Gemini Imagen
Generate images from text prompts using Google's Gemini APIs. This plugin gives Claude Code the ability to generate images directly.
Quick Start
bash
# Generate an image python3 ~/.claude/plugins/gemini-imagen/skills/gemini-imagen/scripts/generate_image.py \ --prompt "A cute cartoon cat" \ --output cat.png
CRITICAL: Exact Model Names
Use ONLY these exact model strings:
| Model String | Speed | Best For |
|---|---|---|
gemini-2.5-flash-image | Fast (2-5s) | Drafts, iterations |
gemini-3-pro-image-preview | Slower (5-15s) | Quality, text rendering, 2K |
Common mistakes:
- •
gemini-2.5-flash-preview-05-20- WRONG (date suffixes are for text models) - •
gemini-2.5-pro-image- WRONG (doesn't exist) - •
gemini-3-flash-image- WRONG (doesn't exist)
Instructions
Step 1: Check API Key
bash
echo "GEMINI_API_KEY is ${GEMINI_API_KEY:+set}"
If not set, tell the user to run /imagen:setup.
Step 2: Install Dependencies
bash
pip install google-genai Pillow
Step 3: Generate Image
bash
python3 ~/.claude/plugins/gemini-imagen/skills/gemini-imagen/scripts/generate_image.py \ --prompt "YOUR PROMPT HERE" \ --output /path/to/output.png
Step 4: Verify Output
bash
ls -la /path/to/output.png
Model Selection
| Use Case | Model | Why |
|---|---|---|
| Iterating on prompts | gemini-2.5-flash-image | Fast feedback (2-5s) |
| Final asset | gemini-3-pro-image-preview | Best quality |
| Game sprites | gemini-2.5-flash-image | Many images, consistent |
| Text in image | gemini-3-pro-image-preview | Better typography |
| Batch generation | gemini-2.5-flash-image | Cost effective |
Post-Processing Options
Remove Watermarks (--remove-watermark)
Removes bright pixels from image corners. Very useful for cleaning up generated images.
Background Transparency (--transparent-bg)
Converts solid-color backgrounds to transparent. Great for sprites and icons.
bash
python3 generate_image.py \ --prompt "Character on gray background" \ --output char.png \ --remove-watermark \ --transparent-bg
Batch Generation
Generate multiple images from a file:
bash
# prompts.txt (one per line) python3 generate_image.py \ --batch prompts.txt \ --output-dir ./images/
Error Handling
| Error | Solution |
|---|---|
GEMINI_API_KEY not set | Run /imagen:setup |
Rate limit (429) | Wait 60s, script auto-retries |
Content policy (400) | Modify prompt |
No image in response | Add more detail to prompt |
Pillow not installed | Run pip install Pillow |
Script Reference
Location: scripts/generate_image.py
| Argument | Required | Description |
|---|---|---|
--prompt | Yes* | Text prompt |
--output | Yes* | Output file path (.png) |
--model | No | Model (default: gemini-3-pro-image-preview) |
--remove-watermark | No | Remove corner watermarks |
--transparent-bg | No | Make background transparent |
--bg-color | No | Background hex color (default: #3a3a3a) |
--batch | No | Prompts file (one per line) |
--output-dir | No | Directory for batch output |
*Required unless using --batch
Exit Codes:
- •0: Success
- •1: Missing API key
- •2: Generation failed
- •3: Invalid arguments
What This Plugin CAN Do
- •Generate images from text prompts
- •Select between fast and quality models
- •Remove watermarks from images
- •Make backgrounds transparent
- •Batch generate multiple images
What This Plugin CANNOT Do
- •Use non-Gemini models (DALL-E, Midjourney, Stable Diffusion)
- •Generate video or audio
- •Bypass content policy restrictions