conceivably, develop on top of.
But the tech-savvy city quickly ran into a problem: Its virtual mesh was huge, accounting for about 700 GB in total.
The city turned to San Francisco-based Umbra to make the data deliverable. The company optimized the 3D rendering using its proprietary automated cloud platform.
"Any 3D data set, be it large point clouds or complex polygon models, can be exported directly to Umbra Composit, where our system automatically divides the input into smaller computation tasks," according to the company. "These tasks are then parallel-processed in the cloud, distributing the work over many cloud instances, allowing for rapid processing of even the largest data sets."
Umbra says its delivery solution has the potential to render and stream at higher resolution than GoogleEarth, including to sub-millimeter accuracy.
"In addition, we are democratizing city-scale 3D scanning," says Shawn Adamek, Chief Strategy Officer at Umbra, "giving cities and developers the opportunity to take ownership of their own datasets, rather than using and licensing existing mapping data."