Text to imageSurreal & fun
The exact text-to-image prompt behind this picture, made with GPT Image 2.5 Sunburst.
Vertical 4:5 portrait frame. An open worn passport on a wooden desk where each colorful stamp bursts upward into the miniature place it stands for: a tiny lighthouse on a cliff, a desert caravan, a canal city with gondolas, macro photography with warm desk lamp light, whimsical and detailed, no readable text.
Prompts are kept in their original English to help reproduce the example. Copy one as it is, or adapt it for your project.
Opens GPT Image 2.5 Sunburst with this prompt and the settings below filled in.
135 tested GPT Image prompts →
The exact price shows before you generate.
macro photography with warm desk lamp light.no readable text. Leave that in, or the model may add random lettering.Inspired by an idea from @iamaiistudio, collected in Evolink-AI/awesome-gpt-image-2-API-Prompts under CC0 1.0. PopNow wrote this prompt and made the example above.
Scene prompt guide
An open, weathered passport transforms into a tabletop diorama where stamped ink sprouts into three-dimensional travel destinations. GPT Image 2.5 Sunburst balances physical tactile textures like textured paper and desk wood with miniature architectural scale. The concept works because macro depth-of-field seamlessly blends the flat stamps with the miniature scenes rising above them.
Vertical 4:5 portrait frame. An open worn passport on a wooden deskThis establishes the physical surface, orientation, and context of the shot. Removing the desk or the worn quality makes the passport look like a sterile digital graphic floating in void space.
where each colorful stamp bursts upward into the miniature place it stands for:This instructs the AI to treat 2D surface ink as the foundation for 3D miniature geometry. Without this mechanic, the generator will simply draw standard flat ink marks on the paper.
a tiny lighthouse on a cliff, a desert caravan, a canal city with gondolas,These concrete subjects give the model diverse micro-environments to render with distinct palettes and shapes. Omitting them leads to ambiguous, generic fantasy mounds rather than readable travel motifs.
macro photography with warm desk lamp light, whimsical and detailed, no readable text.This sets close-focus miniature optics, cozier evening lighting, and suppresses garbled AI letterforms on the passport pages. Altering this can result in flat studio illumination and distracting gibberish typography.
The macro perspective and tabletop illumination establish a cozy, believable miniature illusion.
Replaces the passport with a weathered leather journal and morning window light for a historic expedition feel.
Vertical 4:5 portrait frame. An open antique leather journal on a mahogany desk where vintage hand-drawn maps burst upward into miniature 3D landscapes: a snow-capped mountain pass with pine trees, a stormy coastline with a wooden galleon, macro photography with soft morning window light, whimsical and detailed, no readable text.
Switches to an overhead vantage point and twilight atmosphere, emphasizing how the 3D landmarks rise up toward the lens.
Vertical 4:5 portrait frame. Overhead bird-eye macro view of an open worn passport on a rustic oak desk, colorful stamps rising vertically into miniature dioramas: a tropical island with palm trees, a medieval walled castle, macro photography under cool twilight lamp light, whimsical and detailed, no readable text.
Shifts the time to midnight with glowing miniature landmarks illuminated by fairy lights.
Vertical 4:5 portrait frame. An open worn passport on a dark walnut desk at midnight where luminous stamps burst upward into glowing miniature scenes: an aurora borealis over snowy tundra, a neon-lit Tokyo crosswalk, macro photography illuminated by tiny fairy lights, whimsical and detailed, no readable text.
Keep the ratio at 4:5 portrait mode because vertical framing gives the miniature landmarks room to grow upward without cropping the passport margins. Set quality to high so GPT Image 2.5 Sunburst renders clean architectural micro-facets, tiny stamp lines, and wood grain without blurring. If adapting this concept for wide desktop wallpapers, adjust the aspect ratio to 16:9 while adding empty desk space on either flank.
Adjust the scale adjectives in the prompt. Replace 'miniature' and 'tiny' with 'dollhouse-scale' or 'fingertip-sized' to shrink them, or describe them as 'towering pop-up sculptures rising above passport edges' to make them grander.
The model sometimes groups multiple location descriptions into a single hybrid landmass. To fix this, explicitly assign one vignette to the left page and one to the right page so the AI clearly separates their boundaries.
Yes, you can swap generic descriptions for distinct monuments like the Eiffel Tower or the Taj Mahal. Keep the descriptor 'miniature' before each landmark name so the model maintains the micro-scale diorama aesthetic.
