>new wan ditto model works nicely for style transfer
>Still super inconsistent across individual videos
Wish there were a way to sort of lock in character consistency. But maybe I can just turn down the model noise.