ModelHub Image Generation Plugin

A Dify plugin that generates images using the nanobanana model via ModelHub API. Supports both text-to-image and image-to-image generation with customizable size and aspect ratio options.

Features

• 🎨 Text-to-Image Generation: Create images from text prompts
• 🖼️ Image-to-Image Generation: Transform existing images based on text descriptions
• 📐 Multiple Aspect Ratios: Support for 1:1 (square), 9:16 (portrait), and 16:9 (landscape)
• 🎯 Multiple Resolutions: Choose from 1K, 2K, or 4K image sizes
• 🔒 Secure API Key Storage: API keys are encrypted using PKCS1 OAEP encryption

Requirements

• Dify platform (version compatible with plugin SDK 0.4.0-0.7.0)
• Python 3.12
• ModelHub API Key (obtain from https://api.easyart.cc/register?aff=Uimc)

Installation

1. Package the plugin:

   dify-plugin-windows-amd64 plugin package ./gemini
   

2. Upload to Dify:

   • Go to your Dify workspace
   • Navigate to Plugins section
   • Upload the generated plugin package

3. Configure Credentials:

   • After installation, configure the plugin credentials
   • Enter your ModelHub API Key
   • The API Server URL is pre-configured to https://api.easyart.cc and cannot be modified

Configuration

Credentials

The plugin requires the following credentials:

• API Key (required): Your ModelHub API key obtained from https://api.easyart.cc/register?aff=Uimc
• API Server Url (read-only): Pre-configured to https://api.easyart.cc

Parameters

When using the plugin in a Dify workflow, you can configure the following parameters:

Parameter	Type	Required	Default	Description
prompt	string	✅ Yes	-	Text description of the image you want to generate
image	string	❌ No	-	Input image URL or base64 encoded image data for image-to-image generation
size	select	❌ No	1K	Image size: 1K, 2K, or 4K
aspect_ratio	select	❌ No	1:1	Aspect ratio: 1:1 (square), 9:16 (portrait), or 16:9 (landscape)

Usage Examples

Text-to-Image Generation

Generate an image from a text prompt:

Prompt: "A serene mountain landscape at sunset with snow-capped peaks"
Size: 2K
Aspect Ratio: 16:9

Image-to-Image Generation

Transform an existing image:

Prompt: "Make this image look like a watercolor painting"
Image: [URL or base64 encoded image]
Size: 1K
Aspect Ratio: 1:1

Supported Image Input Formats

• URL: https://example.com/image.png
• Base64 Data URL: data:image/png;base64,iVBORw0KGgo...
• Base64 String: Raw base64 encoded image data

How It Works

1. The plugin receives user input (text prompt and optional image)
2. User preferences (size and aspect ratio) are applied
3. Data is sent to ModelHub API (api.easyart.cc) using the Gemini 3 Pro Image Preview model
4. Generated image is returned to the Dify workflow
5. Image is displayed in the workflow result or saved as a file

Technical Details

• Model: gemini-3-pro-image-preview
• API Endpoint: https://api.easyart.cc/v1beta/models/{model}:generateContent
• Image Formats: PNG, JPEG, GIF, WebP (supported for input and output)
• Encryption: API keys stored with PKCS1 OAEP encryption

Troubleshooting

Common Issues

Error: API Key 未配置

• Solution: Make sure you've configured your API Key in the plugin credentials settings

Error: 提示词不能为空

• Solution: Provide a text prompt when generating images

Error: 无法返回图片文件

• Solution: Check your API Key validity and ensure you have sufficient API credits

Generated images are always landscape

• Note: The aspect ratio parameter is passed to the API, but the model may not always strictly follow it. The plugin also injects aspect ratio instructions into the prompt to improve compliance.

Privacy & Security

• API keys are encrypted before storage
• User prompts and images are transmitted to ModelHub API for processing
• No data is permanently stored by the plugin itself
• See PRIVACY.md for detailed privacy information

Support

• API Key Registration: https://api.easyart.cc/register?aff=Uimc
• Author: easyart
• Version: 0.0.1

License

This plugin is provided as-is. Please refer to ModelHub's terms of service for API usage terms.

Changelog

Version 0.0.1

• Initial release
• Support for text-to-image and image-to-image generation
• Multiple aspect ratios and resolutions
• Secure credential storage