Notes (6/20/25):
I renamed this style after discovering how versatile it is. It works well for stylized realism as well as flat illustrated styles.
The model was trained on ≈800 well-tagged, high-quality images, all AI-generated, sharing, more or less, a flat-colored, cel-shaded, illustrated aesthetic.
I've been using the SDXL and Illustrious versions together on the same generation. That seems to work well with SDXL, Juggernaut XL, RealCartoon XL, and Illustrious checkpoints.
Notes (6/29/25):
Of the 3 Flux versions, V3 is the most flat/illustrated.
Don't be afraid to crank it up if the images you're getting aren't flat enough. Flux is especially difficult to wrangle into animated submission, but it can be done. I've gotten great results with V3 setting it to 1.5 and using it in conjunction with this wonderful Métal Hurlant Comics LoRA by Atomease. at 1-1.2. Above that the line work gets too busy for my taste but it's a really cool style and absolutely does what it's supposed to extremely well.