Model description

This LoRA represents the character NINA, trained for FLUX models to generate a consistent identity across different scenes.

What it does / what it’s for

It allows you to generate NINA in varied outfits, poses, and environments using natural language prompts, while keeping her visual identity stable.

What it’s good at

• Consistent character identity

• Responding well to detailed prompts (clothing, pose, camera)

• Working across different compositions (close-up to full body)

Limitations

• May reflect dataset biases (poses, lighting, style)

• Less reliable with extreme angles or highly stylized prompts

• Identity can weaken with conflicting or vague prompts

Usage

• Use the trigger word NINA

• Prompt in natural language (FLUX-style)

• Explicitly describe non-identity elements (clothes, pose, setting)

Not recommended

• Vague or contradictory prompts

• Styles or scenarios not represented in training