Rude Frog 3D style is on ANIMA / NoobAI-XL V-pred / Epsilon / Illustrious 0.1 now!
I'll add more previews. Sorry I am exhausted.
ADDED TAG LIST FOR ALL CHARACTERS IN LORA! +For V3 update Anima too. CHECK "TRAINING DATA" FILE (zip, .txt inside) UNDER LORA MODEL FILE!
V3: +300 pics (Dec 2025-August 2026). New taglist. Added natural language for some difficult pics in dataset for easier using NL.
Recommended model for this LoRA:
For ANIMA - Anima 1.0 base.
For V-pred version: - NoobAI-XL V-pred 1.0. (Don't use cyberfix version for V-pred, If you can't use rescaling CFG (in forge / ComfyUI) it is better to use my lora for Epsilon.).
For Epsilon-pred version - NoobAI-XL Epsilon 1.1 , Cyberfix versions.
For Illustrious 0.1 version use Illustrious 0.1.
I. Tags:
Main tags:
@rudefrog3d - for Anima and NoobAI V-pred V3
rudefrog3d - for previous versions
Optional:
shiny skin, large breasts, huge breasts, nipples, huge nipples, stomach, wet, sweat, thick lips, red lips, thick eyebrows
Some popular clothes from RF arts:
black choker, white choker, o-ring choker, shrug \(clothing\), bridal gauntlets, thigh boots, highleg panties, black panties, crop top
Recommended negative (or not):
FOR ANIMA and NoobAIV-pred v3: white granular speckles on skin (againt "salt" effect caused by wet and sweat tags), stray hair on forehead (against wet hair effect on forehead).
FOR previous versions: salt (if you don't like rude frog's wet effect), stray hair (side effect of sweat - wet hair on forehead).
SETTINGS FOR ANIMA:
My main guide for Anima gen: https://civarchive.com/articles/30357/about-anima-settings
NoobAI-XL 3D quality settings:
Recommended generation settings:
Euler A SGM Uniform (Also you can try Uniform, Normal, beta schedulers but I prefer SGM), 30 steps.
CFG 4-5! (NOT MORE FOR BOTH VERSIONS).
For V-pred only: rescale CFG 0.5 (Or other rescaling features. Use my guide abot V-pred if your pictures are oversaturated on V-pred version).
CFG++ samplers are not recommended this time.
Please don't forget to rate my LoRA. This motivates me to do more!
Description
NAI Epsilon / Illustrious.
Tested on NoobAI-XL Epsilon 1.1 and Cyberfix of NAI 1.1.
FAQ
Comments (7)
O_O
PERFECT!
Very good
its actually crazy how good ai has gotten at translating characters 1:1 into 3d, noobai is goated beyond belief bro
Yeah it's cool because NoobAI (especially V-pred) listen all your prompt if you use tags from danbooru in comparison with Pony or IL.
But making this lora was difficult. I changed my settings about 8 times before good results. Finally I swapped adam8bit + cosine with restarts on Came + Rex for more agressive training.
Also I changed pics in dataset a few times for more accurate style from last years (old arts were more like pure assets without his style). I spent almost 2 weeks from 0 to posting on civitai. It was a very tiring experience but it was worth it.