V1 updated!
English Description(中文介绍在下方)
V1 version
This LoRA was trained on a revised dataset built upon the Beta version dataset. I added realistic images generated by GPT Image2.5 and removed samples with heavy airbrushed beautification from the original dataset.
What's new?
(1)This version delivers stronger photorealism, with more natural and detailed skin texture.
(2)You can even see fine vellus hairs on the character's skin! Enhanced texture performance for full-body shots as well.
(3)You no longer need lengthy prompts to describe skin details; a simple sentence or just a few words will suffice.
(4)NSFW is supported, though I have not tested scenes depicting sexual intercourse between men and women.
Tips
(1)Recommended strength: 0.7 ~ 1.0. If the effect feels overpowering, lower the strength.You can find the workflow in my sample images.
(2)To better generate images of huge breasts or nude character and etc., I recommend pairing it with the Krea2 TextFusion Refusal-Reduction LoRA at a strength of 1.
Here is the link:
Krea2 TextFusion Refusal-Reduction LoRA - v1.0 | Krea 2 LoRA | Civitai
中文介绍
V1 版本
本 LoRA 基于 Beta 版数据集修订后训练而成。我在数据集中加入 GPT Image2.5 生成的写实图像,并剔除了原数据集中重度磨皮美颜的样本。
更新内容:
(1) 本版本写实质感更强,皮肤纹理更加自然细腻。
(2) 甚至可以看见人物皮肤上细小的绒毛!同时也增强了全身照的质感表现。
(3) 不再需要冗长提示词去描述皮肤细节,只需一句话甚至几个词就足够。
(4) 支持 NSFW 内容,但男女性行为相关场景我尚未测试。
提示
(1) 推荐强度:0.7 ~ 1.0。如果觉得效果过强,可以调低强度。你可以在示例图片中找到工作流。
(2) 如果想要更好生成巨乳或者裸体人物等图像,建议搭配 Krea2 TextFusion Refusal-Reduction LoRA,强度设置为 1。
链接如下:
Krea2 TextFusion Refusal-Reduction LoRA - v1.0 | Krea 2 LoRA | Civitai
BETA version
This is a LoRA optimized for Asian aesthetics and photorealism enhancement. It is currently a beta version, but you are welcome to download and use it!I will further adjust the dataset and train a new version later.
这是一个对于亚洲审美和写实增强的 lora,目前只是 beta 版,不过欢迎下载使用!后续我还会调整数据集继续训练一个版本
Description
This is a LoRA optimized for Asian aesthetics and photorealism enhancement. It is currently a beta version, but you are welcome to download and use it!I will further adjust the dataset and train a new version later.Current limitations: lack of fine details on hair strands, and occasional blurriness.
这是一个对于亚洲审美和写实增强的 lora,目前只是 beta 版,不过欢迎下载使用!后续我还会调整数据集继续训练一个版本。目前存在的不足:发丝不够细节,有时候有点模糊




