>>33943
I'm paying for it; it's cheap and super good.
It's actually a 2-part job. First I generate the "weight gain" with Reference-to-video, then I take the last frame and use Image-to-video.
This was the Reference-to-video prompt:
Muscular Korean Girl@image4 smiling in pleasure as she looks at a Girl, wearing glasses@image3, laying on a couch with her legs spread apart@image5, with a horror expression as she rapidly drinks blood from a hose in her mouth while her belly continuosly distends outward@image6 and overinflates bigger, fatter and rounder@image2.
And this was the Image-to-video prompt:
Muscular Korean Girls smiles as Fat girl's body explodes in every direction, splattering blood and intestines everywhere
The key is to keep the Image-to-video simple so it jumps straight into the bursting, without additional shenanigans.