Both forms keep one native image semantic, required alternative text, and a reserved 16:9 box before the files finish loading.