用了下网站上的模型,感觉都不是很适合我,所以我就练了这个。"kiana"是触发词,“ponytail,bare legs,flower,dress,diamond shaped pupils”获得更稳定的输出,也可以描述其默认服装来增加稳定性。默认服装权重推荐0.8,想要换装可以适当降低权重且保留触发词。训练时为纯黑背景,因此生成图像时需要添加背景描述词。
After using the model of the character on the website, I didn't think it was very suitable for me, so I practiced this. 'kiana' is a trigger word, while 'ponytail, bare legs, flower, dress, diamond shaped pupils' can provide more stable output or describe its default clothing to increase stability. The default clothing weight recommendation is 0.8, and if you want to change clothes, it is recommended to low the weight and keep the trigger word. During training, the background is pure black, so background descriptors need to be added when generating images.
仅作交流学习使用。
For communication and learning purposes only.