On September 14, at the Association Typographique Internationale’s annual typography conference in Warsaw, Poland, representatives of four of the biggest companies behind operating systems, design and the web announced their support for a new standard: OpenType 1.8, featuring OpenType Font Variations (variable fonts). The 1.8 label is deceptive—widespread support for variable fonts is arguably the biggest development in fonts since OpenType was first announced 20 years ago. Its potential impact on design and designers is far greater than that of Monotype acquiring Linotype and of Hoefler & Frere-Jones splitting up.
Variable fonts permit near-infinite variation along specified design axes and offer space-saving packaging of large font families. The former is appealing to designers in all media, including print and screen, and the latter interests web developers and companies involved in web infrastructure and devices with limited storage capacity. This intersection of interests, combined with the need for a common web platform, sparked the alliance of the four headline companies, with support and interest from type foundries and makers of font tools, including from us at FontLab.
We’ve seen very similar technologies in the past—Multiple Master (MM), GX/Apple Advanced Typography (GX)—that were not widely used. But there are a variety of reasons to think that the stars have aligned differently this time, starting with this broad-based alliance instead of a single company.
A design axis is a style for variation in a variable font. The most common axis in the past has been weight, but width is also common. Optical size is another useful axis; fonts can be optimized for use at very small sizes, very large sizes and anything in between. Note that these are typographically savvy variations, not like what you get through automatic algorithms to stretch or embolden. Beyond the everyday, type designers can do almost anything with design axes: go from sans to serif (Penumbra) or achieve countless wood type tweaks (Buffalo Gals).
Each additional axis further multiplies the variations available for a font. This range of possibilities is referred to as a design space. So the design space for a single-axis font can be visualized as a line; for a two-axis font, a square; three axes, a cube. After that, one gets into hypercubes and beyond—challenging to visualize.
A master is a design in the font source data, which might be thought of as a set of outlines at one point in a design space. The output variable font may store these more compactly as deltas to a main design, rather than storing multiple sets of outlines. For example, a variable font with just a weight axis might have an ExtraLight master and a Black master, with interpolation creating everything in between. If needed, masters can occupy intermediate points for the design and to finesse specific glyphs.
Design axes have always been a very efficient way to design large font families, even those of plain ol’ invariant fonts. Many well-known large font families of the past 25 years were designed partly or entirely using axis-based tools such as Superpolator or our own FontLab Studio, whether or not they ever shipped in a related format such as MM or GX.
Font producers can represent large font families very compactly using variable fonts. The ability to store a master in terms of deltas rather than requiring a full set of outlines (as in an older format such as MM) makes for smaller fonts. Plus, with all the separate styles as a single font, OpenType code controlling things such as ligatures, alternates and small caps only needs to be stored once in the variable font, instead of once per style. Hence, a variable font will save considerable file size over that used by a stack of stand-alone fonts.
Older MM/GX fonts updated to the new standard might still store more full sets of outlines than the minimum required, depending on how the translation is done. But even then, they will be smaller than individual fonts: Minion Pro was built from only sixteen masters, but it has three axes and 64 primary “instances,” spots anywhere in the design space of the variable font that work just like a stand-alone family of fonts—except that they are all part of a single font file.
WHAT DOES THE USER SEE?
A family of variable fonts typically consists of a single font file for the upright; if there is an italic, it will be a second, separate font. Variable fonts don’t need more than two font files for an entire family.
What shows up in font menus (given a Variations-savvy operating system) are predefined instances. Even nonsavvy apps should see these on a savvy operating system. And if users are in a sufficiently savvy application, they may also be given sliders (or some equivalent) to choose the level of variant they want on each axis. So in a savvy app with a variable font having weight and width axes, users could just dial in the precise weight and width they want. This is the dream of infinite variability in the hands of the user.
DEVIL IN THE DETAILS
The new features are all very good, but what happens when you install these fonts in operating systems and apps that don’t fully support them yet? The answer is the often unappreciated, “it depends.” The fonts are currently expected to have the same file extension, creating some compatibility benefits—but also potential user confusion.
In old or incompatible environments, an OpenType Variations font with TrueType outlines will display a single default style instead of a family in font menus. Variable fonts with PostScript style (CFF) outlines use CFF2, a new and more compact format. In today’s systems, such CFF2 fonts won’t work at all. Adobe Creative Cloud apps “roll their own” font support—so their support, or lack thereof, may not even match the host operating system.
WILL VARIABLE FONTS BE MORE ACCEPTED THAN PREDECESSORS?
Originally, Apple and Adobe both developed their own version of axis-based fonts independently in the early 1990s. OpenType Font Variations is based substantially on the GX variations font originally created by Apple for its GX Typography, a functional superset of Adobe’s approach for MM fonts.
MM was a variant of PostScript Type 1 fonts and was briefly incorporated as part of the original OpenType specification (1996–1998). But MM was only for fonts with PostScript outlines and was stripped from OpenType before any MM OpenType fonts shipped. The decision to kill MM was a contentious one inside Adobe, justified largely by two concerns: Microsoft wasn’t interested, and bundling an already failing technology with the new OpenType technology might have made OpenType’s success less certain.
GX Typography was never widely supported by font developers, nor outside Apple, for that matter. Although Apple system fonts use the advanced typography part of the technology extensively, any further support by font developers has been minimal.
The variable fonts collaborators may also be at least somewhat chastened by recent events with color fonts, wherein they had the OpenType format enshrine not one, but three of their four distinct, independently developed approaches, none of which have yet seen much real-world traction in the ensuing confusion.
This time around, all the players from the color fonts debacle are united. The typography advantages combine with space saving, exciting all the players. Nobody feels a need to keep things proprietary because the web just doesn’t work that way—a common movement toward broad support has already started and seems likely to continue.
APPLICATION SUPPORT AND THE FUTURE
As of this writing, it is a bit early to predict third-party support. Of course, Google, Microsoft and Apple cover most of the world’s web browsing, so that’s a start. Similarly, Adobe, Microsoft and Apple cover most of the world’s use of major creative design and productivity apps.
There is an important caveat: At large companies like these, there are different divisions with differing agendas. Adobe’s or Microsoft’s or Apple’s fonts team supporting something is not the same as their frontline apps doing so.
In any case, having system-level support for variable fonts on all platforms (Mac, Windows, Android) is an improvement that will make it much easier for everyone. Portions of system-level support are already in place on both Mac and Windows as of this writing.
Web browsers may quickly embrace the new technology, and afterward, web developers should find it easy to do so as well. It offers a seductive promise of things like evening the slight weight differences we are used to seeing between Mac and Windows when viewing the “same” font.
Although not certain, the variable fonts revolution is very likely. But it will take time to rebuild existing fonts and make new ones and for operating system and app support to ramp up. We’re seeing a preview of the future—but its full realization, like that of OpenType before it, could be a decade or more unfolding. Even so, variable fonts make it an exciting time to be a designer. ca