<h3>已联系并征得3d模型作者雨宫24老师同意，特别感谢雨宫24老师提供的帮助</h3>Contact has been made and permission has been obtained from the 3D model author 雨宫24. Thanks to 雨宫24 for the help provided.已上传训练集，需要PSD，jpg及txt原始文件请联系我，B站：鱼泠溟；私信我，暂时只向官方，其他lora训练者或认为我使用了您作品的画师提供The training Data has been uploaded提示词：Azusa,1girl,额外可选提示词：kimono, underboob, fox mask, white pantyhose, short eyebrows , thick eyebrows如果南半球不出现可以使用underboob如果豆豆眉不出现可以使用short eyebrows 或 thick eyebrows可以把white background加入负向提示词跑图建议使用DPM ++2M或DPM ++SDE算法权重根据模型不同可以在0.75-1之间调整；可以配合hire fix使用，放大算法请选择为R-ESRGAN 4x+Anime6B，<s>其实不用也行</s>第一次训练过程使用了25张由雨宫24老师制作的3d模型图像作为训练素材第二次训练过程使用了16张由AI生成并经过修改的图像，+11张3D模型图像进行校正第三次训练使用22张AI生成并经过修改的图像+3张3D模型图像进一步完善，版权归八恰力阿鲁公司所有再次感谢雨宫24老师提供的帮助严禁商用，严禁公开传播由此lora生成nsfw图像及做出任何损害梓宝名誉权的行为，如有不妥之处，请联系我删除此LoraPrompt words: Azusa, 1girl,Additional optional prompt words: kimono, underboob, fox mask, white pantyhose, short eyebrows , thick eyebrowsAdditional optional negative prompt words: white backgroundIt is recommended to use DPM ++2M or DPM ++SDE algorithm. The weight can be adjusted between 0.75-1 according to different models;It is recommended to use with hire fix, please choose R-ESRGAN 4x+Anime6B as the zoom algorithmThe first training process used 25 3D model images made by 雨宫24 as training Data.The second training process used 16 images generated by AI and modified, and 11 3D model images for correction.The third training used 22 images generated by AI and modified, and 3 3D model images for further improvement.The copyright belongs to VirtuaReal .Thanks again for the help provided by 雨宫24..Commercial use is strictly prohibited, and public dissemination of nsfw images generated by this lora and any behavior that damages the reputation of Azusa is strictly prohibited. If there is anything wrong, please contact me to delete this Lora

初始常服|阿梓从小就很可爱|阿梓|Azusa|Azi|网友小z|Vtuber Vup

可以较稳定生成南半球提示词：Azusa,1girl,额外可选提示词：kimono, underboob, fox mask, white pantyhose, short eyebrows , thick eyebrows如果南半球不出现可以使用underboob如果豆豆眉不出现可以使用short eyebrows 或 thick eyebrows可以把white background加入负向提示词跑图建议使用DPM ++2M或DPM ++SDE算法权重根据模型不同可以在0.75-1之间调整；可以配合hire fix使用，放大算法请选择为R-ESRGAN 4x+Anime6B，<s>其实不用也行</s>

Files with SHA256: 3080412a2d128868c5dc9ed0ebec4506a902cbf09bdef815bdcfe98cfe31d878