Skip to content

Media Metadata Generator

The Media Metadata Generator analyzes images in the media library and generates their metadata.

What it does

For an image, the expert generates:

Field Description
title The title of the media content — used as the image alt text
description A short description of the media content — used as the image caption

The image itself is downloaded by the bundle and sent to the sulu.ai platform for visual analysis, so the metadata describes what is actually in the picture. Additional context (collection, tags, licensing, …) can be added via custom context providers.

Metadata translation

Media metadata can also be translated into the other locales of the media, using the Translator — so alt texts and captions stay consistent across languages.

Image loading configuration

Because the bundle has to fetch the binary image data itself, the image_loader configuration matters here:

  • allow_insecure_https — disable TLS certificate verification, useful for local *.wip domains with self-signed certificates
  • authorization — send basic_auth or bearer_token credentials when your (staging) image host is protected
sulu_ai_platform:
    image_loader:
        authorization:
            type: 'basic_auth'
            value: '%env(STAGING_IMAGE_AUTH)%'   # Format: username:password

Permission

Gated by the sulu.ai.media_metadata_generator security context (category Sulu AI). Grant VIEW to roles that should see the generate action in the media library.