
Do not re-publish this model on any for-profit generation service, such as PixAI without my permission!
So wait, another Asuka model???
I know, there are countless Asuka models out there. Besides, many anime checkpoints are capable of drawing Asuka out of box. So why train another Asuka model?
Too many models are overfitting and can only draw Asuka in the red plugsuit; or, they require a long stream of prompt words, resulting in prompt leaks all over the places; or, they are not faithful to the anime art style.
I got frustrated and decided to make a model of my own. The goal is to create an Asuka model that is both faithful to the anime art style (hence, only trained on anime screenshots and official artworks) and also flexible, capable of drawing Asuka in all kinds of outfits. Why? Because a girl like Asuka deserves a long, fulfilling life. We want to see Asuka living her life to the fullest, trying out different professions and experience everything the world has to offer.
The model is trained with anyLoRA at clip 2. It is slightly overtrained, so it's best to use it at 0.9. While I tried to make the model flexible, sometimes you still see the red plugsuit showing up - when that happens, just put "plugsuit" in the negative prompt.
描述:
训练词语: ppasuka
名称: asuka_v1_2.safetensors
大小 (KB): 36983
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success