Update information
2023/4/7
v1.3_dim_16 is uploaded, it is a resized version of v1.3, and it makes the style soft.
2023/4/4
v1.3 is uploaded, it is trained based on SD 1.5, and is continued on the first version of pen sketch LoRA
v1.3 是基于SD第一个版本的直线速写风格的lora继续训练的
For pen_sketch_style_v1.3, I strengthened "hatching" style and improved stablity of it.
在1.3版本中我加强了阴影线的风格效果，并且提升了模型的稳定性,
Recommended settings
Recommended weight for LoRA ：0.6-0.8
推荐LoRA权重：0.6-0.8
To-Do List
1) Training a version with stable pen position-control
训练一个可以控制笔的摆放位置的版本
2) Improving stability of style applying when drawing particular clothes, for example, school uniform, swim suit etc.
提高风格应用的稳定性，特别针对校服，泳装等这些在测试时风格应用不太稳定的服饰,
Difference
v1.3_dim_16 弱 | soft
v1.0 中 | medium
v1.3 强 | hard
Our Community
If you want to join a community of AI drawing, and to learn more about LoRA training tricks, you can scan the QR code below to join our WeCaht group chat.
如果你想加入ai绘画团体，并且对LoRA训练小技巧感兴趣，你可以扫描下方的二维码加入我们的“C站炼丹术分享讨论群”，欢迎大家加入并且交流讨论！！！
捏Ta QQ频道同步欢迎你的加入（复制到qq频道内搜索）
t0o1vs01m0
描述:
All the example images are generated based on anything 4.5.
所有的示例图都是通过anything4.5跑出来的。
I didn't try it on other ckpts, but it should work fine on any sd 1.5 ckpt.
我没有在其他大模型上尝试，不过所以基于sd 1.5 的大模型应该都会适用。
Recommended weight for this LoRa is between 0.7 and 1.
推荐lora权重为0.7到1
When you trying to apply this style into a "strong" character LoRa, ou can also set this style LoRa weight to be 1 and set your character LoRa weight to be 0.8 (I tried this setting several times on my own character LoRa, and it works well for me)
如果你遇到角色风格比较强的lora，你也可以尝试将此风格lora的权重调为1，然后把角色的权重调调低，例如调到0.8，我在自己的训练的风格比较强的角色lora上施加这个风格时，尝试了很多次，发现把角色lora调到0.8效果不错。
Also, I am gonna add more into my dataset, and try to improve the strength and stability of this style LoRa as soon as possible
我也会尽快后续在训练集里面加入更多的参照图，并且尝试把这个风格的强度以及稳定性提升。
训练词语: pen_sketch_style
名称: pen_sketch_lora_v1.0.safetensors
大小 (KB): 73842
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
