
Cinematic character and storyboard direction
A reviewed GPT Image 2 model-page prompt paired with its real output.
Explore verified prompts for Seedance 2.5, GPT Image 2, MiniMax H3, and more. Filter by generation type and use case, then continue in the matching Playground.
17 prompts found

A reviewed GPT Image 2 model-page prompt paired with its real output.

A reviewed GPT Image 2 model-page prompt paired with its real output.

A glossy Cuban salsa box-garden filled with dancers, records, plants, travel keepsakes, and playful ceramic details.

A massive floating fortress appears in a front-facing technical cutaway with dense systems panels and industrial detailing.

Four global cities rise as colorful edible miniatures from sculpted sugar landscapes in a macro diorama grid.

London landmarks wrap around a compact globe-like city world under sweeping clouds and cinematic daylight.

A gourmet pizza slice reconstructed as a crisp exploded stack of crust, sauce, seasoning, toppings, and molten cheese.

A quiet alien desert beneath overlapping blue-green planets, rendered with haze, jagged rock, and restrained twilight color.

A polished Singapore miniature on a stamp-like platform, combining waterfront landmarks, greenery, and soft cloud light.

A premium Paris miniature arranged on a perforated postcard platform with landmarks, waterways, and sunset atmosphere.

A layered paper-sculpture library where an explorer, celestial books, airships, and brass details form a dense fantasy tableau.

A warm classroom portrait shaped by drifting chalk dust, amber window light, and a relaxed editorial pose.

A competition-board composition pairing a regenerative wetland aerial with ecological sections and analytical maps.

A museum-style camel field guide with anatomy, habitat, survival adaptations, and modular reference panels.

A playful miniature factory assembles tiny burgers inside a cutaway sesame bun.

An astronaut tends a bonsai on a ruined classical temple fragment above a planet.

A tiny chibi maintenance robot welds an immense illuminated orbital lattice.
Yes. You can browse and copy every prompt for free. Generating a new result in Playground is billed at the current price of the model you select.
Describe the subject and action first, then the scene, shot sequence, camera movement, timing, sound, and anything that must remain unchanged. Use a clear chronological order instead of a list of vague style words.
Text-to-video defines the whole scene in words. Image-to-video focuses on motion and camera changes while preserving the source image. Multimodal-reference prompts assign a specific role to each image, video, or audio reference.
Specify the subject, composition, camera angle, lighting, materials or visual style, required text and layout, then add constraints such as no logos, watermarks, or distorted hands.
Yes. Open any example to load its prompt, model, and output into the matching Playground, where you can edit variables and generate your own version.
The output lets you judge whether the prompt structure and model fit your goal before spending time or credits on a new generation.
Free prompt library
Open a verified image or video prompt in the matching Flatkey Playground, then adjust the model and settings for your own result.