Google AI Image Generation

Generate images using Google's Imagen and Gemini models with automatic cost tracking.

Quick Start

1. Present Model Options

When a user wants to generate images, first show them the top 3 options:

npx tsx .claude/skills/google-image-creator/scripts/list-models.ts

This displays:

• Gemini 2.5 Flash Image ($0.039/image) - Best value, fastest

• Imagen 4 Fast ($0.020/image) - Photorealistic quality

• Imagen 4 Ultra ($0.060/image) - Highest quality

Let the user choose, or recommend Gemini 2.5 Flash for most use cases.

2. Generate Image

npx tsx .claude/skills/google-image-creator/scripts/generate-image.ts
"prompt here"
"model-id"
"./output.png"

Example:

npx tsx .claude/skills/google-image-creator/scripts/generate-image.ts
"sunset over mountains, photorealistic"
"gemini-2.5-flash-image"
"./sunset.png"

3. Report Cost

The script automatically reports:

• Images generated

• Model used

• Estimated cost

Example output:

✅ Image generated successfully! Model: Gemini 2.5 Flash Image (gemini-2.5-flash-image) Prompt: "sunset over mountains" Saved to: ./sunset.png Images: 1 Cost: $0.0390

Prerequisites

API Key Required:

User must have GOOGLE_API_KEY or GEMINI_API_KEY set:

export GOOGLE_API_KEY="user-api-key-here"

Get key at: https://aistudio.google.com/app/apikey

Note: All Google image generation is paid tier only (no free quota).

Workflow

Standard Workflow

User Request: "Generate an image of X"

Ask Model Preference: "Which model would you like to use?" Then run:

npx tsx .claude/skills/google-image-creator/scripts/list-models.ts

User Chooses: User selects model or you recommend based on needs

Generate: Run generation script with chosen model

Report: Display cost and save location

Quick Workflow (Skip Selection)

If user says "use the cheapest" or doesn't specify:

• Default to gemini-2.5-flash-image

• Generate immediately

• Report cost afterward

Model Selection Guide

Use Gemini 2.5 Flash when:

• User wants lowest cost

• High volume generation

• Speed is priority

• General-purpose images

Use Imagen 4 Fast when:

• User needs photorealism

• Balance of quality and cost

• Simple text-to-image only

Use Imagen 4 Ultra when:

• User explicitly requests highest quality

• Client-facing deliverables

• Budget allows premium pricing

Cost Tracking

Always report costs after generation. For multiple images, sum costs:

Image 1: $0.039 Image 2: $0.039 Image 3: $0.039

Total: $0.117

Error Handling

Common Errors:

Error Solution

"API key not set" Tell user to set GOOGLE_API_KEY

"Unknown model" Run list-models.ts to show valid models

"API Error 401" API key invalid - user needs to verify key

"API Error 429" Rate limit - wait and retry

"Quota exceeded" Gemini image generation requires billing-enabled API key (free tier = 0 quota)

"Imagen API is only accessible to billed users" Enable billing on your Google Cloud project

Advanced: Multiple Images

For generating multiple images, run script multiple times and sum costs:

for i in {1..5}; do npx tsx .claude/skills/google-image-creator/scripts/generate-image.ts
"landscape $i"
"gemini-2.5-flash-image"
"./landscape_$i.png" done

echo "Total: 5 images × $0.039 = $0.195"

Updating Model Information

To update pricing or add new models:

• Check official pricing: https://ai.google.dev/pricing

• Update references/models.md

• Update scripts/generate-image.ts MODEL_PRICING

• Update scripts/list-models.ts TOP_3_MODELS

• Repackage skill

Files to update:

• references/models.md

• Full documentation

• references/api-guide.md

• API patterns

• scripts/generate-image.ts

• MODEL_PRICING object

• scripts/list-models.ts

• TOP_3_MODELS array

References

Detailed Documentation:

• references/models.md - All 6 models with full specs

• references/api-guide.md - API setup and patterns

Official Links:

• Pricing: https://ai.google.dev/pricing

• Imagen Docs: https://ai.google.dev/gemini-api/docs/imagen

• Gemini Image Docs: https://ai.google.dev/gemini-api/docs/image-generation

• Get API Key: https://aistudio.google.com/app/apikey

Examples

Example 1: Basic Generation

User: "Generate a sunset image"

You: "I'll generate a sunset image. Which model would you like?"

npx tsx .claude/skills/google-image-creator/scripts/list-models.ts

User: "Use the recommended one"

You:

npx tsx .claude/skills/google-image-creator/scripts/generate-image.ts
"beautiful sunset over ocean, warm colors, photorealistic"
"gemini-2.5-flash-image"
"./sunset.png"

"✅ Generated sunset.png using Gemini 2.5 Flash Image. Cost: $0.0390"

Example 2: High-Quality Request

User: "Generate a professional product photo, highest quality"

You: "For highest quality, I recommend Imagen 4 Ultra ($0.06/image). Proceed?"

User: "Yes"

You:

npx tsx .claude/skills/google-image-creator/scripts/generate-image.ts
"professional product photo of ceramic mug, studio lighting, high detail"
"imagen-4.0-ultra-generate-001"
"./product.png"

"✅ Generated product.png using Imagen 4 Ultra. Cost: $0.0600"

Example 3: Budget-Conscious

User: "Generate 10 landscape images, keep it cheap"

You: "I'll use Gemini 2.5 Flash ($0.039/image). Total cost: $0.39 for 10 images."

Generate 10 images

Sum costs

"✅ Generated 10 landscapes. Total cost: $0.3900"

Tips

Prompt Engineering:

• Be specific: "sunset over mountains, warm colors, golden hour"

• Mention style: "photorealistic", "watercolor", "3D render"

• Include lighting: "soft natural light", "studio lighting"

• Specify details: "high resolution", "shallow depth of field"

Cost Optimization:

• Use Gemini 2.5 Flash for testing prompts

• Only use Imagen 4 Ultra for final deliverables

• Batch mode available for Gemini models (50% discount)

• Iterate on cheaper models first

File Management:

• Always specify output path

• Use descriptive filenames

• Organize by project/date

• Keep generated images for reference