Prompt template
Istanbul Bosphorus Night Food-Stall Street Photo (Sign-Text Test)
An ultra-realistic 35mm night street photograph of a tiny DIY caravan köfte stall on the Bosphorus waterfront, with a hand-painted neon sign whose name repeats on the waiter's shirt, making it a sign-text stress test. Shared by @ozansihay, who ran the same prompt through Nano Banana 2.1, GPT Image 2.5, Grok Imagine 2.0 and Ideogram 4.5.
Start from an example
Pick an example to load its exact settings, then make it yours.
Make it yours
Only the parts worth changing are shown. The prompt updates as you go.
Your prompt
Ultra-realistic night street photograph on the Istanbul Bosphorus waterfront. A tiny makeshift caravan food stall, slightly battered metal sides, steam rising. Hand-painted neon-style rickety sign reading “Köfteci Buğra Usta” in Turkish, glowing uneven cyan and warm amber, wires and brackets visibly DIY. Inside the open caravan, the usta grilling köfte over charcoal, orange ember glow on his face and apron. One waiter in a simple dark shirt stands in front of the stall; the same “Köfteci Buğra Usta” lettering is printed on the back of his shirt. Plastic stools and small low tables outside, a few locals eating, soft conversation. A plastic water jug (damacana) near the counter. Behind them, dark Bosphorus water with distant city lights and a faint ferry silhouette. Wet asphalt reflections, streetlamp spill, shallow depth of field, natural film grain, shot on 35mm f/1.8, candid documentary look, no CGI look, no text errors on signs.
When to use it
Use when you want a candid documentary-style night street-food scene with legible custom signage, or a fair side-by-side test of how image models render text on signs and clothing.
How to get the best result
- Change the sign and shirt lettering together so the waiter's shirt matches the stall sign; keep the curly quotes around the name.
- Swap location, food and backdrop as a set so the scene stays coherent (stall, cook, waiter, water behind).
- Keep the 35mm f/1.8 / natural film grain / 'no CGI look' tail; it drives the candid documentary feel.
Watch-outs
- Diacritics (ğ, ö, ş) are where models misspell signs; check both the sign and the shirt lettering.
- Models may add extra garbled background signage despite 'no text errors on signs'.
Related templates
- Time-Freeze Diner Spill Escape (Seedance 2.5 Image-to-Video)Use when you have a character reference image or model sheet and want a comedic time-stop moment with timed beats and suspended-liquid physics.
- Titan Shoreline Guardian Summons Alien Creature (4-Shot Kling Sci-Fi)Use when you want a compact multi-shot sci-fi standoff with a creature reveal, written as numbered shots for a text-to-video model.
- Period-Drama Storm Crowd Reveal (Multi-Shot Seedance)Use when you want a multi-shot period-drama storm sequence with documentary-real skin texture and a crowded final reveal framed on the hero.
- Fish-Cam Trevally Into Underwater Neon PartyUse when you want an unbroken fish-mounted action-cam journey from a tropical lagoon into a surreal underwater party reveal.
- One-shot Maglev Titan Valley Train Journey (Seedance 2.5)Use when you want a single unbroken handheld/phone POV train-window journey through escalating fantastical landscapes with sparse timed dialogue and no music.
- Toddler Learns to Milk a Cow — Handheld Farm Home Video (Image-to-Video)Use when you have a character reference sheet of a child in a costume and want a cute, phone-filmed 30-second story clip with a fail-then-success arc and spoken lines.
Source and attribution
- Template source
- https://x.com/ozansihay/status/2107544398531637407
- Example cases by
- Ozan Sihay