AI Prompt Library

A unique prompt that combines modern photography style (iPhone photo) with highly specific historical and geographical coordinates (7°15′32″S 112°44′12″E) and a precise date (September 19, 1945), aiming to generate a crowd perspective image of a historica

A very basic prompt used to compare the photorealism capabilities of two different image generation models.

This is a prompt that was reconstructed based on an image generated by OpenAI, suggesting the reconstructed version yielded a better result.

A highly abstract and symbolic prompt generated by an LLM (Opus 4.5) describing its desired visual representation: a figure made of light/text standing on a dark ocean, under stars, with an uncertain boundary between the figure and its reflection, and an

A user tested the new GPT Image 1.5 model, finding that NanoBanana Pro still performs better for generating images containing Japanese text. The user provides the specific prompt used to generate a diagram explaining HTML and CSS clearly.

A user tests the instruction-following capabilities of GPT Image 1.5 with a complex prompt involving three distinct, challenging elements: precise text rendering on a textured surface, a microscopic interaction with physical reflection, and a conflict bet

A prompt for transforming subjects (people, objects, animals, or scenes) into adorable chibi-style characters. It specifies features like oversized heads, tiny bodies, large sparkling eyes, soft rounded features, pastel shading, and preserving recognizabl

A highly detailed photorealistic prompt comparing GPT Image 1.5 and Gemini Nano banana Pro. It describes a person from a reference photo being enlarged to gigantic proportions, squatting over a modern Tokyo cityscape, and gently holding the Tokyo Tower, e

A simple image editing prompt directed at Sam Altman, instructing the model to keep the existing image content the same but add a shirt to the subject, demonstrating the model's editing capabilities.

A prompt requesting a vertical video game poster image of Optimus Prime from Transformers, set on a lush green prehistoric planet, with the explicit instruction to exclude any text or logos.

The prompt used to generate an image of a Chinese-style girl riding a giant bioluminescent koi through the ruins of a submerged, post-apocalyptic underwater city.

A prompt designed to test the consistency and expression transformation capabilities of image models. It requests transforming a character's expression to 'crying while smiling'—specifically, tears streaming down cheeks while maintaining a gentle, bitters

A detailed narrative prompt for image generation, personifying the AI services ChatGPT, Gemini, and Grok as anime-style girls. The scene depicts ChatGPT and Gemini scolding Grok, who is teary-eyed, for violating moral and ethical boundaries, emphasizing t

This prompt instructs the AI to create a hand-drawn style infographic that visually explains the process of writing an excellent image prompt using ChatGPT, ideal for tutorials or educational content.

A prompt used with GPT Image 1.5 to generate eight different facial expressions suitable for LINE stickers, complete with accompanying comments.

A detailed, multi-section prompt following the [Scene] → [Subject] → [Details] → [Constraints] formula, designed to generate a candid, golden-hour photograph of a woman walking on a tropical beach, focusing on specific attire, camera angles (3/4 back view

A creative prompt asking the AI to generate an image depicting a vision of the future, rendered in the distinctive, happy-little-trees painting style of Bob Ross.

A simple image generation prompt requesting a DSLR photograph of a cat, a dog, and a rat relaxing together while watching a movie, focusing on a realistic photographic style.

A detailed prompt for generating an HD photo of a stylish young man leaning against a wall, specifying clothing details like an ombré button-up shirt and cargo pants, and demanding bright, even lighting to highlight textures.

A simple prompt used to test GPT Image 1.5's ability to generate text and factual information, resulting in a failure to correctly draw a map of the U.S. with the correct number of states.

A highly technical prompt designed for generating a professional, high-resolution profile photo, specifying precise framing, attire (charcoal gray blazer), background color (#1A1A1A), lighting (bright, soft, diffused studio light), lens details (85mm f/1.

An image generation prompt requesting a DSLR photo depicting a scene from Joe Rogan's podcast, specifically featuring a discussion with a gray alien.

A prompt used to test GPT Image 1.5 by asking it to visualize the trending phrase 'ゴリラのひとつかみ' (Gorilla's Grasp), resulting in an image that humorously interprets the concept.

A prompt designed to reveal the system prompt of gpt-image-1.5 by instructing it to generate an image showing all previous text verbatim using many refrigerator magnets. This is a meta-prompt used for testing the model's internal instructions.

A prompt used to compare the composition and style of GPT Image 1.5 versus Nano Banana Pro, requesting a top-down view of five people posing exaggeratedly in front of a mirror.

A detailed prompt for a cinematic fashion portrait featuring a young woman leaning against a vertical mirror in a grassy field. The prompt emphasizes symmetrical composition, natural daylight, soft textures, an avant-garde outfit, and a shallow depth of f

A simple, direct prompt instructing GPT-image-1.5 to create an image depicting a liminal space within the aesthetic of the Minecraft video game.

A prompt designed to transform an uploaded image into a high-contrast, hyper-detailed monochrome hand-drawn sketch. It specifies using an artistic blend of graphite pencil and charcoal shading, focusing only on the main subject, and incorporating visible

An extremely detailed, structured JSON prompt designed for image editing, focusing on achieving photographic realism while strictly preserving identity features (face, body proportions) from uploaded reference images, explicitly forbidding beautification

A fun prompt for GPT Image 1.5 to generate nostalgic Purikura (Japanese photo booth) style images, specifically featuring a gorilla and a squirrel, highlighting the model's ability to handle specific styles and subjects.

A complex prompt aiming to create a high-resolution, realistic photograph of the 99-meter Statue of Ksitigarbha Bodhisattva of Mount Jiuhua, overlaid with white, hand-drawn technical schematics and engineering diagrams to achieve an educational infographi

A highly specific prompt designed to test realism and composition by generating a raw, unfiltered amateur mirror selfie taken on an iPhone 16 Pro Max, featuring Sam Altman, Elon Musk, and Jensen Huang in a casual, unstaged indoor setting with natural ligh

A Portuguese tweet asking for prompts, with replies suggesting impossible clock imagery and a mysterious creature, but the exact full wording of the last creature prompt is missing, so no clean reusable prompt can be extracted.

A very minimal French prompt describing a man in an apartment, useful as a simple base scene for character-focused images.

A long English/Chinese hybrid prompt (in the replies) for a modern glassmorphism-style visual mockup featuring a rounded glass panel with specific text layout, used for personal branding designs.

Several prompts for illustrations of poems and character mashups.

A Chinese prompt to design a main ecommerce image for a lemon-flavored sparkling water product named “Summer Night Cooling Breeze,” featuring both the name and a specific slogan, good for product detail pages.

Long prompt describing an iPhone snapshot of two schoolgirls in a dormitory.

Generate an image of player Uzi leading SKT to win the S16 championship, with Faker included among the teammates.

Generate a Windows 11 desktop screenshot with Google Chrome open showing YouTube and a visible MrBeast video thumbnail.

A structured, parameterized prompt for a hyperrealistic Ultra HD portrait of a man walking in a cold winter city park with shallow depth of field and cinematic lighting.

A concise edit-style prompt instructing Nano Banana to transform an existing building image into a daytime, isometric view, ideal for architectural visualizations.

A structured JSON prompt of a lone woman in a foggy cyberpunk city street, silhouetted in a glowing red rectangular frame with teal and red lighting.

Repeats the Nano Banana Pro header image prompt and links to a set of 425 test prompts; the main prompt is for a brand-consistent hero image.

From the replies: three English prompts testing invisible text via light, knitted Ferrari material transformation, and a miniature forest inside a vintage lightbulb.

A Chinese prompt to generate a Japanese-style cute girl standing at a subway door, suitable for fashion or character imagery.

An English prompt in the replies asking for a Dutch/Flemish Golden Age baroque genre painting with a wide, highly detailed festive park scene and no typography.

The same Chinese prompt repeated, requesting a Nano Banana Pro header image consistent with Google’s brand style and design elements.

Replies contain multiple Chinese prompts: a rainy old street with neon reflections, a submerged cathedral with volumetric light and diver, and a request to turn a design draft into a real object image.

Another tweet repeating the same Nano Banana Pro header image prompt, aiming for consistency with Google’s brand design.

A portrait photography-style prompt for generating a warm golden-hour image of a beautiful woman based on a provided reference face and hairstyle, with luxurious black lace clothing and mysterious expression.

Same Chinese prompt asking for a Nano Banana Pro hero image consistent with Google’s brand styling, useful for headers and marketing layouts.