It's like a magic trick, once you know how it's done, it isn't so amazing afterwards.

Take a video of people holding pictureframes over their faces and then moving them.
Stop on a frame where they have frames in front of them and take a snapshot.

Using Paintshop or whatever, crop all the faces.

On a track above the video, use track motion or pan/crop to position the faces inside the frame and follow the movement of the frames.

Simple.