Miniature Self Portrait

the prompt

Use the uploaded TARGET IMAGE as the exact master reference for the composition and visual concept.

Use the uploaded REFERENCE IMAGE only as the identity reference for the female character.

Create a highly detailed, photorealistic cinematic portrait of the same concept shown in the target image:

A young woman wearing a soft gray knitted sweater, smiling naturally while looking toward a tiny 3D miniature version of herself standing on her open palm. She points toward the miniature with her other hand. The miniature is a cute, polished 3D character version of the same woman.

IDENTITY:
- Use the reference image to accurately reproduce the woman's facial identity.
- Preserve her natural facial structure, eyes, nose, lips, eyebrows, skin tone, and recognizable features.
- Use the reference woman's hairstyle, hair color, hair texture, and overall hair identity.
- Apply the same identity consistently to both the large woman and the miniature character.
- The miniature should clearly look like a tiny stylized version of the same woman.

TARGET PRESERVATION:
- Keep the exact overall composition and camera framing of the target.
- Keep the woman's gray knitted sweater.
- Keep the same hand positions and pointing gesture.
- Keep the miniature standing naturally on the open palm.
- Keep the miniature's gray sweater, blue jeans, and white sneakers.
- Keep the deep blue cinematic background.
- Keep the same lighting direction, soft rim lighting, shadows, depth of field, and realistic photography style.
- Preserve realistic skin texture, natural hair strands, fabric texture, and facial details.
- Keep the vertical portrait format and cinematic advertising-style quality.

IMPORTANT:
The reference image is ONLY for the woman's face and hair identity.

Do NOT copy the reference woman's clothing, background, pose, environment, or accessories.

Do NOT change the target composition.
Do NOT add text, logos, watermarks, or extra objects.
Do NOT over-smooth, beautify, or alter the woman's natural face.

FINAL RESULT:
The output should look like the exact target concept, but with the identity and hairstyle of the woman from the uploaded reference image. Both the large woman and the miniature must have the same recognizable identity.

Image: 2
Use the uploaded TARGET IMAGE as the exact master composition.

Use the uploaded REFERENCE IMAGE ONLY as the identity reference for the person.

Create the same visual concept as the target:
a photorealistic cinematic portrait of a large adult person wearing a gray knitted sweater, looking toward and pointing at a tiny miniature version of themselves standing on their open palm. The miniature person should have the SAME face identity and SAME hairstyle as the reference person, adapted naturally to the miniature scale.

IMPORTANT:
- Replace the main adult person's face and hair with the face and hair from the uploaded reference image.
- The miniature character must also have the same recognizable face identity and matching hairstyle.
- Preserve the target composition, pose, hand positions, body proportions, clothing, sweater texture, miniature character's outfit, facial expressions, camera angle, framing, lighting, shadows, depth of field, and overall photorealistic quality.
- Keep the miniature character standing naturally on the adult's open palm.
- Keep the adult pointing toward the miniature exactly as in the target.
- Keep the dramatic deep-blue studio background and cinematic rim lighting exactly consistent with the target.
- Do NOT copy clothing, background, pose, or body structure from the reference image.
- The reference image is ONLY for face identity and hairstyle.
- Do not add glasses, accessories, logos, text, watermarks, or extra objects unless they already exist in the target.
- Keep realistic skin texture and natural facial details. Do not over-smooth or beautify the face.
- The final result must look like a professionally photographed, photorealistic advertisement.

The uploaded target image is the MASTER.
Only the person's identity/face and hair should come from the uploaded reference image.
Everything else must remain visually faithful to the target image.