Prompt từ kênh
Khung hình a76
Copy prompt, dán vào ChatGPT và đính kèm ảnh của bạn. Nó sẽ chụp lại đúng khung hình này với bạn.

Prompt
Use the uploaded personal reference photos as the only reference for identity, face, facial expression, body proportions, exact hair length, natural hair color, hair density, and real hair texture. The mood image is only for styling, seated park-bench pose, black blazer and white top outfit, light blue wide-leg jeans, black sunglasses, green Parisian park, Eiffel Tower background, bright daylight, lighting, color palette, and composition. Do not copy the mood-image face, expression, body, hair length, hair color, hairstyle, hair texture, or proportions. Create a realistic candid travel photo of the same person sitting naturally on a worn dark-green park bench in Paris with the Eiffel Tower rising large in the background. The photo should feel spontaneous and unplanned, like a friend took it during an ordinary sunny day, not like a professional fashion photoshoot. Place the person slightly to the right side of the vertical frame; let the Eiffel Tower fill the upper background without covering the face. She is sitting casually with one leg bent and raised on the bench, the other leg lowered toward the ground, her hands resting naturally near her lap, shoulders relaxed, and her head turned slightly toward the camera. Expression and emotion must come from the uploaded personal reference photos, not the mood image. Outfit: black blazer with realistic fabric folds, simple white top, light blue wide-leg jeans, black sunglasses, and no visible branding. Preserve the exact identity, facial features, natural skin texture, body proportions, and expression from the personal reference photos. Do not change the face, add makeup, enlarge the lips, reshape the nose, smooth the skin, or make the face more symmetrical. Preserve the exact hair length, natural color, density, structure, and real texture from the uploaded personal reference photos. Hair may fall naturally around the face and shoulders, but it must keep the exact personal length, color, density, and texture. Do not copy the mood-image hairstyle or hair texture. The park scene should include a worn dark-green wooden bench, green grass, trimmed hedges, flower beds, trees, a gravel path, natural garden details, and the Eiffel Tower rising prominently behind the person with realistic metal textures and perspective. Location: a sunny Parisian park with the Eiffel Tower in the background, green lawn, trimmed hedges, flower beds, trees, gravel paths, and a worn dark-green wooden bench. Keep the city park background realistic and atmospheric but secondary, not a studio and not overly staged. Any signs, labels or printed text must be blurred, abstract, or unreadable. Photo quality: highest available quality and maximum detail, natural iPhone 17 Pro Max or compact-camera realism, universal vertical 4:5 crop suitable for Instagram and Pinterest, bright direct daylight, realistic natural shadows, true skin texture, visible natural pores, realistic Eiffel Tower, bench, grass, foliage, blazer, denim and sunglasses details, detailed architectural and garden textures, slight handheld framing, natural smartphone color and exposure, subtle sensor grain, mild imperfections, no haze, no vintage blur, no cinematic grading, no AI-polished look. Negative prompt: no copied mood-image face, no copied expression, no copied body, no copied hair, no changed identity, no changed body proportions, no changed hair length, no altered hair color, no altered hair texture, no beauty filter, no beauty retouching, no skin smoothing, no face reshaping, no body reshaping, no heavy makeup, no fake eyelashes, no enlarged lips, no plastic skin, no waxy skin, no perfect symmetry, no over-posed fashion shot, no studio lighting, no missing Eiffel Tower, no missing green bench, no missing black blazer, no distorted Eiffel Tower, no distorted bench, no distorted hands, no extra fingers, no warped arms, no warped legs, no bags, no backpacks, no wallets, no belts, no bag-shaped objects, no CGI, no illustration, no readable text, no visible logos, no watermark, no screenshot UI.
Prompt bằng tiếng Anh — mô hình hiểu chính xác hơn. Không cần dịch.
Làm gì với nó
- 1
Nhấn «Copy» — toàn bộ prompt vào bộ nhớ tạm.
- 2
Mở ChatGPT và dán vào một đoạn chat mới.
- 3
Đính kèm hai ba tấm ảnh của bạn rồi gửi. Khung hình sẽ có sau một phút.
Thêm prompt
Thư viện còn 429 ảnh
Ba mươi prompt mới mỗi tuần. Prompt nào cũng kèm ảnh mẫu.
Cùng bộ ảnh này





