社交媒体帖子 - Urban Giant Mystery
Prompt previewFull prompt
Original author textThis public page preserves the author's original prompt. Review the preview and any published evaluation, then copy the text or open it as a draft. Choose the model, references and current cost in the creation workspace before generating.
Create an extraordinarily photorealistic cinematic photograph of a {argument name="setting" default="normal modern city at blue hour"}, captured from street level with a professional full-frame camera and a 35mm lens. At first glance, everything looks completely normal: realistic pedestrians walking, taxis and cars moving through traffic, storefronts, streetlights reflecting on slightly wet asphalt, office windows glowing naturally, subtle atmospheric haze. But hidden between the skyscrapers is something absolutely impossible: A {argument name="subject" default="colossal HUMAN-LIKE FIGURE"}, hundreds of meters tall, is partially visible behind the buildings. The figure must look almost completely human rather than monstrous — realistic human skin, subtle pores, tiny imperfections, faint veins, natural facial structure, realistic eyes, slightly messy hair, believable anatomy. Only portions of the enormous person are visible between the buildings, making it initially difficult to understand what the viewer is seeing. One enormous {argument name="interaction" default="hand is resting naturally against the side of a skyscraper"}, with individual fingerprints, wrinkles and skin texture visible. The enormous face is partially obscured by clouds and buildings in the far background, looking calmly down toward the city. CRITICAL: Make the giant blend naturally into the environment. The lighting on the giant must perfectly match the city's lighting, atmospheric perspective, haze, reflections and color temperature. Buildings in front of the figure should partially occlude it. Some atmospheric fog should pass between the giant and the camera. People on the street should NOT be dramatically reacting. Most pedestrians should appear completely unaware, creating an eerie sense that the viewer has discovered something impossible that everyone else somehow missed. Extreme photorealism, physically accurate lighting, realistic global illumination, natural shadows, realistic reflections, atmospheric depth, subtle lens imperfections, realistic skin microtexture, high dynamic range, cinematic depth of field, documentary photography aesthetic, believable scale, extremely detailed architecture. Composition should create an immediate visual mystery: the viewer should need a second look to realize that the shape between the skyscrapers is actually an enormous human being. No fantasy illustration aesthetic, no cartoon appearance, no CGI look, no exaggerated monster features, no glowing eyes, no text, no logos, no watermark. Make it look like an accidentally captured photograph from a real camera — so convincing that it could initially be mistaken for a genuine photograph. 8K-level detail, ultra-high resolution, natural colors, realistic photographic grain, masterpiece-level realism.
Adjust the model, settings and references before generating.
Use prompt →Prefill first, then confirm generation and the current cost.
GPT evaluation
GPT-6.1-sol
AI evaluation for creative referenceModel: GPT-6.1-sol Overall score: 8.4 / 10 Prompt quality: 8.5 / 10 Image quality: 8.7 / 10 Prompt–image alignment: 8 / 10 建议小幅优化后上线:示例图的城市纵深、冷暖光影与巨人尺度具有较强吸引力,但巨手过于显眼、握楼姿态偏强,削弱了提示词要求的“第二眼才发现”的隐秘感。
Community reviews
No creator reviews yet.
View and review →