( Not my Model )
Krea2 v1: per commissioned request, now in Krea2 flavor! This one really starts to shine at higher strength (tested up to 2.0), so I guess not training it with that in mind it's a slider? Another usage tip is to specify the image as a caricature - it'll work without it, but with Krea2 this LoRA seems to really shine when that's put in the prompt even though not really used in the captioning for some reason. The image dataset consisted predominantly of SFW content, it does have some spicier images (male) in there to make it NSFW friendly. Although in that regard, don't expect too much since I was not intending this to be an adult content focused LoRA (it's nightmare fuel enough as it is) - you may need a NSFW merge checkpoint or LoRA to help in that regard if that's what you're looking for.