<h3 id="about-the-lora">About the LoRA</h3>This is a model of the Pink Cafe Art games, which are porn/hentai games. Specifically:<ul><li>Oniga Town of the Dead</li><li>Two Horns - Living In the Town With Ogres</li></ul>It was trained to represent their game style.<hr />This model is:<ul><li>trained on 2D images</li><li>trained on SFW (≈20%) and NSFW data (≈80%)</li><li>trained on actual in-game screenshots (with some cleanups like discarding UI elements)</li><li>also trained on zombie and oni types of characters.</li></ul>This model is not:<ul><li>trained on anthro/non-human characters (while author has them, they weren't currently included; expect some issues while creating them).</li></ul>Tips:<ul><li>you may want to lower the weight to make it a bit more flexible in some cases (0.7-0.9 should do the trick while keeping the character recognizable)</li><li>some prompt tags that may affect the results more:</li></ul>charater types:<pre><code>zombie
oni</code></pre>skin-related:<pre><code>purple skin
dark skin
tan</code></pre>background-related:<pre><code>pink background + heart background
purple background + pink background</code></pre>eyes-related:<pre><code>heart-shaped pupils</code></pre>There are many different clothes-related and NSFW-related tags that may affect the results more, but nothing really unusual to point them out.<hr /><h3 id="if-you-like-this-model">If you like this model</h3><ul><li>don't forget to give it a small like</li><li>share your generations with others and react to their generations</li></ul>

Pink Cafe Art | Style LoRA

<ul><li>Increased dataset size a bit.</li><li>Tweaked existing tags (now it should be more flexible and precise).</li><li>Trained with a new training configuration (should now be more flexible and precise)</li><li>Increased net size (to make it a bit more detailed and handle current dataset better, also means that the file size is bigger).</li></ul>

Files with SHA256: 9ab181cfcf1f82c97e134c9c13b2cfab515246a3d32d3e498701f2b1d70cea2d