UPD Feb 13, 2025: Sadly, I lost my bid. It can happen to anyone, I guess. I'll try my luck again next Friday.
Hello again! This is my first-ever checkpoint merge. My goal was to build a versatile "all-in-one" model that handles as many styles and tools as possible. You might still encounter some typical AI artifacts like extra fingers or messy details and backgrounds in certain generations, but that's just how it works for now. Feel free to experiment and see what this mix can do!
Key Features:
Works very well in a wide range of artistic styles, including anime, furry, semi-realistic and realistic.
Does everything other checkpoints based on NoobAI (Epsilon-pred) and Illustrious XL do, so your usual prompts will work just fine.
Fully compatible with LoRAs for these models, and also works with some LoRAs for Pony XL!
Create your own unique styles by adding artists to the prompt: (by [artist name a]:1.0), (by [artist name b]:0.9), etc.
Models used for the merge (v1.0): Indigo Furry Mix XL, WAI-illustrious-SDXL, WAI-Mature-illustrious.
Recommended Settings:
Sampler: Euler a, Euler
Scheduler: Automatic, Normal, Simple
Steps: 15-40
CFG Scale: 4-7
Clip Skip: 2
Hires upscaler: R-ESRGAN 4x+ Anime6B
Prompt:
masterpiece, best quality, normal quality, absurdres, highres...
Since this is a mix of anime and furry models, it primarily uses tags from Danbooru and e621.
Negative Prompt:
3d, sketch, (worst quality, low quality:1.2), bad anatomy, bad hands, bad fingers, (signature:1.3), (artist name:1.4), (patreon link), jpeg artifacts, (watermark:1.3), twins, clone, public, lowres, username, logo, (text:1.3),