Image Generation Skill

AI-powered visual asset creation for content, marketing, and presentation workflows.

When to Use

Activate this skill when the task involves:

• Creating product imagery or renders

• Generating marketing visuals

• Building infographics and diagrams

• Creating presentation assets

• Producing social media graphics

Capabilities

1. Photorealistic Renders

Generate high-fidelity product and scene imagery:

• Model: Imagen 3 (Google)

• Resolution: Up to 4K (4096x4096)

• Styles: Photorealistic, studio, lifestyle, editorial

2. Marketing Visuals

Create branded marketing assets:

• Social media graphics (1:1, 9:16, 16:9)

• Banner ads and hero images

• Email header graphics

• Campaign keyart

3. Infographics

Generate data-driven visuals:

• Process diagrams

• Comparison charts

• Timeline graphics

• Icon sets

4. Creative Concepts

Produce artistic and conceptual imagery:

• Abstract backgrounds

• Artistic illustrations

• Mood boards

• Style explorations

Execution Pattern

1. BRIEF → Define visual requirements and context
2. PROMPT → Craft detailed generation prompt
3. GENERATE → Execute image generation request
4. REVIEW → Evaluate output quality and accuracy
5. REFINE → Iterate with enhanced prompts if needed
6. DELIVER → Export in required formats and sizes

Prompt Engineering

Structure

[Subject] + [Style] + [Composition] + [Lighting] + [Technical]

Example Prompts

Product Render:

Professional product photography of a modern steam cleaner, studio lighting, white background, 45-degree angle, high-key lighting, sharp focus, 8K resolution, commercial quality

Marketing Visual:

Dynamic hero image for cleaning services website, professional cleaner in action, modern office environment, shallow depth of field, warm natural lighting, contemporary style, aspirational mood

Infographic Element:

Flat design icon set for cleaning industry, minimalist style, consistent line weight, cohesive color palette (blue, green, white), vector-ready, transparent background

Output Format

<image_output> <metadata> <prompt>Original generation prompt</prompt> <model>imagen-3</model> <resolution>2048x2048</resolution> <seed>12345</seed> </metadata>

<files> <file format="png" size="original" path="..." /> <file format="webp" size="web-optimized" path="..." /> <file format="jpg" size="thumbnail" path="..." /> </files>

<variations> <variation id="1" prompt_modifier="..." path="..." /> </variations> </image_output>

Aspect Ratios

Ratio Dimensions Use Case

1:1 1024x1024 Social media, thumbnails

4:3 1024x768 Presentations

16:9 1920x1080 Hero images, YouTube

9:16 1080x1920 Stories, shorts

2:3 800x1200 Pinterest, posters

Integration Points

• Google Slides Storyboard: Supplies presentation assets

• Content Orchestrator: Receives asset generation requests

• Social Commander: Provides social media graphics

Quality Guidelines

Do's

• Use specific, detailed prompts

• Include technical specifications

• Request multiple variations

• Specify lighting and composition

Don'ts

• Avoid vague descriptions

• Don't request copyrighted elements

• Avoid text in images (use overlays instead)

• Don't exceed model capabilities

Brand Consistency

When generating branded assets:

• Reference brand color codes (hex values)

• Specify typography style preferences

• Include brand element descriptions

• Maintain visual language consistency

Error Handling

Error Recovery

Generation fails Simplify prompt, retry

Low quality output Enhance prompt specificity

Wrong style Add explicit style modifiers

Policy rejection Rephrase potentially flagged terms

Cost Considerations

• Fuel Cost: 5-20 PTS per generation

• Optimization:

• Batch similar requests

• Cache reusable assets

• Use lower resolution for drafts