Verified: 2 years ago
SafeTensor
Type | |
Stats | 1,046 23,075 |
Reviews | (166) |
Published | Jun 4, 2023 |
Base Model | |
Training | Steps: 1,320 Epochs: 12 |
Trigger Words | lycaon |
Hash | AutoV2 81125C8010 |
160 images trained on NAI.
Works well on many models with NAI component. Furryvixen2, indigomix, counterfeit3 are recommended.
The trigger word is lycaon, but it does not seem to make much difference.
You should consider adding words like "white fur, eyepatch, red eyes, shirt, vest" to make the appearance stable.
The sample pictures above are from a mixed model, 【indigo16 0.5counterfeit】 0.5【furryvixen2 0.3counterfeit】 which fits this lora well.
/////////////////////
160张图在NAI上训练得到,用的秋叶脚本v4
用novel成分多点的大模型可能效果好点,我自己一般furryvixen2,和counterfeit3混起来也很好看。上面示例图的底模是先indigo16和0.5的counterfeit3混、furryvixen2和0.3的counterfeit3,再两个0.5混起来
几个写上比较好的特征词:white fur, eyepatch, red eyes, shirt, vest。有裸体的训练集,可以指定不穿