I wanted an easy way to generate fun and stylish social avatars,

Before preparing to learn, first of all, 20 ideal pictures were generated with a unified style of cartoon characters, and then let the machine learn the picture structure and posture. Through this learning, everyone can easily produce 512*512 photos of the overall description of the characters, reducing the errors caused by the details of the hands. It is recommended that you use the trigger word: HEADIMAGE

If you like this kind of photos, I will generate more portraits for you, which are suitable for family festivals, social expressions, and making souvenir books. I hope you like them.