——————————————————————————————————————
Warning
All the following content is machine-translated by DeepSeek, or even directly generated by DeepSeek. Do you honestly think I would bother typing this much manually
警告
以下内容均为deepseek机翻,甚至是deepseek写的,你真的认为我会勤奋的打这么多字?
——————————————————————————————————————
This is a 2D anime-style model trained by me and open for download to share and communicate. The model is fine-tuned based on the renowned AI painting model NOOB.
I sequentially trained several models that respectively optimized aspects like character anatomy and lighting/shadow effects, then merged them at proportions I deemed appropriate. Ultimately, I obtained a model with excellent lighting effects and relatively three-dimensional facial features.
It may not suit everyone, so you might need to test it yourself.
If you like it, likes and feedback with generated images would mean a lot to me. (`・ω・´)
由我本人训练,同时开放下载用于分享交流的二次元模型,模型基于著名的ai绘画模型NOOB微调。
我依次训练了几个模型,分别对身体和光影等方面有影响,之后按照我认为合适的比例进行了模型融合,最后我得到了一个光影不错,有较为立体五官的模型。
它不一定适合所有人,你可能需要自己测试它是不是适合你。
如果你喜欢,点赞,返图,这对我很重要。(`・ω・´)
——————————————————————————————————————
NOTES
Base model: NoobAI-XL Eps-pred 110%
Recommended settings:
Sampler: Euler a
Sampling steps: 20+
CFG scale: 5.5
Resolution: ≥512x780
Positive/negative prompts can reference the example images.
Read the manual for detailed parameters before using NOOB models.
For accurate tags, consult Danbooru.
For character/style references, use this site: [website]
注意事项
本模型基于NoobAI-XL Eps-pred 110%
采样推荐Euler a 迭代次数20以上 cfg5.5
推荐图片分辨率不低于512x780
正面或者反面tag可参考图片实例。
使用noob模型前可以阅读手册学习模型详细参数
获取准确的tags可以参考Danbooru
当然如果你需要角色外貌和画师画风参考可以使用这个网站
——————————————————————————————————————
PROMPT RECOMMENDATIONS (Reference Only)
Quality tags:
highres, absurdres, newest, very awa, masterpiece, best quality,
Lighting tags:
chiaroscuro, Natural Light, Mood Lighting, cinematic lighting, volumetric lighting, ray tracing, dramatic shadows, dappled light,
Artist blend