Type | |
Stats | 466 55 |
Reviews | (71) |
Published | Dec 14, 2023 |
Base Model | |
Usage Tips | Clip Skip: 2 Strength: 1 |
Hash | AutoV2 B2AC6EB0E1 |
基于 AnythingV5 Ink 进行训练
这是一个可以生成简约风格的 2D 动漫风格 LORA,拥有简单的线条与干净的颜色。
V2 与 V1 风格接近,但 V1 的画风更简约一些,看你喜好选择。
V2 版本:
1. 相比 V1 版本,提高了些许亮度和对比度,颜色更鲜艳一些;
2. 去除了 V1 版本素材中的光环,出图不会再出现光环了;
--
V1 版本:
1. 更简约的画风,高亮度低对比度,低饱和;
2. 因使用过多蔚蓝档案作为素材,部分出图会出现人物头上顶着一个光环。
--
因为基于 AnythingV5 Ink 进行训练,使用时请使用基于 Anything V5 lnk 训练的模型或者合并模型,否则可能会出现图片崩坏的情况。
For training based on AnythingV5 Ink
This is a 2D anime style LORA that can generate a minimalist style, with simple lines and clean colors.
V2 is similar in style to V1, but V1 has a simpler style. It depends on your preference.
V2 version:
1. Compared with the V1 version, the brightness and contrast are slightly improved, and the colors are more vivid;
2. The halo of "Blue Archive" in the V1 version material has been removed, and the halo will no longer appear in the picture;
--
V1 version:
1. A more simple style of painting, with high brightness, low contrast, and low saturation;
2. Due to the use of too many "Blue Archive" as material, some pictures will appear with a halo on the head of the character.
--
Because the training is based on Anything V5 Ink, please use a model trained based on Anything V5 Ink or merge the models when using it, otherwise the image may collapse.