Enome from Futoku no Guild

V2 Update

I retrained this LORA with the settings from my Hanabata LORA (mostly for consistency), as well as changed up the dataset to separate the monster scenes and repeat them more for better training. This version has the same level of quality, but the monsters look and generate a lot more accurately, specifically the rock cows.

If you would like to generate the monsters, add these: "cow, animal, snout" and "snake, monster"

Trained mostly from screenshots from the anime. Unlike all my other LORAs, this one is trained on NoobAI-EPS (not VPred). It works great and gets the anime's artstyle down extremely well. However, compared to some of my tests with an Illustrious 0.1 version I trained before, it does seem to lose accuracy sometimes like with hair color when left unprompted. I've trained this LORA a few times now and it's an issue present in all versions, so I don't know how to fix it. I think it's a good tradeoff though.

It is also trained on NSFW scenes from the anime, so you are sorta able to recreate them with the right tags and some patience. This means there are some naughty tags in the dataset, you have been warned.

Trigger: enome, large breasts, capelet, collared shirt, frilled shirt, necktie, underbust, pleated skirt, brown boots

Use pink bra and pink panties for her underwear, their details are baked in.

Hair and eye color should be baked in, but use "blonde hair" and "blue eyes" if needed. You might need to include these regardless if you dont use her usual outfit, along with "single side bun" for her hairstyle. You can also add "black outline" to get a more prominent outline like in the anime.

I currently have work-in-progress datasets for other Futoku no Guild characters, Enome is just the first. However, I don't know if I will be able to upload them to Civit due to their designs. It's not like they haven't been posted on here before, but I'd rather not deal with Civit bringing the hammer down on me, so we'll see. Let me know which character you'd want to see next! I'm spending a lot more time than usual gathering images for the datasets so don't expect them soon.