OCT 2025
I am back with the early Halloween edit 🎃.
Same structure, same underlying piece of stock footage, now given a spooky transformation with a new tool: Mago
Mago allows you to get a little deeper under the hood without going fully into ComfyUI. It offers 3 different models as well as controlnet type selection: depth, normal map, pose (used most for this video), soft edge, pose+depth.
Using the pose controlnet allowed me to push more stylized characters like Dracula (with cape), Werewolf (abnormal body shape) and Medusa (snake head) through. When I tried these types of characters with WAN on Higgsfield it wanted to force the stylized characters into a more human form, which looked bad.
It is a lot of fun to be able to consistently drive all these different character types while even getting some of that dynamic motion like the Dracula cape flowing etc.
(Refer to my last post for a more comprehensive breakdown.)