好久没有炼丹了有点生疏了......
首推版本是6000步,当然,我上传了其他几个步数的版本,您可以根据图表自行对比下载合适的版本。
The primary recommended version is for 6,000 steps. I’ve also uploaded versions with other step counts, so you may compare them using the charts and download the one that suits your needs.
这是一个以东亚年轻女性人像美学为核心的写实摄影风格 LoRA,使用我从 MidJourney 中收集并筛选的 180 张图片进行训练。
整体风格偏向细腻、柔和且富有氛围感的电影摄影美学,着重表现自然清透的东亚面部特征、柔和的肤色过渡、细腻的皮肤质感,以及具有情绪表现力的光影和色彩。
无论是日常生活中的自然光、都市夜景中的霓虹色彩,还是具有戏剧性的电影布光,都希望能够呈现出一种介于真实与梦幻之间的独特视觉氛围。
需要注意的是,由于训练集主要由年轻东亚女性构成,LoRA 可能会对人物的面部特征产生一定影响,部分情况下也可能出现面容相似的问题。
This is a realistic photography style LoRA focused on young East Asian female portrait aesthetics, trained on a curated dataset of 180 images collected from MidJourney.
The style emphasizes delicate, soft, and atmospheric cinematic photography, featuring natural East Asian facial characteristics, smooth skin-tone transitions, refined skin textures, and expressive lighting and color.
From gentle natural daylight and vibrant urban neon scenes to dramatic cinematic lighting, this LoRA aims to capture a distinctive visual atmosphere that feels somewhere between reality and a dream.
Please note that since the dataset primarily consists of young East Asian women, the LoRA may influence facial characteristics and occasionally produce similar-looking faces.
推荐参数 | Recommended Settings.
基础模型 / Base Model: Krea2 Raw
触发词 / Trigger Word: MJAR_Style
推荐 LoRA 强度 / Recommended LoRA Weight: 0.8–1.0
推荐采样步数 / Recommended Sampling Steps: 25–30
推荐 CFG / Recommended CFG Scale: 4
以上参数仅供参考,可以根据实际生成效果调整。
These settings are suggestions and can be adjusted according to your results.
其他说明 | Additional Notes
这是我在尝试新的写实摄影风格方向时制作的一款 LoRA。相比之前偏向国风美学的 MJCN_Style,这次更加注重年轻东亚女性的自然美感、摄影氛围以及不同环境下的光影表现。
未来我也会考虑将这一风格移植到其他模型,包括 MOSS-Image,但目前还没有确定具体安排。
如果在使用过程中遇到任何问题,或者对模型效果有任何建议,欢迎随时向我反馈。
感谢大家一直以来的支持!❤️
This LoRA is part of my exploration of realistic photographic aesthetics. Compared with my previous MJCN_Style, which focuses more on Chinese-inspired aesthetics, this project places greater emphasis on natural East Asian feminine beauty, atmospheric photography, and cinematic lighting across different environments.
I may also consider bringing this style to other models, including MOSS-Image, in the future, although no specific release plans have been finalized.
If you encounter any issues or have suggestions, please feel free to share your feedback.
Thank you all for your continued support! ❤️












