<h3 id="if-you-have-the-means-please-do-support-the-original-artist's-creative-work!">⭐If you have the means, please do support the original artist's creative work! </h3>A LoRA that reproduces the drawing style of the artist ミコジン (mikozinn).Artist page: <a target="_blank" rel="ugc" href="https://www.pixiv.net/users/767724">https://www.pixiv.net/users/767724</a>Trigger word: <code>@mikozinnstyle</code> — put it at the very beginning of your prompt<h3 id="moss-version-is-hear!">MOSS VERSION IS HEAR!</h3><s>However, it seems to be slightly underfitting. I will submit a new version later~</s> MOSS v1.1 is available！with more accurate facial features and colors. However, the MOSS model itself tends to add highlights, so there is still a gap compared to the anima version. You can try adding style prompts to optimize it.Base model: anima-base-v1.0 | MOSS image baseThis is my first LoRA upload — feedback and suggestions are very welcome.Training data240 images from the artist's posts (2025 onward), auto-tagged with WD14 + Perform manual full-scale cleaning on the tags.The dataset is dominated by plain backgrounds (~82% white background), so backgroundrendering leans on the base model itself. In scenes completely outside the domain (urban nightscapes, neon lights, rain), the background details tend to be flat and simplistic. It is recommended to pair with LoRA for background/detail enhancement. It generalizes well on original characters and performs well in indoor scene background styles.- ⛔ Do not use this model to generate sexual content involving any minor character--- Base Model License &amp; Attribution 中文： ⭐如果各位有能力的话，请务必支持原画师的创作工作！ 一个旨在还原画师 ミコジン 绘画风格的 LoRA。画师主页：<a target="_blank" rel="ugc" href="https://www.pixiv.net/users/767724">https://www.pixiv.net/users/767724</a>触发词<code>@mikozinnstyle</code> —— 建议放在提示词最前面<h3 id="moss">moss版本来啦！</h3><s>不过看上去似乎有点欠拟合，后续我会再提交新版本~</s>MOSS v1.1 来了，脸部特征和颜色更加还原，但是MOSS模型本身很喜欢加高光，所以和anima版本仍有差距，可以尝试加上风格提示词来优化。底模：anima-base-v1.0 ｜ MOSS image初次上传 LoRA，还有很多没做好的地方，欢迎各位的指正与交流。 训练素材画师 2025 年至今投稿中，抽选 240 张，WD14 自动打标+手动全量清洗。素材以纯色/简单背景为主（白背景约 82%），因此背景表现更多依赖底模本身；实测在完全域外的场景（都市夜景、霓虹、雨）中，背景细节会变得平涂且简单，建议搭配背景/细节增强的 LoRA 。在原创角色上泛化良好，室内场景背景风格渗透较好。- ⛔ 禁止用于生成任何未成年角色的性内容---

Files with SHA256: 82e1281caecaa8f75812620a2b0f139b75f7b39ad2d00b4a2c717ed205c6d7a9