A provider-agnostic guide for crafting effective AI image generation prompts. Covers illustrations, covers, infographics, and general imagery.
CRITICAL: grep references/ for detailed style/layout/palette specs and prompt templates.
Core Principles
Good image prompts follow structure first, aesthetics second:
- Layout/Composition — Describe spatial arrangement, zones, flow direction
- Specific Content — Use actual data, labels, terms (not vague descriptions)
- Visual Relationships — How elements connect and relate
- Semantic Colors — Meaning-based color choices (red=warning, green=success)
- Style Characteristics — Line treatment, texture, mood
- Aspect Ratio + Quality — End with dimensions and detail level
Universal Prompt Suffix
Always append to prompts:
Clean composition with generous white space. Simple or no background. Main elements centered or positioned by content needs.
What to Avoid
- Vague descriptions ("a nice image", "something beautiful")
- Literal metaphor illustrations (visualize the concept, not the metaphor)
- Missing concrete labels/annotations
- Photorealistic human faces (use simplified silhouettes)
- Overly complex backgrounds
Two-Dimensional Design System
All image generation uses two independent dimensions that combine freely:
| Dimension | Controls | Question |
|---|---|---|
| Type/Layout | Information structure | How is content organized? |
| Style | Visual aesthetics | What does it look like? |
Example: infographic (type) × blueprint (style) = technical data visualization with engineering aesthetics.
Quick Reference: Type Selection
| Content | Recommended Type |
|---|---|
| Data, metrics, comparisons | infographic |
| Stories, emotions, atmosphere | scene |
| Processes, workflows, how-to | flowchart |
| A vs B, pros/cons | comparison |
| Architecture, models, concepts | framework |
| History, evolution, milestones | timeline |
| Product launch, announcements | hero (cover) |
| Multiple topics overview | bento-grid (infographic) |
| Abstract concept visualization | conceptual (cover) |
Quick Reference: Style Selection
| Content Tone | Recommended Style |
|---|---|
| Knowledge, tutorials, tech | vector-illustration or blueprint |
| SaaS, productivity | notion (minimal line art) |
| Personal, lifestyle | warm or watercolor |
| Opinion, editorial, cultural | screen-print or editorial |
| Education, teaching | chalkboard or craft-handmade |
| Business, professional | elegant |
| Fun, casual, playful | flat-doodle or kawaii |
| Retro, nostalgic | pixel-art or retro |
Quick Reference: Color Palette
| Mood | Palette | Key Colors |
|---|---|---|
| Friendly, warm | warm |
Orange #ED8936, Golden #F6AD55, Cream bg |
| Professional | elegant |
Slate #475569, Gold #D4AF37, Ivory bg |
| Technical | cool |
Blue #2563EB, Teal #0D9488, Light bg |
| Premium, cinematic | dark |
Charcoal #1A1A2E, Purple #7C3AED, Dark bg |
| Natural, organic | earth |
Green #059669, Brown #92400E, Warm bg |
| Bold, energetic | vivid |
Red #EF4444, Cyan #06B6D4, White bg |
| Gentle, creative | pastel |
Pink #FCA5A5, Lavender #C4B5FD, Soft bg |
| Zen, minimal | mono |
Grays + one accent, White bg |
Scenario Templates
Infographic / Data Visualization
[Title] - Data Visualization
Layout: [grid/radial/hierarchical]
ZONES:
- Zone 1: [data point with specific values]
- Zone 2: [comparison with metrics]
- Zone 3: [summary/conclusion]
LABELS: [specific numbers, percentages, terms]
COLORS: [semantic color mapping]
STYLE: [style characteristics]
ASPECT: 16:9
Scene / Atmospheric
[Title] - Atmospheric Scene
FOCAL POINT: [main subject]
ATMOSPHERE: [lighting, mood, environment]
MOOD: [emotion to convey]
COLOR TEMPERATURE: [warm/cool/neutral]
STYLE: [style characteristics]
ASPECT: 16:9
Flowchart / Process
[Title] - Process Flow
Layout: [left-right/top-down/circular]
STEPS:
1. [Step] - [description]
2. [Step] - [description]
CONNECTIONS: [arrow types, decision points]
STYLE: [style characteristics]
ASPECT: 16:9
Comparison
[Title] - Comparison View
LEFT SIDE - [Option A]:
- [Point 1]
- [Point 2]
RIGHT SIDE - [Option B]:
- [Point 1]
- [Point 2]
DIVIDER: [visual separator]
STYLE: [style characteristics]
ASPECT: 16:9
Cover Image
[Title] - Cover Image
TYPE: [hero/conceptual/typography/metaphor/scene/minimal]
FOCAL ELEMENT: [main visual]
TITLE TEXT: [exact title, language]
TEXT LEVEL: [none/title-only/title-subtitle]
PALETTE: [color scheme]
RENDERING: [flat-vector/hand-drawn/painterly/digital]
MOOD: [subtle/balanced/bold]
ASPECT: 16:9
Composition Principles
| Principle | Guideline |
|---|---|
| White space | 40-60% breathing room |
| Visual anchor | Main element centered or offset left |
| Characters | Simplified silhouettes, NOT realistic humans |
| Text | Large, prominent, handwritten fonts preferred |
| Title | Use exact title from source, never invent |
| Hierarchy | Size = importance |
References
references/styles.md— 28+ visual styles with descriptions, color palettes, and compatibility matrixreferences/layouts.md— 21 infographic layouts with structure, best-for, and visual elementsreferences/palettes.md— 10 color palettes with hex values, decorative hints, and use casesreferences/templates.md— Detailed prompt templates with style-specific overridesreferences/auto-selection.md— Content signal → dimension mapping rulesreferences/sd-styles.csv— 1062 image generation style presets (name, prompt, negative_prompt) merged from 7 sources (Douleb 850+, Fooocus 200+, BBrother, rocketingdatascience, twri/sdxl_prompt_styler, jheidt)references/nano-banana.md— Nano Banana (Gemini 2.5 Flash) prompt guide: core principles, 7 style templates (3D chibi, product ads, retro, effects, style transfer, design, creative scenes), structured JSON prompts, mixed-style techniquesreferences/nano-banana-cases.csv— 100 real-world Nano Banana/GPT-4o prompt cases with zh/en bilingual prompts (from awesome-nano-banana)