LOCO-flux（Real texture portrait）: V3: locoFluxRealTexture_v3.safetensors

Flux.1 D Standard Model

Daily update: limb optimization

If you want to directly output Asian faces, please add the trigger words: MUCHEN+ARang

At the same time, dual clip+vae is provided, which is better than the original output to a certain extent (avoiding NSFW and limb mutilation),

But there is a risk of stylistic transformation and little change in output, welcome to test.

Note: clip type is sd3

★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★

When using this model, please abide by the terms of Black Forest Labs:

License: flux-1-dev-non-commercial-license

★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★

Friends are welcome to test and post pictures! Thank you!

WeChat: locolabs

基于上一版本Flux-SeriesX微调量化的垂直人像模型,

颗粒度更细腻,真实感更强,适合出男女肖像写真,商业时尚大片等.

本模型是基于Flux架构下Finetune的模型，为fp8版本.

在极高审美质量的时尚摄影(部分欧美杂志大片)和胶片写真上进行训练调优，

在构图美感、出图细节上都有了及其显著的提升。

相较于上一代SeriesX，肤质表现更加细腻且真实!

V2为V1美化版，融合了muchen系列未公开的第一版LORA,质感-1，肢体+1，美感+1，

★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★

如果画面不佳，请添加起手式：MUCHEN+ARang

同时提供了双clip＋vae,相较于原始出图在一定程度上更能(避免NSFW),

但有风格化转变和出图不佳的风险,欢迎测试.

注意:clip类型为sd3

★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★

使用该模型时，请遵守黑森林实验室的条款：

License：flux-1-dev-non-commercial-license

★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★

欢迎各位朋友测试并晒图！感谢各位！