Creating a typeface

Minimalist typeface, work in progress.

I like typography.  I don’t quite like like it; not because of misgivings, I’ve just never gotten that deep into learning about it.  There’s a lot going on there, a lot of history and art and science and discipline, that I’ve only skimmed through in fits and starts over the years.  I know just enough to know that I don’t know much at all.

But I like it.  And I like the idea of designing type, and now and then I end up needing to create or manipulate a cohesive set of alphabetic characters (mostly in a pixel art context), so I’ve thought a little bit about it a few times.

And one of the thoughts I’ve had is about the idea of designing a typeface with an aggressively minimal set of reusable sub-components.

Which is part of what most typefaces do already to an extent: the various letter forms in e.g. the lowercase alphabet of any given typeface will share common shapes, common details. A lowercase b and d and p and q will have a lot in common, will echo one another in general proportion even as the smaller details may vary.  The u and the n, the n and the r and the h.  Family resemblance, a consonance of form, hold the whole set of glyphs together and make it more readable when strung into a sentence than a haphazard agglutination of otherwise individually legible letter forms would be.

A circular arc and a straight line segment.


But what if you took it further than that?  What if, instead of saying “there’s a consistent sensibility to these glyph designs”, you got really bullheaded and said “these all need to be made out of identical copies of the same two pieces”?

Which is what I’m going for with the typeface I’m working on.  Aggressive, stubborn reuse of only horizontal/vertical straight line segments and 90 degree circular arcs.  Explicitly repeating forms between glyphs through horizontal or vertical flipping.  No partial rotations, no scaling bits up or down.  Sacrificing detail and legibility for simpler forms and repetition.

A stencil variation; this can be cut out (figuratively or, as in this case, literally) without losing the interior pieces of letters that would otherwise have closed loops leading to large interior blots.

So far, I’m happy with how it’s coming. I started yesterday with the lower case glyphs, then created a stencil-friendly variant that removes segments from any glyphs that had enclosed loops/islands. This morning I did likewise for an upper case set.  There’s a lot more characters I need to tackle yet to get even a minimally workable typeface going — numerals and basic punctuation at the very least — but it’s a start.


But it is just a start, too; it’s a first draft and there are a lot of characters about which I’m still wrestling with myself on questions of logic vs. utility.

The lowercase s and z, for example, are obnoxious by design but, also, immediate sore spots for folks I’ve shown the work in progress to.  And There certainly are ways to shorten the vertical footprint a little or a lot from that unusual ascender+descender profile for those characters; see above some z variations that still use only the line segment and arc components but fit into a smaller profile.

But all of those z variants feel unsatisfactory to me in the way that they add new kinds of shapes to the set.  Too many sharp corners and/or deviation from the overwhelming circular perimeter driving most of the glyphs.  They just jump out at me as noisy, angular.  And while I could see choosing to allow a z with a flourish to end the set, that doesn’t change the problem with the s to which I’m trying to twin it; I like these even less reversed for that glyph.

So ultimately the appeal of the conspicuous pried-open-circle s and z forms in this draft may win out anyway.

But this is all essentially sketching; if I want to take this from designs to an actual working font file I can use in general contexts, I’ll need to learn more about the mechanics of type.  Kerning, leading, tracking, weights, lots of stuff I’ve run into in lay contexts but never really learned about.

And I’ll need to learn some software.

FontForge, with a blank slate.

Googling shows there to exist a number of web-based font-conversion services that will take vector and even raster image files and turn them automagically into font files.  They all seem a little handwavy, though, and I don’t learn much doing things that way even if the results were to turn out okay.

A friend suggested instead taking a look at FontForge, an open-source font creation/editing application with a robust toolset.  It looks like has a lot of engine under the hood, which is promising.

But it also has what feels at first blush like a very open-source approach to user interface and learning curve; I got as far as downloading, installing, launching, and getting utterly intimidated by it yesterday before deciding to slow down and ease into its mess of features and X11-ish UI and control scheme.  So I’ll come back to it.


But for now, I’ll just keep working out glyph designs and playing around with the results by manually duplicating and manipulating glyphs in Inkscape; tedious for any significant amount of text layout work but fine for e.g. replacing the futura in my blog header with one of the new typeface’s variants.

Author: Josh Millard

I manage and help moderate the community website MetaFilter, where I go by "cortex"; in my spare time I get up to all sorts of creative nerdery on the internet and in Portland, Oregon.

5 thoughts on “Creating a typeface”

  1. Overall, I really like it. I think it has a few problems though…

    * The Q is ok, but I like the stencil version much better.

    * The X looks too much like a fancy I. How about taking the right half of K and mirroring it?

    * The Z z are barely recognizable. I know you like the reversals, but I think people would see it as a weird S/s. Z’s are almost always pointy. Also maybe revisit this one after you’ve tackled 5 and 2.

    * The v is going to be a kerning problem, at least: it creates some unpleasant white space in the middle of a word. Plus it may be recognizable as v when you’re doing an alphabet only because we see it after u. Maybe try words like love, moue, Evan, Eustace to make sure it isn’t too close to u.

    A quick check of my old Letraset catalog suggests that you take a look at Cirkulus or Pump for ideas.

    Also it’d be interesting to see what happens if you allow a little cheating. I think a lot of the letterforms would be improved if you could modify the grid or circle size by a tiny amount. A hint of curve at the bottom of the j would be nice, for instance.

  2. I dig all those thoughts, yeah, though whether I’ll specifically act on ’em for this I don’t know. I’m probably gonna stick with the core stubborn aesthetic for now, with a mind to put out an alternate version with less-purist but more readable/well-behaved replacement glyphs.

Leave a Reply

Your email address will not be published. Required fields are marked *