(日本語説明は後半にあります)
v7.0 Update : Made detailed adjustments and modifications at each block(IN, OUT, MID) and text encoder to improve background quality, composition, and overall finish.
About this model
"IllusionBreed" is an Illustrious-based SDXL checkpoint, pursuing "kawaii", cuteness.
There are 2 variations for v7.0 :
• Basic 2D painting style
• "Slender" : 2D painting and more slender body style
• To download each types, switch version tabs ↑ above.
I feel that model merging is similar to the breeding of flowers (to improve their appearance), so named "Breed".
These models have special features :
• Include many originally trained data to adjust painting style (saturation, tone), eye shape, and quality (finely detailed pupils, finely detailed outlines, fine painting, reducing grain noise etc).
• Detailed backgrounds.
• Powerful prompting based by Illustrious (and NoobXL).
More various other LoRA's and Checkpoints are released on my CivitAI models list or HuggingFace repository. Please check it out.
Recommended output settings
• CFG Scale : 4-7 (recommended : 5)
• Sampler : Eular A
• Sampling steps : 20 - 60 (recommended : 40)
• Resolution : mainly tested on 768x768, 768x1152, 832x1216, 800x1200 and 1024x768
• Using A-Detailer (for face etc.) is highly recommended. (inpaint denoising strength : 0.3 for face, 0.5 for breasts, etc.)
• HiRes Fix : use SwinIR_4x, R-ESRGAN 4x+ or 4x-UltraSharp with denoise strength 0.3-0.7.
• Quality prompt / Negative prompt : See below.
• VAE : embedded.
Quality prompts sample
This and base checkpoints are derived from Illustrious, so use quality prompts for Illustrious models.
masterpiece, best quality, amazing quality, very aesthetic, absurdres, perfect anatomy, perfect symmetrical face, cute face, perfect hands, detailed eyes, detailed hair, slender, (PUT YOUR PROMPTS HERE)
To improve background quality, detailed background, photo background, blurry background prompt is recommended.
Negative prompts sample
lowres, bad