How to Write AI Image Prompts: 12 Examples That Actually Work (2026)
The difference between a muddy, generic AI image and a stunning one usually isn't the model — it's the prompt. The good news: writing great prompts is a skill, not a talent, and you can learn the core of it in about ten minutes. This guide gives you a repeatable formula, 12 prompts you can copy and run right now, and the small mistakes that quietly sabotage beginners.
Every prompt below works on the free PixelMind generator. You get 10 free credits every day, so you can test all of them without spending a cent. Open the generator in another tab and follow along.
The 5-part prompt formula
Most strong prompts answer five questions in order. You don't need all five every time, but thinking through them prevents the vague, half-finished images beginners get.
- Subject — who or what is in the image? Be specific: not 'a dog' but 'a golden retriever puppy'.
- Action / pose — what is the subject doing? 'sitting in tall grass', 'mid-jump', 'looking over its shoulder'.
- Setting — where is it? 'a misty pine forest at dawn', 'a neon-lit Tokyo alley'.
- Style & medium — the look: 'photorealistic', 'watercolor', 'oil painting', 'anime cel shading', '3D render'.
- Lighting & detail — the finishing touch: 'golden hour light', 'soft studio lighting', 'sharp focus, high detail'.
Subject + Action + Setting + Style + Lighting. Read your prompt back and check each part is present. Missing parts are where the AI 'guesses' — and guesses are where images go wrong.
Putting it together
Watch a weak prompt become a strong one by adding each part:
- ❌ 'a cat' — the AI has to invent everything.
- 🙂 'a black cat sitting on a windowsill' — subject + action + setting.
- ✅ 'a sleek black cat sitting on a wooden windowsill, rainy city street outside, photorealistic, soft afternoon light, sharp focus' — all five parts.
12 prompts that actually work
Each of these follows the formula and is ready to run. Tap any prompt to open it in the generator with the right style pre-selected, then tweak the words to make it yours.
Portraits & people
Close-up portrait of an elderly fisherman, weathered face, golden hour light, photorealistic, sharp detail, shallow depth of fieldTry →A cheerful anime girl with pink twin-tails in a blooming cherry orchard, soft cel shading, expressive eyes, detailed backgroundTry →Oil painting portrait of a young woman in Renaissance dress, dramatic chiaroscuro lighting, rich brushstrokes, old-master styleTry →Landscapes & scenes
A misty pine forest at dawn, sun rays through fog, photorealistic, ultra detailed, atmospheric, wide shotTry →Watercolor painting of a quiet Italian coastal village at sunset, soft washes, loose brushwork, warm pastel paletteTry →A neon-lit Tokyo alley in the rain, glowing signs, reflections on wet pavement, cyberpunk, cinematic, moody atmosphereTry →Products & objects
A steaming bowl of ramen on a rustic wooden table, soft window light, photorealistic, food photography, shallow depth of fieldTry →A pair of futuristic sneakers floating on a clean gradient background, studio lighting, 3D render, product shot, high detailTry →Characters & concepts
Cyber-samurai standing on a neon rooftop at night, glowing katana, anime style, dynamic pose, detailed armorTry →A friendly robot barista serving coffee in a cozy cafe, 3D render, soft lighting, warm colors, Pixar-like styleTry →An ancient dragon curled around a crumbling stone tower, stormy sky, fantasy concept art, dramatic lighting, epic scaleTry →Minimalist flat illustration of a city skyline at dusk, bold shapes, limited color palette, clean vector styleTry →5 mistakes beginners make
- Being too vague. 'A beautiful landscape' gives the AI nothing to work with. Name the place, time of day, and mood.
- Stacking too many ideas. Ten subjects in one prompt confuses the model. One clear scene beats a crowded one.
- Forgetting the style word. Without 'photorealistic', 'watercolor', etc., you get a generic default look. Always state the medium.
- Ignoring lighting. Lighting is 50% of the mood. 'Golden hour', 'soft studio light', and 'moody backlight' transform the same scene.
- Not iterating. Your first result is a draft. Change one part at a time — swap the lighting, then the style — and you'll learn fast what each word does.
Pro tip: change only ONE part of your prompt between generations. That's how you learn which word did what — and how you build an instinct for prompting.
Quick reference: useful style & lighting words
Style words
- Realism: photorealistic, hyperrealistic, cinematic, film photography
- Painted: oil painting, watercolor, ink wash, impressionist
- Illustration: anime, comic, flat design, pop art, low poly
- 3D & digital: 3D render, octane render, isometric, pixel art
Lighting words
- Warm & soft: golden hour, soft window light, candlelight
- Dramatic: chiaroscuro, rim light, moody backlight, volumetric light
- Clean: studio lighting, softbox, even diffused light
- Atmospheric: foggy, hazy, god rays, neon glow
Start creating
The fastest way to get good at prompting is to run a lot of prompts. Copy any example above, change one detail, and see what happens. Browse the full set of AI art styles to see how the same idea looks across realistic, anime, oil painting and more.
Ready to create your own?
Start free with 20 credits every day — no credit card, no commitment.
Open the generator →