How to Write Better AI Image Prompts: 10 Tips

Mar 14, 2026

Why Prompts Matter

The difference between a mediocre AI image and a stunning one often comes down to one thing: the prompt. A well-crafted prompt acts as precise instructions for the AI, guiding it to produce exactly what you envision.

Whether you're using Nano Banana 2, Flux 2 Pro, or any other model, these 10 tips will dramatically improve your results.

Tip 1: Be Specific About Your Subject

Vague prompts produce vague results. Instead of describing what you want in general terms, paint a clear picture.

Before: "a cat"

After: "a fluffy orange tabby cat curled up on a vintage leather armchair, soft afternoon sunlight streaming through a window"

The more details you provide about the subject, the closer the output will match your vision.

Tip 2: Define the Art Style

AI models can produce a wide range of styles, but you need to tell them which one you want.

Popular style keywords:

  • Photorealistic — looks like a real photograph
  • Digital art — clean, modern illustration style
  • Oil painting — textured, classical look
  • Watercolor — soft, flowing aesthetic
  • Anime/Manga — Japanese animation style
  • 3D render — CGI-quality output
  • Pixel art — retro game aesthetic (try our Pixel Art Generator)

Tip 3: Specify Lighting and Mood

Lighting transforms everything. Include lighting details to set the mood of your image.

Lighting keywords:

  • Golden hour, sunset lighting
  • Studio lighting, soft box
  • Dramatic side lighting, chiaroscuro
  • Neon glow, cyberpunk lighting
  • Overcast, diffused natural light
  • Backlit, silhouette

Example: "portrait of a woman, dramatic Rembrandt lighting, warm tones, moody atmosphere"

Tip 4: Include Camera and Composition Details

Think like a photographer. Camera terms help the AI understand framing and perspective.

Useful terms:

  • Close-up / macro — tight framing on details
  • Wide angle — expansive view
  • Bird's eye view — looking down from above
  • Low angle — looking up, makes subjects imposing
  • Bokeh background — blurred background, sharp subject
  • Rule of thirds — balanced composition

Tip 5: Use Negative Prompting

Many models support negative prompts — telling the AI what you don't want. This is surprisingly effective.

Common negative prompts:

  • "no text, no watermark" — for clean images
  • "no blurry, no distorted" — for sharp results
  • "no extra fingers, no deformed hands" — for accurate anatomy

Tip 6: Leverage Reference Images

Text-to-image is powerful, but image-to-image gives you even more control. Upload a reference image to guide composition, color palette, or style, then refine with your prompt.

This is especially useful for:

  • Maintaining consistency across multiple images
  • Applying a specific style to your existing content
  • Iterating on a concept with small adjustments

Try Image-to-Image Mode

Upload a reference image and let AI transform it based on your prompt.

Tip 7: Master Text in Images

Generating accurate text within AI images used to be nearly impossible. With models like Nano Banana 2, text rendering has become remarkably reliable.

Tips for text generation:

  • Put the text in quotation marks: a poster with the text "HELLO WORLD"
  • Keep text short — single words or short phrases work best
  • Specify the font style: "bold sans-serif", "elegant script", "handwritten"
  • Mention placement: "centered text at the top"

Tip 8: Control Color Palettes

Don't leave colors to chance. Specify the palette you want.

Examples:

  • "pastel color palette, soft pinks and blues"
  • "monochrome, black and white with high contrast"
  • "earth tones, warm browns and greens"
  • "vibrant neon colors on dark background"
  • "muted desaturated tones, vintage feel"

Tip 9: Build Prompts in Layers

Structure your prompt in logical layers for best results:

  1. Subject — what's in the image
  2. Action/Pose — what the subject is doing
  3. Setting/Background — where it takes place
  4. Style — art style and medium
  5. Lighting — mood and atmosphere
  6. Technical — camera, quality, aspect ratio

Template: "[Subject] [action] in [setting], [style], [lighting], [technical details]"

Example: "a samurai warrior standing on a cliff overlooking a misty valley, digital art, golden hour lighting, cinematic composition, 4K detailed"

Tip 10: Iterate and Refine

Your first prompt rarely produces the perfect result. Treat prompting as an iterative process:

  1. Start with a basic prompt
  2. Analyze the output — what's good, what's off
  3. Add or adjust details in your prompt
  4. Use image-to-image to refine promising outputs
  5. Try different models — Nano Banana 2 for speed, Flux 2 Pro for artistic quality

Prompt Templates to Get You Started

Product Photography

"[Product] on a clean white marble surface, professional product photography, soft studio lighting, high detail, 4K"

Social Media Post

"[Subject], vibrant and eye-catching, modern graphic design style, bold colors, clean composition, square format"

Portrait

"portrait of [subject], [expression], [lighting style], shallow depth of field, professional photography, 85mm lens"

Landscape

"[location/scene], panoramic view, [time of day] lighting, ultra wide angle, high detail, nature photography"

Fantasy Art

"[character/creature] in [setting], epic fantasy art, dramatic lighting, intricate details, concept art style"

Put These Tips Into Practice

Try all models and see which one works best for your style.

Advanced: Post-Processing with AI Tools

After generating your image, take it further with these tools:

  • Image Upscale — Enhance resolution to 4K for print or large displays
  • Background Remover — Isolate subjects for compositing or product shots
  • Face Swap — Swap faces for creative or branding purposes

Combining generation with post-processing tools gives you a complete creative pipeline — all in one platform.

Start Creating Now

Sign up and get free credits to experiment with all models and tools.

Nano Banana Team