My first approach to make a new and completely fictional character. A beautiful south-east asian woman. Trained with other people and objects to get more variation. Tested with several checkpoints and other LoRAs - works as intended :-) Please let me know if not - and please post images.

I learned - less is more for character training, so I trained with 26 synthetic images and 1024 resolution with the standard settings of ai-toolkit and "shianvu" as the trigger word.