Deepfakes have already existed for several years, and let you paste any face onto any video, in a much more convincing manner than Stable Diffusion, which is very limited in video2video domains.