Two versions are included, one with bolder lines, another...is just very cute.
Trained on a balanced dataset, so the eyes and mouth wouldn't be taken over by other expressions. This model already comes with a range of default expressions which works in high accuracy. But you can find some LoRAs for non-typical expressions to get amazing results.
Training images are also uncropped, so you wouldn't need to wonder where is the lower body.
You get what you prompt. it can be used on boys and animals as well, the dataset is quite diverse on training subjects, and won't just generate a specific clothing and hairstyle.
If you like this model and want to see more of them, please follow me and leave a positive review. Thanks