I may have mistakenly trained on x0.1 instead of x1.0, which sucks ass, but its seem to work nonetheless. Gotta fix it as soon as possible, over 10 loras trained in a single day 😵
Recommended Model in resources, scroll down the page.
Trigger Words:
(nezunezu artist), large female, hyper anus, hyper pussyAbout Tags: The difference between this and all other nezunezu is that I've put the correct tags instead of forcing the style over, so you can adjust proportions, text and all fancy stuff.
My Settings:
Sampler: Euler A / DIMM / UniPC
Scheduler: Normal / Karras / Beta
Steps 20 - 25 + ERSGAN 4x Upscaler
CFG: 5 - 6
LoRa Weight: 0.4 - 1
About: It has been trained on 100 images, hand-picked and manually edited to proper training, it will excel on both human and non-human generations.
Issues: None that I'm aware of.
TODO:
Retraining on 1.0 or little fix for multiple characters.
Description
FAQ
Comments (7)
If possible, please create an illustrious ver. model LoRA!
It is Illustrious. I don't train on pony cuz its adherence and concept sucks and there are merges of illustrious with pony already, it should work with these as well.
Unfortunately, without illustrious base data, the reproduction quality suffers significantly. You, who know the effort involved in LoRA generation, should understand this. To ensure that effort pays off, I urge you to consider this.
@dtsESÂ What do you mean? I think I trained this model on IL 0.1, I have to upload the version trained on IL 1.0 first
always love seeing more nezumi loras,
been my favorite artist for over 15 years. lol
@se7sinblackbird795Â the lora is mixed with ill build-in nezuzu dataset oppsie. The new version if lora only
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.