Media Queries Level 5

Editor’s Draft,

More details about this document
This version:
https://drafts.csswg.org/mediaqueries-5/
Latest published version:
https://www.w3.org/TR/mediaqueries-5/
Previous Versions:
Feedback:
CSSWG Issues Repository
Inline In Spec
Editors:
Dean Jackson (Apple)
Florian Rivoal (Invited Expert)
Tab Atkins Jr. (Google)
Daniel Libby (Microsoft)
Luke Warlow (Igalia)
Suggest an Edit for this Spec:
GitHub Editor

Abstract

Media Queries allow authors to test and query values or features of the user agent or display device, independent of the document being rendered. They are used in the CSS @media rule to conditionally apply styles to a document, and in various other contexts and languages, such as HTML and JavaScript.

Media Queries Level 5 describes the mechanism and syntax of media queries, media types, and media features. It extends and supersedes the features defined in Media Queries Level 4.

CSS is a language for describing the rendering of structured documents (such as HTML and XML) on screen, on paper, etc.

Status of this document

This is a public copy of the editors’ draft. It is provided for discussion only and may change at any moment. Its publication here does not imply endorsement of its contents by W3C. Don’t cite this document other than as work in progress.

Please send feedback by filing issues in GitHub (preferred), including the spec code “mediaqueries” in the title, like this: “[mediaqueries] …summary of comment…”. All issues and comments are archived. Alternately, feedback can be sent to the (archived) public mailing list www-style@w3.org.

This document is governed by the 03 November 2023 W3C Process Document.

The following features are at-risk, and may be dropped during the CR period:

“At-risk” is a W3C Process term-of-art, and does not necessarily imply that the feature is in danger of being dropped or delayed. It means that the WG believes the feature may have difficulty being interoperably implemented in a timely manner, and marking it as such allows the WG to drop the feature if necessary when transitioning to the Proposed Rec stage, without having to publish a new Candidate Rec without the feature first.

1. Introduction

This section is not normative.

In 1997, HTML4 [HTML401] defined a mechanism to support media-dependent style sheets, tailored for different media types. For example, a document may use different style sheets for screen and for print. In HTML, this can be written as:

<link rel="stylesheet" type="text/css" media="screen" href="style.css">
<link rel="stylesheet" type="text/css" media="print" href="print.css">

CSS adapted and extended this functionality with its @media and @import rules, adding the ability to query the value of individual features:

Inside a CSS style sheet, one can declare that sections apply to certain media types:
@media screen {
  * { font-family: sans-serif }
}

Similarly, stylesheets can be conditionally imported based on media queries:

@import "print-styles.css" print;

Media queries can be used with HTML, XHTML, XML [xml-stylesheet] and the @import and @media rules of CSS.

Here is the same example written in HTML, XHTML, XML, @import and @media:
<link media="screen and (color), projection and (color)"
      rel="stylesheet" href="example.css">

<link media="screen and (color), projection and (color)"
      rel="stylesheet" href="example.css" />

<?xml-stylesheet media="screen and (color), projection and (color)"
                 rel="stylesheet" href="example.css" ?>

@import url(example.css) screen and (color), projection and (color);

@media screen and (color), projection and (color) { … }

1.1. Module interactions

This module extends and supersedes [MEDIAQUERIES-4] and its predecessor [MEDIAQUERIES-3], which themselves built upon and replaced Cascading Style Sheets Level 2 Revision 1 (CSS 2.1) Specification § media#q7.0.

1.2. Values

Value types not defined in this specification, such as <integer>, <number> or <resolution>, are defined in [CSS-VALUES-4]. Other CSS modules may expand the definitions of these value types.

1.3. Units

The units used in media queries are the same as in other parts of CSS, as defined in [CSS-VALUES-4]. For example, the pixel unit represents CSS pixels and not physical pixels.

Relative length units in media queries are based on the initial value, which means that units are never based on results of declarations.

Note: For example, in HTML, the em unit is relative to the initial value of font-size, defined by the user agent or the user’s preferences, not any styling on the page. Note that this will also take into account additional restrictions the user might apply, such as minimum font sizes.

1.4. Prefers-* Media Features Security and Privacy

Information about a user can be used as an active fingerprinting vector. Analysis of impact pending, more information to be provided before spec is published.

User agents and developers implementing this specification need to be aware of this vector and take it into consideration when deciding whether to use the feature. Specifically `prefers-reduced-motion`, `prefers-color-scheme`, `prefers-reduced-transparency` and `prefers-reduced-data` are currently of concern for exploitation.

2. Media Queries

A media query is a method of testing certain aspects of the user agent or device that the document is being displayed in. Media queries are (almost) always independent of the contents of the document, its styling, or any other internal aspect; they’re only dependent on “external” information unless another feature explicitly specifies that it affects the resolution of Media Queries.

The syntax of a media query consists of an optional media query modifier, an optional media type, and zero or more media features:

media condition only not media type and media condition

A media query is a logical expression that is either true or false. A media query is true if:

Statements regarding media queries in this section assume the syntax section is followed. Media queries that do not conform to the syntax are discussed in § 3.2 Error Handling. I.e. the syntax takes precedence over requirements in this section.

Here is a simple example written in HTML:
<link rel="stylesheet" media="screen and (color)" href="example.css" />

This example expresses that a certain style sheet (example.css) applies to devices of a certain media type (screen) with certain feature (it must be a color screen).

Here is the same media query written in an @import-rule in CSS:

@import url(example.css) screen and (color);

User agents must re-evaluate media queries in response to changes in the user environment that they’re aware of, for example if the device is tiled from landscape to portrait orientation, and change the behavior of any constructs dependent on those media queries accordingly.

Unless another feature explicitly specifies that it affects the resolution of Media Queries, it is never necessary to apply a style sheet in order to evaluate expressions.

2.1. Combining Media Queries

Several media queries can be combined into a comma-separated media query list.

media query ,

A media query list is true if any of its component media queries are true, and false only if all of its component media queries are false.

For example, the following media query list is true if either the media type is screen and it’s a color device, or the media type is projection and it’s a color device:
@media screen and (color), projection and (color) { … }

An empty media query list evaluates to true.

For example, these are equivalent:
@media all { … }
@media { … }

2.2. Media Query Modifiers

A media query may optionally be prefixed by a single media query modifier, which is a single keyword which alters the meaning of the following media query.

2.2.1. Negating a Media Query: the not keyword

An individual media query can have its result negated by prefixing it with the keyword not. If the media query would normally evaluate to true, prefixing it with not makes it evaluate to false, and vice versa.

For example, the following will apply to everything except color-capable screens. Note that the entire media query is negated, not just the media type.
<link rel="stylesheet" media="not screen and (color)" href="example.css" />

2.2.2. Hiding a Media Query From Legacy user agents: the only keyword

The concept of media queries originates from HTML4 [HTML401]. That specification only defined media types, but had a forward-compatible syntax that accommodated the addition of future concepts like media features: it would consume the characters of a media query up to the first non-alphanumeric character, and interpret that as a media type, ignoring the rest. For example, the media query screen and (color) would be truncated to just screen.

Unfortunately, this means that legacy user agents using this error-handling behavior will ignore any media features in a media query, even if they’re far more important than the media type in the query. This can result in styles accidentally being applied in inappropriate situations.

To hide these media queries from legacy user agents, the media query can be prefixed with the keyword only. The only keyword has no effect on the media query’s result, but will cause the media query to be parsed by legacy user agents as specifying the unknown media type “only”, and thus be ignored.

In this example, the stylesheet specified by the <link> element will not be used by legacy user agents, even if they would normally match the screen media type.
<link rel="stylesheet" media="only screen and (color)" href="example.css" />

Note: Note that the only keyword can only be used before a media type. A media query consisting only of media features, or one with another media query modifier like not, will be treated as false by legacy user agents automatically.

Note: At the time of publishing this specification, such legacy user agents are extremely rare, and so using the only modifier is rarely, if ever, necessary.

2.3. Media Types

A media type is a broad category of user-agent devices on which a document may be displayed. The original set of media types were defined in HTML4, for the media attribute on <link> elements.

Unfortunately, media types have proven insufficient as a way of discriminating between devices with different styling needs. Some categories which were originally quite distinct, such as screen and handheld, have blended significantly in the years since their invention. Others, such as tty or tv, expose useful differences from the norm of a full-featured computer monitor, and so are potentially useful to target with different styling, but the definition of media types as mutually exclusive makes it difficult to use them in a reasonable manner; instead, their exclusive aspects are better expressed as media features such as grid or scan.

As such, the following media types are defined for use in media queries:

all
Matches all devices.
print
Matches printers, and devices intended to reproduce a printed display, such as a web browser showing a document in “Print Preview”.
screen
Matches all devices that aren’t matched by print.

In addition, the following deprecated media types are defined. Authors must not use these media types; instead, it is recommended that they select appropriate media features that better represent the aspect of the device that they are attempting to style against.

User agents must recognize the following media types as valid, but must make them match nothing.

Note: It is expected that all of the media types will also be deprecated in time, as appropriate media features are defined which capture their important differences.

2.4. Media Features

A media feature is a more fine-grained test than media types, testing a single, specific feature of the user agent or display device.

Syntactically, media features resemble CSS properties: they consist of a feature name, a colon, and a value to test for. They may also be written in boolean form as just a feature name, or in range form with a comparison operator.

( feature name : feature value feature name range form (see below) )

There are, however, several important differences between properties and media features:

If a media feature references a concept which does not exist on the device where the UA is running (for example, speech UAs do not have a concept of “width”), the media feature must always evaluate to false.

The media feature device-aspect-ratio only applies to visual devices. On an speech device, expressions involving device-aspect-ratio will therefore always be false:
<link media="speech and (device-aspect-ratio: 16/9)"
      rel="stylesheet" href="example.css">

2.4.1. Media Feature Types: “range” and “discrete”

Every media feature defines its “type” as either “range” or “discrete” in its definition table.

“Discrete” media features, like pointer take their values from a set. The values may be keywords or boolean numbers (0 and 1), but the common factor is that there’s no intrinsic “order” to them—​none of the values are “less than” or “greater than” each other.

“Range” media features like width, on the other hand, take their values from a range. Any two values can be compared to see which is lesser and which is greater.

The only significant difference between the two types is that “range” media features can be evaluated in a range context and accept “min-” and “max-” prefixes on their name. Doing either of these changes the meaning of the feature—​rather than the media feature being true when the feature exactly matches the given value, it matches when the feature is greater than/less than/equal to the given value.

A (width >= 600px) media feature is true when the viewport’s width is 600px or more.

On the other hand, (width: 600px) by itself is only true when the viewport’s width is exactly 600px. If it’s less or greater than 600px, it’ll be false.

2.4.2. Evaluating Media Features in a Boolean Context

While media features normally have a syntax similar to CSS properties, they can also be written more simply as just the feature name, like (color).

When written like this, the media feature is evaluated in a boolean context. If the feature would be true for any value other than the number 0, a <dimension> with the value 0, the keyword none, or a value explicitly defined by that media feature to evaluate as false in a boolean context, the media feature evaluates to true. Otherwise, it evaluates to false.

Some media features are designed to be written like this.

For example, update is typically written as (update) to test if any kind of updating is available, or not (update) to check for the opposite.

It can still be given an explicit value as well, with (update: fast) or (update: slow) equal to (update), and (update: none) equal to not (update).

Some numeric media features, like width, are rarely if ever useful to evaluate in a boolean context, as their values are almost always greater than zero. Others, like color, have meaningful zero values: (color) is identical to (color > 0), indicating that the device is capable of displaying color at all.
Only some of the media features that accept keywords are meaningful in a boolean context.

For example, (pointer) is useful, as pointer has a none value to indicate there’s no pointing device at all on the device. On the other hand, (scan) is just always true or always false (depending on whether it applies at all to the device), as there’s no value that means “false”.

2.4.3. Evaluating Media Features in a Range Context

Media features with a “range” type can be alternately written in a range context that takes advantage of the fact that their values are ordered, using ordinary mathematical comparison operators:

( feature name/value > <= < = >= feature value/name value < <= feature name < <= value value > >= feature name > >= value )

Note: This syntax is new to Level 4 of Mediaqueries, and thus is not as widely supported at the moment as the min-/max- prefixes.

The basic form, consisting of a feature name, a comparison operator, and a value, returns true if the relationship is true.

For example, (height > 600px) (or (600px < height)) returns true if the viewport height is greater than 600px.

The remaining forms, with the feature name nested between two value comparisons, returns true if both comparisons are true.

For example, (400px < width < 1000px) returns true if the viewport width is between 400px and 1000px (but not equal to either).

Some media features with a "range" type are said to be false in the negative range. This means that negative values are valid and must be parsed, and that querying whether the media feature is equal to, less than, or less or equal than any such negative value must evaluate to false. Querying whether the media feature is greater, or greater or equal, than a negative value evaluates to true if the relationship is true.

Note: If negative values had been rejected at parse time instead, they would be treated as unknown based on the error handling rules. However, in reality, whether a device’s resolution is -300dpi is not unknown, it is known to be false. Similarly, for any visual device, the width of the targeted display area is known to be greater than -200px The above rule reflects that, making intuition match what UAs do.

The following examples result in a green background on all visual devices:
@media not (width <= -100px) {
  body { background: green; }
}
@media (height > -100px) {
  body { background: green; }
}
@media not (resolution: -300dpi) {
  body { background: green; }
}
This is a behavior change compared to Media Queries Level 3 [MEDIAQUERIES-3], where negative values on these properties caused a syntax error. In level 3, syntax errors—including forbidden values—resulted in the entire media query being false, rather than the unknown treatment defined in this level. Implementations updating from level 3 should make sure to change the handling of negative values for the relevant properties when they add support for the richer syntax defined in § 2.5 Combining Media Features, to avoid introducing unintended semantics.

2.4.4. Using “min-” and “max-” Prefixes On Range Features

Rather than evaluating a “range” type media feature in a range context, as described above, the feature may be written as a normal media feature, but with a “min-” or “max-” prefix on the feature name.

This is equivalent to evaluating the feature in a range context, as follows:

Note: because “min-” and “max-” both equate to range comparisons that include the value, they may be limiting in certain situations.

For instance, authors trying to define different styles based on a breakpoint in the viewport width using “min-” and “max-” would generally offset the values they’re comparing, to ensure that both queries don’t evaluate to true simultaneously. Assuming the breakpoint is at 320px, authors would conceptually use:
@media (max-width: 320px) { /* styles for viewports <= 320px */ }
@media (min-width: 321px) { /* styles for viewports >= 321px */ }

While this ensures that the two sets of styles don’t apply simultaneously when the viewport width is 320px, it does not take into account the possibility of fractional viewport sizes which can occur as a result of non-integer pixel densities (e.g. on high-dpi displays or as a result of zooming/scaling). Any viewport widths that fall between 320px and 321px will result in none of the styles being applied.

One approach to work around this problem is to increase the precision of the values used for the comparison. Using the example above, changing the second comparison value to 320.01px significantly reduces the chance that a viewport width on a device would fall between the cracks.

@media (max-width: 320px) { /* styles for viewports <= 320px */ }
@media (min-width: 320.01px) { /* styles for viewports >= 320.01px */ }

However, in these situations, range context queries (which are not limited to “>=” and “<=” comparisons) offer a more appropriate solution:

@media (width <= 320px) { /* styles for viewports <= 320px */ }
@media (width > 320px) { /* styles for viewports > 320px */ }

“Discrete” type properties do not accept “min-” or “max-” prefixes. Adding such a prefix to a “discrete” type media feature simply results in an unknown feature name.

For example, (min-grid: 1) is invalid, because grid is a “discrete” media feature, and so doesn’t accept the prefixes. (Even though the grid media feature appears to be numeric, as it accepts the values 0 and 1.)

Attempting to evaluate a min/max prefixed media feature in a boolean context is invalid and a syntax error.

2.5. Combining Media Features

Multiple media features can be combined together into a media condition using full boolean algebra (not, and, or).

It is invalid to mix and and or and not at the same “level” of a media query. For example, (color) and (pointer) or (hover) is illegal, as it’s unclear what was meant. Instead, parentheses can be used to group things using a particular joining keyword, yielding either (color) and ((pointer) or (hover)) or ((color) and (pointer)) or (hover). These two have very different meanings: if only (hover) is true, the first one evaluates to false but the second evaluates to true.

3. Syntax

Informal descriptions of the media query syntax appear in the prose and railroad diagrams in previous sections. The formal media query syntax is described in this section, with the rule/property grammar syntax defined in [CSS-SYNTAX-3] and [CSS-VALUES-4].

To parse a <media-query-list> production, parse a comma-separated list of component values, then parse each entry in the returned list as a <media-query>. Its value is the list of <media-query>s so produced.

Note: This explicit definition of <media-query-list> parsing is necessary to make the error-recovery behavior of media query lists well-defined.

Note: This definition of <media-query-list> parsing intentionally accepts an empty list.

Note: As per [CSS-SYNTAX-3], tokens are ASCII case-insensitive.

<media-query> = <media-condition>
             | [ not | only ]? <media-type> [ and <media-condition-without-or> ]?
<media-type> = <ident>

<media-condition> = <media-not> | <media-in-parens> [ <media-and>* | <media-or>* ]
<media-condition-without-or> = <media-not> | <media-in-parens> <media-and>*
<media-not> = not <media-in-parens>
<media-and> = and <media-in-parens>
<media-or> = or <media-in-parens>
<media-in-parens> = ( <media-condition> ) | <media-feature> | <general-enclosed>

<media-feature> = ( [ <mf-plain> | <mf-boolean> | <mf-range> ] )
<mf-plain> = <mf-name> : <mf-value>
<mf-boolean> = <mf-name>
<mf-range> = <mf-name> <mf-comparison> <mf-value>
           | <mf-value> <mf-comparison> <mf-name>
           | <mf-value> <mf-lt> <mf-name> <mf-lt> <mf-value>
           | <mf-value> <mf-gt> <mf-name> <mf-gt> <mf-value>
<mf-name> = <ident>
<mf-value> = <number> | <dimension> | <ident> | <ratio>
<mf-lt> = '<' '='?
<mf-gt> = '>' '='?
<mf-eq> = '='
<mf-comparison> = <mf-lt> | <mf-gt> | <mf-eq>

<general-enclosed> = [ <function-token> <any-value>? ) ] | [ ( <any-value>? ) ]

The <media-type> production does not include the keywords only, not, and, or, and layer.

Note: The exclusion of layer is because it would otherwise be ambiguous when when used in the @import url() layer; syntax for the sake of cascade layers. See [CSS-CASCADE-5].

No whitespace is allowed between the “<” or “>” <delim-token>s and the following “=” <delim-token>, if it’s present.

Note: Whitespace is required between a not, and, or or keyword and the following ( character, because without it that would instead parse as a <function-token>. This is not made explicitly invalid because it’s already covered by the above grammar. It’s fine to have whitespace between a ) and a following keyword, however.

When parsing the <media-in-parens> production, the <general-enclosed> branch must only be chosen if the input does not match either of the preceding branches. <general-enclosed> exists to allow for future expansion of the grammar in a reasonably compatible way.

3.1. Evaluating Media Queries

Each of the major subexpression of <media-condition> or <media-condition-without-or> is associated with a boolean result, as follows:

<media-condition>
<media-condition-without-or>
The result is the result of the child subexpression.
<media-in-parens>
The result is the result of the child term.
<media-not>
The result is the negation of the <media-in-parens> term. The negation of unknown is unknown.
<media-in-parens> <media-and>*
The result is true if the <media-in-parens> child term and all of the <media-in-parens> children of the <media-and> child terms are true, false if at least one of these <media-in-parens> terms are false, and unknown otherwise.
<media-in-parens> <media-or>*
The result is false if the <media-in-parens> child term and all of the <media-in-parens> children of the <media-or> child terms are false, true if at least one of these <media-in-parens> terms are true, and unknown otherwise.
<general-enclosed>
The result is unknown.

Authors must not use <general-enclosed> in their stylesheets. It exists only for future-compatibility, so that new syntax additions do not invalidate too much of a <media-condition> in older user agents.

<media-feature>
The result is the result of evaluating the specified media feature.

If the result of any of the above productions is used in any context that expects a two-valued boolean, “unknown” must be converted to “false”.

Note: This means that, for example, when a media query is used in a @media rule, if it resolves to “unknown” it’s treated as “false” and fails to match.

Media Queries use a three-value logic where terms can be “true”, “false”, or “unknown”. Specifically, it uses the Kleene 3-valued logic. In this logic, “unknown” means “either true or false, but we’re not sure which yet”.

In general, an unknown value showing up in a formula will cause the formula to be unknown as well, as substituting “true” for the unknown will give the formula a different result than substituting “false”. The only way to eliminate an unknown value is to use it in a formula that will give the same result whether the unknown is replaced with a true or false value. This occurs when you have “false AND unknown” (evaluates to false regardless) and “true OR unknown” (evaluates to true regardless).

This logic was adopted because <general-enclosed> needs to be assigned a truth value. In standard boolean logic, the only reasonable value is “false”, but this means that not unknown(function) is true, which can be confusing and unwanted. Kleene’s 3-valued logic ensures that unknown things will prevent a media query from matching, unless their value is irrelevant to the final result.

3.2. Error Handling

A media query that does not match the grammar in the previous section must be replaced by not all during parsing.

Note: Note that a grammar mismatch does not wipe out an entire media query list, just the problematic media query. The parsing behavior defined above automatically recovers at the next top-level comma.

@media (example, all,), speech { /* only applicable to speech devices */ }
@media &test, speech           { /* only applicable to speech devices */ }

Both of the above media query lists are turned into not all, speech during parsing, which has the same truth value as just speech.

Note that error-recovery only happens at the top-level of a media query; anything inside of an invalid parenthesized block will just get turned into not all as a group. For example:

@media (example, speech { /* rules for speech devices */ }

Because the parenthesized block is unclosed, it will contain the entire rest of the stylesheet from that point (unless it happens to encounter an unmatched “)” character somewhere in the stylesheet), and turn the entire thing into a not all media query.

An unknown <media-type> must be treated as not matching.

For example, the media query unknown is false, as unknown is an unknown media type.

But not unknown is true, as the not negates the false media type.

Remember that some keywords aren’t allowed as <media-type>s and cause parsing to fail entirely: the media query or and (color) is turned into not all during parsing, rather than just treating the or as an unknown media type.

An unknown <mf-name> or <mf-value>, or a feature value which does not matches the value syntax for that media feature, results in the value “unknown”. A <media-query> whose value is “unknown” must be replaced with not all.

<link media="screen and (max-weight: 3kg) and (color), (color)"rel="stylesheet" href="example.css" />

As max-weight is an unknown media feature, this media query list is turned into not all, (color), which is equivalent to just (color).

@media (min-orientation:portrait) { … }

The orientation feature does not accept prefixes, so this is considered an unknown media feature, and turned into not all.

The media query (color:20example) specifies an unknown value for the color media feature and is therefore turned into not all.
Note that media queries are also subject to the parsing rules of the host language. For example, take the following CSS snippet:
@media test;,all { body { background:lime } }

The media query test;,all is, parsed by itself, equivalent to not all, all, which is always true. However, CSS’s parsing rules cause the @media rule, and thus the media query, to end at the semicolon. The remainder of the text is treated as a style rule with an invalid selector and contents.

4. Viewport/Page Characteristics Media Features

4.1. Width: the width feature

Name: width
For: @media
Value: <length>
Type: range

The width media feature describes the width of the targeted display area of the output device. For continuous media, this is the width of the viewport (as described by CSS2, section 9.1.1 [CSS2]) including the size of a rendered scroll bar (if any). For paged media, this is the width of the page box (as described by CSS2, section 13.2 [CSS2]).

<length>s are interpreted according to § 1.3 Units.

width is false in the negative range.

For example, this media query expresses that the style sheet is used on printed output wider than 25cm:
<link rel="stylesheet" media="print and (min-width: 25cm)" href="http://…" />
This media query expresses that the style sheet is used on devices with viewport (the part of the screen/paper where the document is rendered) widths between 400 and 700 pixels:
@media (400px <= width <= 700px) { … }
This media query expresses that style sheet is used if the width of the viewport is greater than 20em.
@media (min-width: 20em) { … }

The em value is relative to the initial value of font-size.

4.2. Height: the height feature

Name: height
For: @media
Value: <length>
Type: range

The height media feature describes the height of the targeted display area of the output device. For continuous media, this is the height of the viewport including the size of a rendered scroll bar (if any). For paged media, this is the height of the page box.

<length>s are interpreted according to § 1.3 Units.

height is false in the negative range.

4.3. Aspect-Ratio: the aspect-ratio feature

Name: aspect-ratio
For: @media
Value: <ratio>
Type: range

The aspect-ratio media feature is defined as the ratio of the value of the width media feature to the value of the height media feature.

4.4. Orientation: the orientation feature

Name: orientation
For: @media
Value: portrait | landscape
Type: discrete
portrait
The orientation media feature is portrait when the value of the height media feature is greater than or equal to the value of the width media feature.
landscape
Otherwise orientation is landscape.
The following media query tests for “portrait” orientation, like a phone held upright.
@media (orientation:portrait) { … }

4.5. Block-Axis Overflow: the overflow-block feature

Name: overflow-block
For: @media
Value: none | scroll | paged
Type: discrete

The overflow-block media feature describes the behavior of the device when content overflows the initial containing block in the block axis.

none
There is no affordance for overflow in the block axis; any overflowing content is simply not displayed. Examples: billboards
scroll
Overflowing content in the block axis is exposed by allowing users to scroll to it. Examples: computer screens
paged
Content is broken up into discrete pages; content that overflows one page in the block axis is displayed on the following page. Examples: printers, ebook readers

Media that match none or scroll are said to be continuous media, while those that match paged are said to be paged media

Note: Additional values for this media feature may be added in the future to describe classes of user agents with a hybrid behavior combining aspects of continuous and paged media. For example, the Presto layout engine (now discontinued) shipped with a semi-paginated presentation-mode behavior similar to continuous except that it honored forced page breaks. Not knowing of any currently-shipping user agent with this type of behavior, the Working Group has decided not to add such a value in this level to avoid mischaracterizing any such user agent. Anyone implementing a user agent not adequately described by any of the values specified above is encouraged to contact the Working Group so that extensions to this media feature may be considered.

4.6. Inline-Axis Overflow: the overflow-inline feature

Name: overflow-inline
For: @media
Value: none | scroll
Type: discrete

The overflow-inline media feature describes the behavior of the device when content overflows the initial containing block in the inline axis.

none
There is no affordance for overflow in the inline axis; any overflowing content is simply not displayed.
scroll
Overflowing content in the inline axis is exposed by allowing users to scroll to it.

Note: There are no known implementations of paged overflow of inline-overflowing content, and the very concept doesn’t seem to make much sense, so there is intentionally no paged value for overflow-inline.

4.7. Horizontal Viewport Segments: the horizontal-viewport-segments feature

Name: horizontal-viewport-segments
For: @media
Value: <integer>
Type: range

The horizontal-viewport-segments media feature describes the number of logical segments of the viewport in the horizontal direction.

The horizontal-viewport-segments media feature is false in the negative range.

When the viewport is split by one or more hardware features (such as a fold or a hinge between separate displays) that act as a logical divider, segments are the regions of the viewport that can be treated as logically distinct by the page.

4.8. Vertical Viewport Segments: the vertical-viewport-segments feature

Name: vertical-viewport-segments
For: @media
Value: <integer>
Type: range

The vertical-viewport-segments media feature describes the number of logical segments of the viewport in the vertical direction.

The vertical-viewport-segments media feature is false in the negative range.

This media query detects a viewport that has exactly two segments that are side-by-side.
@media (horizontal-viewport-segments: 2) and (vertical-viewport-segments: 1) { … }

4.9. Display Modes: the display-mode media feature

Name: display-mode
For: @media
Value: fullscreen | standalone | minimal-ui | browser | picture-in-picture
Type: discrete

The display-mode media feature describes the mode in which the current browsing context is currently being presented to the end user. In child browsing contexts, the display mode must match that of the top-level browsing context.

This feature is primarily used to determine which display mode the user agent has applied to an application context. As such, the values of this feature correspond to the display modes defined in [APPMANIFEST]. However, it can also be used in non-application contexts to determine whether the viewport is in other modes, such as fullscreen or picture-in-picture.

fullscreen
The browsing context is displayed with browser UI elements hidden and takes up the entirety of the available display area. The fullscreen context may have been caused by the fullscreen display mode in the application manifest, by the requestFullscreen() method of the Fullscreen API, or through some other means (such as the user manually activating fullscreen mode using the user agent’s built-in controls).

Corresponds to the fullscreen display mode.

standalone
The standalone display mode is in use. Only applicable in an application context.
minimal-ui
The minimal-ui display mode is in use. Only applicable in an application context.
browser
The browsing context is displayed using the platform-specific convention for opening hyperlinks in the user agent (e.g., in a browser tab or web browser window with controls such as an address bar). This should be used for non-application contexts where no other display mode is appropriate.

Corresponds to the browser display mode.

picture-in-picture
This mode allows users to continue consuming media while they interact with other sites or applications on their device. The browsing context is displayed in a floating and always-on-top window. A user agent may include other platform specific UI elements, such as "back-to-tab" and "site information" buttons or whatever is customary on the platform and user agent.
For example, the application manifest can request the standalone display mode as follows:
{
  "display": "standalone"
}

This media query can be used to determine whether the user agent has actually applied the standalone mode:

@media (display-mode: standalone) {}

The user agent could set display-mode to any of the other values, depending on the actual mode currently in use.

The fullscreen display mode is distinct from the Fullscreen API.

The fullscreen value for display-mode is not directly related to the CSS :fullscreen pseudo-class. The :fullscreen pseudo-class matches an element exclusively when that element is put into the fullscreen element stack. However, a side effect of calling the requestFullscreen() method on an element using the Fullscreen API can be that the browser enters a fullscreen mode at the OS-level, in which case both :fullscreen and (display-mode: fullscreen) will match.

On some platforms, it is possible for a user—​or a Web Application Manifest—​to put a web application into fullscreen without invoking the Fullscreen API. When this happens, the :fullscreen pseudo class will not match, but (display-mode: fullscreen) will match. This is exemplified in CSS code below:
/* applies when the viewport is fullscreen */
@media (display-mode: fullscreen) {}

/* applies when an element is fullscreen */
#game:fullscreen {}

Note: Additional values for this media feature may be added in the future to match new display modes added to [APPMANIFEST].

5. Display Quality Media Features

5.1. Display Resolution: the resolution feature

Name: resolution
For: @media
Value: <resolution> | infinite
Type: range

The resolution media feature describes the resolution of the output device, i.e. the density of the pixels, taking into account the page zoom but assuming a pinch zoom of 1.0.

The resolution media feature is false in the negative range

When querying media with non-square pixels, resolution queries the density in the vertical dimension.

For printers, this corresponds to the screening resolution (the resolution for printing dots of arbitrary color). Printers might have a different resolution for grayscale printing.

For output mediums that have no physical constraints on resolution (such as outputting to vector graphics), this feature must match the infinite value. For the purpose of evaluating this media feature in the range context, infinite must be treated as larger than any possible <resolution>. (That is, a query like (resolution > 1000dpi) will be true for an infinite media.)

This media query simply detects “high-resolution” screens (those with a hardware pixel to CSS px ratio of at least 2):
@media (resolution >= 2dppx)
For example, this media query expresses that a style sheet is used on devices with resolution greater than 300 dots per CSS in:
@media print and (min-resolution: 300dpi) { … }

This media query is equivalent, but uses the CSS cm unit:

@media print and (min-resolution: 118dpcm) { … }
<resolution> does not refer to the number of device pixels per physical length unit, but the number of device pixels per css unit. This mapping is done by the user agent, so it is always known to the user agent.

If the user agent either has no knowledge of the geometry of physical pixels, or knows about the geometry physical pixels and they are (close enough to) square, it would not map a different number of device pixels per css pixels along each axis, and the would therefore be no difference between the vertical and horizontal resolution.

Otherwise, if the UA chooses to map a different number along each axis, this would be to respond to physical pixels not being square either. How the UA comes to this knowledge is out of scope, but having enough information to take this decision, it can invert the mapping should the device be rotated 90 degrees.

5.2. Display Type: the scan feature

Name: scan
For: @media
Value: interlace | progressive
Type: discrete

The scan media feature describes the scanning process of some output devices.

interlace
CRT and some types of plasma TV screens used “interlaced” rendering, where video frames alternated between specifying only the “even” lines on the screen and only the “odd” lines, exploiting various automatic mental image-correction abilities to produce smooth motion. This allowed them to simulate a higher FPS broadcast at half the bandwidth cost.

When displaying on interlaced screens, authors should avoid very fast movement across the screen to avoid “combing”, and should ensure that details on the screen are wider than 1px to avoid “twitter”.

progressive
A screen using “progressive” rendering displays each screen fully, and needs no special treatment.

Most modern screens, and all computer screens, use progressive rendering.

For example, the “feet” of letters in serif fonts are very small features that can provoke “twitter” on interlaced devices. The scan media feature can be used to detect this, and use an alternative font with less chance of “twitter”:
@media (scan: interlace) { body { font-family: sans-serif; } }

Note: At the time of writing, all known implementations match scan: progressive rather than scan: interlace.

5.3. Detecting Console Displays: the grid feature

Name: grid
For: @media
Value: <mq-boolean>
Type: discrete

The grid media feature is used to query whether the output device is grid or bitmap. If the output device is grid-based (e.g., a “tty” terminal, or a phone display with only one fixed font), the value will be 1. Otherwise, the value will be 0.

The <mq-boolean> value type is an <integer> with the value 0 or 1. Any other integer value is invalid. Note that -0 is always equivalent to 0 in CSS, and so is also accepted as a valid <mq-boolean> value.

Note: The <mq-boolean> type exists only for legacy purposes. If this feature were being designed today, it would instead use proper named keywords for its values.

Here is an example that detects a narrow console screen:
@media (grid) and (max-width: 15em) { … }

Note: At the time of writing, all known implementations match grid: 0 rather than grid: 1.

5.4. Display Update Frequency: the update feature

Name: update
For: @media
Value: none | slow | fast
Type: discrete

The update media feature is used to query the ability of the output device to modify the appearance of content once it has been rendered. It accepts the following values:

none
Once it has been rendered, the layout can no longer be updated. Example: documents printed on paper.
slow
The layout may change dynamically according to the usual rules of CSS, but the output device is not able to render or display changes quickly enough for them to be perceived as a smooth animation. Example: E-ink screens or severely under-powered devices.
fast
The layout may change dynamically according to the usual rules of CSS, and the output device is not unusually constrained in speed, so regularly-updating things like CSS Animations can be used. Example: computer screens.
For example, if a page styles its links to only add underlines on hover, it may want to always display underlines when printed:
@media (update) {
  a { text-decoration: none; }
  a:hover, a:focus { text-decoration: underline; }
}
/* In non-updating UAs, the links get their default underline at all times. */

5.5. Detecting the display technology: the environment-blending feature

Name: environment-blending
For: @media
Value: opaque | additive | subtractive
Type: discrete

The environment-blending media feature is used to query the characteristics of the user’s display so the author can adjust the style of the document. An author might choose to adjust the visuals and/or layout of the page depending on the display technology to increase the appeal or improve legibility.

The following values are valid:

opaque
The document is rendered on an opaque medium, such as a traditional monitor or paper. Black is dark and white is 100% light.
additive
The display blends the colors of the canvas with the real world using additive mixing. Black is fully transparent and white is 100% light.

For example: a head-up display in a car.

subtractive
The display blends the colors of the canvas with the real world using subtractive mixing. White is fully transparent and dark colors have the most contrast.

For example: an LCD display embedded in a bathroom mirror.

Is there a need for the subtractive value?

body { background-color: white; }
p { color: black; }

@media(environment-blending: additive) {
    body { background-color: black; }
    p { color: white; font-size: 16px; font-weight: 1000; }
}

6. Color Media Features

6.1. Color Depth: the color feature

Name: color
For: @media
Value: <integer>
Type: range

The color media feature describes the number of bits per color component of the output device. If the device is not a color device, the value is zero.

color is false in the negative range.

For example, these two media queries express that a style sheet applies to all color devices:
@media (color) { … }
@media (min-color: 1) { … }
This media query expresses that a style sheet applies to color devices with at least 8 bits per color component:
@media (color >= 8) { … }

If different color components are represented by different number of bits, the smallest number is used.

For instance, if an 8-bit color system represents the red component with 3 bits, the green component with 3 bits, and the blue component with 2 bits, the color media feature will have a value of 2.

In a device with indexed colors, the minimum number of bits per color component in the lookup table is used.

Note: The described functionality is only able to describe color capabilities at a superficial level. color-gamut, is generally more relevant to authors’ needs. If further functionality is required, RFC2879 [RFC2879] provides more specific media features which may be supported at a later stage.

6.2. Paletted Color Screens: the color-index feature

Name: color-index
For: @media
Value: <integer>
Type: range

The color-index media feature describes the number of entries in the color lookup table of the output device. If the device does not use a color lookup table, the value is zero.

color-index is false in the negative range.

For example, here are two ways to express that a style sheet applies to all color index devices:
@media (color-index) { … }
@media (color-index >= 1) { … }
This media query expresses that a style sheet applies to a color index device with 256 or more entries:
<?xml-stylesheet media="(min-color-index: 256)"
  href="http://www.example.com/…" ?>

6.3. Monochrome Screens: the monochrome feature

Name: monochrome
For: @media
Value: <integer>
Type: range

The monochrome media feature describes the number of bits per pixel in a monochrome frame buffer. If the device is not a monochrome device, the output device value will be 0.

monochrome is false in the negative range.

For example, this is how to express that a style sheet applies to all monochrome devices:
@media (monochrome) { … }
Express that a style sheet applies to monochrome devices with more than 2 bits per pixels:
@media (monochrome >= 2) { … }
Express that there is one style sheet for color pages and another for monochrome:
<link rel="stylesheet" media="print and (color)" href="http://…" />
<link rel="stylesheet" media="print and (monochrome)" href="http://…" />

6.4. Color Display Quality: the color-gamut feature

Name: color-gamut
For: @media
Value: srgb | p3 | rec2020
Type: discrete

The color-gamut media feature describes the approximate range of colors that are supported by the UA and output device. That is, if the UA receives content with colors in the specified space it can cause the output device to render the appropriate color, or something appropriately close enough.

Note: The query uses approximate ranges for a few reasons. Firstly, there are a lot of differences in display hardware. For example, a device might claim to support "Rec. 2020", but actually renders a significantly lower range of the full gamut. Secondly, there are a lot of different color ranges that different devices support, and enumerating them all would be tedious. In most cases the author does not need to know the exact capabilities of the display, just whether it is better than sRGB, or significantly better than sRGB. That way they can serve appropriate images, tagged with color profiles, to the user.

srgb
The UA and output device can support approximately the sRGB gamut or more.

Note: It is expected that the vast majority of color displays will be able to return true to a query of this type.

p3
The UA and output device can support approximately the gamut specified by the Display P3 [Display-P3] Color Space or more.

Note: The p3 gamut is larger than and includes the srgb gamut.

rec2020
The UA and output device can support approximately the gamut specified by the ITU-R Recommendation BT.2020 Color Space or more.

Note: The rec2020 gamut is larger than and includes the p3 gamut.

The following table lists the primary colors of these color spaces in terms of their color space chromaticity coordinates, as defined in [COLORIMETRY].

Color Space White Point Primaries
Red Green Blue
xW yW xR yR xG yG xB yB
srgb 0.3127 0.3290 0.640 0.330 0.300 0.600 0.150 0.060
p3 0.3127 0.3290 0.680 0.320 0.265 0.690 0.150 0.060
rec2020 0.3127 0.3290 0.708 0.292 0.170 0.797 0.131 0.046

Note: The table above does not contains enough information to fully describe the color spaces, but is sufficient to determine whether an output device approximately covers their respective gamuts. See [SRGB] for more information on sRGB, [Display-P3] for more information on Display P3, and [ITU-R-BT-2020-2] for more information on ITU-R Recommendation BT.2020.

For example, this media query applies when the display supports colors in the range of Display P3:
@media (color-gamut: p3) {}

Note: An output device can return true for multiple values of this media feature, if its full output gamut is large enough, or one gamut is a subset of another supported gamut. As a result, this feature is best used in an "ascending" fashion—​set a base value when (color-gamut: srgb) is true, then override it if (color-gamut: p3) is true, etc.

Note: Some output devices, such as monochrome displays, cannot support even the srgb gamut. To test for these devices, you can use this feature in a negated boolean-context fashion: not (color-gamut).

6.5. Dynamic Range: the dynamic-range feature

Name: dynamic-range
For: @media
Value: standard | high
Type: discrete

dynamic-range represents the combination of max brightness, color depth, and contrast ratio that are supported by the user agent and output device.

high
The user agent and the output device fulfill all of the following criteria:

Note: Some devices have high dynamic range capabilities that are not always on, and that need to be activated (sometimes programmatically, sometimes by the user, sometimes based on the content). This media feature does not test whether such a mode is active, just whether the device is capable of high dynamic range visuals.

standard
This value matches on any visual device, and not on devices lacking visual capabilities.

Note: More than one value of this media feature can match simultaneously: a user agent matching high will also match standard.

6.5.1. Determining contrast and brightness of display

Peak brightness refers to how bright the brightest point a light-emitting device such as an LCD screen can produce, or in the case of a light reflective device such as paper or e-ink, the point at which it least absorbs light.

Note: Some devices can only produce their peak brightness for brief periods of time or on a small portion of their surface at any given time.

The contrast ratio is the ratio of the luminance of the brightest color to that of the darkest color that the system is capable of producing.

This specification does not define precise ways by which these qualities can be measured; it also lets the user agent determine what counts as a high contrast ratio and as a high peak brightness. User agents must nonetheless attempt to conform to the following intent: a device capable of high peak brightness can display “brighter than white” highlights, and a simultaneous ability to do so while also presenting deep blacks (rather than an overall bright but washed out image) is indicative of a high contrast ratio.

Note: The determination for dynamic-range and video-dynamic-range will be vary depending on the user agent, but is expected to have broadly dependable semantics.

6.6. Detecting inverted colors on the display: the inverted-colors feature

Name: inverted-colors
For: @media
Value: none | inverted
Type: discrete

The inverted-colors media feature indicates whether the content is displayed normally, or whether colors have been inverted.

Note: This is an indication that the user agent or underlying operating system has forcibly inverted all colors, not a request to do so. This is sometimes provided as a simple accessibility feature, allowing users to switch between light-on-dark and dark-on-light text. However, this has unpleasant side effects, such as inverting pictures, or turning shadows into highlights, which reduce the readability of the content.

none
Colors are displayed normally.
inverted
All pixels within the displayed area have been inverted.

This value must not match if the user agent has done some kind of content aware inversion such as one that preserves the images.

Note: This is because the goal of this media feature is to enable authors to mitigate the undesirable effects of the non content aware approach that invert all the pixels. If the author were to take counter measures even in the content-aware cases, their counter measures and the UA’s would be at risk of cancelling each other.

Depending on their style sheet, authors may wish to invert images and videos:
@media (inverted-colors) {
  img:not(picture > img), picture, video {
    filter: invert(100%);
  }
}

Authors may also invert images injected via CSS (such as backgrounds), or disable shadows:

@media (inverted-colors) {
  * {
    text-shadow: none !important;
    box-shadow: none !important;
  }
}

7. Interaction Media Features

The “interaction” media features reflect various aspects of how the user interacts with the page.

Typical examples of devices matching combinations of pointer and hover:
pointer: none pointer: coarse pointer: fine
hover: none keyboard-only controls, sequential/spatial (d-pad) focus navigation smartphones, touch screens basic stylus digitizers (Cintiq, Wacom, etc)
hover: hover Nintendo Wii controller, Kinect mouse, touch pad, advanced stylus digitizers (Surface, Samsung Note, Wacom Intuos Pro, etc)

The pointer and hover features relate to the characteristics of the “primary” pointing device, while any-pointer and any-hover can be used to query the properties of all potentially available pointing devices.

Note: While this specification does not define how user agents should decide what the “primary” pointing device is, the expectation is that user agents should make this determination by combining knowledge about the device/environment they are running on, the number and type of pointing devices available, and a notion of which of these is generally and/or currently being used. In situations where the primary input mechanism for a device is not a pointing device, but there is a secondary – and less frequently used – input that is a pointing devices, the user agent may decide to treat the non-pointing device as the primary (resulting in 'pointer: none'). user agents may also decide to dynamically change what type of pointing device is deemed to be primary, in response to changes in the user environment or in the way the user is interacting with the UA.

Note: The pointer, hover, any-pointer and any-hover features only relate to the characteristics, or the complete absence, of pointing devices, and can not be used to detect the presence of non-pointing device input mechanisms such as keyboards. Authors should take into account the potential presence of non-pointing device inputs, regardless of which values are matched when querying these features.

While pointer and hover can be used to design the main style and interaction mode of the page to suit the primary input mechanism (based on the characteristics, or complete absence, of the primary pointing device), authors should strongly consider using any-pointer and any-hover to take into account all possible types of pointing devices that have been detected.

7.1. Pointing Device Quality: the pointer feature

Name: pointer
For: @media
Value: none | coarse | fine
Type: discrete

The pointer media feature is used to query the presence and accuracy of a pointing device such as a mouse. If multiple pointing devices are present, the pointer media feature must reflect the characteristics of the “primary” pointing device, as determined by the user agent. (To query the capabilities of any available pointing devices, see the any-pointer media feature.)

none
The primary input mechanism of the device does not include a pointing device.
coarse
The primary input mechanism of the device includes a pointing device of limited accuracy. Examples include touchscreens and motion-detection sensors (like the Kinect peripheral for the Xbox.)
fine
The primary input mechanism of the device includes an accurate pointing device. Examples include mice, touchpads, and drawing styluses.

Both coarse and fine indicate the presence of a pointing device, but differ in accuracy. A pointing device with which it would be difficult or impossible to reliably pick one of several small adjacent targets at a zoom factor of 1 would qualify as coarse. Changing the zoom level does not affect the value of this media feature.

Note: As the UA may provide the user with the ability to zoom, or as secondary pointing devices may have a different accuracy, the user may be able to perform accurate clicks even if the value of this media feature is coarse. This media feature does not indicate that the user will never be able to click accurately, only that it is inconvenient for them to do so. Authors are expected to react to a value of coarse by designing pages that do not rely on accurate clicking to be operated.

For accessibility reasons, even on devices whose pointing device can be described as fine, the UA may give a value of coarse or none to this media query, to indicate that the user has difficulties manipulating the pointing device accurately or at all. In addition, even if the primary pointing device has fine pointing accuracy, there may be additional coarse pointing devices available to the user. Authors may wish to query the any-pointer media feature to take these other coarse potential pointing devices into account.

/* Make radio buttons and check boxes larger if we have an inaccurate primary pointing device */
@media (pointer:coarse) {
  input[type="checkbox"], input[type="radio"] {
    min-width:30px;
    min-height:40px;
    background:transparent;
  }
}

7.2. Hover Capability: the hover feature

Name: hover
For: @media
Value: none | hover
Type: discrete

The hover media feature is used to query the user’s ability to hover over elements on the page with the primary pointing device. If a device has multiple pointing devices, the hover media feature must reflect the characteristics of the “primary” pointing device, as determined by the user agent. (To query the capabilities of any available pointing devices, see the any-hover media feature.)

none
Indicates that the primary pointing device can’t hover, or that there is no pointing device. Examples include touchscreens and screens that use a basic drawing stylus.

Pointing devices that can hover, but for which doing so is inconvenient and not part of the normal way they are used, also match this value. For example, a touchscreen where a long press is treated as hovering would match hover: none.

hover
Indicates that the primary pointing device can easily hover over parts of the page. Examples include mice and devices that physically point at the screen, like the Nintendo Wii controller.
For example, on a touch screen device that can also be controlled by an optional mouse, the hover media feature should match hover: none, as the primary pointing device (the touch screen) does not allow the user to hover.

However, despite this, the optional mouse does allow users to hover. Authors should therefore be careful not to assume that the ':hover' pseudo class will never match on a device where 'hover:none' is true, but they should design layouts that do not depend on hovering to be fully usable.

For accessibility reasons, even on devices that do support hovering, the UA may give a value of hover: none to this media query, to opt into layouts that work well without hovering. Note that even if the primary input mechanism has 'hover: hover' capability, there may be additional input mechanisms available to the user that do not provide hover capabilities.

/* Only use a hover-activated drop down menu on devices that can conveniently hover. */
@media (hover) {
  .menu > li        {display:inline-block;}
  .menu ul          {display:none; position:absolute;}
  .menu li:hover ul {display:block; list-style:none; padding:0;}
  /* ... */
}

7.3. All Available Interaction Capabilities: the any-pointer and any-hover features

Name: any-pointer
For: @media
Value: none | coarse | fine
Type: discrete
Name: any-hover
For: @media
Value: none | hover
Type: discrete

The any-pointer and any-hover media features are identical to the pointer and hover media features, but they correspond to the union of capabilities of all the pointing devices available to the user. In the case of any-pointer, more than one of the values can match, if different pointing devices have different characteristics.

any-pointer and any-hover must only match none if all of the pointing devices would match none for the corresponding query, or there are no pointing devices at all.

any-pointer is used to query the presence and accuracy of pointing devices. It does not take into account any additional non-pointing device inputs, and can not be used to test for the presence of other input mechanisms, such as d-pads or keyboard-only controls, that don’t move an on-screen pointer. 'any-pointer:none' will only evaluate to true if there are no pointing devices at all present.
On a traditional desktop environment with a mouse and keyboard, 'any-pointer:none' will be false (due to the presence of the mouse), even though a non-pointer input (the keyboard) is also present.
'any-hover:none' will only evaluate to true if there are no pointing devices, or if all the pointing devices present lack hover capabilities. As such, it should be understood as a query to test if any hover-capable pointing devices are present, rather than whether or not any of the pointing devices is hover-incapable. The latter scenario can currently not be determined using any-hover or any other interaction media feature. Additionally, it does not take into account any non-pointing device inputs, such as d-pads or keyboard-only controls, which by their very nature are also not hover-capable.
On a touch-enabled laptop with a mouse and a touchscreen, 'any-hover:none' will evaluate to false (due to the presence of the hover-capable mouse), even though a non-hover-capable pointing device (the touchscreen) is also present. It is currently not possible to provide different styles for cases where different pointing devices have different hover capabilities.
Designing a page that relies on hovering or accurate pointing only because any-hover or any-pointer indicate that at least one of the available input mechanisms has these capabilities is likely to result in a poor experience. However, authors may use this information to inform their decision about the style and functionality they wish to provide based on any additional pointing devices that are available to the user.
A number of smart TVs come with a way to control an on-screen cursor, but it is often fairly basic controller which is difficult to operate accurately.

A browser in such a smart TV would have coarse as the value of both pointer and any-pointer, allowing authors to provide a layout with large and easy to reach click targets.

The user may also have paired a Bluetooth mouse with the TV, and occasionally use it for extra convenience, but this mouse is not the main way the TV is operated. pointer still matches coarse, while any-pointer now both matches coarse and fine.

Switching to small click targets based on the fact that (any-pointer: fine) is now true would not be appropriate. It would not only surprise the user by providing an experience out of line with what they expect on a TV, but may also be quite inconvenient: the mouse, not being the primary way to control the TV, may be out of reach, hidden under one of the cushions on the sofa...

By contrast, consider scrolling on the same TV. Scrollbars are difficult to manipulate without an accurate pointing device. Having prepared an alternative way to indicate that there is more content to be seen based on (pointer: coarse) being true, an author may want to still show the scrollbars in addition if (any-pointer: fine) is true, or to hide them altogether to reduce visual clutter if (any-pointer: fine) is false.

Name: nav-controls
For: @media
Value: none | back
Type: discrete

The nav-controls media features allows authors to know whether the user agent is providing obviously discoverable navigation controls as part of its user interface.

Note: Traditional browsers typically do provide such controls and web pages typically have not needed to concern themselves with that, but in some contexts, web applications are run through so-called web-views, which do not always feature a full-fledged user interface. It is thus useful for authors to know what is being supplied by the user agent, so that they can consider whether they need to provide an easily discovered alternative.

In this context, obviously discoverable refers to controls which are either directly visible in the user interface, such as buttons, or some other form of control which is typical of the user interface of that device and trivially identifiable by the user. In the case of visual user interfaces, this would typically a visible control, although it could be something else in the case of an audio or tactile user interface. Importantly, this is not about keyboard shortcuts or gestures; as convenient as these can be, these are not obviously discoverable by just looking at (in the case of a visual UI) the user agent.

The following values are valid:

none
The user agent does not have any obviously discoverable navigation controls, and in particular none that cause the user agent to move back one page in the joint session history.
back
The user agent provides navigation controls, including at least an obviously discoverable control causing the user agent to move back one page in the joint session history (typically, a “back” button).
Authors can include a back button in their web application, and then conditionally hide it if the user agent already offers that functionality:
@media (nav-controls: back) {
  #back-button {
    display: none;
  }
}

As this media feature can be used in a boolean context, the same example can be written with shorter syntax:

@media (nav-controls) {
  #back-button {
    display: none;
  }
}

Note: Theoretically, the two are not strictly equivalent, as there could be new values in a future extension of this media feature other than back that could match when back doesn’t. In that case, using the nav-controls feature in a boolean context could be misleading. However, given that navigation back is arguably the most fundamental navigation operation, the CSS Working Group does not anticipate user interfaces with explicit navigation controls but no back button, so this problem is not expected to occur in practice.

Whether obviously discoverable controls are active does not impact the evaluation of this media feature.

If there is no previous page in the joint session history, a user agent with a “back” button could toggle it to a disabled state that cannot be interacted with until there actually is history that can be navigated back to. In such a case, @media (nav-controls: back) { … } would still be expected to match.

8. Video Prefixed Features

Some user agents, including many TVs, render video and graphics in two separate "planes" (bi-plane) with distinct screen characteristics. A set of video-prefixed features is provided to describe the video plane.

Any bi-plane implementation must return values based on the video plane for the following features: video-color-gamut; video-dynamic-range. All other features must return values based on the graphics plane.

Non bi-plane implementations must return the same values for video-prefixed features and their non-prefixed counterparts.

8.1. Video Color Display Quality: the video-color-gamut feature

Name: video-color-gamut
For: @media
Value: srgb | p3 | rec2020
Type: discrete

The video-color-gamut media feature describes the approximate range of colors that are supported by the UA and output device’s video plane. That is, if the UA receives content with colors in the specified space it can cause the output device to render the appropriate color, or something appropriately close enough.

Value and color space definitions are the same as color-gamut

8.2. Video Dynamic Range: the video-dynamic-range feature

Name: video-dynamic-range
For: @media
Value: standard | high
Type: discrete

video-dynamic-range represents the combination of max brightness, color depth, and contrast ratio that are supported by the UA and output device’s video plane.

Supported values are the same as dynamic-range.

9. Scripting Media Features

9.1. Scripting Support: the scripting feature

Name: scripting
For: @media
Value: none | initial-only | enabled
Type: discrete

The scripting media feature is used to query whether scripting languages, such as JavaScript, are supported on the current document.

enabled
Indicates that the user agent supports scripting of the page, and that scripting in the current document is enabled for the lifetime of the document.
initial-only
Indicates that the user agent supports scripting of the page, and that scripting in the current document is enabled during the initial page load, but is not supported afterwards. Examples are printed pages, or pre-rendering network proxies that render a page on a server and send a nearly-static version of the page to the user.

Should there be an explicit minimum threshold to meet before a UA is allowed to claim initial-only? Having one would mean authors would know what they can depend on, and could tailor their scripts accordingly. On the other hand, pinpointing that threshold is difficult: if it is set too low, the scripting facilities that authors can depend on may be to constrained to be practical, even though actual UAs may potentially all support significantly more. But trying to set it higher may cause us to exclude UAs that do support scripting at loading time, but restrict it in some cases based on complex heuristics. For instance, conservative definitions likely include at least running all inline scripts and firing the DOMContentLoaded event. But it does not seem useful for authors to constrain themselves to this if most (or maybe all) initial-only UAs also load external scripts (including async and defer) and fire the load event. On the other hand, requiring external scripts to be loaded and the load event to be fired could exclude UAs like Opera mini, which typically do run them, but may decide not to based on timeouts and other heuristics. [Issue #503]

none
Indicates that the user agent will not run scripts for this document; either it doesn’t support a scripting language, or the support isn’t active for the current document.

Some user agents have the ability to turn off scripting support on a per script basis or per domain basis, allowing some, but not all, scripts to run in a particular document. The scripting media feature does not allow fine grained detection of which script is allowed to run. In this scenario, the value of the scripting media feature should be enabled or initial-only if scripts originating on the same domain as the document are allowed to run, and none otherwise.

Note: A future level of CSS may extend this media feature to allow fine-grained detection of which script is allowed to run.

10. Custom Media Queries

When designing documents that use media queries, the same media query may be used in multiple places, such as to qualify multiple @import statements. Repeating the same media query multiple times is an editing hazard; an author making a change must edit every copy in the same way, or suffer from difficult-to-find bugs in their CSS.

To help ameliorate this, this specification defines a method of defining custom media queries, which are simply-named aliases for longer and more complex media queries. In this way, a media query used in multiple places can instead be assigned to a custom media query, which can be used everywhere, and editing the media query requires touching only one line of code.

A custom media query is defined with the @custom-media rule:

@custom-media = @custom-media <extension-name> [ <media-query-list> | true | false ] ;

The <extension-name> can then be used in a media feature. It must be used in a boolean context; using them in a normal or range context is a syntax error. If a <media-query-list> is given, the custom media query evaluates to true if the <media-query-list> it represents evaluates to true, and false otherwise. If true or false is given, the custom media query evaluates to true or false, respectively.

The custom media query is evaluated logically, not treated as a textual substitution. Take the following code snippet for instance:
/* --modern targets modern devices that support color or hover */
@custom-media --modern (color), (hover);

@media (--modern) and (width > 1024px) {
  .a { color: green; }
}

It is equivalent to:

@media ((color) or (hover)) and (width > 1024px) {
  .a { color: green; }
}

Processing it as if it meant the following would be incorrect:

@media (color), (hover) and (width > 1024px) {
  .a { color: green; }
}

A @custom-media rule can refer to other custom media queries. However, loops are forbidden, and a custom media query must not be defined in terms of itself or of another custom media query that directly or indirectly refers to it. Any such attempt of defining a custom media query with a circular dependency must cause all the custom media queries in the loop to fail to be defined.

If multiple @custom-media rules declare the same <extension-name>, the truth value is based on the last one alone, ignoring all previous declarations of the same <extension-name>.

Note: For error handling purposes, an undefined media feature is different from a media feature that evaluates to false. See Media Queries 4 § 3.2 Error Handling for details.

For example, if a responsive site uses a particular breakpoint in several places, it can alias that with a reasonable name:
@custom-media --narrow-window (max-width: 30em);

@media (--narrow-window) {
  /* narrow window styles */
}
@media (--narrow-window) and (script) {
  /* special styles for when script is allowed */
}
/* etc */

10.1. Script-based Custom Media Queries

Define a map of names to values for JS. Values can be either a MediaQueryList object or a boolean, in which case it’s treated identically to the above, or can be a number or a string, in which case it’s treated like a normal MQ, and can use the normal or range context syntax. Like:
<script>
CSS.customMedia.set('--foo', 5);
</script>
<style>
@media (_foo: 5) { ... }
@media (_foo < 10) { ... }
</style>

11. User Preference Media Features

11.1. Detecting the desire for less motion on the page: the prefers-reduced-motion feature

Name: prefers-reduced-motion
For: @media
Value: no-preference | reduce
Type: discrete

The prefers-reduced-motion media feature is used to detect if the user has requested the system minimize the amount of non-essential motion it uses.

no-preference
Indicates that the user has made no preference known to the system. This keyword value evaluates as false in the boolean context.
reduce
Indicates that user has notified the system that they prefer an interface that removes or replaces the types of motion-based animation that either trigger discomfort for those with vestibular motion sensitivity, or distraction for those with attention deficits.

11.2. Detecting the desire for reduced transparency on the page: the prefers-reduced-transparency feature

Name: prefers-reduced-transparency
For: @media
Value: no-preference | reduce
Type: discrete

The prefers-reduced-transparency media feature is used to detect if the user has requested the system minimize the amount of transparent or translucent layer effects it uses.

no-preference
Indicates that the user has made no preference known to the system. This keyword value evaluates as false in the boolean context.
reduce
Indicates that user has notified the system that they prefer an interface that minimizes the amount of transparent or translucent layer effects.

How does this interact with preferences around e.g. pattern fills and backgrounds? They’re not about transparency, but they also interfere with shape recognition.

11.3. Detecting the desire for increased or decreased color contrast from elements on the page: the prefers-contrast feature

Name: prefers-contrast
For: @media
Value: no-preference | less | more | custom
Type: discrete

The prefers-contrast media feature is used to detect if the user has requested more or less contrast in the page. This could be responded to, for example, by adjusting the contrast ratio between adjacent colors, or by changing how much elements stand out visually, such as by adjusting their borders.

no-preference
Indicates that the user has made no preference known to the system. This keyword value evaluates as false in the boolean context.
less
Indicates that user has notified the system that they prefer an interface that has a lower level of contrast.
more
Indicates that user has notified the system that they prefer an interface that has a higher level of contrast.
custom
Indicates that the user has indicated wanting a specific set of colors to be used, but the contrast implied by these particular colors is such that neither more nor less match.

Note: This value will match for users of forced colors mode who have picked a palette that is neither particularly high nor low contrast. See § 11.4 Detecting Forced Colors Mode: the forced-colors feature.

A user calling for cyan text over a rust background is not—​at least in terms of luminosity—​expressing a need for particularly high or low contrast, but this is not a lack of a preference either.
Note: Authors can respond to specific user preferences for more or less contrast using prefers-contrast: more or prefers-contrast: less, as appropriate.

Using an unqualified @media (prefers-contrast) {} to apply high contrast styles is incorrect and user-hostile, as it would also impose high contrast styles to people who have requested the exact opposite.

However, it is also common to reduce visual clutter and color complexity in response to both high and low contrast preferences. In that case, it is appropriate to use @media (prefers-contrast) {} without specifying more or less, to do things like replacing background images with plain colors, turning off decorative gradients, or replacing border images or box shadows with simple solid borders. As prefers-contrast: custom—​like prefers-contrast: more or prefers-contrast: less—​evaluates to true in a boolean context, such simplifications would also benefit users of forced colors mode, even when their colors of choice do not result in a particularly high or low contrast. This is desirable, as the reduced palette enforced by forced colors mode calls for some visual simplification of the page.

Preference for more or less contrast may arise from a variety of different situations. Here are some examples:

This list should be refined and expanded.

11.4. Detecting Forced Colors Mode: the forced-colors feature

Name: forced-colors
For: @media
Value: none | active
Type: discrete
active
Indicates that forced colors mode is active: the user agent enforces a user-chosen limited color palette on the page, The UA will provide the color palette to authors through the CSS system color keywords. See CSS Color Adjustment 1 § 3 Forced Color Palettes for details.
This does not necessarily indicate a preference for more contrast. The colors have been forcibly adjusted to match the preference of the user, but that preference can be for less or more contrast, or some other arrangement that is neither particularly low or high contrast.

In addition to forced-colors: active, the user agent must also match one of prefers-contrast: more or prefers-contrast: less if it can determine that the forced color palette chosen by the user has a particularly high or low contrast, and must make prefers-contrast: custom match otherwise.

Similarly, if the forced color palette chosen by the user fits within one of the color schemes described by prefers-color-scheme, the corresponding value must also match.

none
Forced colors mode is not active.
When forced colors mode is active, the only colors that are available to the author are system colors. The user agent will enforce this limited palette automatically, but the author may choose a different way of using these colors, using the forced-colors media feature to detect when it is appropriate to do so.

11.5. Detecting the desire for light or dark color schemes: the prefers-color-scheme feature

Name: prefers-color-scheme
For: @media
Value: light | dark
Type: discrete

The prefers-color-scheme media feature reflects the user’s desire that the page use a light or dark color theme.

light
Indicates that user has expressed the preference for a page that has a light theme (dark text on light background), or has not expressed an active preference (and thus should receive the "web default" of a light theme).
dark
Indicates that user has expressed the preference for a page that has a dark theme (light text on dark background).

Note: The values for this feature might be expanded in the future (to express a more active preference for light color schemes, or preferences for other types of color schemes like "sepia"). As such, the most future-friendly way to use this media feature is by negation such as (prefers-color-scheme: dark) and (not (prefers-color-scheme: dark)), which ensures that new values fall into at least one of the styling blocks.

The method by which the user expresses their preference can vary. It might be a system-wide setting exposed by the Operating System, or a setting controlled by the user agent.

Note: User preferences can also vary by medium. For example, a user may prefer dark themes on a glowing screen, but light themes when printing (to save ink and/or because inked text on blank paper prints better than blank letterforms knocked out of an inked background). UAs are expected to take such variances into consideration so that prefers-color-scheme reflects preferences appropriate to the medium rather than preferences taken out of context.

If evaluated in an embedded SVG document using the "Secure Animated" embedding mode, the preferred color scheme must reflect the value of the used color scheme on the embedding node in the embedding document.

Why do this?

While the outermost document needs to get the user’s preference directly, it’s more useful for an embedded document to use the color scheme of its surrounding embedding context, so it matches the surrounding content.

However, this enables communication from the embedding document to the embedded document, so it’s currently restricted to SVG’s using the "Secure Animated" mode, which can’t load external resources or run script, and thus can’t respond to the color scheme in any way observable to the outside world.

Whether or not to do similar for iframes, and under what conditions, is being discussed in Issue 7213.

This feature, like the other prefers-* features, previously had a no-preference value to indicate an author not expressing an active preference. However, user agents converged on expressing the "default" behavior as a light preference, and never matching no-preference.

If a future user agent wishes to expose a difference between "no preference" and "really wants a light display", please contact the CSSWG to discuss this.

11.6. Detecting the desire for reduced data usage when loading a page: the prefers-reduced-data feature

This feature may be an undesired source of fingerprinting, with a bias towards low income with limited data. [Issue #10076]

Name: prefers-reduced-data
For: @media
Value: no-preference | reduce
Type: discrete

The prefers-reduced-data media feature is used to detect if the user has a preference for being served alternate content that uses less data for the page to be rendered.

no-preference
Indicates that the user has made no preference known to the system. This keyword value evaluates as false in the boolean context.
reduce
Indicates that user has expressed the preference for lightweight alternate content.

The method by which the user expresses their preference can vary. It might be a system-wide setting exposed by the Operating System, or a setting controlled by the user agent. User agents may consider setting this based on the same user or system preference as they use to set the Save-Data HTTP request header.

Note: User agents are encouraged to use their own user centered discretion when handling a toggle of this value, whether it’s toggled post page load or during page load. A primary goal could be to not download unnecessary data. Consider, if a page is already loaded with high quality assets and the user changes their preference to reduced, the page could perhaps not update the document right away, instead wait for an explicit page reload invocation from the user. Consider also, if a page is taking a long time to download and a user changes their preference to reduced, this could be a nice point to save the user’s bandwidth and immediately switch to downloading smaller assets. User agents are free to implement logic as they see appropriate for situations like these, and also ought to be aware the situations exist.

For example, a site could honour the preference of a user who has turned on data-saving mode by serving a smaller image.
.image {
  background-image: url("images/heavy.jpg");
}

@media (prefers-reduced-data: reduce) {
  /* Save-Data: On */
  .image {
    background-image: url("images/light.jpg");
  }
}

11.7. Automatic handling of User Preferences

User agents may have explicit settings allowing users to indicate their preferences or may make the determination based on settings in the underlying operating system. User agents may also automatically infer the preferences of the user based on knowledge about the device, the environment, etc. In such case, it is recommended that they also offer a way for users to opt out of or override the automatically determined preferences.

In addition to allowing users to explicitly choose between a preference for a light or dark color scheme, a user agent could have a mode where the determination is automatically made based on the current time, expressing a preference for dark between sunset and dawn.
Depending on the type of display used, changes in the ambient light level may make the reading experience difficult or uncomfortable.

For instance, liquid crystal displays can be washed out and very hard to read in brightly lit environments. A device with such a screen and with an ambient light sensor could automatically switch prefers-contrast to more when it detects conditions that would make the screen difficult to read. A user agent on a device with an e-ink display would not make the same adjustment, as such displays remain readable in bright daylight.

In the opposite situation, user agents running of device with a light-emitting screen (LCD, OLED, etc.) and an ambient light sensor could automatically switch prefers-contrast to less and prefers-color-scheme to dark when used in a dim environment where excessive contrast and brightness would be distracting or uncomfortable to the reader.

A user agent could automatically switch between prefers-reduced-data: no-preference and reduce depending on whether the network connection in use allows for unlimited data or is on a metered plan.

12. Script Control of User Preferences

It is common for website authors to want to respect the user’s system preferences while also allowing those preferences to be overridden. To help with this, this specification defines a way for authors to override the § 11 User Preference Media Features using the PreferenceManager interface.

This override allows the preference to integrate with various platform features that are affected by these preferences.

12.1. Extensions to the Navigator interface

[Exposed=Window, SecureContext]
partial interface Navigator {
  [SameObject] readonly attribute PreferenceManager preferences;
};

12.1.1. preferences attribute

When getting the preferences attribute always return the same instance of the PreferenceManager object.

12.1.2. PreferenceManager interface

[Exposed=Window, SecureContext]
interface PreferenceManager {
  readonly attribute PreferenceObject colorScheme;
  readonly attribute PreferenceObject contrast;
  readonly attribute PreferenceObject reducedMotion;
  readonly attribute PreferenceObject reducedTransparency;
  readonly attribute PreferenceObject reducedData;
};

12.1.3. colorScheme attribute

The colorScheme attribute is a PreferenceObject used to override the user’s preference for the color scheme of the site. This is modeled after the § 11.5 Detecting the desire for light or dark color schemes: the prefers-color-scheme feature.

The get valid values for colorScheme algorithm, when invoked, must run these steps:
  1. Let validValues be a new empty sequence.

  2. Add light to validValues.

  3. Add dark to validValues.

  4. Return validValues.

If an override is set for this preference:

12.1.4. contrast attribute

The contrast attribute is a PreferenceObject used to override the user’s preference for the contrast of the site. This is modeled after the § 11.3 Detecting the desire for increased or decreased color contrast from elements on the page: the prefers-contrast feature.

The get valid values for contrast algorithm, when invoked, must run these steps:
  1. Let validValues be a new empty sequence.

  2. Add more to validValues.

  3. Add less to validValues.

  4. Add no-preference to validValues.

  5. Return validValues.

If an override is set for this preference:

Note: Unlike the media feature this preference is NOT able to be set to custom as this is tightly coupled to the § 11.4 Detecting Forced Colors Mode: the forced-colors feature.

12.1.5. reducedMotion attribute

The reducedMotion attribute is a PreferenceObject used to override the user’s preference for reduced motion on the site. This is modeled after the § 11.1 Detecting the desire for less motion on the page: the prefers-reduced-motion feature.

The get valid values for reducedMotion algorithm, when invoked, must run these steps:
  1. Let validValues be a new empty sequence.

  2. Add reduce to validValues.

  3. Add no-preference to validValues.

  4. Return validValues.

If an override is set for this preference:

Note: An example of a UA feature that is affected by this preference could be disabling smooth scrolling, or pausing marquee elements.

12.1.6. reducedTransparency attribute

The reducedTransparency attribute is a PreferenceObject used to override the user’s preference for reduced transparency on the site. This is modeled after the § 11.2 Detecting the desire for reduced transparency on the page: the prefers-reduced-transparency feature.

The get valid values for reducedTransparency algorithm, when invoked, must run these steps:
  1. Let validValues be a new empty sequence.

  2. Add reduce to validValues.

  3. Add no-preference to validValues.

  4. Return validValues.

If an override is set for this preference:

12.1.7. reducedData attribute

The reducedData attribute is a PreferenceObject used to override the user’s preference for reduced data usage on the site. This is modeled after the § 11.6 Detecting the desire for reduced data usage when loading a page: the prefers-reduced-data feature.

The get valid values for reducedData algorithm, when invoked, must run these steps:
  1. Let validValues be a new empty sequence.

  2. Add reduce to validValues.

  3. Add no-preference to validValues.

  4. Return validValues.

If an override is set for this preference:

12.1.8. PreferenceObject interface

[Exposed=Window, SecureContext]
interface PreferenceObject : EventTarget {
  readonly attribute DOMString? override;
  readonly attribute DOMString value;
  readonly attribute FrozenArray<DOMString> validValues;

  undefined clearOverride();
  Promise<undefined> requestOverride(DOMString? value);

  attribute EventHandler onchange;
};
12.1.8.1. override attribute
The override attribute, when accessed, must run these steps:
  1. Let preference be the preference object’s name.

  2. Let override be null.

  3. If an override for preference exists, set override to the value of that override.

  4. Return override.

12.1.8.2. value attribute
The value attribute, when accessed, must run these steps:
  1. Let preference be the preference object’s name.

  2. Let value be null.

  3. If an override for preference exists, set value to the value of that override.

  4. If value is null, set value to the UA value of the preference.

  5. Return value.

12.1.8.3. validValues attribute
The validValues attribute, when accessed, must run these steps:
  1. Let preference be the preference object’s name.
  2. Switch on preference:
    "colorScheme"
    Return the result of get valid values for colorScheme.
    "contrast"
    Return the result of get valid values for contrast.
    "reducedMotion"
    Return the result of get valid values for reducedMotion.
    "reducedTransparency"
    Return the result of get valid values for reducedTransparency.
    "reducedData"
    Return the result of get valid values for reducedData.
12.1.8.4. onchange event handler attribute

The onchange attribute is an event handler IDL attribute for the onchange event handler, whose event handler event type is change.

Whenever the user agent is aware that the state of a PreferenceObject instance value has changed, it runs the PreferenceObject update steps:
  1. Let preference be the PreferenceObject object that value is associated with.

  2. If this's relevant global object is a Window object, then:

    1. Let document be preference’s relevant global object's associated Document.

    2. If document is null or document is not fully active, terminate this algorithm.

  3. Fire an event named change at preference.

12.1.8.5. requestOverride() method
The requestOverride(value) method, when invoked, must run these steps:
  1. Let result be a new promise.

  2. Let allowed be false.

  3. Set allowed to the result of executing a UA defined algorithm for deciding whether the request is allowed.

  4. If allowed is false, return a promise rejected with a "NotAllowedError" DOMException.

  5. Let value be the method’s argument.

  6. Let result be a new promise.

  7. If value is null or the empty string:

    1. Run clearOverride.

    2. Resolve and return result.

  8. Let currentValue be the preference object’s value.

  9. Let validValues be null.

  10. Switch on preference:

    "colorScheme"
    Set validValues to the result of get valid values for colorScheme.
    "contrast"
    Set validValues to the result of get valid values for contrast.
    "reducedMotion"
    Set validValues to the result of get valid values for reducedMotion.
    "reducedTransparency"
    Set validValues to the result of get valid values for reducedTransparency.
    "reducedData"
    Set validValues to the result of get valid values for reducedData.
  11. If value is not in validValues:

    1. Reject result with a "TypeError" DOMException.

    2. Return result.

  12. Let previousOverride be null.

  13. If an override for preference exists, set previousOverride to the value of that override.

  14. If value is different from previousOverride:

    1. Set the preference override for preference to value.

  15. If previousOverride is null, then:

    1. If value is the same as currentValue, then:

      1. Fire an event named change at this.

  16. Resolve and return result.

This algorithm needs more detail on what exactly setting the preference override does.

Is TypeError correct here?

Note: The `change` event is fired when the computed value changes, but when a new override is set it is also fired if the value hasn’t changed.

12.1.8.6. clearOverride() method
The clearOverride() method, when invoked, must run these steps:
  1. Let preference be the preference object’s name.

  2. Let override be null.

  3. If an override for preference exists, set override to the value of that override.

  4. If override is null, then return.

  5. Clear the override for preference.

  6. Let newValue be the preference object’s value.

  7. If newValue is equal to override, then:

  8. Fire an event named change at this.

Note: The `change` event is fired when the computed value changes, but when an override is cleared it is also fired if the value hasn’t changed.

Appendix A: Deprecated Media Features

The following media features are deprecated. They are kept for backward compatibility, but are not appropriate for newly written style sheets. Authors must not use them. User agents must support them as specified.

To query for the size of the viewport (or the page box on page media), the width, height and aspect-ratio media features should be used, rather than device-width, device-height and device-aspect-ratio, which refer to the physical size of the device regardless of how much space is available for the document being laid out. The device-* media features are also sometimes used as a proxy to detect mobile devices. Instead, authors should use media features that better represent the aspect of the device that they are attempting to style against.

device-width

Name: device-width
For: @media
Value: <length>
Type: range

The device-width media feature describes the width of the rendering surface of the output device. For continuous media, this is the width of the Web-exposed screen area. For paged media, this is the width of the page sheet size.

device-width is false in the negative range.

@media (device-width < 800px) { … }

In the example above, the style sheet will apply only to screens less than 800px in length. The px unit is of the logical kind, as described in the Units section.

Note: If a device can be used in multiple orientations, such as portrait and landscape, the device-* media features reflect the current orientation.

device-height

Name: device-height
For: @media
Value: <length>
Type: range

The device-height media feature describes the height of the rendering surface of the output device. For continuous media, this is the height of the Web-exposed screen area. For paged media, this is the height of the page sheet size.

device-height is false in the negative range.

<link rel="stylesheet" media="(device-height > 600px)" />

In the example above, the style sheet will apply only to screens taller than 600 vertical pixels. Note that the definition of the px unit is the same as in other parts of CSS.

device-aspect-ratio

Name: device-aspect-ratio
For: @media
Value: <ratio>
Type: range

The 'device-aspect-ratio media feature is defined as the ratio of the value of the device-width media feature to the value of the 'device-height media feature.

For example, if a screen device with square pixels has 1280 horizontal pixels and 720 vertical pixels (commonly referred to as “16:9”), the following media queries will all match the device:
@media (device-aspect-ratio: 16/9) { … }
@media (device-aspect-ratio: 32/18) { … }
@media (device-aspect-ratio: 1280/720) { … }
@media (device-aspect-ratio: 2560/1440) { … }

Appendix B: Privacy Considerations

This section is not normative.

this section is incomplete

The prefers-reduced-data media feature may be an undesired source of fingerprinting, with a bias towards low income with limited data.

The PreferenceManager object allows querying some user-preference media features. This is not a privacy leak, as that information is already trivially available by using media features themselves.

The PreferenceManager object also allows overriding these user-preference media features; this is also neither a privacy nor accessibility regression, as the media features were already ignorable by simply not querying them.

Appendix C: Security Considerations

This section is not normative.

this section is incomplete

The display-mode media feature allows an origin access to aspects of a user’s local computing environment and, particularly when used together with an application manifest display member [APPMANIFEST], allows an origin some measure of control over a user agent’s native UI. Through a CSS media query, a script can know the display mode of a web application. An attacker could, in such a case, exploit the fact that an application is being displayed in fullscreen to mimic the user interface of another application.

Changes

This section is not normative.

Changes Since the 18 December 2021 Working Draft

In addition to editorial changes and minor clarifications, the following changes and additions were made to this module since the 2021-12-18 Working Draft:

Changes Since the 2020-07-31 Working Draft

In addition to editorial changes and minor clarifications, the following changes and additions were made to this module since the 2020-07-31 Working Draft:

Changes Since the 2020-07-15 Working Draft

The following additions were made to this module since the 2020-07-15 Working Draft:

  • Added a UA style sheet rule for inverted-colors.
  • Added the prefers-contrast: forced value.
  • Remove the light-level media feature as it is redundant with prefers-contrast and prefers-color-scheme; add examples of how these media features may be automatically inferred by the user agent based on the same factors light-level was expected to respond to.

Changes Since the 2020-06-03 Working Draft

The following additions were made to this module since the 2020-06-03 Working Draft:

  • Merged the content of level 4 into this specification. It previously was maintained as a delta over level 4.
  • Made a few editorial tweaks.

Changes Since the 2020-03-18 Working Draft

The following additions were made to this module since the 2020-03-18 Working Draft:

  • Added video-* and dynamic-range media features
  • Removed 'prefers-color-scheme: no-preference'

Changes Since the First Public Working Draft

The following additions were made to this module since the 2020-03-03 First Public Working Draft:

  • Highlight some known issues inline in the document.

Changes for FPWD (since the Media Queries Level 4)

The following additions were made to create the First Public Working Draft of this module, since the Media Queries Level 4:

  • Reinstate the light-level, inverted-colors, and Custom Media Queries sections from earlier Media Queries Level 4 drafts.
  • Added prefers-reduced-motion, prefers-reduced-transparency, prefers-contrast, prefers-color-scheme, and forced-colors media features.

Acknowledgments

This section is not normative.

This specification is the product of the W3C Working Group on Cascading Style Sheets.

Comments from Adam Argyle, Amelia Bellamy-Royds, Andreas Lind, Andres Galante, Arve Bersvendsen, Björn Höhrmann, Chen Hui Jing, Chris Lilley, Chris Rebert, Christian Biesinger, Christoph Päper, Elika J. Etemad (fantasai), Emilio Cobos Álvarez, François Remy, Frédéric Wang, Fuqiao Xue, Greg Whitworth, Ian Pouncey, James Craig, Jay Harris, Jinfeng Ma, Kivi Shapiro, L. David Baron, Masataka Yakura, Matt Giuca, Melinda Grant, Michael Smith, Nicholas C. Zakas Patrick H. Lauke, Philipp Hoschka, Rick Byers, Rijk van Geijtenbeek, Rik Cabanier, Roger Gimson, Rossen Atanassov, Sam Sneddon, Sigurd Lerstad, Simon Kissane, Simon Pieters, Steven Pemberton, Susan Lesch, Tantek Çelik, Thomas Wisniewski, Vi Nguyen, Xidorn Quan, Yves Lafon, akklesed, and 張俊芝 improved this specification.

Privacy Considerations

No new privacy considerations have been reported on this specification.

Security Considerations

No new security considerations have been reported on this specification.

Conformance

Document conventions

Conformance requirements are expressed with a combination of descriptive assertions and RFC 2119 terminology. The key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “MAY”, and “OPTIONAL” in the normative parts of this document are to be interpreted as described in RFC 2119. However, for readability, these words do not appear in all uppercase letters in this specification.

All of the text of this specification is normative except sections explicitly marked as non-normative, examples, and notes. [RFC2119]

Examples in this specification are introduced with the words “for example” or are set apart from the normative text with class="example", like this:

This is an example of an informative example.

Informative notes begin with the word “Note” and are set apart from the normative text with class="note", like this:

Note, this is an informative note.

Advisements are normative sections styled to evoke special attention and are set apart from other normative text with <strong class="advisement">, like this: UAs MUST provide an accessible alternative.

Tests

Tests relating to the content of this specification may be documented in “Tests” blocks like this one. Any such block is non-normative.


Conformance classes

Conformance to this specification is defined for three conformance classes:

style sheet
A CSS style sheet.
renderer
A UA that interprets the semantics of a style sheet and renders documents that use them.
authoring tool
A UA that writes a style sheet.

A style sheet is conformant to this specification if all of its statements that use syntax defined in this module are valid according to the generic CSS grammar and the individual grammars of each feature defined in this module.

A renderer is conformant to this specification if, in addition to interpreting the style sheet as defined by the appropriate specifications, it supports all the features defined by this specification by parsing them correctly and rendering the document accordingly. However, the inability of a UA to correctly render a document due to limitations of the device does not make the UA non-conformant. (For example, a UA is not required to render color on a monochrome monitor.)

An authoring tool is conformant to this specification if it writes style sheets that are syntactically correct according to the generic CSS grammar and the individual grammars of each feature in this module, and meet all other conformance requirements of style sheets as described in this module.

Partial implementations

So that authors can exploit the forward-compatible parsing rules to assign fallback values, CSS renderers must treat as invalid (and ignore as appropriate) any at-rules, properties, property values, keywords, and other syntactic constructs for which they have no usable level of support. In particular, user agents must not selectively ignore unsupported component values and honor supported values in a single multi-value property declaration: if any value is considered invalid (as unsupported values must be), CSS requires that the entire declaration be ignored.

Implementations of Unstable and Proprietary Features

To avoid clashes with future stable CSS features, the CSSWG recommends following best practices for the implementation of unstable features and proprietary extensions to CSS.

Non-experimental implementations

Once a specification reaches the Candidate Recommendation stage, non-experimental implementations are possible, and implementors should release an unprefixed implementation of any CR-level feature they can demonstrate to be correctly implemented according to spec.

To establish and maintain the interoperability of CSS across implementations, the CSS Working Group requests that non-experimental CSS renderers submit an implementation report (and, if necessary, the testcases used for that implementation report) to the W3C before releasing an unprefixed implementation of any CSS features. Testcases submitted to W3C are subject to review and correction by the CSS Working Group.

Further information on submitting testcases and implementation reports can be found from on the CSS Working Group’s website at http://www.w3.org/Style/CSS/Test/. Questions should be directed to the public-css-testsuite@w3.org mailing list.

Index

Terms defined by this specification

Terms defined by reference

References

Normative References

[APPMANIFEST]
Marcos Caceres; et al. Web Application Manifest. URL: https://w3c.github.io/manifest/
[COLORIMETRY]
Colorimetry, Fourth Edition. CIE 015:2018. 2018. URL: http://www.cie.co.at/publications/colorimetry-4th-edition
[CSS-CASCADE-5]
Elika Etemad; Miriam Suzanne; Tab Atkins Jr.. CSS Cascading and Inheritance Level 5. URL: https://drafts.csswg.org/css-cascade-5/
[CSS-COLOR-ADJUST-1]
Elika Etemad; et al. CSS Color Adjustment Module Level 1. URL: https://drafts.csswg.org/css-color-adjust-1/
[CSS-CONDITIONAL-3]
Chris Lilley; David Baron; Elika Etemad. CSS Conditional Rules Module Level 3. URL: https://drafts.csswg.org/css-conditional-3/
[CSS-EXTENSIONS-1]
CSS Extensions. Editor's Draft. URL: https://drafts.csswg.org/css-extensions-1/
[CSS-SYNTAX-3]
Tab Atkins Jr.; Simon Sapin. CSS Syntax Module Level 3. URL: https://drafts.csswg.org/css-syntax/
[CSS-VALUES-4]
Tab Atkins Jr.; Elika Etemad. CSS Values and Units Module Level 4. URL: https://drafts.csswg.org/css-values-4/
[CSS-WRITING-MODES-4]
Elika Etemad; Koji Ishii. CSS Writing Modes Level 4. URL: https://drafts.csswg.org/css-writing-modes-4/
[CSS2]
Bert Bos; et al. Cascading Style Sheets Level 2 Revision 1 (CSS 2.1) Specification. URL: https://drafts.csswg.org/css2/
[CSSOM-VIEW-1]
Simon Pieters. CSSOM View Module. URL: https://drafts.csswg.org/cssom-view/
[DOM]
Anne van Kesteren. DOM Standard. Living Standard. URL: https://dom.spec.whatwg.org/
[FULLSCREEN]
Philip Jägenstedt. Fullscreen API Standard. Living Standard. URL: https://fullscreen.spec.whatwg.org/
[HTML]
Anne van Kesteren; et al. HTML Standard. Living Standard. URL: https://html.spec.whatwg.org/multipage/
[INFRA]
Anne van Kesteren; Domenic Denicola. Infra Standard. Living Standard. URL: https://infra.spec.whatwg.org/
[MEDIAQUERIES-3]
Florian Rivoal. Media Queries Level 3. URL: https://drafts.csswg.org/mediaqueries-3/
[MEDIAQUERIES-4]
Florian Rivoal; Tab Atkins Jr.. Media Queries Level 4. URL: https://drafts.csswg.org/mediaqueries-4/
[RFC2119]
S. Bradner. Key words for use in RFCs to Indicate Requirement Levels. March 1997. Best Current Practice. URL: https://datatracker.ietf.org/doc/html/rfc2119
[SAVEDATA]
Save Data API. Editor's Draft. URL: https://wicg.github.io/savedata/
[USER-PREFERENCE-MEDIA-FEATURES-HEADERS]
User Preference Media Features Client Hints Headers. Draft Community Group Report. URL: https://wicg.github.io/user-preference-media-features-headers/
[WEBIDL]
Edgar Chen; Timothy Gu. Web IDL Standard. Living Standard. URL: https://webidl.spec.whatwg.org/

Informative References

[CSS-FONTS-4]
Chris Lilley. CSS Fonts Module Level 4. URL: https://drafts.csswg.org/css-fonts-4/
[Display-P3]
A; et al. Display P3. 2022-02. URL: https://www.color.org/chardata/rgb/DisplayP3.xalter
[HTML401]
Dave Raggett; Arnaud Le Hors; Ian Jacobs. HTML 4.01 Specification. 27 March 2018. REC. URL: https://www.w3.org/TR/html401/
[ITU-R-BT-2020-2]
Parameter values for ultra-high definition television systems for production and international programme exchange. October 2015. URL: https://www.itu.int/rec/R-REC-BT.2020/en
[RFC2879]
G. Klyne; L. McIntyre. Content Feature Schema for Internet Fax (V2). August 2000. Proposed Standard. URL: https://www.rfc-editor.org/rfc/rfc2879
[SELECTORS-4]
Elika Etemad; Tab Atkins Jr.. Selectors Level 4. URL: https://drafts.csswg.org/selectors/
[SRGB]
Multimedia systems and equipment - Colour measurement and management - Part 2-1: Colour management - Default RGB colour space - sRGB. URL: https://webstore.iec.ch/publication/6169
[XML-STYLESHEET]
James Clark; Simon Pieters; Henry Thompson. Associating Style Sheets with XML documents 1.0 (Second Edition). 28 October 2010. REC. URL: https://www.w3.org/TR/xml-stylesheet/

Property Index

No properties defined.

@media Descriptors

Name Value Initial Type
any-hover none | hover discrete
any-pointer none | coarse | fine discrete
aspect-ratio <ratio> range
color <integer> range
color-gamut srgb | p3 | rec2020 discrete
color-index <integer> range
device-aspect-ratio <ratio> range
device-height <length> range
device-width <length> range
display-mode fullscreen | standalone | minimal-ui | browser | picture-in-picture discrete
dynamic-range standard | high discrete
environment-blending opaque | additive | subtractive discrete
forced-colors none | active discrete
grid <mq-boolean> discrete
height <length> range
horizontal-viewport-segments <integer> range
hover none | hover discrete
inverted-colors none | inverted discrete
monochrome <integer> range
nav-controls none | back discrete
orientation portrait | landscape discrete
overflow-block none | scroll | paged discrete
overflow-inline none | scroll discrete
pointer none | coarse | fine discrete
prefers-color-scheme light | dark discrete
prefers-contrast no-preference | less | more | custom discrete
prefers-reduced-data no-preference | reduce discrete
prefers-reduced-motion no-preference | reduce discrete
prefers-reduced-transparency no-preference | reduce discrete
resolution <resolution> | infinite range
scan interlace | progressive discrete
scripting none | initial-only | enabled discrete
update none | slow | fast discrete
vertical-viewport-segments <integer> range
video-color-gamut srgb | p3 | rec2020 discrete
video-dynamic-range standard | high discrete
width <length> range

IDL Index

[Exposed=Window, SecureContext]
partial interface Navigator {
  [SameObject] readonly attribute PreferenceManager preferences;
};

[Exposed=Window, SecureContext]
interface PreferenceManager {
  readonly attribute PreferenceObject colorScheme;
  readonly attribute PreferenceObject contrast;
  readonly attribute PreferenceObject reducedMotion;
  readonly attribute PreferenceObject reducedTransparency;
  readonly attribute PreferenceObject reducedData;
};

[Exposed=Window, SecureContext]
interface PreferenceObject : EventTarget {
  readonly attribute DOMString? override;
  readonly attribute DOMString value;
  readonly attribute FrozenArray<DOMString> validValues;

  undefined clearOverride();
  Promise<undefined> requestOverride(DOMString? value);

  attribute EventHandler onchange;
};

Issues Index

Information about a user can be used as an active fingerprinting vector. Analysis of impact pending, more information to be provided before spec is published.

User agents and developers implementing this specification need to be aware of this vector and take it into consideration when deciding whether to use the feature. Specifically `prefers-reduced-motion`, `prefers-color-scheme`, `prefers-reduced-transparency` and `prefers-reduced-data` are currently of concern for exploitation.

Is there a need for the subtractive value?
Should there be an explicit minimum threshold to meet before a UA is allowed to claim initial-only? Having one would mean authors would know what they can depend on, and could tailor their scripts accordingly. On the other hand, pinpointing that threshold is difficult: if it is set too low, the scripting facilities that authors can depend on may be to constrained to be practical, even though actual UAs may potentially all support significantly more. But trying to set it higher may cause us to exclude UAs that do support scripting at loading time, but restrict it in some cases based on complex heuristics. For instance, conservative definitions likely include at least running all inline scripts and firing the DOMContentLoaded event. But it does not seem useful for authors to constrain themselves to this if most (or maybe all) initial-only UAs also load external scripts (including async and defer) and fire the load event. On the other hand, requiring external scripts to be loaded and the load event to be fired could exclude UAs like Opera mini, which typically do run them, but may decide not to based on timeouts and other heuristics. [Issue #503]
Define a map of names to values for JS. Values can be either a MediaQueryList object or a boolean, in which case it’s treated identically to the above, or can be a number or a string, in which case it’s treated like a normal MQ, and can use the normal or range context syntax. Like:
<script>
CSS.customMedia.set('--foo', 5);
</script>
<style>
@media (_foo: 5) { ... }
@media (_foo < 10) { ... }
</style>
How does this interact with preferences around e.g. pattern fills and backgrounds? They’re not about transparency, but they also interfere with shape recognition.
This list should be refined and expanded.
This feature may be an undesired source of fingerprinting, with a bias towards low income with limited data. [Issue #10076]
This algorithm needs more detail on what exactly setting the preference override does.
Is TypeError correct here?
this section is incomplete
this section is incomplete
CanIUse

Support:Android Browser131+Baidu Browser13.52+Blackberry BrowserNoneChrome68+Chrome for Android131+Edge79+Firefox62+Firefox for Android132+IE (limited)9+IE Mobile (limited)10+KaiOS Browser2.5+Opera55+Opera Mini (limited)AllOpera Mobile12.1+QQ Browser14.9+Safari16.0+Safari on iOS16.0+Samsung Internet10.1+UC Browser for Android15.5+

Source: caniuse.com as of 2024-12-06

CanIUse

Support:Android Browser131+Baidu Browser13.52+Blackberry BrowserNoneChrome41+Chrome for Android131+Edge12+Firefox64+Firefox for Android132+IENoneIE MobileNoneKaiOS Browser3.0+Opera28+Opera MiniNoneOpera Mobile80+QQ Browser14.9+Safari9+Safari on iOS9.0+Samsung Internet5.0+UC Browser for Android15.5+

Source: caniuse.com as of 2024-12-06