<h3 id="heading-97">2023/7/15 update</h3><p>Update dataset. Retrain the model using new craft.</p><h3 id="heading-533">2023/6/5 update</h3><p>Update dataset.</p><h3 id="heading-534">2023/5/25 update</h3><p>Update dataset.</p><h3 id="heading-535">2023/5/21 experimental update</h3><p>By analyzing the impact of the weights of each layer of the model on the generation results, adjust the proportion of learning rates for each layer to train the model in an attempt to improve the generation quality. <strong>If some weird problems occur during the generation process, pls use the original 1.5 version.</strong></p><h3 id="heading-536">2023/5/21 update</h3><p>Update dataset.</p><h1 id="heading-537">Introduction</h1><p>A LoRA model of new character from Goddess of Victory: NIKKE called Blanc. <strong>The trigger tag is 'blanc \(nikke\)'.</strong> <strong>The key features are 'playboy bunny, rabbit ears, pantyhose, fishnet legwear, white leotard'.</strong> The 784mb VAEs (NovelAI, Anythingv3, <a target="_blank" rel="ugc" href="https://huggingface.co/WarriorMama777/OrangeMixs"><strong>Orangemix</strong></a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/4468/counterfeit-v25"><strong>Counterfeit</strong></a>) are recommended. 0.6 ~ 0.8 weights are good. <strong>The preview images are generated using 1.6 version (original) </strong><a target="_blank" rel="ugc" href="https://civarchive.com/models/9139/yesmix"><strong>YesMix.</strong></a></p><p></p><h1 id="heading-538">Cautions (Must Read)</h1><h2 id="heading-539"><strong>For anime model,</strong></h2><h2 id="heading-540"><strong>pls disable 'face restoration' during generating.</strong></h2><h2 id="heading-541"><strong>pls disable 'face restoration' during generating.</strong></h2><h2 id="heading-542"><strong>pls disable 'face restoration' during generating.</strong></h2><ul><li><p>Some preview images are generated using 'Hires Fix' and upscaled using 'SD Upscale'. Thus, <strong>if you wanna generate the preview images, pls download the original preview images and analyze them using 'PNG INFO' in webui to get all generating information.</strong></p><p></p></li></ul><h3 id="heading-543"><strong><u>BTW, make sure set this option in 'Stable Diffusion' settings to 'CPU' to successfully regenerate the preview images with the same seed.</u></strong></h3><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c1e77cab-093f-44e5-803b-1afcf2d42df4/width=525/c1e77cab-093f-44e5-803b-1afcf2d42df4.jpeg" /></p><p></p><ul><li><p><strong>The training resolution of this model is 1024x1024, so generating resolution must not exceed this range. </strong>Try to use 'hires fix' to prevent this problem and help to generate higher-resolution images.</p><p></p></li><li><p>For 784mb VAEs (NovelAI, Anythingv3, Orangemix, Counterfeit) users, <strong>make sure add '--no-half-vae' in command line of webui to prevent generating black images.</strong> See <a target="_blank" rel="ugc" href="https://rentry.org/voldy"><strong>voldy's log</strong></a> for the details of edit method. <strong>For high-performance GPU, make sure 'medvram' and 'lowvram' mode are disabled.</strong></p></li></ul>

【Character】Blanc (NIKKE)

<p>By analyzing the impact of the weights of different layers of the model on the generation results, adjust the proportion of learning rates for each layer to train the model in an attempt to improve the generation quality. </p>

Files with SHA256: 55c5588034cf1ecd29374b772d7e263958abc5999d63589a7e498ea4baa98162