Try size: 768 x 1024
Weights: 0.5~1.0
Clip skip: 2
Training material from bing and AI generate pics
Pictures from kantanmix_v16LCMTurbo
Note: This is the final version, but there are still some issues, such as two ATMs often appearing when two people are together. However, the number of hats appearing has decreased, and red hair and red clothes have also decreased. Finally, the girl facing left and right is closer to balance. (v1.0 Final)
Note: The girl facing left and right more evenly, and the appearance of warm colored is reduced. The disadvantage may be that there will be occasional strange pictures, such as close-ups of hands, so the description of the girl could be increased. 1boy's hat appears more frequently and you can describe the his hair color. (v1.0 Fix)
Note: I didn't train on scenes outside of stores, so if you want to simulate an ATM on the roadside, please lower the weight of LoRA or increase the description of prompt. The girl is usually facing the viewer's left side, because of the English font, I can't flip it horizontally. If you want the girl facing to right side. and reducing the LoRA weight maybe also improve this. (v1.0)