Skip to main content

Describe Image Node

The Describe Image node provides descriptions for given images using reference text and selected AI models. This powerful node enables image-to-text conversion, making it ideal for generating image captions, accessibility descriptions, content analysis, and automated image documentation.

Describe Image node

Basic Usage​

Use Text tool, Image Input, Display Text, and Describe Image Node for your process.

Inputs​

The Describe Image node accepts the following inputs:

Input​

  • Type: Image to be described
  • Mandatory: Required
  • Works best with: Image Input, File Upload, Generate Image output

Reference​

  • Type: Reference text to guide the description style or focus
  • Mandatory: Optional
  • Works best with: Text Input, Text node

Use reference text to specify what aspects of the image to focus on or what style of description is needed.

Overwrite System Prompt​

  • Type: Custom system prompt to replace default behavior
  • Mandatory: Optional
  • Works best with: Text Input, Text node

Use this to completely customize how the AI analyzes and describes images.

Outputs​

Output​

  • Type: Text description of the image
  • Works best with: Display Text, Document Download, AI General Prompt

The output provides a detailed description of the image based on the selected model and any reference text provided.

Configuration​

Model Selection​

Select the AI model to use for image description. The node supports a wide range of vision-capable models:

GPT-4o (Default in screenshot)​

GPT-4o is OpenAI's advanced multimodal model with excellent vision capabilities for detailed and accurate image descriptions.

Available Models​

Vision Models:

  • Mistral OCR: Specialized for optical character recognition and text extraction from images
  • GPT-5 Nano: Compact model for quick image analysis
  • Gemini 2.5 Flash Lite: Google's lightweight vision model for fast descriptions
  • GPT-4o-mini: Efficient model for general image descriptions
  • GPT-5 Mini: Compact OpenAI model with vision capabilities
  • Claude 3 Haiku: Anthropic's fast and efficient vision model
  • Claude 3.5 Haiku: Enhanced version of Claude Haiku
  • Gemini 2.5 Flash: Google's fast vision model
  • GPT-5 Codex: Specialized for code and technical diagrams
  • Gemini 2.5 Pro: Google's professional-grade vision model

Advanced Models:

  • GPT-5: Advanced OpenAI model with superior vision understanding
  • GPT-5 Chat Latest: Latest conversational model with vision
  • Grok 4: Advanced vision and reasoning capabilities
  • Claude Sonnet 4: Anthropic's balanced vision model
  • Claude Sonnet 4.5: Enhanced Sonnet with improved vision
  • GPT-4o: OpenAI's flagship multimodal model
  • Claude Opus 4.1: Anthropic's most powerful vision model
  • Claude Opus 4: High-capability vision analysis
  • Claude 3 Opus: Anthropic's advanced vision model
  • GPT-5 Pro: Professional-grade vision analysis
  • GPT-4.5 Preview: Preview of next-generation GPT vision
  • GPT-4 Vision: OpenAI's vision-specialized model
  • GPT-4.1 Mini: Compact advanced vision model
  • GPT-4.1 Nano: Ultra-compact vision model
  • Mathpix OCR: Specialized for mathematical notation and equations

Example Workflows​

Accessibility Image Descriptions​

Scenario: Generate detailed accessibility descriptions for images in educational content.

Describe Image Example

Steps to Create the Flow:

  1. Add a Start Node.

  2. Add and connect an Image Input or File Upload for the image to describe.

  3. Add and connect a Text node with reference instructions.

    • Example reference text:
    Provide a detailed accessibility description suitable for screen readers. Include:
    - Main subject and composition
    - Colors and visual elements
    - Text content if present
    - Spatial relationships
    - Important details for understanding

  4. Add and connect a Describe Image Node.

    • Select Model: GPT-4o for detailed descriptions
    • Connect Image Input to Input
    • Connect Text node to Reference

  5. Add and connect a Display Text to show the description.

Result: Images are automatically described with detailed, accessibility-friendly text suitable for screen readers and visually impaired users.