
Photorealistic magical old-town street at night, wide 16:9 panoramic composition, one-point perspective, cinematic, ultra-detailed. Edge to edge, no frame, no border. STREET: the entire lower half is a mosaic cobblestone pavement of irregular rounded stones separated by dark joints, split into a central raised walkway with a low kerb on the left. Every stone glows from within like stained glass or opal — saturated turquoise and emerald flowing into cobalt blue, magenta, violet, warm amber and orange, each stone a separate luminous cell with wet cracks and bright glossy reflective highlights. TOWERS: at the vanishing point, dead centre, stands a grand blue-grey church with a large ribbed dome and a slender arcade, and beside it a tall Baroque bell tower of four stacked tiers with arched openings, a cupola and a gold finial, its windows glowing warm amber from inside. Pale warm streetlamps burn along the facade. LEFT: a curved row of ornate old-European buildings in warm brown stone with lit arched windows, wrought-iron balconies and carved cornices. A black cast-iron railing runs along the kerb. A dark angular lamppost with a glowing lantern stands at the left edge, and a bare crooked tree with sparse brown leaves rises above the rooftops. RIGHT: a mirrored row of tall classical facades in the same warm stone, with a row of streetlamps on curved arms marching into the distance, glowing amber, and deep arched doorways lit gold from within. SKY: the upper third is a deep cosmic teal and turquoise night sky, richly nebulous and swirling, scattered with countless tiny stars. A pale meteor trail streaks diagonally across the upper right. Low warm clouds glow peach and amber behind the towers at the horizon. PALETTE: turquoise, cobalt, emerald, magenta, violet, warm amber, brown stone, near-black shadow. Sharp focus, glossy wet highlights, high dynamic range, clean and unmarked.
This image was generated with GPT Image 2.5 by OpenAI. The prompt on this page is the exact text used to create it.
The example was generated with GPT Image 2.5 using: 34:19 · Cinematic · Architecture · Fantasy. Reusing these settings gives the closest match.
Copy the prompt as-is first, then replace only the subject while keeping the style, lighting, and composition keywords. Generate a few variations — outputs differ between runs.
Yes. Prompts work best as starting points — adjust descriptors, aspect ratio, or style tags to make the output your own.