IDENTITY PRESERVATION INSTRUCTION:
Using the uploaded photo as reference, preserve the person's exact facial features, face shape, skin tone, and unique characteristics while applying the eye color transformation. Maintain the original eye shape, eyelashes, and surrounding skin texture to ensure the person remains fully recognizable.

TRANSFORMATION DESCRIPTION:
Change the eye color to [insert desired color, e.g., vivid emerald green, deep sapphire blue, warm amber, or icy gray]. The color change should focus exclusively on the iris, leaving the pupils dark and natural. The new color should look vibrant yet realistic, with natural variations and depth within the iris texture. Ensure the transition from the sclera (white of the eye) to the iris is smooth and anatomically correct.

STYLE AND AESTHETIC PARAMETERS:
- Art style: Photorealistic, high-definition portrait photography.
- Lighting: Maintain the original lighting conditions of the photo. Add subtle catchlights in the eyes that reflect the new color, enhancing depth and realism.
- Color palette: The new eye color should harmonize with the subject's skin undertone and hair color. The surrounding facial colors should remain unchanged.
- Background: Keep the background identical to the original photo to maintain context.
- Texture: High detail on iris patterns (stroma, crypts, and furrows) to make the color look natural and not like a flat overlay.

TECHNICAL QUALITY PARAMETERS:
- High resolution, 4K, detailed
- Sharp focus on face and specifically the eyes
- Professional quality, studio-grade retouching
- Aspect ratio: Match the original photo's aspect ratio (or default to 1:1 for portraits)
- Natural skin texture preserved; no smoothing or plastic look

NEGATIVE PROMPT GUIDANCE:
Avoid: distorted eye features, mismatched eye color (e.g., color bleeding onto skin or sclera), unrealistic proportions, blurry face, loss of likeness, unnatural flat color overlays, reflections that don't match the lighting, double pupils, asymmetrical eyes.

USER CUSTOMIZATION:
{additional_context}

---

### Technical Execution Tips for AI Generators:

**For Midjourney:**
Use the image upload feature with the prompt. Add parameters like `--ar 2:3` or `--ar 1:1` depending on your photo. Use `--style raw` for more photorealistic results.

**For Stable Diffusion (WebUI/ComfyUI):
Use the img2img mode with a low denoising strength (0.3-0.5) to preserve identity. Include the negative prompt provided above. Use a checkpoint model optimized for portraits (e.g., Realistic Vision, Juggernaut).

**For DALL-E 3 (via ChatGPT or API):
Upload the photo and paste the prompt. DALL-E is good at following natural language, so emphasize "keep the face exactly the same except for the eye color."

**For Flux:**
Use the image-to-image feature. Flux handles lighting and identity preservation exceptionally well. Keep guidance scale moderate (3.5-4.5).