On this article, we’ll discover the 4 broad classes of CSS sizing models. We’ll have a look at what the sizing models are for, the place they work greatest, and the way to decide on the very best ones in every situation, in order that our layouts will probably be optimized throughout a spread of media and gadget dimensions.

Desk of Contents
  1. About CSS Sizing Unit
  2. Absolute Units
  3. Font-relative Units
  4. Viewport-relative Units
  5. Container-relative Unit
  6. Conclusion

About CSS Sizing Items

CSS affords a number of methods to specify the scale or size of parts — some extra intuitive than others. CSS models fall into 4 broad classes:

  • absolute models, similar to cm and px
  • font-relative models, similar to em and ch
  • viewport-relative models, similar to vw and vmin
  • container-relative models, similar to cqw and cqh

We’ll have a look at every sort of CSS unit on this piece.

Earlier than persevering with, let’s refresh your reminiscence about some phrases you’ll see on this piece: specified worth, computed worth, and used worth.

  • Specified worth is the worth of a CSS property as indicated within the doc’s stylesheet.
  • Computed worth is the worth of a property after the browser applies the foundations of the cascade, inheritance, and the property’s definition.
  • A used worth is the worth of a property after the browser makes its last changes and conversions. Throughout this course of, relative models get transformed to absolute ones. For screened media (that’s, units with screens), bodily models get transformed to their pixel equivalents.

You’ll see these phrases a number of instances on this article.

Absolute Items

Absolute models are anchored to particular, media-dependent measurements. For bodily media similar to paper, absolute CSS models are anchored to their corresponding bodily models. For screened media, absolute models are anchored to pixels. One pixel is roughly 1/96th of an inch.

Absolute models embrace in, cm, mm, and Q, or inches, centimeters, millimeters, and quarter-millimeters, respectively. Level (pt) and pica (laptop) are additionally absolute models. They’ve their roots in bodily typesetting and desktop publishing. Every pt equals 1/72th of an inch, whereas 1pc equals 1/sixth of an inch. Desk 1 exhibits absolute models and their equivalents.

UnitIdentifyEqual to
cmcentimeters1cm 37.8px
mmmillimeters1mm 3.78px
Qquarter-millimeters1Q 0.944px
ininches1in = 96px
laptoppicas1pc = 16px (1/6 of 1 inch)
ptfactors1pt 1.33px (1/72th of 1 inch)
pxpixels1px = 1/96th of 1 inch

When the required width of a component is 2in, its printed width will probably be two inches. On screens, nevertheless, 2in leads to a computed worth of 192px.

Absolute models usually are not affected by font metrics, inherited property values, or the viewport. They work greatest when you recognize bodily properties of the output medium, as with paged media.

Keep away from utilizing absolute values with the font-size property. Some low-vision internet customers enhance the default font measurement of their browser to enhance legibility. Absolute values, together with px, don’t scale with that change. As an alternative, use font-relative models. We’ll focus on them within the subsequent part.

Font-relative Items

Font-relative models use font metrics to calculate the scale of a component. This can be the computed worth of the font-size, or line-height properties. Or they could be computed relative to the scale of a specific glyph, as with the ch, ex and ic models.

A phrase of warning when utilizing font-relative models: they will set off a font obtain if the font isn’t already loaded. This will trigger format shifts on sluggish networks or networks with intermittent availability.

Font-relative models may be categorized into two sorts: native and root-relative.

  • Native font-relative models calculate measurement relative to the computed worth of the font-size property for the component. For the reason that font-size property is an inherited property, this often means it’s relative to the font-size property worth of the closest ancestor component.

  • Root-relative models calculate measurement relative to the doc’s root component — usually the font-size worth for the html component.

em and rem

You’re in all probability aware of the em unit and its root-relative counterpart rem. The em unit represents a proportion of the computed worth of the font-size property for the component. For instance, 1em is 100% of the worth of font-size. A worth lower than 1, similar to 0.5em works out to 50% or half the worth of font-size. Values better than 1 act as a multiplier.

Within the previous instance, the computed font measurement for h1 is 48 pixels. Its dad or mum component, article, has a specified font-size worth of 24px. The h1 inherits that worth, however 2em tells the browser to make the font measurement of the h1 twice the proportion of article.

The rem unit, alternatively, calculates measurement relative to the font-size worth of the root component.

Right here, the h1 has a computed font measurement of 32 pixels. Altering the font-size worth for article doesn’t change the scale of the h1, though it’s a descendant.

In case you want a refresher on em and rem models, attempt The Power of em Units in CSS and Rem in CSS: Understanding and Using rem Units.

Each em and rem sizes are lengths calculated relative to the doc’s default font measurement. The ch, ex, and ic models and their root-relative counterparts rch, rex, and ic are calculated relative to the scale of the zero, lowercase x, and 水 glyphs respectively.

What’s a glyph?

A glyph is the visible illustration of a personality — actually, the form of the letter, quantity or punctuation mark utilized by a font. A zero character could also be represented by in any variety of methods, as illustrated by the next picture.

Various representations of the 0 character

Glyph dimensions can fluctuate fairly a bit between fonts; 1ch could also be 5 pixels or 50 pixels relying on the metrics of your chosen font. Consequently, specified values could also be very totally different from used values for ch, ic, and ex models and their root-relative counterparts, rch, ric, and rex. Maintain that in thoughts when utilizing a number of fonts.

Zero-width models ch and rch

The ch and rch models are primarily based on the superior measure — the width or peak — of the zero glyph within the font used to render it. When the inline axis of the doc is horizontal, the calculation relies on its width. When the inline axis is vertical, the calculation relies on the peak of the zero glyph. If the browser can’t decide the measure of the 0glyph, the ch unit behaves as if the zero glyph is 0.5em vast by 1em tall.

Much like rem models, rch models use the superior measure of the zero glyph for the foundation component’s font.

X-height and cap peak models: ex/rex and cap/rcap

In typography, the x-height refers back to the peak of the lowercase letter x glyph, measured from its baseline.

The words vox populi, with the text's x-height highlighted in pink

Sizes set utilizing ex models are calculated relative to the used x-height of the primary obtainable font. The rex unit works equally, however calculates measurement relative to the ex unit of the foundation component as an alternative of the closest ancestor.

Cap peak, alternatively, refers back to the distance from the baseline to the highest of capital or uppercase letters — usually the peak of letters with flat tops. Pointed or rounded capital letters similar to A, O, and S could have barely taller cap heights in some fonts.

The text Twelfth Night, with the cap height indicated with a pink background

Cap-height models (cap) calculate lengths relative to the used cap peak of the primary obtainable font for a component. Root-relative rcap models use the cap worth of the foundation component as a foundation for calculating lengths. Sadly, cap unit assist is presently restricted to Firefox, whereas rcap models aren’t but supported by any browser.

Some fonts do a poor job of exposing font metrics to the browser, or lack dependable metrics. Different fonts could lack a lowercase x glyph, or use a non-Latin script similar to Arabic. When the x-height can’t be decided from the font itself, browsers use a fall again x-height of 0.5em.

Letters b d f h k x, with the ascenders highlighted with a pink background

When the browser can’t decide cap peak from the font, it makes use of the font’s ascender value. The ascender is the portion of a lowercase letter, such because the stem of h or b, that extends above the x-height.

Ideograph models: ic and ric

The ic unit works greatest with Chinese language, Japanese, and Korean character units. It calculates lengths primarily based on the used superior measure of the 水, or water ideograph, of the font used to render it. The 水 ideograph is widespread to all three character units.

Glyphs in Chinese language, Japanese, and Korean fonts usually have the identical width and peak. Consequently, ic models can work effectively to restrict textual content to a specific variety of glyphs per line for these character units. Within the demo under, the inline measurement for every paragraph is 20ic. That accommodates about 20 glyphs per line, relying on the font.

Though 水 is a shared ideograph throughout Chinese language, Japanese, and Korean, not each font accommodates a glyph representing it. When the browser can’t decide the superior measure of 水, it assumes a measure of 1em.

As with different font-relative models, ic models are calculated relative to the computed worth of dad or mum parts, and ric models are calculated relative to the computed worth of the foundation component.

Line peak models: lh and rlh

You too can set lengths utilizing the line-height relative unit — lh — and its root-relative sibling rlh. An lh unit is the same as the computed worth of the line-height property of the component on which it’s used. It’s calculated relative to the component’s rapid ancestor. The rlh unit calculates lengths relative to the line-height of the doc’s root component.

When the worth of the line-height property is regular, the peak of every line relies on the font’s personal metrics. When the worth is a quantity — similar to line-height: 1.3 — the road peak is the product of font-size and the multiplier, as expressed in pixels. If the worth of line-height is a share, the computed worth of line-height is the proportion worth multiplied by the computed font measurement, in pixels.

For instance, if the consumer’s minimal font measurement is 18px and the required worth of line-height is 1.5, the computed line peak is 27px. This computed line peak is one lh or rlh unit. A declaration of inline-size: 10lh leads to a component that’s 270 pixels vast (or tall, if the inline axis is vertical).

Root-relative line peak models — rlh models — calculate lengths utilizing the used line peak of the doc’s root component. Native line peak, or lh models, inherit the line-height worth of ancestor parts.

Items similar to ex, cap, ic, and lh are notably helpful when your mission makes use of a number of typefaces and/or languages. You possibly can keep vertical rhythm and measurement ratios, even when the consumer modifications their font settings.

Font-relative models are affected by the writing-mode, text-orientation and text-transform properties amongst others. Chances are you’ll, for instance, discover that CJK glyphs of some fonts occupy extra pixels when the writing mode is horizontal versus vertical. Chapter 6 of CSS Master, 3rd Edition explains how writing mode impacts format. It’s obtainable from Pylogix Premium.

Up to now, we’ve lined absolute lengths and font-relative models. Nevertheless, CSS additionally helps two extra sorts of measurement models: viewport-relative models and container-relative models.

Viewport-relative Items

Viewport-relative models, because the identify suggests, rely on the scale of the browser window, iframe, or gadget dimensions. They’re calculated relative to the scale of the initial containing block — both the viewport or web page within the case of paged media. One viewport share unit equals 1 p.c of the preliminary containing block. That’s totally different from percentages, which set dimensions as a proportion of the dad or mum component’s width or peak.

Viewport share models are a bit tough to grasp, partly as a result of they’re primarily based on 4 notions of the viewport:

  • UA-default viewport, which can be equal to both the big or small viewport, or an intermediate measurement
  • Giant viewport, or the obtainable measurement when retractable parts of the browser interface are retracted
  • Small viewport, which assumes that the retractable parts of the browser’s interface are expanded
  • Dynamic viewport, which exists whether or not or not the browser’s interface is expanded or retracted and grows or shrinks to fill the obtainable house

Large viewport units include the portion of the browser window occupied by the browser's user interface. Small viewport units exclude the browser's user interface

Safari on iOS, for instance, hides the again button, tab menu and different controls as you scroll down from the highest of the web page and divulges them once more as you scroll up.

Every of those conceptual viewports has a corresponding set of viewport models. UA-default viewport models embrace vw, vh, vmin, and vmax. Giant, small, and dynamic viewport models comply with an analogous naming conference, with an l, s, or d prefix — that’s, lvw, or dvmin.

The *vw and *vh models equal 1 p.c of the preliminary containing block’s width and peak, respectively. The *vi and *vb models work equally. Every *vi unit equals 1 p.c of the preliminary containing block alongside the inline axis, whereas every *vb unit equals 1 p.c of the preliminary containing block alongside the block axis. Inline and block axes rely on the worth of the writing-mode property. When the doc makes use of a vertical writing mode, the inline axis is vertical and the block axis is horizontal. For horizontal writing modes, the inline axis is horizontal and the block axis is vertical.

Within the case of *vmin models, the size is calculated as a proportion of the smaller of *vw or *vh. If the UA default viewport is 390px by 844px, then a specified worth of 10vmin turns into a used worth of 39 pixels (or 10 p.c of 390).

Equally, *vmax models are calculated as a proportion of the bigger of *vw or *vh. A specified worth of 10vmax, interprets to a used worth of 84.4 pixels, for viewport that measures 390px by 844px.

Giant, small, and default viewport sizes are steady values. They solely change when the viewport itself modifications, similar to by rotating from portrait to panorama mode. In case you use svw or svi models to measurement a component, its measurement is not going to increase when the browser interface retracts. Alternatively, in the event you use lvh or lvb models, elements of your content material could also be hidden by the browser’s controls after they increase.

Dynamic viewport sizes, alternatively, usually are not steady. They could change when the orientation modifications, or when the consumer scrolls. For instance, a component with a peak worth of 100dvmax modifications measurement when the browser interface impacts the scale of the viewport. You possibly can see this impact within the video under.

Right here, the light-blue field expands vertically as soon as the browser’s controls retract, and it shrinks when the controls change into seen.

Viewport models may be helpful for creating full-width, full-height interface parts, similar to a slideshow that takes up your complete width and peak of the display.

Viewport models additionally work properly for creating fluid typography that expands or shrinks with the scale of the viewport. Mix it with the clamp() operate to stop sort that’s too small or too giant, as proven under.

Use warning with dynamic viewport models, nevertheless. Customers could expertise format shifts or textual content measurement modifications as they scroll.

CSS Viewport Units: vh, vw, vmin, and vmax affords extra examples of how you should utilize viewport relative models.

Container-relative Items

Whereas viewport-relative models apply to the obtainable house of the browser window, container-relative models are calculated relative to the scale of a component’s containment context. Meant to be used with container queries, container-relative models are presently outlined within the CSS Containment Module Level 3 specification as an alternative of the CSS Values and Units Module Level 4 one. In case you’re new to container queries, An Introduction to Container Queries in CSS will deliver you in control.

Container relative models are additionally known as container question size models. Every unit is the same as 1 p.c of the container measurement alongside both the horizontal or vertical axis, relying on the unit. For instance, the cqw and cqh models are equal to 1 p.c of the container width and peak, respectively.

To assist a number of languages and scripts in your layouts, use the cqi and cqb models. A cqi unit is the same as 1 p.c of the inline measurement of the container, whereas the cqb unit is the same as 1 p.c of the block measurement. Very similar to the vi and vb models, cqi and cqb are affected by the writing-mode property.

Lastly, we now have the cqmin and cqmax models. The cqmin unit, just like vmin, will get evaluated relative to the smaller of cqi or cqb. The cqmax unit, alternatively, is evaluated to the bigger of cqi or cqb. Every cqmin unit represents 1 p.c of the smaller dimension. Every cqmax unit represents 1 p.c of the bigger dimension.

Container-relative models allow you to create elements that work in a number of contexts. Within the instance under, the cqi unit offers the picture the identical proportions whatever the container’s inline measurement.

Take a deeper dive into container relative models by studying Unleashing the Power of CSS, obtainable on Pylogix Premium.


Understanding measurement models is the important thing to creating CSS layouts that work effectively throughout a spread of media and gadget dimensions. Choosing the proper unit can enhance the legibility, usability, and accessibility of your web site. Use absolute models when you recognize the bodily dimensions of your output medium. Font-relative and viewport-relative models are well-suited to creating layouts that adapt to a number of display sizes. Container-relative models are excellent for creating reusable elements that adapt to quite a lot of layouts.