How to Generate Stunning Images
The Importance of Careful Writing
The quality and relevance of the visual result directly depend on the precision of your wording.
🚫
Spelling errors or confusing wording can mislead the artificial intelligence and generate an image far from your initial vision.
✅
To ensure the effectiveness of your instruction, three elements are essential: careful spelling, clear syntax, and coherent organization.
Optimization also involves successive adjustments:
- switching to English,
- eliminating superfluous wording,
- structuring with bullet points...
Each AI model has its own specificities. Nevertheless, a methodical approach based on a few key questions allows you to structure your thinking and develop an instruction that is simultaneously explicit, precise, and relevant.
Describing Your Vision in a Structured Way
⬇️
An effective prompt is based on five key steps:
- Major Components - Clearly identify who or what constitutes the heart of the image
- Visual Styles and Artistic Inspirations - Specify the desired aesthetic: movement, era, or artistic reference
- Lighting and Effects - Define the light atmosphere and textures to bring the scene to life
- Angles and Perspectives - Choose the optimal viewpoint to showcase your subject
- Image Format - Specify the appropriate dimensions and proportions
This hierarchical structure ensures consistent results that conform to your creative vision.
Major Components
- Characters: gender, age, emotion, position
- Objects: distinctive accessories, decorative elements
- Scene: precise setting (jungle, cozy living room, Parisian street)
- Movement: still or in action?
Visual Styles and Inspirations
Photography
- Realistic: detailed photography, high definition
- Black and White: pronounced contrast, dramatic lighting
Classical Paintings
- Watercolor: fluid colors, transparency
- Impressionism: fragmented brushstrokes, light capture (Monet)
- Van Gogh: swirling whorls, generous impasto
- Surrealism: dreamlike, distorted reality, unbridled imagination (Dalí)
- Cubism: spatial deconstruction, multiple perspectives (Picasso)
Decorative Arts
- Gothic: mysterious darkness, monumental architecture
- Baroque: sumptuous ornaments, gilding, theatricality
- Art Nouveau: organic curves, Mucha floral motifs
- Art Deco: stylized geometry, 1920s elegance
Graphic Arts
- Minimalist: refined composition, elegant sobriety
- Pop Art: saturated colors, comic book codes, Warhol style
- Street Art: urban expression, spray paint, Banksy style
Animation
- Anime/Manga: expressive lines, vivid colors, Japanese style
- Studio Ghibli: visual poetry, soft light, dreamlike atmosphere
- Pixel Art: retro-gaming aesthetic, 8-bit rendering
Imaginary Worlds
- Cyberpunk: futuristic, bright neons, technological dystopia
- Steampunk: Victorian-tech, copper gears, steam era
- Fantasy: imaginary universes, legendary creatures
Lighting and Effects
- Lighting: natural, dramatic, cinematic
- Warm tones: red/orange/yellow (energy)
- Cool tones: blue/green/purple (calm)
- Pastel: muted softness (lightness)
- Effects and textures: contrast, saturation, grain, artistic blur
Angles and Perspective
- Close-up: detailed facial expressions
- Aerial view: panorama, overall staging
- Fish-eye: immersive effect, dynamic distortion
- Wide shot: extended scene, ideal for landscapes/environments
- Low angle: grandeur, power, view from below
- Backlight: silhouettes, cinematic effect
Image Format
- Orientation: portrait, landscape
- Ratio:
--ar 1:1(square),--ar 2:3,--ar 3:2,--ar 4:5,--ar 5:4,--ar 4:3,--ar 3:4,--ar 16:9or--ar 9:16
⚠
Pitfalls to Avoid:
- Information overload: too much information overwhelms the AI and generates incoherent images
- Lack of precision: a vague prompt produces unpredictable and random results
- Contradictory instructions: the AI cannot resolve conflicting directives
- Poor prioritization: always place essential elements at the beginning of the prompt
Improvement Through Iteration and Reprompting
After writing your prompt, test it and adjust it based on the results.
3 tips to become a pro at image generation:
🥉
Modify one parameter at a time to progressively refine the generated image.
🥈
Ask the AI to refine your prompt for greater clarity and fluidity.
This approach can result in a particularly powerful custom mammouth.
This approach can result in a particularly powerful custom mammouth.
🥈
Reprompt with another model!
41% of image queries are reprompted by our users (vs 34% for text).
Why? Images are inherently subjective: visual style, composition, and artistic interpretation vary greatly from one model to another. Comparison becomes almost indispensable.
41% of image queries are reprompted by our users (vs 34% for text).
Why? Images are inherently subjective: visual style, composition, and artistic interpretation vary greatly from one model to another. Comparison becomes almost indispensable.