Crystal Crown
This Illustrious model is designed to have stylistic similarity to my Crystal Opal PonyXL model, but with more anime style strengths. The goal is semi-realistic visually striking, anime-inspired images. The model is versatile, capable of handling a wide range of genres and settings, though it occasionally faces challenges with contextual coherence and stylistic uniformity.
Color and Lighting: It produces vivid, well-balanced color palettes and realistic lighting effects.
Detail and Texture: The model renders intricate details—such as the texture of fur, fabric, hair, and eyes.
Generation
While the model generally maintains coherence, it occasionally falters in distant details, but I have a high success rate in most cases.
Sampling Method: DPM++ 2M, Euler a, DPM++ SDE - works with most common schedulers
CFG Scale: 4-10 (but it can go higher or lower if you increase the step count)
Steps: 6-∞
Very flexible with LoRA compatibility for NoobAI and Illustrious models. It can also use some standard SDXL and PonyXL LoRA to a lesser degree.
Let me know what you think. I welcome feedback.
Description
Version 1
FAQ
Comments (4)
well done sir, this is the best for couple of girls in the same shot with different outfits, Thanks so much!
Thank you for your feedback!
My most used checkpoint for over a year now. Big fan of your works - most of my success came from studying your prompts. Do you have plans to do a similar checkpoint for Flux 2/Qwen? Would be absolute Day 1 download.
At one point I was working toward training a Qwen model, but I stopped because of the scope of the work. I suppose I could pick it up again. Krea2 looks promising as a model. I haven't thought much about Flux 2, but I'd consider that. Appreciate the feedback!













