Christopher Mitchell, a PhD candidate in NYU’s Computer Science program, is building a 1:1 scale model of Manhattan in Minecraft, with faithful, handmade reproductions of each of the island’s skyscrapers. He’s relying on data from diverse sources, including Google Earth, and the model to date is 277m^2, with 71Bm^3 of volumetric detail, running on a 200 core cluster with 200GB of RAM. It’s part of a larger project (!), called Sparseworld, through which Mitchell is combining data from diverse geographical and architectural systems to faithfully model the physical world.
Where to get replicable data for every building in New York City? “Completion is reliant on getting models for every building on every street, and to my knowledge, only Google has that much information,” Mitchell said of Google’s Earth and Maps combination of products. Here.com, which used to be Nokia Ovi Maps, and Microsoft’s Bing are other possible sources.“Dad’s favorite pastime shouldn’t treat girls like second-class citizens.”
The problem is that these companies’ data is under license and encrypted, which Mitchell doesn’t want to mess with under the table. “I’ve considered reverse-engineering the encrypted format that Google Earth uses to fetch building models from the server and just keep the building models to myself,” Mitchell said, but he still worries about violating the Google Earth license. He’s had trouble getting in touch with the right people at Google to discuss accessing the data and using it for the academic pursuit of a Minecraft Manhattan. He has also reached out to the Here.com and Bing team
Manhattancraft: The quest to make a full-size city of Minecraft blocks [Casey Johnston/Ars Technica]