Best Practices for AI Image Prompts

Master the art of writing prompts for AI image generation across different platforms.

Best Practices for AI Image Prompts: Create Stunning Visuals

The difference between a mediocre AI image and a stunning one often comes down to the prompt. Here's a comprehensive guide to writing prompts that get results.

Prompt Structure Fundamentals

Basic Formula:


[Subject] + [Action/Pose] + [Environment] + [Style] + [Lighting] + [Camera/Angle] + [Quality Modifiers]

Example Evolution:

Level 1 (Basic): "A cat"

Level 2 (Better): "A fluffy orange cat sitting on a windowsill"

Level 3 (Good): "A fluffy orange tabby cat sitting on a sunny windowsill, looking outside, soft natural light"

Level 4 (Great): "A fluffy orange tabby cat sitting on a white painted windowsill, gazing thoughtfully outside at falling autumn leaves, warm afternoon sunlight streaming in, shallow depth of field, professional pet photography style, 85mm lens"

Key Elements in Detail

Subject Description

Be specific:

  • Not "a dog" but "a golden retriever puppy"
  • Not "a woman" but "a woman in her 30s with curly red hair"
  • Not "a building" but "a Victorian-era brownstone with bay windows"
  • Include:

  • Age/size
  • Color/texture
  • Distinctive features
  • Emotional state
  • What they're wearing/holding
  • Action and Pose

    Static poses:

  • "Standing confidently"
  • "Sitting cross-legged"
  • "Leaning against a wall"
  • "Looking over shoulder"
  • Dynamic actions:

  • "Running through"
  • "Reaching toward"
  • "Dancing gracefully"
  • "Laughing while"
  • Environment and Setting

    Specify:

  • Location (specific > general)
  • Time of day
  • Weather/season
  • Surrounding objects
  • Atmosphere
  • Examples:

  • "In a cozy coffee shop with exposed brick"
  • "On a misty mountain trail at dawn"
  • "In a futuristic city with neon signs"
  • "In a sunlit meadow during golden hour"
  • Style and Medium

    Photography Styles

    Portrait:

    
    portrait photography, professional headshot, studio lighting,
    85mm lens, shallow depth of field, catchlights in eyes
    

    Landscape:

    
    landscape photography, dramatic sky, golden hour,
    wide angle lens, high dynamic range, rule of thirds
    

    Product:

    
    commercial product photography, white background,
    soft box lighting, sharp focus, high key, advertising style
    

    Street:

    
    street photography, candid moment, natural light,
    black and white, documentary style, decisive moment
    

    Art Styles

    Digital Art:

    
    digital painting, concept art, detailed, vibrant colors,
    fantasy art style, trending on ArtStation
    

    Traditional:

    
    oil painting, visible brushstrokes, classical composition,
    Renaissance style, museum quality, canvas texture
    

    Illustration:

    
    editorial illustration, flat design, limited color palette,
    modern graphic style, clean lines
    

    Artist References

    Use carefully and ethically:

  • "In the style of Studio Ghibli"
  • "Reminiscent of Monet's water lilies"
  • "Art deco poster style"
  • "Bauhaus design influence"
  • Lighting Techniques

    Natural Light:

  • Golden hour (warm, long shadows)
  • Blue hour (cool, soft)
  • Overcast (soft, diffused)
  • Direct sunlight (harsh, high contrast)
  • Dappled light (through trees/windows)
  • Studio Light:

  • Soft box (diffused, flattering)
  • Rim lighting (edge highlights)
  • Butterfly lighting (shadows under nose)
  • Dramatic spotlight
  • High key (bright, few shadows)
  • Low key (dark, moody)
  • Atmospheric:

  • Volumetric lighting (god rays)
  • Neon lights
  • Candlelight
  • Firelight
  • Bioluminescence
  • Camera and Composition

    Angles:

  • Eye level (neutral)
  • Low angle (powerful, imposing)
  • High angle (vulnerable, overview)
  • Dutch angle (tension, unease)
  • Bird's eye (pattern, context)
  • Worm's eye (dramatic, sky emphasis)
  • Focal Length:

  • 14mm: Dramatic wide angle, distortion
  • 35mm: Natural wide, environmental
  • 50mm: Natural perspective
  • 85mm: Portrait, compression
  • 200mm: Telephoto, background blur
  • Composition:

  • Rule of thirds
  • Centered/symmetrical
  • Leading lines
  • Frame within frame
  • Negative space
  • Quality and Technical Modifiers

    Resolution/Detail:

  • "Highly detailed"
  • "4K", "8K"
  • "Ultra high resolution"
  • "Intricate details"
  • "Sharp focus"
  • Render Quality:

  • "Octane render"
  • "Unreal Engine"
  • "Ray tracing"
  • "Global illumination"
  • "Physically based rendering"
  • General Quality:

  • "Professional"
  • "Award-winning"
  • "Masterpiece"
  • "Museum quality"
  • "Commercial quality"
  • Platform-Specific Tips

    Midjourney

    Best Practices:

  • More artistic, interpretive
  • Style references work well (--sref)
  • Parameters matter (--ar, --v, --style)
  • Shorter prompts often better
  • Parameters:

  • --ar 16:9 (aspect ratio)
  • --v 6 (version)
  • --s 100-1000 (stylization)
  • --c 0-100 (chaos/variation)
  • DALL-E 3

    Best Practices:

  • Detailed descriptions work well
  • Good at following specific instructions
  • Best for text in images
  • Natural language works
  • Stable Diffusion

    Best Practices:

  • Weight important terms: (important:1.3)
  • Use negative prompts
  • Model-specific keywords
  • More technical prompts work
  • Negative Prompts:

    
    (bad quality:1.4), (worst quality:1.4), blurry, watermark,
    text, logo, distorted, deformed, ugly, duplicate
    

    Common Mistakes

    Too Vague: "A nice landscape" → "Rolling hills of Tuscany at golden hour with cypress trees and a medieval village in the distance"

    Contradictory: "A dark bright image" → Choose one mood or describe contrast intentionally

    Too Many Subjects: "A cat, dog, bird, and fish playing together" → Focus on fewer subjects for better results

    Conflicting Styles: "Photorealistic anime" → Choose one or specify blend intentionally

    Missing Context: "A person" (no setting, style, or purpose) → Add environment, style, and purpose

    Iteration Strategy

    Process:

  • Start with core concept
  • Generate first batch
  • Identify what's working
  • Add specific details
  • Adjust style elements
  • Refine until satisfied
  • When Results Miss:

  • Too generic → Add specific details
  • Wrong style → Strengthen style keywords
  • Wrong mood → Adjust lighting/color terms
  • Missing elements → Emphasize in prompt
  • Unwanted elements → Use negative prompts or reframe

Quick Reference Templates

Portrait:


[Subject description] portrait, [setting],
[lighting type], [camera specs], [style],
professional photography, detailed, [mood]

Landscape:


[Location/scene], [time of day], [weather],
[style] photography, [composition], dramatic sky,
[quality terms]

Product:


[Product] on [surface], [background], commercial
photography, studio lighting, sharp focus, high key,
advertising style, professional

Concept Art:


[Subject/scene], concept art, digital painting,
[art style], detailed, vibrant colors, dramatic
lighting, trending on ArtStation

The best AI image prompts are specific, structured, and iterative. Start with a clear vision, describe it thoroughly, and refine based on results.

Share this article: