这里看到一句很形象的话:
出处:https://www.bilibili.com/video/BV1Pg4y1t74E
就是魔法卷轴,别人把魔法阵刻上去你直接激活就能用,不用念长长的咒语了.
其实有点类似封装?
embedding其实就是快速添加prompt.
首先在C站过滤一下Model类型,选择Textual Inversion
image.png
然后找一下想要的模型.
例如下载Style Sylva Magic.
https://civitai.com/models/7523/style-sylva-magic
在CivitAI中,下载这个模型.
image.png
下载后会在embedding文件夹下.
image.png
然后点击 show extra networks,Textual Inversion就会出现.
image.png
这里随便使用一个AIPRM for ChatGPT生成的prompt.
A rising sun over a misty forest, casting a warm golden light on the trees and foliage. The mist is slowly lifting, creating a hazy and ethereal atmosphere.
The environment is quiet, with only the sound of rustling leaves and chirping birds.
The atmosphere is mystical and dreamy, with a sense of calmness and wonder.
3D rendering, with a focus on creating a realistic forest environment with detailed textures and lighting, using a soft color palette to capture the misty and ethereal atmosphere.
随便选择一个,例如EasyNegative(这里是自带的,不是刚刚下载的Sylva Magic),把光标移到到Negative prompt中,然后点击EasyNegative,即可快速添加到Negative prompt中.
image.png
在保持使用seed值为3136315253,对比如下.
未使用embedding.png 使用embedding.png
使用style-sylvamagic的话,就是点击对应的选项,在prompt中快速添加单词.
style-sylvamagic 使用style-sylvamagic
参考链接:
【基础03】stable diffusion的embedding是个啥?有啥用?怎么用?有啥效果?小白也速通!独家分享
网友评论