Skip to content
Nappai Documentation

Describe Image

The Describe Image component lets you upload an image or provide a link to an image, and it uses an AI language model to produce a natural‑language description of what the image contains. This can be useful for adding context to images, generating alt text, or creating summaries for visual content.

How it Works

When you drop an image into the component or give a URL, the component first prepares the image data. It then sends the image to the ImageAnalyzer, which uses the selected language model to run a prompt (default: “Describe the image.”) on the image. The model returns a text description, which the component outputs as a Data object. You can also choose to resize the image to 255 × 255 or 512 × 512 pixels before analysis, or keep the original size.

Inputs

  • Image Data: The image you want described. It can be a direct image file, a text string, an object containing a URL key, or a list of such objects.
  • Model: The AI language model that will generate the description. This must be connected to a LanguageModel component.
  • Image/s Path: A path or URL pointing to the image(s) to be described. If you provide a URL, the component will fetch the image from that location.
  • Prompt Text: The prompt that will be sent to the language model. By default it says “Describe the image.”, but you can customize it to ask for specific details.
  • Resize Image: Choose whether to keep the original image size or resize it to 255 × 255 or 512 × 512 pixels before sending it to the model.

Outputs

  • Data: A list of Data objects containing the AI‑generated description of each image. You can use this output in subsequent components, such as a Text Display or a Database Insert.

Usage Example

  1. Drag the Describe Image component onto the canvas.
  2. Connect an Image Loader (or provide a URL) to the Image Data input.
  3. Connect a Language Model (e.g., GPT‑4) to the Model input.
  4. Leave the Prompt Text as the default or change it to “Give a detailed description of the product in the photo.”
  5. Keep the Resize Image option set to “keep original size.”
  6. Run the workflow. The component will output a Data object with the description, which you can then feed into a Text Display component to show the alt text on your dashboard.
  • Image Loader – Load images from local files or URLs.
  • Text Display – Show text output on the dashboard.
  • Database Insert – Store the description in a database for later use.

Tips and Best Practices

  • Use the Resize Image option if you want faster processing; smaller images reduce the amount of data sent to the model.
  • Keep the Prompt Text concise to avoid overly long descriptions.
  • If you’re describing sensitive images, double‑check the output for privacy compliance before publishing.

Security Considerations

  • Images may contain personal or confidential information. Ensure that any image data you process complies with your organization’s privacy policies.
  • The component sends image data to the selected language model; if the model is hosted externally, be aware of data transmission over the network. Use secure connections (HTTPS) whenever possible.