触发词【不需要的元素可以自行删除】&lt;lora:YQ_pony_nuannuan16_V4:0.8&gt;,q1528437228,1girl,solo,black hair,dress,jewelry,hat,necklace,purple eyes,lace gloves,high heels,pantyhose,white headwear,ribbon,long hair,hair bun,sidelocks,single hair bun,hair ornament,feet,braid,crying,tears,模型说明：只推荐我首页的两个生图底模，两个底模的使用参数请看下面的参数推荐【云落-国漫Pony2.5D生图大模型【推荐】】参数推荐：【采样方法(Sampler)】 &nbsp; &nbsp;Euler【调度类型】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; SGM Uniform【采样步数】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 20-30 &nbsp; &nbsp; &nbsp; &nbsp; 节省算力20步也是可以的【CFG】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 3-7 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 建议为3到4，cfg越高画面饱度越高【宽高】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;1024*1024 &nbsp; &nbsp;建议为宽高皆为64的倍数（玄学？）如：1024*1024、832*1216等【高分辨率修复】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 建议开启高分辨率修复以达到最佳的效果【放大算法】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;4x-UltraSharp &nbsp; / &nbsp; R-ESRGAN 4x+ &nbsp;（前者使画面更细节，后者使画面更平滑，其他放大模型需自行尝试）【重绘步数】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;14-20 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;节省算力15步也是可以的【重绘幅度】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;0.35-0.64 &nbsp; &nbsp; &nbsp; &nbsp;建议为0.45左右【放大倍数】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;1.5-2 &nbsp; &nbsp; &nbsp; &nbsp;建议为1.5 （倍数越高重绘幅度要相应降低，不然容易产生肢体畸变）【云落-国漫Pony低步数加速大模型【推荐】】参数推荐：【采样方法(Sampler)】 &nbsp; &nbsp;Euler【调度类型】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; SGM Uniform【采样步数】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 10【CFG】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 1.5-2 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 建议为1.8左右【宽高】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;1024*1024 &nbsp; &nbsp;建议为宽高皆为64的倍数（玄学？）如：1024*1024、832*1216等【高分辨率修复】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 建议开启高分辨率修复以达到最佳的效果【放大算法】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;4x-UltraSharp &nbsp; / &nbsp; R-ESRGAN 4x+ （前者使画面更细节，后者使画面更平滑，其他放大模型需自行尝试）【重绘步数】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;10【重绘幅度】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;0.4-0.64 &nbsp; &nbsp; &nbsp; &nbsp;建议为0.45左右【放大倍数】 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;1.5-2 &nbsp; &nbsp; &nbsp; &nbsp;建议为1.5 （倍数越高重绘幅度要相应降低，不然容易产生肢体畸变）提示词起手式：正向&nbsp; (score_9),反向&nbsp; (deep_skin:0.9),(brown_skin:0.8),(yellow_skin:0.8),blurry,（其他提示词可自行测试，不过不建议加入过多的提示词，过多提示词可能会降低模型效果）警告：使用该模型即默认遵守以下用户协议，有违者期间发生的任何不良事态需您自行承担后果1、不允许私自分享开源此模型2、 您不得实施以下行为，也不得为任何违反法律法规的行为提供便利，包括但不限于：散布淫秽、**、赌博、暴力和违法内容。散布谣言，扰乱社会秩序，破坏社会稳定。包含法律、行政法规禁止的其他内容。因您的数据的产生使用等相关事项中如存在违反法律法规的情况，您需自行承担因此造成的全部后果和责任。版权声明与使用须知本模型基于互联网公开素材训练，仅供学术研究及个人学习AI技术使用，禁止一切商业用途。模型所涉原创内容知识产权归属原作者/版权方所有。模型收费仅用于覆盖模型训练、数据整理及技术服务等基础成本，不含任何版权许可费用。版权方特别说明如相关版权方认为内容涉及侵权，请通联系删除。

Files with SHA256: 5a30be4fcd2ae4bd10cbba79209ddeb459e87de05c39b47e4ac704f580beeb96