The arms are in three different positions with the processing composing three different frames based on which frame/composite it thought that “person” looked best.