Video models work by taking randomly generated visual static and progressively “de-noising” it by adding order to the chaos ...