<h3 id="heading-57"><strong>原模型：</strong><a target="_blank" rel="ugc" href="https://civarchive.com/user/TanQiu"><span data-type="mention" class="mantine-1yiar0p" data-id="mention:79125" data-label="TanQiu">@TanQiu</span></a>，模型修复：<a target="_blank" rel="ugc" href="https://civarchive.com/user/Yuno779">Yuno779</a></h3><p><strong>——【元素法典】</strong>：谈秋&amp;Yuno779</p><h2 id="heading-58">说明 | specification</h2><ul><li><p>群友的模型，我拿来练手用的，模型是通过训练的方式来修复。</p><p>A QQ group friend model, I use it to practice.</p></li></ul><h3 id="heading-59">[2023/4/18]</h3><ul><li><p>BlackCat-<strong>TrainFix</strong></p><p>黑猫盛宴-修复版</p></li><li><p>下图是原版模型和修复版模型的同参数对照图</p><p>Below is a comparison of the original model and the restored model with the same parameters</p><p><img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/4de808d0-c726-4c1c-3b49-efe62fa73900/width=525/4de808d0-c726-4c1c-3b49-efe62fa73900.jpeg" /></p></li></ul><h2 id="heading-60">说明 | specification</h2><ul><li><p>模型无需外挂VAE。</p></li><li><p>VAE does not need to be VAE externally applied to the model.</p></li><li><p>部分情况下可能出现过度拟合，如有此类情况请和我进行反馈</p><p>Overfitting may occur in some cases. Please give me feedback if there are such cases</p></li><li><p>不建议继续进行训练，因为本质还是群友的融合模型</p><p>Do not recommend for Fine-tuning.</p></li><li><p>请不要添加“masterpiece”等起手式，除非画面出现各种问题</p><p>Please do not add "masterpiece" or other prompt words</p></li><li><p>CFG可以推荐设置在3~30之间</p><p>CFG can be set between 3 and 30, AI painting has infinite possibilities</p></li><li><p>个人制作能力有限，模型并不会非常完美，并且可能与人们的期望不同</p><p>My crafting skills are limited, the model will not be perfect, and it may not be what people expect</p></li></ul><h2 id="heading-61">局限性 | Limitations</h2><ul><li><p>可能不适合搭配使用例如badhand或者EasyNegative等</p><p>May not be suitable for use together such as badhand or EasyNegative etc</p></li><li><p>部分未经训练的部分可能依旧会出现无法准确反映tag的情况</p><p>Some parts not involved in training still cannot accurately reflect the tag information</p></li><li><p>请根据自己的情况使用高清修复，有的时候开启高清修复效果反而并不如直接出图</p><p>Hires. fix according to your own situation. Sometimes, opening Hires. fix results are not as straightforward as drawing</p></li></ul><h2 id="heading-62">其他说明 | other</h2><p>模型可以和其他模型一样随意使用，但是请遵守所在地区的法律法规，以免造成麻烦（我们并不负责）</p><p>The model can be used as freely as other models, but please abide by the local laws and regulations, so as not to cause trouble (we are not responsible for it).</p><h3 id="heading-63">特别鸣谢：</h3><ul><li><p><a target="_blank" rel="ugc" href="https://civarchive.com/user/TanQiu"><span data-type="mention" class="mantine-1yiar0p" data-id="mention:79125" data-label="TanQiu">@TanQiu</span></a>：原本的模型</p></li><li><p><a target="_blank" rel="ugc" href="https://github.com/bmaltais/kohya_ss">GitHub - bmaltais/kohya_ss</a>：训练UI</p></li><li><p><a target="_blank" rel="ugc" href="https://github.com/XavierXiao/Dreambooth-Stable-Diffusion">GitHub - XavierXiao/Dreambooth</a></p></li><li><p><a target="_blank" rel="ugc" href="https://github.com/AUTOMATIC1111/">Automatic1111</a>: SD-WebUI</p></li><li><p>Danbooru：打标器，帮助标准训练集</p></li><li><p>元素法典作者组：在这里让我有了这个想法，并且帮忙测试了模型</p></li></ul><h3 id="heading-64">祝大家玩得开心＜3</h3>

Files with SHA256: d46ff5f7914f54da0b55f955a1a290477cf311a0549e8ab9e75da00fef442496