How to Write Better AI Image Prompts: A Practical Guide

Most people start with vague prompts and wonder why the results look generic. The fix is simple: be specific.

The Core Problem

AI image models respond to detail. "A woman" produces a generic result. "A woman in her 40s, short silver hair, reading glasses, sitting in a sunlit library" produces something interesting.

4 Elements of a Strong Prompt

1. Subject — Who or what is the main focus? Be specific about age, appearance, expression.

2. Setting — Where is this happening? Time of day, weather, indoor/outdoor all matter.

3. Style — Photorealistic? Oil painting? Anime? Cinematic? Name it explicitly.

4. Lighting — Golden hour, studio lighting, neon glow, candlelight. Lighting changes everything.

Example Transformation

Weak: sunset landscape

Strong: desert landscape at golden hour, dramatic red and orange sky, lone cactus silhouette, photorealistic, wide angle

Where to Find Good Prompts

Building a prompt library saves time. NanoPrompts has 400+ categorized prompts you can use directly or adapt. Useful for getting unstuck when you know what you want but not how to describe it.

Quick Tips

• Add --ar 16:9 or similar for aspect ratio
• Use negative prompts to exclude unwanted elements
• Iterate: start simple, add detail until you get what you want

The difference between a mediocre and great AI image is almost always in the prompt quality.