It's simply a matter of focus...what's in focus in one picture becomes out of focus as YOUR focus changes...it's very well balanced. If you have a focusable lens handy (camera, binoculars, etc.), look at the picture through it, and then change the focus -- you'll see the "switch" as it occurs.
Funny reading all the posters that needed to back up to see the switch. I actually had to move closer to the screen to see the switch. At normal viewing distance, the faces were just blobs.