Trained on images from <code>haiz ai</code><hr /><h2 id="moss">MOSS</h2>Near perfect recreation of the style, not much to say, this model learns extremely well.Lower strength if you want less effect.If you find the effect to be too small on out of distribution prompts, adding <code>Minimalist flat illustration</code> helps a lot<h2 id="anima">Anima</h2>Strength 1.25 / 1.5 might give more stylized results.

Style > Haiz ai

Training Settings: batch size 4, mixed res from 512px to 1216px, ~251 images per epoch, optimizer is <a target="_blank" rel="ugc" href="https://github.com/Koratahiu/Advanced_Optimizers/tree/goemtric_wd">SinkSGD_adv</a>

Style > Haiz ai Anima-base1 v1.0

Style > Haiz ai Anima-base1 v1.0

MOSS

Anima