How to Write AI Image Prompts (2026 Guide)

The difference between a mediocre AI image and a stunning one is almost never the tool. It's the prompt. Knowing how to write AI image prompts is the single most important skill for getting consistent, professional results from any generator. Two people using the same version of Midjourney will get wildly different results based on how they describe what they want.

Prompt writing is a learnable skill, not a talent. Once you understand the structure that works across Midjourney, DALL-E, Stable Diffusion, and other generators, you can consistently produce professional-quality images instead of crossing your fingers and hoping for the best.

This guide breaks down exactly how to write prompts that work, with the structure, the keywords that matter, and 20+ examples you can use and adapt.

The Anatomy of a Good AI Image Prompt

Every effective AI image prompt has the same core structure. Think of it as a creative brief: the more specific you are, the closer the output matches your vision.

You don't need every element in every prompt, but including 4-6 of these consistently produces better results than a one-line description.

The second prompt gives the AI specific decisions to make: breed, position, setting details, lighting, style, and feeling. The first leaves everything to chance.

If you're already generating images with ChatGPT and want to level up your prompts, our guide on how to use ChatGPT to create images covers the image generation basics. Once you've mastered still images, you can take things further by learning to animate your AI art with video generation tools.

Subject and Detail

The subject is the most important part of your prompt. Be specific about what you're depicting, and include physical details that matter.

For people:

• Age range, gender, ethnicity, hair color/style, clothing, expression, body language
• "A woman in her 40s with silver-streaked dark hair, wearing a linen blazer" vs "a woman"

For objects:

• Material, color, condition, size relative to surroundings
• "A matte black ceramic coffee mug on a weathered oak table" vs "a coffee mug on a table"

For animals:

• Breed/species, coloring, size, action
• "A Maine Coon cat with amber eyes, lounging on a velvet cushion" vs "a cat on a cushion"

For scenes:

• Time of day, season, weather, location specifics
• "A narrow cobblestone street in Prague during a light snowfall at dusk" vs "a street in Europe"

The more specific your subject description, the less the AI has to guess, and AI guessing is what produces generic results.

Developing this kind of precision in your prompts is a learnable skill. the AI Academy teaches prompt engineering across image, text, and video tools with real projects you can practice on.

Style and Medium Keywords

This is where you control the visual aesthetic. Style keywords tell the AI what kind of image to create.

Photography styles:

• Portrait photography, street photography, product photography, editorial photography, documentary photography
• "Taken with a Canon EOS R5, 85mm lens" adds photorealistic camera-specific quality

Art mediums:

• Oil painting, watercolor, pencil sketch, charcoal drawing, digital illustration, gouache, ink wash, pastel
• "Oil painting in the style of the Dutch Golden Age" gives period-specific aesthetics

Design styles:

• Flat illustration, vector art, isometric, pixel art, low poly, paper cut, collage
• "Flat vector illustration with clean lines and solid colors"

Cinematic references:

• Movie still, cinematic lighting, film noir, Wes Anderson color palette, 1970s film grain
• "Movie still, anamorphic lens, dramatic lighting" creates cinematic looks

Era-specific:

• Art Nouveau, Art Deco, Mid-century modern, Brutalist, Vaporwave, Cyberpunk
• "Art Deco poster design with geometric patterns and gold accents"

Lighting Keywords

Lighting transforms an image more than almost any other element. Specific lighting keywords produce dramatically different moods.

Natural light:

• Golden hour, blue hour, overcast soft light, harsh midday sun, dappled sunlight through trees, backlit

Studio light:

• Rembrandt lighting, butterfly lighting, rim lighting, split lighting, high-key, low-key

Atmospheric:

• Volumetric light, god rays, neon glow, candlelight, bioluminescent, moonlit

Dramatic:

• Chiaroscuro, dramatic shadows, silhouette lighting, spotlight

vs

A portrait of an elderly man with deep wrinkles, flat fluorescent office lighting, cool tones

Same subject, completely different feeling.

Understanding how individual prompt elements transform the output is what our AI Academy focuses on -- building intuition through structured practice, not just reading about it.

Camera and Composition

Composition keywords tell the AI how to frame the shot. These are especially useful for photorealistic images.

Camera angles:

• Bird's eye view, worm's eye view, eye level, Dutch angle, overhead shot, low angle

Framing:

• Close-up, extreme close-up, medium shot, wide shot, establishing shot, full body

Depth:

• Shallow depth of field (bokeh background), deep focus, tilt-shift, macro

Composition rules:

• Rule of thirds, centered composition, symmetrical, leading lines, negative space

Aspect ratios:

• Portrait (2:3), landscape (3:2), square (1:1), cinematic widescreen (16:9 or 21:9)
• In Midjourney, use --ar 16:9 to set the aspect ratio

Negative Prompts for AI Image Generation

Negative prompts tell the AI what to exclude. This is essential for Stable Diffusion and useful in other tools.

Common negative prompts:

• "No text, no watermark, no signature" (prevents unwanted text artifacts)
• "No extra fingers, no deformed hands" (addresses common anatomy issues)
• "No blurry, no low quality, no pixelated" (enforces quality)
• "No border, no frame" (produces edge-to-edge images)

In Stable Diffusion: Add negative prompts in the dedicated negative prompt field.

In Midjourney: Use --no followed by what to exclude: --no text --no watermark

In ChatGPT/DALL-E: Include exclusions naturally in your prompt: "Do not include any text or watermarks in the image."

Tool-Specific Tips

Each AI image generator interprets prompts differently. Here's what works best for each.

Midjourney V7

Midjourney prefers short, high-signal phrases over long paragraphs. Front-load the most important details.

• Use reference images with Omni-Reference for consistency
• Parameters: --ar (aspect ratio), --no (negative), --s (stylize, 0-1000), --c (chaos, 0-100)
• Higher --s values produce more artistic/stylized results
• Higher --c values produce more varied/unexpected results
• Keep prompts to 40-60 words for best results

ChatGPT (GPT-4o)

ChatGPT understands natural language better than any other generator. Write prompts as detailed descriptions, like briefing a designer.

• You can iterate conversationally: "Make the background darker" or "Move the text to the left"
• Handles text in images well; specify font style and placement
• Upload reference images and ask for modifications
• Works best with paragraph-style descriptions, not keyword lists

Stable Diffusion

Stable Diffusion rewards structured, weighted prompts. Use parentheses for emphasis and colons for weight.

• (masterpiece, best quality:1.4) (weights of 1.0-1.5 for important elements)
• Comma-separated keywords work better than sentences
• Negative prompt field is critical, so always use it
• Model choice matters as much as prompting (SDXL, SD3, custom models)
• LoRA models can define specific styles or characters

20+ AI Image Prompt Examples

Product Photography

A sleek glass perfume bottle on a black marble surface, soft studio lighting from the left, water droplets on the glass, shallow depth of field, luxury product photography, dark moody atmosphere

Social Media Graphics

Flat illustration of a person working on a laptop in a cozy home office, plants on the windowsill, warm afternoon light, pastel color palette with mint green and soft peach, clean modern style, square format

Portrait

Professional headshot of a man in his 30s, wearing a charcoal suit, confident expression, neutral gray studio background, Rembrandt lighting, shallow depth of field, photorealistic

Landscape

Dramatic aerial view of the Norwegian fjords at sunset, deep blue water reflecting orange and purple sky, snow-capped mountains, cinematic photography, wide angle lens, 16:9 aspect ratio

Food Photography

A rustic sourdough loaf on a wooden cutting board, flour dusted surface, warm kitchen background blurred, steam rising, golden morning light from a nearby window, editorial food photography

Fantasy Art

A massive dragon perched on a crumbling castle tower during a thunderstorm, lightning illuminating its scales, rain pouring, dark epic fantasy illustration, detailed, dramatic composition, volumetric lighting

Architecture

A minimalist Japanese tea house surrounded by a zen garden, raked gravel patterns, single cherry blossom tree, misty morning light, shot from a low angle, architectural photography, serene and contemplative

Children's Book

A small fox wearing rain boots jumping in a puddle on a rainy day, colorful autumn leaves on the ground, warm watercolor children's book illustration, soft edges, playful and joyful mood

Fashion Editorial

A model in a flowing red silk dress standing on the edge of a cliff overlooking the ocean, wind blowing the fabric dramatically, golden hour backlighting, fashion editorial photography, shot on medium format film

Abstract/Conceptual

An abstract representation of time passing, melting clocks flowing like liquid across a surreal landscape, inspired by Salvador Dali, oil painting style, dreamlike atmosphere, muted earth tones with gold accents

Vintage/Retro

A classic American diner at night, neon signs glowing in pink and blue, chrome details, wet pavement reflecting the lights, 1950s aesthetic, cinematic photography, moody and nostalgic

Sci-Fi Concept Art

Interior of a space station corridor, floor-to-ceiling windows showing Earth below, holographic displays on the walls, a lone astronaut walking toward the light, concept art, cinematic lighting, blue and white color scheme

Minimalist Design

A single red umbrella on a wet city street, everything else in black and white, selective color photography, rainy day, reflections on pavement, minimal composition, centered

Macro Photography

Extreme macro shot of a honeybee covered in pollen on a lavender flower, every detail of the compound eyes visible, shallow depth of field, natural sunlight, nature photography

These prompts work across Midjourney, ChatGPT, and Stable Diffusion. Adapt the language to each tool: shorter and keyword-focused for Midjourney, more conversational for ChatGPT, weighted and structured for Stable Diffusion.

Building Your Prompt Library

The fastest way to improve is to keep a prompt library, a document or spreadsheet where you save every prompt that produced a good result, along with the output image and which tool you used.

Over time, you'll develop your own set of "power phrases" that consistently produce the results you want. Phrases like "editorial lighting," "shot on Kodak Portra 400," or "clean vector illustration" become reliable building blocks you can mix and match.

If you want an accelerated path to building that prompt library and creative intuition, the AI Academy provides guided exercises across every major AI image tool.

For more on using AI-generated images in your marketing and content work, check out our guides on generative AI for content creation and ChatGPT for marketing. If you work in Photoshop, our guide on using generative AI in Photoshop shows how to apply these prompting skills inside Adobe's editing tools.

If you're also interested in monetizing the images you create, our guide on how to make money with AI art covers seven proven methods.

What's Next

Prompt engineering for images is a skill that compounds. Every image you generate teaches you something about how the model interprets language, and within a few weeks of regular practice, you'll be writing prompts that consistently nail what you envision.

FAQ

Want to go deeper into AI image generation and creative workflows? Start your free 14-day trial →