Machine Vision Surreality

How does Google Earth form these “ground level” images? My guess is it’s via compositing aerial photographs taken at different angles and attempting to basically detect object boundaries then apply the images to them as textures.

It works okay for some things like buildings standing out in the open, not obscured by trees.

For other things, umm…. Well, I particularly like the floaty treeblobs. They give you all the shade of a normal tree, but without any of that pesky risk of bumping into the trunk.

I keep feeling like they should integrate some really stupid auto racing game into this view with amusingly retro controls (like Pole Position)…!

A street corner in Disney's Hollywood Studios
A street corner in Disney’s Hollywood Studios

For whatever reason, the tower's right on, but the rest...
For whatever reason, the tower’s right on, but the rest…

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.