Creates a hogtie with ropes and gags and everything.

There are enough samples for you to understand how it is working.

Trained it with the same data set as flux (499 images, extended joycaption corrected by hand) with the parameters by @joschek (0.0002, 64/64), here on civitai (went for 15 epochs, but it failed after the 7th, but the 7th looks excellent).

IMHO, it's great! I was surprised. Probably ZIT needs a big dataset (>250) for complex poses. But this one rocks! (And I will never say a bad word for ZIT)