[new]
V5.0 has been released.
The influence of previous anime style drawings has been reduced.
If you enter [anime] in the prompt, you will get the traditional version.
I haven't tried it, but it may be that the pattern has less of an effect when crossing with other Loras.
The learning file has been modified so that the nose is depicted slightly larger.
The lips are also now pulled upwards. I haven't experimented with creating many images yet, so there may be many bugs.
It supports faces from the front and at a slight angle.
Triger word is [nosehook]
3way nose hook Ver.1 has been released.
I have set a trigger word for this [3waynosehook]
Front and side views are shown. Most of the time, it doesn't draw well, probably due to a lack of learning.
The problem is that if you turn to the side, your ears won't be covered.
I'm thinking of upgrading when the environment is ready.
Please try changing the strength between 0.7 ~ 1.3
The best distance for me is around 0.8.
The farther the subject is, the less accurate it will be.
I made it for SDXL. It was made by an amateur. I don't know if it will work well.
The file size is large.
Some versions require triggers and some do not. If it is not reflected, please check carefully.
The learning model is
[Counterfeitxl_v25]<-v1.0
[animagineXLV3_v30]<-v3.0 and 3way v1.0
[animazinexlv3.1]<-v4.0
I haven't been able to use it well myself.
If you create a good work, please upload it here.
描述:
训练词语:
名称: SDXL_nosehook_anime_v3-000012.safetensors
大小 (KB): 501462
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
