Removes (positive side) or Add (negative side) censorship from genitals.

Works nearly always on mosaic censoring.

Might need to add a bit more bar censorship examples to the dataset so it performs better on those.

Placeholders replaced during training in captions:

[slider]
placeholder = "<CENSOR_AMOUNT>"
positive = "exposed, uncensored"
negative = "mosaic censoring, bar censor, censored"
neutral = ""