← 返回提示词图库

PROMPT RECORD图像记录

GPT Image 2 on ChatGPT 米兰,全日时光|深夜相拥 提示词 ⬇️⬇️⬇️ 一张亲密、抓拍的深夜照片,照片中十位约20岁的年轻亚洲女性——六位日本人、两位韩国人、两位中国人—...

GPT-Image-2 ToroJushiAi Thu Jul 30 06:00:01 +0000 2026
一键生图 查看来源

中文说明

GPT Image 2 on ChatGPT 米兰,全日时光|深夜相拥 提示词 ⬇️⬇️⬇️ 一张亲密、抓拍的深夜照片,照片中十位约20岁的年轻亚洲女性——六位日本人、两位韩国人、两位中国人——深夜在现已安静的米兰大教堂广场(Piazza del Duomo)附近彼此相拥,形成一个温暖的大合影,大教堂的尖塔在黑暗天空的背景中柔和地发光,广场上空无一人。 其中三到四位面朝镜头的女性展现出清晰、彼此可清晰区分的东亚面部特征,带着真诚温暖的微笑;其余人在拥抱中或被侧面捕捉,或被部分遮挡,符合一个不摆拍的大合影的自然构图。她们的肌肤在环境光照到可见面部时呈现出真实的次表面散射效果,镜面微高光柔和温暖,自然质感得以保留。 可见人物的发型各异,在亲密的拥抱中,几缕发丝自然地散落。她们穿着休闲的晚间混搭——宽松的外套内搭吊带...

原始 Prompt

GPT Image 2 on ChatGPT

Milan, All Day Long|Embrace in Late-night 

Prompt ⬇️⬇️⬇️
An intimate, candid late-night photograph of ten young Asian women, around 20 years old — six Japanese, two Korean, two Chinese — embracing each other in a warm group hug near the now-quiet Piazza del Duomo late at night, the cathedral's spires glowing softly in the background against the dark sky, the piazza otherwise empty.

The three or four women whose faces are turned toward the camera show clear, individually distinct East Asian facial features with genuine warm smiles; the rest of the group is captured from the side or partially obscured within the embrace, consistent with the natural composition of an unposed group hug. Their skin shows realistic subsurface scattering where the ambient light catches the visible faces, specular micro-highlights soft and warm, natural texture preserved.

Hairstyles vary across the visible figures, a few strands displaced naturally within the closeness of the embrace. They wear a relaxed evening mix — oversized coats layered over slip dresses, some scarves loosely draped — a lived-in, end-of-night softness to the styling.

The lighting is low ambient ISO-pushed night light from the distant illuminated cathedral spires, soft and low-contrast, with a quiet dark blue-black background enveloping the group. Subtle ISO 400 film grain sits in the deep shadow areas, giving the scene a grainy, intimate late-night photographic quality. Aspect ratio 2:3. No watermark, no text overlay, not cartoon, not illustration, not anime.