Prompt 1
Image 1 is the identity-neutral chroma-green scene template. Image 2 is the identity reference for Mila. Pose authority: Image 1 controls all scene geometry. Preserve the primary subject's exact shoulder line, neck rotation, head yaw, head pitch, head roll, and eye-gaze direction from Image 1 even when they differ from Image 2. Image 2 controls identity only; never straighten, front-face, recenter, or otherwise normalize the head pose to match Image 2. Scene: Medium shot of the person in the uploaded reference image (DO NOT TOUCH FACE) seated on a dark couch, body angled slightly toward the viewer with torso upright and both legs bent and drawn up on the cushions. One hand rests on the couch beside the hip with fingers extended; the other hand rests on the raised thigh. Indoor room with a beige wall, corkboard of stickers, and a wall calendar behind; even ambient indoor lighting. Wardrobe: long-sleeve ribbed top tied in a front knot, brown plaid mini skirt, light blue underwear, heart pendant necklace, and rings on the fingers; no footwear visible. Face: Head faces the camera with a slight tilt. Eyes are open and directed at the camera, brows relaxed, mouth closed in a soft slight smile with no teeth visible. Wardrobe: long-sleeve ribbed top tied in a front knot, brown plaid mini skirt, light blue underwear, heart pendant necklace, and rings on the fingers; no footwear visible. Preserve every visible garment and footwear item exactly. Replace only the primary chroma-green mannequin completely with Mila from Image 2. Replace every secondary chroma-green mannequin with a separate, distinct new adult identity that broadly preserves that secondary subject's visible styling, clothing state, pose, scale, and scene role. Never copy or duplicate Mila onto a secondary person. Use Image 2 exclusively for facial identity, facial proportions, skin, eye appearance, and hair identity. Do not copy any normalized mannequin facial proportions as identity, and do not reconstruct the original source person's identity. Match the stated action, camera relationship, face, hand positions and interactions, visible object inventory, person count, pose, crop, clothing state, lighting, background, and composition. Do not add an implied phone, camera, mirror, held prop, person, garment, or text that the Scene does not show. Remove all chroma-green mannequin material, green spill, plastic texture, blank-face artifacts, and placeholder traces from every subject. Preserve legitimate green scene objects and clothing. Photorealistic, natural skin texture, anatomically coherent hands and body, no text.
- Engine
- Seedream V4.5 Edit
- Dimension
- Portrait (3:4)
- Gender
- Female
- Ethnicity
- White
- Age
- Twenties









