中文 | English
仿yeurei的作画风格的LoRA模型。脸型会非常的“V”,比较适合方形的图片生成。
如果你用起来效果不是很好,请提高CFG到14~16。
权重建议值:1 (or 0.6~1.4)
基于模型:Anything-5RE
训练信息:筛选出共计232张原画(含裁剪图),迭代1160步,多分辨率迭代训练。
其他的一些建议:使用词条upper_body(上半身),close_shot(近景,视情况使用)。
LoRA model with a drawing style inspired by yeurei. The facial structure is very "V" shaped, and it is more suitable for generating square images.
If you are not satisfied with the results, please increase CFG to 14~16.
Recommended weight value: 1 (or 0.6~1.4)
Based on the model: Anything-5RE
Training information: Selected a total of 232 original drawings (including cropped images), iterated for 1160 steps, and trained with multi-resolution iteration.
Some other suggestions: Use the term "upper_body" for upper body shots, and "close_shot" for close-up shots depending on the situation.