The model I think is fairly self explanatory.

I suppose that confirms my proposition of artists leaving 'consistent' watermarks. If the only solution to avoid the watermark is to reduce the strength, then essentially using ai for the full style is at a consequence, and a lower strength would just be a pale imitation of an already pale imitation of the artists skill and talent.

Works well at a strength of 1.

Previews made with novaFurryXL_illustriousV30.