2.0版本
加强了一些面部细节，稍微提升了融合其他艺术家风格的效果
Prompt：al1star, red eyes, cow horns, facial hair, beard, nose ring, scar, pelvic curtain, muscular,
Description
FAQ
Comments (6)
it always seems pretty hard to nail down a lora for alistar that looks close to canon without too much null-ghost does it? i would've used his in-game models personally but i'm too lazy and stacked to train some
我在训练2.0版本时，已经使用了Wild Rift的alistar模型，因为我喜欢Wild Rift中alistar的胡子，但是过多的使用模型素材，会导致lora更加接近3d风格，并且身体比例和皮肤质感都很差，最终我将模型素材的比例控制在25%，null-ghost控制在40%，我想这也是其他创作者使用null-ghost风格的原因
@riiwil_ai717 does it make a difference if you tag the 3D pics with the 3D tag? i used to do that for some of my LORAs that use a 3D model of dataset
@oh_well 我不清楚，我没尝试过。
@riiwil_ai717 ngl i think you should try it nexttime. i've been looking up the danbooru tags a lot more recently and they have a lot of tags for different mediums like 3d, painting, digital, etc
@oh_well 也许以后有机会我会尝试这个办法。