Page tree
Skip to end of metadata
Go to start of metadata

We need to move our images off of Flickr onto a system that we have more control over and that provides better metadata about the images hosted. This would be a standalone application

Discover for an application

MVP requirements

  • accepts uploads of images, and provides an embeddable URL
    • should enforce image dimensions/aspect ratio to conform to style guidance
    • image should be made available as a static resource, ideally on a CDNed location
    • invalidates any CDN cache for changed images (see s3_website gem for an example)
    • image URL should include its ID to make reverse lookup in Daguerreotype easier (and avoid namespace clashes for images with similar/same names)
  • on upload, ImageMagick should create a thumbnail for previews in Daguerreotype
  • on upload, records who uploaded the image, and when
  • on upload, information should be entered about the image, including
    • who took it/where it was sourced (i.e. Nic D.B, Shutterstock, etc)
    • default alt-text
    • description of the image (example: "Parade during the University of Bath in 2017"
  • images should be replaceable without changing the URL (so changing an image does not necessitate a republish of pages that embed it)
  • images should be searchable/filterable by the users who last interacted with them (upload, update)

Desirable features

  • automated image resizing to generate images for different breakpoints in a src-set
  • use TypeCase user table to look up which orgs a user belongs to, so we can filter images uploaded by department
  • filterable search by content keywords, date added, whodunnit
  • image gallery creation and embedding
  • integration with Googl's Vision API for image descriptions/cataloging?

Features available if an API is added to TypeCase and TypeCase for Courses

There are a lot of useful features that would require interaction with the other two applications.

  • reporting of which pages an image is used on (see below)
  • reporting (and filtering) for images not currently in use on the site (and theoretically available for deletion)
    • automated deletion of images that haven't been used in a while?
  • upload of images, and return of embed and alt-text, directly from the TypeCase editor interface

Reporting of which pages embed an image

This would require exposing an API on both Daguerreotype and TypeCase (and TC for Courses). Every time a page is published, or deleted (and possibly saved to draft), TypeCase should send a message to Daguerreotype including the image ID (probably extracted from the embed URL), the owning org of the page, and possibly the whodunnit.

Upload from TypeCase editor

This would require exposing an API on Daguerreotype that would accept all the required information for an image, and respond with the alt-text and embed URL.

Even better would be if it automatically entered those into the right fields in TypeCase.

Original brainstorm stuff!

  • No labels

1 Comment

  1. "Daguerreotype" is a bit of a tall order for our users to spell and type repeatedly. I was going to suggest "dgrotype", but that doesn't read the same, really.