Files

GPT Image 2 Tool Plugin

A QwenPaw tool plugin that enables image generation and editing using OpenAI's GPT Image 2 model.

Features

  • Generate images from text prompts
  • Edit/generate images using reference images (1-16 images)
  • Support for multiple image sizes (1024x1024, 1024x1536, 1536x1024, auto)
  • Quality options: low, medium, high, auto
  • High fidelity image processing (automatic for gpt-image-2)
  • Pure backend implementation - no frontend code required

Installation

qwenpaw plugin install /path/to/gpt-image2

Or from ZIP:

qwenpaw plugin install gpt-image2-tool.zip

Configuration

  1. Start QwenPaw application
  2. Navigate to Agent Settings → Tools
  3. Find the GPT Image 2 tools:
    • generate_image_gpt (🎨 icon) - Generate images from text
    • edit_image_gpt (🖼️ icon) - Edit images with reference images
  4. Click "Configure" button for each tool
  5. Enter your OpenAI API Key (get it from https://platform.openai.com/api-keys)
  6. Save configuration
  7. Enable the tools you want to use

Usage

Once configured and enabled, the Agent can automatically call these tools when asked to generate or edit images.

Example 1: Generate Image from Text

User: Please generate an image of a serene mountain landscape at sunset

Agent: [Calls generate_image_gpt tool with appropriate parameters]

Example 2: Edit Image with Reference

User: I have a photo at /path/to/my/photo.jpg, please make it look like a watercolor painting

Agent: [Calls edit_image_gpt tool with reference image and prompt]

Example 3: Generate from Multiple References

User: Use these product images to create a gift basket: image1.png, image2.jpg, https://example.com/image3.png

Agent: [Calls edit_image_gpt tool with multiple reference images]

Tool Parameters

generate_image_gpt

Generate an image using OpenAI GPT Image 2 model from text prompt only.

Parameters:

  • prompt (str, required): Text description of the image to generate
  • size (str, optional): Image size, one of "1024x1024", "1024x1792", "1792x1024" (default: "1024x1024")
  • quality (str, optional): Quality level, one of "low", "medium", "high", "auto" (default: "auto")

Returns:

  • ImageBlock with the generated image
  • TextBlock with generation metadata

edit_image_gpt

Edit or generate image using reference images with OpenAI GPT Image 2 model.

Parameters:

  • prompt (str, required): Text description of the desired image edit or generation
  • reference_images (List[str], required): List of 1-16 reference images. Each can be:
    • Local file path (e.g., /path/to/image.png, ./photo.jpg)
    • Web URL (e.g., https://example.com/image.png)
    • Note: Local files are automatically converted to base64
  • size (str, optional): Image size, one of "1024x1024", "1024x1536", "1536x1024", "auto" (default: "1024x1024")
  • quality (str, optional): Quality level, one of "low", "medium", "high", "auto" (default: "auto")

Note: GPT Image 2 always processes images at high fidelity and does not support the input_fidelity parameter.

Returns:

  • ImageBlock with the edited/generated image
  • TextBlock with editing metadata

Supported Image Formats:

  • PNG (.png)
  • JPEG (.jpg, .jpeg)
  • WebP (.webp)

Requirements

  • QwenPaw >= 1.1.6
  • httpx >= 0.24.0
  • Valid OpenAI API key with access to GPT Image 2

Pricing

GPT Image 2 usage is billed by OpenAI. See https://openai.com/pricing for current pricing.

Troubleshooting

Tool not showing up

  • Ensure the plugin is installed: qwenpaw plugin list
  • Check QwenPaw logs: ~/.qwenpaw/logs/qwenpaw.log
  • Restart QwenPaw after installation

API errors

  • Verify your API key is correct
  • Check your OpenAI account has sufficient credits
  • Ensure you have access to GPT Image 2 model

Configuration not saving

  • Check file permissions in ~/.qwenpaw/plugins/
  • Review logs for error messages

Development

This is a pure backend plugin. To modify:

  1. Edit tool.py for tool logic
  2. Edit plugin.py for registration logic
  3. Edit plugin.json for metadata
  4. Reinstall with --force flag

License

Same as QwenPaw

Support

For issues and feature requests, please use the QwenPaw issue tracker.