Although image formats can be used to create fixed layouts, doing so can greatly reduce the accessibility
of the publication. When using images, keep the following guidelines in mind:
Avoid empty files that set an image to render using the CSS background-image property. Not only does it leave the image without a description, but
any user without a CSS-aware reading system will see an empty document (note that even if the
system does not handle fixed layout metadata, it will still render the publication). See the section on background images for more information.
Avoid embedding text into images. Doing so removes access from assistive technologies. Use the CSS
z-index property to define layering and position of text as needed
on the image.
Whenever possible, use figures and captions to add images to the fixed layout document (including the
description in the caption). If you are using fixed layouts to facilitate the rendering of image
galleries, plates and similar, you can use scipting to enahnce the default experience and remove the
caption from visual display. The caption will be accessible to those who need and can also be
presented generally in interesting ways (see the Samples section for an
interesting implementation using this technique).
Add a text serialization of the work if it is completely visual in nature. It's not true that users
who are blind or have low vision are not interested in reading these works, but fully representing
the publication typically requires more than image descriptions.