**Creation: King Pingxi.
Champion drawing requirements: like myself, pretty, long legs, high ponytail, red basketball uniform, etc.
There are three elements of creation: the basic model, the LoRa model, and the prompt words (prompt words and reverse words).
Here's a quick rundown of the 7 creative steps.
1.Collect Champions**.
Requirements: Clear face map, multiple angles, 10-30 photos** is appropriate.
Two main images are given below**.
2.Create a dataset.
Upload the champion avatar** to the dataset.
3.Train the champion LoRa model.
Purpose: The generated diagram is more in line with the characteristics of the champion himself.
Key points: The basic model selection is the same as that used for self-plotting.
In this example, the base model is selected as the blind box model.
Subject Selection: Character.
4.Start making a picture.
Model: Select the blind box base model.
LoRa model: Select the champion model you trained.
Prompt words: basketball, fair-skinned, delicate dress, long hair in a high ponytail, black hair, full body, female, real, cute.
Reverse words: wrong hand index, bad hand, uncoordinated proportions.
The renderings are as follows.
Note: There is a gap between the generated ** and the championship requirements (and my portrait, tall stature, long legs).
5.Optimized prompt words.
Prompt words: Chinese women's basketball team, (red basketball uniform: 1.)2), Basketball shorts, size 13 basketball uniform, tall figure, coordinated body proportions, long legs, coordinated body proportions, delicate dress, front, delicate face, ponytail, black hair, female.
Reverse words: wrong hand index, bad hand, uncoordinated proportions.
Instructions:12 is the weight of the prompt word "red basketball uniform".
The renderings are as follows.
Description: Tall figure, long legs meet the requirements. But his face is flawed, and he doesn't look like him.
6.Face refinement.
Train the model using Stable Diffusion EasyPhoto.
7.In the end, the picture was satisfied.
One person cheers for the graduate school entrance examination