How to Write AI Image Prompts: 10+ Examples That Actually Work

The difference between a mediocre AI image and a stunning one almost always comes down to the prompt. Same model, same settings — but a well-crafted prompt produces dramatically better results.

The good news: prompt writing isn't magic. It's a learnable skill with clear patterns. Here's everything you need to know, with real examples you can use today.

The Prompt Formula

Great prompts generally follow this structure:

[Subject] + [Style/Medium] + [Composition] + [Lighting] + [Mood/Atmosphere] + [Details]

You don't need every element every time, but the more specific you are, the better your results. Let's see this in action.

Before & After: 10 Prompt Improvements

1. Portrait Photography

❌ Before: "a woman in a coffee shop"

✅ After: "candid portrait of a young woman reading at a sunlit café table, warm morning light streaming through large windows, shallow depth of field, shot on 85mm lens, soft bokeh background with other patrons, natural skin tones"

What changed: Added lighting direction, lens specification, depth of field, and atmosphere. The AI now has a clear visual reference instead of a vague concept.

2. Product Shot

❌ Before: "headphones on a table"

✅ After: "matte black over-ear headphones resting on a polished walnut desk, soft studio lighting from the left, subtle reflection on the desk surface, minimalist composition, white background gradient, commercial product photography style"

What changed: Specified material, surface, lighting direction, and explicitly called out the photography style.

3. Landscape

❌ Before: "a mountain landscape"

✅ After: "dramatic mountain range at golden hour, jagged snow-capped peaks piercing low-hanging clouds, alpine meadow with wildflowers in the foreground, warm amber light on the peaks contrasting with cool blue shadows in the valleys, wide-angle perspective, National Geographic style"

What changed: Time of day, weather conditions, foreground interest, color contrast, and a style reference the AI can anchor to.

4. Abstract/Conceptual

❌ Before: "technology concept"

✅ After: "abstract visualization of data flowing through neural networks, streams of luminous teal and violet particles forming interconnected nodes, dark background with subtle grid pattern, depth of field creating layers of focus, clean and modern aesthetic, 4K digital art"

What changed: Specific colors, visual metaphor, background treatment, and artistic direction. Abstract concepts need more specificity, not less.

5. Food Photography

❌ Before: "a plate of pasta"

✅ After: "overhead flat-lay of fresh handmade fettuccine with porcini mushroom cream sauce, garnished with shaved parmesan and fresh thyme, rustic ceramic plate on a dark slate surface, natural window light from the top-left, steam rising gently, food magazine editorial style"

What changed: Camera angle, specific ingredients, surface material, light direction, and that crucial "steam rising" detail that makes food look alive.

6. Illustration Style

❌ Before: "a cat drawing"

✅ After: "whimsical illustration of a fluffy orange tabby cat wearing a tiny astronaut helmet, floating in space among colorful planets, children's book illustration style, soft watercolor textures, warm pastel color palette, playful and charming mood"

What changed: Specific breed/color, scenario, artistic medium, color palette, and emotional tone.

7. Architecture

❌ Before: "a modern building"

✅ After: "brutalist concrete apartment building at twilight, geometric balconies with warm interior lights glowing against the blue hour sky, a single tree in the foreground, shot from a low angle looking up, architectural photography, urban atmosphere"

What changed: Architectural style, time of day, the contrast between warm interiors and cool sky, camera angle, and a compositional element (the tree).

8. Fashion

❌ Before: "a person in nice clothes"

✅ After: "editorial fashion photograph, model in an oversized camel wool coat walking down a rain-slicked city street at dusk, motion blur on the edges, streetlights creating golden halos in the mist, cinematic color grading with desaturated blues and warm highlights, Vogue style"

What changed: Garment specificity, environmental context, movement, weather conditions, and a clear publication style reference.

9. Flat Design / UI

❌ Before: "a rocket icon"

✅ After: "minimal flat design rocket icon, geometric shapes, limited color palette of coral red and midnight blue on a white background, clean vector style, subtle long shadow, suitable for a tech startup app icon, 1:1 aspect ratio"

What changed: Design style, specific colors, visual treatment (long shadow), intended use, and aspect ratio.

10. Surreal / Creative

❌ Before: "something surreal"

✅ After: "a grand piano floating in the middle of a still ocean at sunset, mirror-like water surface reflecting the sky, a single beam of light illuminating the piano from above, scattered sheet music pages floating in the air around it, hyperrealistic rendering with dreamlike atmosphere, Salvador Dalí meets Magritte"

What changed: Went from zero specificity to a complete scene with multiple visual elements, lighting, and artistic references.

Tips for Myjourney's Modes

Different generation modes on Myjourney respond differently to prompts:

Standard Mode

Best for most use cases. Follows prompts closely and produces polished, reliable results.

Tip: Be specific about style and composition. Standard mode excels when you give it clear direction.

Good for: Product shots, portraits, marketing visuals, blog illustrations.

Raw Mode

Produces more artistic, less "processed" results. Better for fine art and editorial work.

Tip: Use more evocative, emotional language. Raw mode responds well to mood descriptors like "melancholy," "ethereal," or "gritty." Lean into artistic references — mention specific photographers, painters, or film directors.

Good for: Editorial photography, fine art, conceptual work, unique artistic styles.

Draft Mode

Fast and cheap ($0.03/image). Great for exploring ideas before committing to higher-quality modes.

Tip: Use Draft for brainstorming. Try 10 variations of a prompt at $0.03 each ($0.30 total), pick the direction you like, then regenerate your favorite in Standard or Raw mode.

Good for: Rapid prototyping, concept exploration, high-volume content, thumbnails.

Advanced Prompt Techniques

Negative Prompting

Tell the AI what you don't want: "...without text, no watermarks, no extra fingers, avoid cluttered background"

Aspect Ratio

Specify dimensions for your intended use: "16:9 aspect ratio" for YouTube thumbnails, "9:16" for Instagram Stories, "1:1" for profile pictures.

Style Mixing

Combine unexpected references: "cyberpunk aesthetic meets Art Nouveau, Blade Runner lighting with Mucha-style organic curves"

Camera Language

AI models understand photography terms: "shot on 35mm film," "f/1.4 aperture," "long exposure," "tilt-shift miniature effect," "fisheye lens distortion"

Specificity Over Length

A focused 20-word prompt often beats a rambling 100-word one. Every word should add visual information. Cut filler like "beautiful," "amazing," or "high quality" — these are vague and add little.

The Iteration Workflow

Great images rarely come from the first prompt. Here's the workflow that gets results:

1. Start broad in Draft mode — test the concept with 3–4 variations ($0.09–$0.12)
2. Refine the prompt — add specifics based on what's working and what's missing
3. Test refined versions — another 3–4 drafts with the improved prompt
4. Upgrade to Standard/Raw — take your best prompt to a higher quality mode
5. Fine-tune — adjust lighting, composition, or style words for the final version

This entire process costs $1–$3 and takes 15–20 minutes. Compare that to hours of searching stock photo libraries.

Common Mistakes to Avoid

Being too vague. "A cool image" gives the AI nothing to work with. What makes it cool? Be specific.

Contradictory descriptions. "Bright sunny day with dark moody shadows" confuses the model. Pick a consistent mood.

Overloading with keywords. "4K, HD, ultra-realistic, masterpiece, best quality, award-winning" is keyword stuffing. One or two quality descriptors is plenty.

Forgetting composition. Where is the subject in the frame? What's in the background? Without this, you get centered subjects on generic backgrounds.

Start Generating

The best way to learn prompt writing is to practice. Myjourney's Draft mode at $0.03 per image makes experimentation practically free — generate 30 images for less than a dollar and you'll develop an intuition for what works.

Bookmark this guide and reference it next time you're crafting prompts. And if you're wondering about the cost breakdown or whether you can use your creations commercially, we've got you covered.

Now go write some prompts. Your first few will be rough. Your tenth will be good. Your fiftieth will be great.