<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Mozudoll®(魔族人形®)[EN]It features the aesthetic of Anime-style Soft-body Moe Doll, not BJD or MJD dolls, though both belong to harmless doll art.The character design in the images is original to the mozudoll brand.⚠️This LoRA is officially authorized by the mozudoll brand owner.⚠️For personal, non-commercial learning and use only. Resale and any unauthorized use are strictly prohibited.[CN]它具有二次元萌系软体娃娃的美学风格，并非bjd或mjd，虽然均属于无害的人偶艺术。图片素材的形象来自于 mozudoll 品牌的原创设计。⚠️本 LoRA 已获得 mozudoll 品牌的官方授权。⚠️仅限个人学习、非商用使用，禁止倒卖及违规用途。Recommended Settings(推荐设置):Sampler(采样器)：Euler aQuality tags: Not required不需要正面的质量提示词。I finished training this new LoRA on July 3rd, and it shows obvious improvements in all aspects compared to the last version. I'm terribly sorry for the delayed release, as I've been swamped recently.Special thanks to 魔族娘 - 萨利希尔 for supplying more ultra-high-res training images. They greatly helped push the overall quality of this LoRA to the next level.Upon review, the last version used rep2, which was excessively high. Worse still, I carelessly set the mature style of training materials to rep3 by mistake.With expanded dataset and lowered rep values, this version has far lower overfitting risks. Most noticeably, the model responds much better to character name triggers.我在7月3日完成了这个新LoRA的训练，各维度效果都比上一版有不小的提升，但因为挤不出时间，一直拖到现在才在这里发布，非常抱歉。感谢魔族娘-萨利希尔提供的一百余张超高清的新素材，这使新lora的综合效果进一步提升。回头看上期的repeat2确实还是太高了，尤其是我还手滑把素材的mature部分弄成repeat3。这次素材的增加和repeat的减少进一步减小了这个版本过拟合的风险，尤其是对角色名的响应有明显好转。————————个人组建的兴趣小组Q群：1083538641

Mozudoll-Anima

[EN]I finished training this new LoRA on July 3rd, and it shows obvious improvements in all aspects compared to the last version. I'm terribly sorry for the delayed release, as I've been swamped recently.Special thanks to 魔族娘 - 萨利希尔 for supplying more ultra-high-res training images. They greatly helped push the overall quality of this LoRA to the next level.Upon review, the last version used rep2, which was excessively high. Worse still, I carelessly set the mature style of training materials to rep3 by mistake.With expanded dataset and lowered rep values, this version has far lower overfitting risks. Most noticeably, the model responds much better to character name triggers.[CN]我在7月3日完成了这个新LoRA的训练，各维度效果都比上一版有不小的提升，但因为挤不出时间，一直拖到现在才在这里发布，非常抱歉。感谢魔族娘-萨利希尔提供的一百余张超高清的新素材，这使新lora的综合效果进一步提升。回头看上期的repeat2确实还是太高了，尤其是我还手滑把素材的mature部分弄成repeat3。这次素材的增加和repeat的减少进一步减小了这个版本过拟合的风险，尤其是对角色名的响应有明显好转。

Files with SHA256: cea1e31f6e650540601110692c64359403f573c2c0ed94e1d0e976fb799d7eff