I’ve been jaded about 3D buildings in virtual globes for a while, but the recently introduced blanket 3D modelling of Manhattan in Google Earth is just stunning in its scope and detail.
Upon closer scrutiny it is immediately obvious that the models are unique, and each one faithful to the real thing. Sides of buildings are not filled with repeating patterns but with an actual picture, even for repetitive-looking housing projects. An apartment I grew up in on the upper east side had its elaborate terraces modeled accurately to within the meter. The view from my 6th floor walkup towards the World Trade Center on 9/11 can be replicated perfectly:
In fact, you can even figure out which buildings have been built since that picture was taken.
Google Earth credits Sanborn with the buildings layer, but doesn’t divulge just how it was made. It can’t have been an off-the-shelf fully automated system: Just stroll through the canyon than is Wall Street and you’ll notice that many of the buildings there received individual attention. Flying a plane in a gridlike pattern over Manhattan and taking georeferenced photogrammetic photos doesn’t cut it over Midtown and Wall Street, where much of a building would be obscured from view. (This is one of the downsides of the otherwise impressive technique used by Sweden’s C3.) Possibly the layer was made from a combination of different techniques, including taking photos from the ground, perhaps combined with laser scanning — something which Sanborn does.
There are a couple more. Why might that be? Could there possibly be copyright issues with overzealous architects or owners? Was the layer just not finished in time? Or are these buildings currently undergoing renovations, and are thus hidden in scaffolding? I’m not in NYC at the moment, so can’t check the later theory, but am guessing it is the correct one.
Finally, it’s also immediately obvious that this solution scales well. It is very responsive on my Mac in a way that the recently released Ancient Rome layer emphatically is not.