The new web-based Google Earth is based on a completely new graphics engine. So, we wondered whether or not Google had implemented any compression algorithms or other mechanisms to reduce the amount of data that Google Earth needs to download. An example of such a mechanism is Google’s own Draco 3D compression library which they open sourced in January.
To run our tests we used a technique we previously used to try and estimate the size of the Google Earth database. We created Google Earth Tours that follow a zigzag pattern across an area, going slow enough so that all the imagery that Google Earth needs is downloaded. We cleared the caches then ran the tour in both Google Earth Classic and in New Earth. We chose some areas over regions with 3D imagery and one where there is only satellite imagery.
Zigzag 1: Near Old Bridge Township, New Jersey, USA.
Google Earth Classic: 60 MB
Google Earth Web: 60 MB
Zigzag 2: Maceió, Brazil
Google Earth Classic: 66 MB
Google Earth Web: 63 MB
Zigzag 3: Longyangxia Dam Solar Park, China
Google Earth Classic: 6.6 MB
Google Earth Web: 8 MB
At this point it looked like everything was going well and there was very little difference between the two platforms. But then we thought of trying an area with lots of tall buildings. So we tried Hong Kong and got very different figures.
Zigzag 4: Hong Kong China
Google Earth Classic: 170 MB
Google Earth Web: 106 MB
After some experimenting, we found that the size of the view port in Google Earth matters a lot. If we changed the size of the side bar in Google Earth Classic to roughly match the placemarks sidebar in Google Earth Web it downloaded a mere 75 MB, less than half the amount of data than it did with a slightly narrower side bar and thus larger view port. So it would appear our overall methodology is flawed as it is difficult to ensure that the view ports are identical between the two products.
Despite the problems, we think that we can conclude that there are no major new compression algorithms in the new Google Earth at this time and you can expect similar internet usage when viewing a given area.
While creating the tours, we came across a maze someone had created in a field in New Jersey which was captured in the 3D imagery:
Here are the tours we used for testing.