use to 1 ~ 0.8Since it was trained with very few images, overfitting will occur.However, you can try different outfits and hair by writing prompts.Using just the keyword (the first prompt) works fine.if the picture is not generated properly, enter additional prompts.1.Normal<pre><code>iris_yuma, pleated_skirt,</code></pre>or<pre><code>iris_yuma, pleated_skirt, coat</code></pre>2.Soulworker<pre><code>soulworker, fur_trim,</code></pre>3.Desireworker<pre><code>desire, belt,</code></pre>

Files with SHA256: 51b6be4cd4ed822f94cd38b715d15c682ec0e7ffaeaf79fdb62df045e640da40