This style is a request trained on a dataset provided by a discord friend. It's a My Hero Academia anime style trained on synthetic data. I don't think it took quite as well as the Anima version, but that is to be expected given Krea2 nature. If you want to do spicy stuff with it, then you'll need supporting loras, and there aren't many that are suitable for non-realism work yet. I'll train a Moss version soon-ish and I'm expecting better out of that. Despite the shortfalls, Krea2 is amazing and you can do a lot with this. Running this at 0.75 strength is plenty. It won't help much past that, but might increase some artifacts of the style.
Tigger: MHAAnime style image of
Slightly useful: Anime screenshot style,
Recommended Strength: 0.75