Candidate 06 – Z-Image Turbo Character LoRA

An original, fully AI-generated South Asian woman. She isn’t based on any real person. The LoRA is built for a consistent face and identity across scenes, outfits, angles and lighting.

Trigger word:  cand06_zit 

Recommended settings

• Base model: Z-Image Turbo

• LoRA weight: 0.8 – 1.0

• Steps: 8 – 10

• CFG: 1

• Sampler: Euler

• Resolution: 864×1152 (portrait) or 1024×1024

Prompt tips

• Start with:  cand06_zit, a young woman, small black bindi on her forehead, ... 

• She has long dark wavy hair, warm brown skin, dark brown eyes and a slim natural build.

• The bindi is part of her signature look. Leave it out of the prompt if you don’t want it.

• She works well in close-up portraits, three-quarter shots, profiles, sarees and casual wear, both indoors and out.

Training data

• 26 curated synthetic images: front, three-quarter and profile angles, a range of expressions, and indoor, outdoor and studio lighting.

• Every image is hand-captioned with the trigger word.

Example prompt

 cand06_zit, a young woman, small black bindi on her forehead, wearing a red silk saree, standing in a garden, three-quarter body shot, soft natural light, photorealistic 

Notes

• For a different outfit or setting, describe it clearly. The LoRA keeps the face and leaves styling to your prompt.

• Lower the weight to 0.7 if images look over-baked or stiff.