The <image> data type can be represented with any of the following:

• An image denoted by the url() data type
• A <gradient> data type
• A part of the webpage, defined by the element() function
• An image, image fragment or solid patch of color, defined by the image() function
• A blending of two or more images defined by the cross-fade() function.
• A selection of images chosen based on resolution defined by the image-set() function.

CSS can handle the following kinds of images:

• Images with intrinsic dimensions (a natural size), like a JPEG, PNG, or other raster format.
• Images with multiple intrinsic dimensions, existing in multiple versions inside a single file, like some .ico formats. (In this case, the intrinsic dimensions will be those of the image largest in area and the aspect ratio most similar to the containing box.)
• Images with no intrinsic dimensions but with an intrinsic aspect ratio between its width and height, like an SVG or other vector format.
• Images with neither intrinsic dimensions, nor an intrinsic aspect ratio, like a CSS gradient.

CSS determines an object's concrete size using (1) its intrinsic dimensions; (2) its specified size, defined by CSS properties like width, height, or background-size; and (3) its default size, determined by the kind of property the image is used with:

The concrete object size is calculated using the following algorithm:

• If the specified size defines both the width and the height, these values are used as the concrete object size.
• If the specified size defines only the width or only the height, the missing value is determined using the intrinsic ratio, if there is any, the intrinsic dimensions if the specified value matches, or the default object size for that missing value.
• If the specified size defines neither the width nor the height, the concrete object size is calculated so that it matches the intrinsic aspect ratio of the image but without exceeding the default object size in any dimension. If the image has no intrinsic aspect ratio, the intrinsic aspect ratio of the object it applies to is used; if this object has none, the missing width or height are taken from the default object size.

Browsers do not provide any special information on background images to assistive technology. This is important primarily for screen readers, as a screen reader will not announce its presence and therefore convey nothing to its users. If the image contains information critical to understanding the page's overall purpose, it is better to describe it semantically in the document.

• MDN Understanding WCAG, Guideline 1.1 explanations
• Understanding Success Criterion 1.1.1 | W3C Understanding WCAG 2.0

<image> = 
  <url>       |
  <gradient>  

<url> = 
  url( <string> <url-modifier>* )  |
  src( <string> <url-modifier>* )  

• <gradient>
• element() Experimental
• image()
• image-set()
• cross-fade()