A semirealistic style LoRA, it isn't consistent on all models. See the thumbnails for how to use it. You may have to adjust strength accordingly for what model you use, as always.
V2.0:
Thumbnail images are from "Va'nsyx V2 - Base (https://civarchive.com/models/1393785?modelVersionId=1658591)" if you use that model as well I suggest using these negatives "worst quality, artist sign, artist name, username, signature, logo" and no positives and to put the trigger word at the beginning of the prompt:
V1.0:
The thumbnail images for V1.0 were tested using NoobAI Epsilon V1.1. There is also No Trigger Word.
⚠️ IF YOU'RE GOING TO STEAL/SCRAPE MY MODELS, MAKE IT SO THE DOWNLOAD IS ENABLED. MY MODELS WILL ALWAYS WORK BETTER/AS INTENDED LOCALLY. - Arctenox ⚠️
Description
A revised attempt at training this on illustrious, so I can possibly use it on almost any model. With around 245 images and a cleaned up dataset. :-P
There is a Trigger Word unlike V1.0.
If I make a V3.0 it'll be a more cleaned up dataset. Disclaimer: Don't know if this does semirealism for most models in this version. Also make sure to put the trigger at the start of your prompt otherwise it will still do anime on most models.