z_image_base_cos

此模型的训练初衷是希望融合coser的服装、妆造、场景灯光与私房摄影的美感。训练数据中已强制清洗了NSFW内容。事实上，纯粹的暴露图像并不会激发人类最原始的欲望，相反，朦胧含蓄的氛围才是大多数人所喜爱的。

本模型已对女性隐私部位进行强制清洗，希望屏蔽这些内容后，大家能回归更原始、纯粹的审美状态~ 祝大家玩得开心！

此模型是coser模型的第二个版本。在继续使用White Base模型作为训练基础的前提下，优化了训练参数，并在8张RTX 5090显卡上训练了72小时。

触发词：

提示词：默认使用中文！

只需正常使用Z-Image-Turbo模型及工作流即可！

About This Model

The original intention behind training this model was to combine the costumes, makeup/styling, and scene lighting of cosplayers with the aesthetic feel of boudoir photography. NSFW content has been forcibly cleaned from the training data. In reality, purely explicit imagery doesn't evoke humanity's most primal desires—on the contrary, a more subtle, hazy aesthetic is what most people prefer.

This model has undergone mandatory cleaning of female intimate areas. I hope that by filtering out this aspect, everyone can return to a more primal state of appreciation. Have fun!

This is the second version of the cosplay model. Compared to the first version, while continuing to use the White Base model as the training foundation, the training parameters have been optimized. It was trained for 72 hours on 8× RTX 5090 GPUs.

Trigger word: <pond>

Prompt: Default to Chinese!

Simply use the Z-Image-Turbo model and workflow as normal!