Updated this one, turns out I just overfit. Prompting should now be much easier and you should get better images much more easily.
THIS IS ALL FOR V1. CHECK V2 VERSION NOTES FOR UPDATES.
EXTREMELY EXPERIMENTAL PLEASE READ:
rating safe images are easy, just add the witcher_hag,monster tags and you're generally good to go with weight 1:
NSFW GENERATION INSTRUCTIONS:
This LoRa was trained on exactly ZERO NSFW images, yet still has NSFW capability thanks to how robust PONY is for NSFW image generation. However, you have to be somewhat careful with your prompting to get good results every 1 in 3 images roughly. All in all I am still pretty happy with it, this is not a monster that you can really find NSFW images like this for.
Regional Prompter helps alot, this is one of those lora where high weight will transform human characters into monsters even with it, though its still finicky.
NO HIGHER THAN WEIGHT 0.7, experiment with weight 0.6-0.7
Without regional prompter, it is much more difficult and RNG dependent, I have both regional prompter and non regional prompter example images that you can try for inspiration. All around for NSFW image compsition is highly inconsistent.
emphasize the human parts like adding ((1girl)), ((human)) etc, to try to ensure that monsterification does not occur.
if you want the hag to have a penis you need to add the testicles and penis tags and hope the image gives them to the hag and not the human. I don't know how to do M on Monster, not my thing but if it is yours feel free to experiment.
to use my regional prompter examples you should have "use common prompt" and "use common negative prompt" ticked in regional prompter. I used a 1,1 1024x1024 matrix, with generation mode set to attention
Description
FAQ
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.