<h1 id="heading-1863">Introduction</h1>A Lycoris model trained with images of Eve from Stellar Blade. Due to the current scarcity of training images, there tends to be a phenomenon where the model fidelity decreases as the prompts increases. Appropriately increasing the weights can alleviate this issue. <h1 id="heading-1864">Usage</h1><ul><li>Trigger tag: <code>eve \(stellar blade\)</code></li><li>Key Features: ponytail, blunt bangs, bodysuit, mechanical parts, barcode, barcode tattoo</li><li>The dataset is classified using rating tags: <code>safe, general, sensitive, questionable, nsfw, explicit</code>, which can control the results. You also can use <code>realistic</code> tag to control anime and realistic style of results (default is anime style for anime base model).</li></ul><h2 id="heading-1865">Generating Settings</h2><ul><li>Base Model: Anime SDXL model is recommended (Do not use this Lycoris with any SD1.5 model).</li><li>Sampler: DPM++ samplers, Euler a are recommended.</li><li>Eta noise delta: 31337 or 0</li><li>Upscale Method: HiresFix+<a target="_blank" rel="ugc" href="https://huggingface.co/konohashinobi4/4xAnimesharp/tree/main">4xAnimeSharp</a></li><li>Max Resolution without Upscaling: 1024x1024</li><li>VAE: No need to load any external VAEs.</li><li>LORA Weights: 0.7~1 are recommended.</li></ul>

Files with SHA256: df163fa930cd28392780878bb37be6df770b4108574f4598105b59528057649a