YobboIllustrious — Furry + Anime-Ready SDXL Merge (Illustrious-Style)
🔹 Description:
YobboIllustrious is the foundational release in a premium NSFW/SFW furry/anime merge series built for Illustrious-style prompting.
Merge Info
Models merged: bananaSplitz_vee4pointOh + novaFurryXL_illustriousV9a + Tigermilk Illustrious
bananaSplitz_vee4pointOh: punchy color, expressive anatomy, smooth shadingnovaFurryXL_illustriousV9a: detailed fur, semi-realistic anatomy, cinematic lightingTigermilk Illustrious: refined texture depth, enhanced lighting balance, stronger semi-realistic detail
VAE: ✅ Baked-in SDXL VAE (no external VAE needed)
🔸Expect smooth anatomy, expressive depth, species-flexible—great for pinups or cinematic scenes. Illustrious LoRA-friendly and anime-ready (prompt-switchable), spanning cel/cartoony to semi-realistic; built for ComfyUI, A1111, and wildcard workflows.
🔥 YobboIllustrious excels at
High-detail furry & anime characters — clean forms, stable proportions, crisp fur/skin/line detail
Wide species & cast coverage — anthro (canines, felines, deer, hybrids, fantasy) plus human/anime characters
Complex scenic compositions — beaches, bedrooms, clubs, dungeons; steady lighting, depth, and occlusion
Style agility — prompt-switchable semi-realistic furry ⇄ anime/cel (Illustrious tag chains, minimal prompt changes)
🧬 LOrA Info:
Baked-in LoRA (with permission): Furry Enhancer
Creator: freek22
LoRA: Furry Enhancer
Big thanks to freek22 for permission to use this LoRA in my merge!
LoRA compatibility: Illustrious-trained LoRAs (poses, body shapes, outfits, expressions) and anime/lineart/cel LoRAs; use modest weights (~0.6–0.9).
🌱 Future Versions:
Focused on quality improvements without major reworks. Expect small, seed-friendly updates that enhance fidelity and usability while keeping existing prompts effective.
Render clarity: sharper detail, cleaner edges, reduced haloing/banding
Fur/skin micro-texture: better specular balance; less waxiness in close-ups
Lighting behavior: steadier rim/volumetric response; fewer blown highlights
Anime path polish: crisper lineart, smoother cel shading, fewer jaggies
Prompting tolerance: reliable results from shorter prompts; refreshed negatives/presets
Multi-character stability: improved occlusion/spacing at mid–high resolutions
Still includes a free car wash. 🚗
🔹 Recommended Settings:
Sampler: DPM++ 2M Karras or Euler a (DDPM + AYS schedulers also work)
Steps: 25–40 (close-ups: 32–40)
CFG: 5–7
Clip skip: 2 (-2 for Comfy workflows)
Base sizes: 1024×1536 (portrait) • 1152×896 (landscape); upscale after if needed
VAE: Not needed – SDXL VAE is already baked in
Description
Illustrious model blend.
FAQ
Comments (5)
I don't know what it is about furry models, but they keep surprising me with how good they create non-furry images (not a furry).
Incredibly good prompt adherence. Thankfully it seems to not force you to create furry stuff, and the non-furry images are beautiful
Thank you for the image posts! I know I don't market it as anything other than furry, but I'm glad you're able to use it to fit your needs. Perhaps in the next version, I'll make a note to say that it does more than furry very well.
@Agri_Yob You should definitely mention it - I've reached the same conclusions as @clueless_engineer across several models (FurryVoidnova, reedFurry and IndigoVoid VPRED, for instance). I'm curious if the structured-yet-unexpectedness of furries vs. anime or realistic characters helps with the training process, improving non-furry generation?
Regardless - great work.
@sendmefreestuffa8246 I'm not sure about the training process, as I'm new to this and simply merging what's aesthetically pleasing to me and hoping it works in some way. I'll have to look into the models you mentioned in your comment, as I'm partial to the more "popular" models (not saying those aren't, I've just never heard of them). I'm always looking for and trying models that might look good in my next version, but the pool is so deep, it's a slog. And thank you for the encouragement, it's greatly appreciated and needed!
I've been very happily surprised by the results I've been able to obtain on first passes of any given prompt. It's certainly given me plenty of alternate ideas to explore.