The model prohibits any form of reproduction and secondary publication. It is for learning, reference and experience exchange. Please abide by local laws and regulations and do not use it for commercial or other illegal purposes.
v0:动漫类型的融模,效果为强调线条感,但由于过分强调也导致直出图片会糊图,结合高清修复使用效果良好,属于是其他子版本的基础模型。
v1-Dline:(旧命名为2.5D)基于v0进行融合调整,一定程度上修复了v0直出糊图的问题,效果上强调线条感(detailed line),特色挺强,检验过读tag能力和图片质量可适用于各种用法。此模型可以文生图直出不会糊图,兼容各种采样器,无需捆绑采样器和高修算法使用。此模型内置anime的vae。注意,不适合使用840000的vae,会导致糊图以及轮廓破坏的问题,如选择外挂clip建议使用anime或者是clear的vae。
v1-SlightFlat:这个版本的效果有些许偏向平涂风,并且兼容了少女幼态(没有做过关于男性的测试),能出可爱类型的人物。
v1-FlatCoating:基于SlightFlat进一步调整,相比于SlightFlat进一步强化了平涂风格,是纯纯的平涂风格动画类模型,同样兼容可爱类型的人物出图。
(目前就更这么多,不同效果的v1子分支在测试中,后续更新会放出来)
The Dline fusion model emphasizes the sense of line in effect, has a similar sharpening feeling, has strong characteristics, and can be applied to various uses after testing the ability to read tags and picture quality. This model can be straight out of the text without paste, compatible with various samplers, without the need to bundle samplers and advanced repair algorithms. This model has a built-in VAE for anime, note that the VAE of the 840000 is not suitable.
(PS: v0 Need to use Hires. fix)