XL_v2.1模型使用精简优化的100+张图片作为训练集，并增加了几个角色的标签绑定，整体效果更接近藤岛康介手绘复古风格。

训练图片主要来自藤岛康介的原画集，还有一部分来自动画版原画师：松原秀典。

FLUX版本：https://civarchive.com/models/722342

illustriousXL 版本：https://civarchive.com/models/987904

Animagine XL V3 和 SD1.5 版本：https://civarchive.com/models/73822

--- --- --- --- --- --- --- ---

The XL_v2.1 model uses 100 streamlined and optimized images as the training set, and adds Trigger Words for several characters.

The overall effect is closer to the hand-painted retro style of Kosuke Fujishima.

The training pictures are mainly from the original painting collection of Kosuke Fujishima, and some are from the original artist of the animation version: Hidenoria Matsubara.

FLUX version：https://civarchive.com/models/722342

illustriousXL version：https://civarchive.com/models/987904

Animagine XL V3 and SD1.5 version：https://civarchive.com/models/73822