Cruddy first version of KF1 lora.
Generation tags are "kf1, kf1 style" and I suggest adding "shading" to the negative prompts. Trained on anything v3.
描述:
Dataset is deepdanbooru tagged screencaps of season 1. Model seemed to stop working well after the first checkpoint (afterwards would just generate noise) and the dataset was like 2,000 or so images.
Also ignore that the astolfo example was generated with 1girl i forgot to change that tag
训练词语: kf1,kf1 style
名称: kf1style-000002.pt
大小 (KB): 37116
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: **Detected Pickle imports (3)**
```
torch.HalfStorage
torch._utils._rebuild_tensor_v2
collections.OrderedDict
```
病毒扫描结果: Success
