I tried the deepfake software a few years ago to see how it worked. It's interesting because you need a crapload of source material, basically face shots of the person you want to duplicate at every angle and facial expression possible. They are scripts that you can run that will auto grab and crop the person's face out of movie scenes and interviews off youtube videos. Then in the target video you want to override the face you select the person in it, reference the folder with the thousands of face shots and set the program to go. But there's some catches... the person you swap the face with has to look very similar to begin with, bone structure, eyes, check bones, eyebrows, etc. Because the final product is more of a face merge than swap. The closer the person looks like the image the more accurate it'll look like a face swap. That Luke one looked as good as it did because the cg Luke looked pretty damn spot on. Basically Disney could do the cg de-aging they do now and then do the deep fake as a final layer and it'd be perfect. If a deepfake person tried to put lukes face on qui gon jin in episode 1 it'd most likely look like ass.
The other issue when I tried to make a deepfake was processing time. Because it's an AI script you have to let it do its thing. It takes an ass load of gpu processing power and alot of time. I have a gtx 1060 with 6gb ram at the time and that was pretty good, but even after letting the script run over night for a week 6 hours a night the deep fake I was trying still looked like crap.