AI Prompt Library

A highly detailed, cinematic image generation prompt focusing on a specific location (Galata Bridge, Istanbul) and time (sunset). It specifies composition, mood, lighting, color palette, and even camera equipment settings to achieve an atmospheric, award-

A photorealistic prompt for a wide drone shot featuring a colossal man, whose face and body are based on a reference image, casually sitting across a London street. The scene emphasizes the scale contrast between the giant figure and the tiny surrounding

A simple test prompt for GPT Image 1.5 to check its ability to render complex text accurately, specifically requesting the opening lines of the Japanese traditional story 'Jyumon' on white paper in black text.

A prompt designed to test the model's ability to generate a 3x3 grid of images, all depicting the same scene but from 9 different angles, and crucially, requiring the model to place the cell number in the bottom left corner of each scene.

A highly specific image generation prompt designed to test the model's ability to handle complex textures, architectural styles, and precise details. It requests a photorealistic, folded-paper blueprint of King Kong, emphasizing geometric folds, blueprint

A prompt implied by the context of generating a custom Yu-Gi-Oh card, although the exact text is not provided, the goal is clear: to create a custom card based on a text description.

A complex prompt designed to test GPT Image 1.5's ability to render specific text accurately within a dystopian poster design, featuring a character riding a broken scooter in a desert.

A complex, ultra-realistic documentary photography prompt set on the rain-soaked Galata Bridge in Istanbul. It focuses on an elderly fisherman laughing while holding a fish out to a street cat, specifying details like weathered face, coarse clothing, heav

A prompt used in a comparison test, focusing on maintaining a specific geometric shape (a bird's triangular shape) during image generation. The prompt text is visible in the accompanying screenshot.

A simple, functional prompt used for frame extraction from a previously generated image grid, specifically targeting the image located at row 1, column 2.

A prompt used to generate an image of a soccer player, testing the model's ability to handle specific athletic subjects and composition. The prompt text is visible in the accompanying screenshot.

A prompt designed to test the model's capability in rendering complex text and diagrams accurately, often a challenge for image generation models. The prompt text is visible in the accompanying screenshot.

A prompt used to test GPT Image 1.5's capabilities in portrait retention, instruction following, and feature migration, likely involving an uploaded image for transformation.

A simple text-to-image prompt describing a curious cat peeking out from a pile of autumn leaves, likely used to test the general image generation capabilities of GPT Image 1.5.

A prompt designed to test GPT Image 1.5's ability to maintain distinct features across multiple people in a complex scene, which resulted in a failure according to the user.

A simple style modifier prompt instructing GPT Image 1.5 to render the output image in the aesthetic of a 2000s film camera photo.

A detailed prompt for generating a technical illustration of an iPhone in an exploded view. It requires precise alignment of all internal components, technical annotations, an isometric engineering perspective, and photorealistic textures on metallic and

A highly detailed prompt focused on generating an ultra-photorealistic image of a domestic cat, specifying minute details like fur texture, eye reflections, skin imperfections, camera settings (85mm lens, f/1.8), and lighting to achieve studio-quality rea

A creative prompt requesting an image of two rival Indian politicians looking extremely bored in parliament, accompanied by a tired dog, rendered in the aesthetic of an early 2000s film camera.

A simple prompt to generate an image of a person holding a newspaper, where the headline clearly reads a specific comparison between two AI models.

A comparison prompt used to test the new GPT Image 1.5 model against Nano Banana Pro, instructing the AI to generate an image of an existing character eating ramen. This tests the model's ability to maintain character consistency while performing an actio

A simple French prompt used to test the image editing capabilities of GPT Image 1.5, specifically instructing the model to rotate an existing image 45 degrees to the left.

A prompt requesting a clean, Apple-style website design for Nike, focusing on powerful design elements and a specific aspect ratio.

An edit prompt used to transform the subject of a previously generated image (a photorealistic candid portrait of an everyday person) into a specific fictional character, testing the model's ability to handle identity and detail adherence.

A test of GPT Image 1.5's ability to generate a specific emotion and pose for a character based on an original image, while maintaining the original art style. The prompt instructs the model to generate the character looking excited with clenched fists.

A prompt designed to create a visual aid—specifically an infographic flow chart—for Dunder Mifflin sales associates, featuring the 'reference man' character in a cartoon style, demonstrating the model's ability to handle complex visual instructions and sp

A meta-prompt designed to instruct an AI model to take existing content and transform it into a highly dynamic, visually exciting, and landscape-oriented presentation slide, giving the model complete design freedom beyond the original constraints.

A prompt for generating a polarized selfie photo featuring Elon Musk and Sam Altman drinking Coca-Cola while wearing Christmas caps.

A simple prompt requesting the generation of an infographic about the features and capabilities of ChatGPT Image 1.5, useful for creating visual summaries or educational content.

A two-part prompt intended to first restore and enhance an old photo to modern high-end camera quality, and then apply a specific edit: making the subject bald. This tests the model's multi-step instruction following and editing precision.

A prompt for generating a hyper-realistic, full-shot image of an elderly Buddhist monk meditating on a rugged mountain peak. It specifies detailed textures for robes and rocks, dramatic sunrise lighting, and a cinematic atmosphere with snow-capped mountai

A highly structured prompt defining an aerial, hyperrealistic photograph of a gigantic Santa Claus sitting casually on a New York City avenue. It specifies camera setup, subject details (maintaining reference likeness), wardrobe, environment, and atmosphe

A prompt for generating a poster of deep-sea creatures, visualized within a vertical ocean cutaway, rendered in a beautiful and detailed Japanese anime style, testing the model's ability to handle scientific visualization combined with a specific artistic

A simple, direct prompt used to test GPT Image 1.5's ability to generate recognizable landmarks from an aerial perspective.

A highly detailed image generation prompt designed for image editing mode, focusing on preserving realistic skin texture, freckles, and pores in an extreme close-up portrait. The scene is set outdoors in winter with natural cold daylight, aiming for a raw

A simple text prompt requesting a multi-panel image designed to illustrate basic phrases for ordering food in Arabic, likely intended for language learning or visual aids.

A straightforward prompt requesting the generation of an infographic detailing the history of OpenAI, specifically including its progress up to the year 2025.

A highly detailed, ultra-photorealistic image generation prompt designed for editorial fashion realism, focusing on a specific, complex pose captured immediately after losing balance. The prompt specifies every detail from the subject's ethnicity and body

A prompt requesting an extremely realistic and minimalistic image suitable for a marriage announcement or portrait, based on an implied reference image of a couple.

An image editing prompt requiring the model to combine elements from two reference images: the cloud from image 1 and the face of the person from image 2, while maintaining the overall style of image 1. This tests complex image manipulation and style adhe

A prompt requesting a modification of the GitHub logo to appear red, dark, and evil, presented within a screenshot context.

A detailed zero-shot prompt for generating a modern tech-influencer poster design with a strong contrast and editorial style. It features a stylish young man in grayscale against a vibrant royal blue background, incorporating digital UI elements like a fl

A creative prompt asking for a photorealistic image of a hamburger where the right half is entirely made of glass, contrasting with the normal left half.

A prompt in Japanese requesting a celebratory illustration that includes the text 'GPT-image-1.5 has been released today,' used to test the model's ability to handle Japanese text rendering.

An iterative prompt sequence used to refine an image generation result, starting with a request for a realistic photo and then demanding even greater realism with specific details like skin texture, emphasizing a non-artistic, human look.

A straightforward prompt requesting an image that looks like a Fortnite screenshot, specifically showing the skin selector menu with Donald Trump selected as a playable skin.

A simple prompt instructing the model to generate an image of a man styled as the album cover for Michael Jackson's 'Thriller', testing the model's ability to interpret and apply specific cultural references and styles.

This prompt is used to generate a highly photorealistic image using GPT Image 1.5, demonstrating the model's capability to create visuals that are easily mistaken for real photographs. The actual prompt is found in the ALT text of the image attached to th

A concise prompt used for image manipulation or image-to-image generation, instructing the model to combine two provided reference photos.

A highly specific prompt for a macro photorealistic diorama built on the reverse side of a copper US Penny. It details miniature architecture (Lincoln Memorial), tiny scale-model tourists, cherry blossom trees, warm sunset lighting, and tilt-shift photogr

A creative prompt designed to test GPT Image 1.5's ability to embed a specific, hard-to-see hidden word ('zsakib') within a complex image—a pride of lions—that becomes obvious once spotted.

A detailed prompt designed to generate a split-screen YouTube thumbnail comparing 'Wasted Effort' (human) and '0 Seconds with AI' (AI robot). The prompt specifies composition, color schemes (sepia and futuristic blue), text overlays in Japanese, and cinem