该模型使用了超过100张高质量的原始图片进行训练。
素材主要来自《电影少女》《I"s》、《D.N.A2》,接近其手绘复古画风。
TV动画版图片素材没有混入其中,得到纯净的桂正和风格LORA。
FLUX版本:https://civarchive.com/models/910169
Pony XL版本:https://civarchive.com/models/375401
illustriousXL 版本:https://civarchive.com/models/892089
碎碎念:
XL_v1.01版本更换底模为Animagine XL V3.1,整体效果更精致了,但是对比XL_v1.0版本缺了一些复古的味道,仅供娱乐吧。
推荐权重:0.6~1.0
推荐CFG:3~5
推荐模型:Animagine XL V3
--- --- --- --- --- --- --- ---
The model used more than 100 high-quality original pictures for training.
The material mainly comes from 《Video Girl》、《I"s》、《D.N.A2》, be like hand drawn retro style.
The TV version animation picture materials are not mixed in, so as to obtain the pure artist style LORA.
FLUX版本:https://civarchive.com/models/910169
Pony XL版本:https://civarchive.com/models/375401
illustriousXL 版本:https://civarchive.com/models/892089
One more thing:
The XL_v1.01 version replaces the base model with Animagine XL V3.1, and the overall effect is more refined. However, compared with the XL_v1.0 version, it lacks some retro flavor. It is just for entertainment.
Recommendation Weight:0.6~1.0
Recommendation CFG:3~5
Recommendation Model:Animagine XL V3
Description
FAQ
Comments (4)
哈哈,up的模型质量是真的高,但这个模型感觉也是脸超级稳定。
对了,想问下您为什么底模都选的特定一个模型呢?
可能是因为个人使用下来,这个模型不论还原画风效果,还是学习率都比较不错吧。
有试过NovelAI的原始底模,虽然兼容性是最好的,但是还原画风的效果还是稍差一些,后来就习惯性用这个scmix_NSM.ckpt作为底模了~
太棒了,秒回童年!大概是95年前后吧?



















