I wonder if this sort of thing will become trivial w/ generative models like gans/llms and such. To paraphrase Arnold...If you can parameterize (facial expres

I wonder if this sort of thing will become trivial w/ generative models like gans/llms and such. To paraphrase Arnold...If you can parameterize (facial expressions - already not too hard w/ motion capture software), you can generate it?