中文简介：
v1触发词：ohou_style
v2触发词:ohouhou_style
我把我在sd模型跑的一些图训练成了这个风格lora，训练素材图像都是我跑的熟女画作，如果想要你绘制其他体型的人物，可能会有一些问题，以下是个人推荐参数（这个lora我没怎么测试，推荐参数仅供参考，可以多尝试不同的参数）：
步数：40
采样器：er_sde
调度器：beta
我的pixiv主页，偶尔会发我跑的图：哦齁齁 - pixiv
v2版本ora仅为版本一lora变种，两个lora无优劣之分，看你的需求与喜好使用，同时使用两个版本的lora也没问题
English description：
v1 trigger word：ohou_style
v2 trigger word: ohouhou_style
I created this style LoRA using images I generated in Stable Diffusion. Since the training dataset consists entirely of mature female, it might struggle a bit if you try to generate characters with different body types or ages.
Recommended Parameters:
(Please note: I haven't tested this LoRA extensively yet, so these are just for reference. Feel free to experiment with different settings!)
Steps: 40
Sampler: er_sde
Scheduler: Beta
My Pixiv:
I occasionally post my AI-generated art here:
https://www.pixiv.net/users/119142162
v2 lora is just a variation of V1, and there is no superiority between the two. Feel free to choose based on your personal preference. Also, using both V1 and V2 LoRAs together works fine!
Description
第二个版本的lora:同样是来自我在il模型跑的图，第一版lora画风的变种，训练素材的画风变更，更加凸显角色的马甲线，皮肤更有光泽一些，与第一版lora的画风相比，皮肤没有那么细腻
这个lora仅为版本一lora变种，两个lora无优劣之分，看你的需求与喜好使用，同时使用两个版本的lora也没问题
LoRA V2 Introduction:
This is a stylistic variant of my V1 LoRA. The dataset also comes from images I generated using the "il" model, but with an adjusted art style in the training materials.
Features of V2:
More defined abs / midriff lines.
Glossier / shinier skin.
Note: The skin texture is not as delicate/soft as in V1.
Usage Note:
This is just a variation of V1, and there is no superiority between the two. Feel free to choose based on your personal preference. Also, using both V1 and V2 LoRAs together works fine!
FAQ
Comments (1)
大佬，能不能做一个il版本的，il也想哦齁