What does it take to process 3D imagery?

Google has recently been pushing out 3D imagery of various Brazilian cities in preparation for the Olympics. The interesting part, however, is that the imagery isn’t new. The imagery of Rio de Janeiro, for example was captured around February 2013 in preparation for the FIFA World Cup that took place in Brazil in 2014. At the time, Google did release some 3D imagery of Brazil, sometimes limited to the football stadiums involved in the world cup and small areas around them.

Google Earth does not provide dates for when 3D imagery was captured, so it must be compared with available historical imagery to determine the approximate date. The easiest way to do this is to look for a construction area. Depending on the availability of historical imagery the date can be determined within a month in some cases but only to within a couple of years in others.

Here are the dates that some of the 3D imagery, released in the last few weeks, was actually captured:
Rio de Janeiro – circa February 2013
Belo Horizonte, Brazil – August 2014
Manaus, Brazil – circa August 2015
Durban, South Africa – August 2015
Soweto, South Africa, between September 2015 and March 2016
Arecibo, Puerto Rico – after January 2016
Valencia, Spain – between July 2014 and August 2015
Daytona Beach, Florida – between January 2014 and February 2016

All this points to Google still having to do quite a lot of manual work when it comes to processing and releasing 3D imagery and they have a team that can only process a given amount in a given time. This is also borne out by the relatively regular release of 3D imagery.

So how much work do they need to do? There was the case of Lexington, Kentucky that we looked at, where several buildings were missing in the 3D imagery (it has now been corrected). If someone had to manually check every building in Rio de Janeiro, it could take years!

Let’s take a look around Valencia, Spain and see if we can identify signs of manual intervention in the 3D imagery generation:

If you don’t have it already, download our KML map of 3D areas to find out where new imagery has been released as well as the extent of existing imagery.