以下是V2.5与V2和V1的对比(V1,V2,V2.5):
Here's how V2.5 compares to V2 and V1 (V1,V2,V2.5):
以下是三个版本图生图的对比:
Here's a comparison of the img2img between these Versions:
模型最开始是用来生成二次元正太的(φ(゜▽゜*)♪),但是意外
发现对于图生图特别友好,尤其是照片之类的,下面是两个示例:
(The model was orignally used to generate anime boys((′д` )…彡…彡,but it was unexpected have goods results about img 2 img,here is some example)
If there is a translation error, please let me know in the comments
(tags:{{masterpiece}},best quality,illustration,1 male kid,prince,rose,floating short hair,yellow hair,long yellow scarf,green clothes,rose,closeup,chibi)
针对图生图的示例如下:
There are some example of img2img
(以下所有真人照片均出自this-person-does-not-exist)
(All of the following live photos are from this-person-does-not-exist)
推荐的图生图参数:steps:30,cfg:11,denoising strength:0.52,使用clip进行反推文本
Recommended img2img parameter: steps:30,cfg:11,denoising strength:0.52,Use clip to reverse text