This model is trained on images from Envy Cute Mix 03. It's not exactly the same, but I think it's good where it is. It leans pretty heavily toward realistic shading as opposed to a cel shaded look. I'll probably release a cel shaded version later, or maybe something that can do both.

Image style on SDXL is a lot more sensitive to prompts, so check my examples for the positive and negative prompts to see what I did. Images where generated on Envy Overdrive XL v1.1.