Here’s an experiment with Google’s latest video generation model (Veo 3):
I took a top-half picture of myself, outpainted it to full-body via Nano Banana, then used that outpainted image and a prompt to seed an 8-second video.
The result is funny and mesmerizing, at least for me. Notice how my shirt folds when I move. The only two things that reveal this is fake: my unnaturally happy smile and those smooth moves.
I don’t dance like that.