An AI image of Wakana Umezawa I created from a LoRA trained on 100 copies of the same image.
1
Since the dataset image has usual AI mistakes, it kinda got reflected in the LoRA model as well, especially the curtain. I used the same base model I used to make the original images.
umezawawakana2
Show Parameters
- Size
- 512 x 768
- Sampling Steps
- 12
- Sampling Method
- Euler a
- CFG Scale
- 6
- Seed
- 1
- Negative
- worst quality, large head, low quality, extra digits, bad eye, EasyNegativeV2, ng_deepnegative_v1_75t
