In more urban areas where there is a lot of imagery (multiple days per year, multiple years), you have a lot more data for a reconstruction and can get more detailed models.

In areas with little imagery, my guess is that Google generates fairly sparse point clouds (maybe 10-20 points for a tree canopy) and then interpolates the rest - which is why you can have some seriously fugly looking tree canopies. It would all be done through aerial imagery - either USGS, NAIP, or private entities. It's quite remarkable if that's enough to be able to trace individual power lines or, apparently, the space under the canopy between trees (though I suppose that specific aspect could be fake).Īny chance the info sources indicated in the screen corner of Maps are more than just plain 2D photos (Landsat/Copernicus, Data SIO, NOAA, US Navy, GEBCO )?įor the 3d reconstructions, they wouldn't be using any satellite based platforms as the pixel resolutions are too coarse. Doesn't Photosynth need a lot of images to create good 3D? Are there that many satellite/aerial photos for small rural towns in Europe?

3d view maps