OCT 2025
"Oh hell there he go again messin with other people's IP!"
So many WAN-Animate experiments are rolling through my feed but so few are focused on full-body transformation. Everyone is showing waist-up, hands and face. Why? That has already been out. WAN is better than Act-2 at waist up, but imo the unique part of WAN-Animate is that it can do full body. With unique hair motion! Here is my experiment. One piece of stock footage, body-transferred into 10 different people in 10 different outfits, all wearing the same Nike Vaporfly 4's. The shot was broken into shorter clips, start frames were generated using Nanobanana, run through WAN-animate, and then cut back together in AE with some light comp work (mostly to fix differences in the regenerated bg). I think this would be quite hard to do using any other means: maybe shoot 10 people running on a track and get them all to maintain the exact same body position/foot placement etc through the entire sequence? Or how about reskinning the same CG rig with 10 characters with different hair, skin, outfits etc? Sounds hard. At the end I pulled stills of the Vaporfly 4's and did some Kling start frame / end frame gens to get a kinetic product shot at the end. The logo break / reform at the end is also Kling with AE time displacement on top.
***The Mago beta looks amazing for this stuff, I probably should redo this whole project tomorrow 😭