Tried to do a Wan Lora with my adjusted dataset of my previous models, added just a few videos into it. Noticed that the movement of the videos is strongly embedded in the process, so I will focus more on the video input for my next version. Hope you like it !
