Skip to content

Image Generation

Generate AI images using Google Gemini and Imagen models.

Overview

Image generation uses generateContent API method (synchronous) and supports text-to-image generation with various aspect ratios and resolutions.

Supported Models

Model Quality Cost per Image
gemini-3-pro-image-preview Best (Default) $0.134
imagen-3 Good $0.03
imagen-4 Good $0.04

Default Model

gemini-3-pro-image-preview is the default model for best quality.

Basic Usage

# Simple image generation
python image_gen.py "A futuristic city at sunset"

# With options
python image_gen.py "A cat sitting on a couch" \
  --aspect-ratio 16:9 \
  --image-size 2K

Command Line Options

Option Values Default Description
--aspect-ratio 1:1, 16:9, 9:16, 4:3, 3:4 1:1 Image aspect ratio
--image-size 1K, 2K 1K Image resolution
--model gemini-3-pro-image-preview Default Model to use
--estimate-cost Flag - Show cost without generating
--project Project name default Config project to use

Aspect Ratios

Ratio Best For
1:1 Social media, avatars, icons
16:9 Widescreen, desktop wallpaper
9:16 Mobile, stories, vertical video
4:3 Traditional photo format
3:4 Portrait photos

Examples

Social Media

# Instagram square
python image_gen.py "A coffee cup with latte art" --aspect-ratio 1:1

# Instagram story
python image_gen.py "Mountain landscape" --aspect-ratio 9:16

Desktop Wallpaper

python image_gen.py "Abstract geometric patterns, purple and blue" \
  --aspect-ratio 16:9 \
  --image-size 2K

Documentation/Diagrams

python image_gen.py "Architecture diagram for microservices" \
  --doc README.md

Cost Estimation

Preview cost before generating:

python image_gen.py "A beautiful sunset" --estimate-cost

Output: 

💰 Cost Estimate:
   Model: gemini-3-pro-image-preview
   Count: 1
   Price: $0.134 per image
   Total: $0.134

Output

Images are saved to +generated_images/ directory:

+generated_images/
├── image_20260123_123456.png
├── image_20260123_123500.png
└── ...

Using with Documents

Generate images based on document content:

python image_gen.py "Diagram of the system" --doc design.md

The script reads the document and incorporates context into the prompt.

API Differences from Video

Aspect Image Video
API Method generateContent predictLongRunning
Response Synchronous Asynchronous (polling)
Free Tier ✅ Gemini 2.0 Flash ❌ None
Generation Time ~5-15 seconds ~60-180 seconds

Troubleshooting

"Unknown provider"

Some models may not be available:

# Use default model instead
python image_gen.py "prompt"  # Uses gemini-3-pro-image-preview

"Rate limit exceeded"

Wait a few seconds between requests or reduce frequency.

Image not generated

  • Check API key is valid
  • Ensure prompt follows content guidelines
  • Try simplifying the prompt

Learn More