Okay, the name of the article really bears very little on what the heck the video is about, but...

Basically, some very smart guys figured out a very slick way to decide what parts of an image can be dropped or expanded in order to resize an image without noticeable distortion, and still keep all the components in the image more or less intact.

What this means in a more specific fashion is that if you have a picture of two people, and you resize it to half the original width, instead of getting two very anorexic-looking people in the image, they will simply look like they're standing closer together.

That is just slick.