My opinion: One camera on a -copter drone that is programmed with specific GPS waypoints for specific times. I think they very likely filmed it at half speed (or less) and speeded it up in post to get everything smooth.

Pretty darn impressive.

This sequence -- the video was done in one continuous shoot taking approximately 50 to 60 tries to get it right -- opens with abandoned warehouse in the city of Chiba, about 45 minutes outside of Tokyo. The video was then filmed in double-time to evoke Hollywood choreographer Busby Berkeley's signature manic style.

To coordinate everyone moving in synchrony to the parking lot for the final setup, Harano set up enormous speakers and played "I Won't Let You Down" at half-speed, which "allowed for more precise movements even with the complicated choreography," he says.

To capture all those Japanese schoolgirls flashing colored umbrellas in the shapes of lyrics and OK Go members' faces in perfect synchronicity, Harano used a specialized drone camera controlled both by GPS and by hand from an altitude of nearly half a mile.

http://www.billboard.com/articles/news/6296760/ok-go-i-wont-let-you-down-video-honda

Thanks all for help. I think she will give up challenging me and I won't have to reply to her " what do you read in there? ! -;) any more .

It looks like a series of stills. Possibly stop frame animation technique used in a unique way.