physical violence

weapon violence

wide hips

revealing clothes

thick thighs

downblouse

convenient censoring

huge breasts

pg-13

corpses

suggestive

oral invitation

pg13

sexy

sexual situations

male nudity

disturbing

male swimwear or underwear

female swimwear or underwear

partial nudity

undressed

female nudity

breasts out

exposed female nipple

breast out

lingerie

male underwear

hair over breasts

female swimwear

gigantic breasts

no panties

graphic violence or gore

covered nipples

huge butt

strapless leotard

sitting on face

emaciated bodies

one breast out

nsfw

female underwear

nude

graphic male nudity

adult toys

illustrated explicit nudity

nudity

graphic female nudity

hentai

futanari

porn

sexual intent

genitals

peeing

vore

oral

sexual activity

anal

blowjob

dildo riding

incest

hanging

hate symbols

nazi party

white supremacy

diapers

scat

self injury

hate speech

urine

extremist

child on child

latex clothing

swimwear

bukkake

fellatio

cumshot

implied fellatio

eat_cum

cumdrip

cum in pussy

cum on face

after fellatio

cum on hair

cum on body

cum on tongue

cum on hands

cum in mouth

triple fellatio

autofellatio

fucked silly

cum on pussy

Trained using OneTrainer on an RTX 4070 SUPER 12gb, using 200+ illustrations by sushi yuusha toro from black souls 2 as well as the bonus contentUse the sushi yuusha toro tag for the style. (also "sketch" might help, as well as removing some negative/quality tags)All the images were auto tagged by <a target="_blank" rel="ugc" href="https://huggingface.co/p1atdev/wd-swinv2-tagger-v3-hf">wd-swinv2-tagger-v3-hf</a>

trained using diffusion-pipe for 2610 steps, 5e-5 LR

chromav50-v1

experimental flux version, trained on-site

flux-1.0

trained for illustrious 0.1, no idea if anyone even uses it anymore

illustrious-0.1

a wai nsfw illustrious version, as requested

wai-nsfw-illustrious

using a logit normal timestep distribution, should stick closer to the style

vpred-1.0-v4

apologies for another update, i'm testing timestep shift with a value of 2consider giving feedback on which version works best for you

vpred-1.0-v3

once again using prodigy, now with DoRa enabled and debiased loss estimation it comes pretty close to the style, still looks a lot aesthetically better than the original images, without overfitting on the dataset

vpred-1.0-v2

a test version using the prodigy optimizer and text augmentation, constant loss and DoRa disabledbatch size 4, 2 image variations (1 repeat) 4 text variations, 10 epochs, might be a bit underbaked

vpred-1.0-prodigy

epspred-1.1

vpred-1.0

[NoobAI | Illustrious | Chroma] black souls / sushi yuusha toro style

00484-2085447829.png