This Model is trained in Flux, I’m experiments on the training settings, trying to get a version I’m happy with.

This is the baseline for a character I created: Su-Young. She is from a fictional, cyberpunk world. Based on some creative writing I did, ChatGPT created this prompt:

“A high-resolution portrait of a young East Asian woman with a blend of approximately 75% Korean and 25% Thai ancestry. She has a slender, heart-shaped face, soft but defined cheekbones, almond-shaped dark brown eyes, a smooth complexion with natural warmth, and straight black hair that falls past her shoulders. Her expression is neutral or gently contemplative, with a calm, introspective gaze that suggests quiet strength and emotional depth. She wears a fitted charcoal gray tank top and appears in a well-lit environment with a neutral white background. Her features are balanced and symmetrical, projecting a grounded, authentic beauty shaped by both resilience and sensitivity.”

From that prompt we generated the training images for Su-Young. I used those images to train this LoRA. My intent is to be able to use Su-Young in more violent, or intimate scenes than ChatGPT allows.