How to use Describe

1. Drop in an image or a video, or paste one.
2. Pick Short or Detailed. Switch off any fact you don't want under Include.
3. Read it over, fix names and context only you know, then copy it into the alt text field.

Whose guidelines does it follow?

Alexa Heinrich's Accessible Social. It skips "photo of" and "image of" because screen readers already say it's an image, names the medium when it isn't a photo, leads with the viewpoint, quotes every word of text in the image, leaves out emoji and ™ © ® marks, and calls people "person" rather than guessing who they are.

Is it AI?

Not the kind that writes. Fixed vision models count objects, read text and pick labels from a set list, and templates turn those facts into sentences. Accessible Social's advice still applies: an automatic description is a starting point, so add what only you know.

Source: Accessible Social by Alexa Heinrich