
A model created by accident in the making of an image of a certain style. Of course it's just a bit special, but it's not stable,
Regarding the direction of mixed model creation: My goal is to mix models that are unique enough, so I did not use any Lora in the process of mixing. All the special features of my models are obtained through mixing. In fact, the joy of model fusion lies in the mixing and discovery.
You can tell me the combination of Lora and prompt phrases you need to test. I will test it during the mixing process and demonstrate it upon release. This will allow you to have a more direct understanding of whether you need the model or not.
About model updates:
I prefer the fusion model to the generated image, so I don't know the results of the model any better than you do. If you think my model is flawed, please leave a comment and let me know where it needs to be improved, and I'll try to update and optimize it.
Known faults: not nice to furry people, not nice to male characters, not nice to people who like feet. Lora requires a lower weight
In fact, the current SD model is much more stable than the original SD model, and a large number of useless prompts will affect the overall expression. The use of multiple Loras can seriously affect the overall picture.
The role of the negative prompt in the current role of the quality of the prompt is far greater.
Here are the negative tips I use:
(low quality,worst quality:1.2),poorly drawn,(poorly drawn face),(deformed,disfigured),(bad anatomy,wrong anatomy),bextra lim,missing lim,floating lim,disconnected lim,(mutated hands and fingers),ugly|fingernails,blurry,(signature|watermark|text),(mutant feet and toes),Mix the feet together,Mixed limbs,
Anyway, if you like my model, please let me know so I can put more enthusiasm into it. By the way, I like to see the best pictures you can generate
描述:
模型浓度:中等 Model Concentration: Medium
模型纯度:稍高 Model Purity: Slightly high
这些信息是展示这个模型的扩展性,纯度代表着它的主要属性,浓度代表这个模型在混合时的表现。
These pieces of information demonstrate the scalability of this model, purity represents its primary attribute, and concentration represents the performance of this model when mixed.
训练词语:
名称: kmMIXSpecialAnime_v10.safetensors
大小 (KB): 2087900
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success