Fans of Claus Eggers Sørensen’s ‘Playfair Display’ — which has long been one of the classier and better crafted serif types in the Google library — might be interested in this. Sørensen has now reworked Playfair as a pair of variable fonts with three axes (weight, width and optical), and a full complement of ‘sorts’ for every purpose. The range of these fonts is huge (Wakamai Fondue lists an incredible 120 named instances), and they make it possible to use Playfair as a readable text font, a hyper-legible ‘caption’ font for tiny text, and a stylish headline type with superfine hairlines. It’s a magnificent piece of work. Still in beta, but working well — it will still need some judicious subsetting to strip out the features you don’t need.
1 Like
Interesting. Although with 3 axes and a full complement of ‘sorts’ for every purpose, it would need a good amount of sub setting to use on a web site.
1 Like
And Cyrillic!
Wow. He has even included the Glyphs source files so anyone with a Glyphs 3 license should be able to modify and export their own version.
This is also a great resource to study the works of an advanced font designer.
Well spotted! I’ve downloaded the source files and subsetted and re-exported the fonts in Glyphs 3 (modified Latin-1) and it makes a big difference. The Roman is down to 228kb in .woff2 (no doubt it could have been more, but I’ve kept smallcaps, alternatives, ligatures and subs/sups as I use these all the time). Once I’ve tested them, I’ll post a link if anyone wants to download and play with them.
Still very big IMHO.
Sure, but it depends what you intend to use if for. The VFs that I made from Vernon Adams’ Nunito Sans were 24kb each (and I added small caps to Roman and Italic). But that is a one axis font, and this has three axes (so eight masters for roman and eight for italic). I suspect a three axis font will only really become Googlicious when dynamic subsetting becomes available, as Google demonstrate themselves with Noto (and Playfair 2.0 could lose the width axis and still be a really versatile font).
Sure. In my experimentation, I found that a good subsetted variable font, ideal for web use, adds about 25Kb per axes.
So a variable font with slant and weight, can be about 50Kb. If you are only using 1 weight and 1 italic, typically 2 non variable fonts for the same font, will be around 15-20Kb each. So size wise,not much is to be gained by using variable fonts. However when you start to use more weights with their italic counterparts, the variable fonts far outweight their non variable versions in size with the beauty of being able to fine tune every weight.
Indeed. The nature and drawing of the font also makes a big difference — twice as many bezier points per character is effectively the same as twice as many characters (thus sans serif fonts will always tend to be lighter than serifs).
I think the difficulty here is that whilst we can make huge performance improvements to fonts for our own purposes (for instance, if we only ever use English and have no need to include words from other languages, we can ruthlessly strip out all accented characters). And case by case tweaking of fonts looks like it’s going to be increasingly important in web design. But if we’re making a font for others to use, then it’s much more problematic. Thus in stripping down Sørensen’s font, it was an easy decision to get rid of all the Cyrillic. As a minimum for others to experiment with, though, it needs support for the principal Western European languages.
1 Like