<h2 id="heading-1340">Usage</h2><ul><li><p>Base model: Animagine v3.1</p></li></ul><ul><li><p>Tags:</p><ul><li><p>Basic tags: hinoshita kaho, medium hair, orange hair, cyan eyes</p></li><li><p>School uniform: school uniform, serafuku, brown dress (or white dress), white colar, red bowtie</p></li></ul></li><li><p>Recommended strength: 0.9</p></li></ul><p>All sample images are generated with no postprocessing.</p><h2 id="heading-1341">Quirks</h2><p>No. Generalize quite well, including non-safe.</p><p></p><h2 id="heading-1342">Training reference</h2><p>74 images (repeated: 154 images) from Link Live cards as of May 2024. Cropped to 1024x1536 (one bucket only).</p><p>20 epochs, total 1000 steps.</p><p>Final training loss: 0.804</p><p>Config: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/1771/how-to-create-near-perfect-character-and-style-loras-for-sdxl-10-latest-update-07092023">here</a></p>

Files with SHA256: 4563a4d29baacdcbacffe559ec1bffe1b8b52bb812f74e9b5192e14c280f6d12