Avería: an "average" font

Dan Sayers ("I am not a type designer") decided to explore "generative" type-design by seeing what happened when he "averaged out" a large number of fonts. Once he got his teeth into the problem, he realized that "averaging out" is a complicated idea when it comes to shapes, and came up with a pretty elegant way of handling the problem, which, in turn, yielded a rather lovely face: Avería, "the average font."

Then it occurred to me: since my aim was to average a large number of fonts, perhaps it would be best to use a very simple process, and hope the results averaged out well over a large number of fonts. So, how about splitting each letter perimeter into lots of (say, 500) equally-spaced points, and just average between the corresponding positions of each, on each letter? It would be necessary to match up the points so they were about the same location in each letter, and then the process would be fairly simple

Having found a simple process to use, I was ready to start. And after about a month of part-time slaving away (sheer fun! Better than any computer game) – in the process of which I learned lots about bezier curves and font metrics – I had a result. I call it Avería – which is a Spanish word related to the root of the word ‘average’. It actually means mechanical breakdown or damage. This seemed curiously fitting, and I was assured by a Spanish friend-of-a-friend that “Avería is an incredibly beautiful word regardless of its meaning”. So that's nice.

Avería – The Average Font (via Waxy)


    1. Fontographer? As in the software?
      I think perhaps you mean ‘type designer’ or even ‘typographer’.
      In my 30 years as a graphic designer, I had never heard the term ‘fontographer’ until today.
      And this font doesn’t displease me. It’s just not really that interesting a result, although the idea behind it is pretty cool. It looks a lot like other ‘distressed’ fonts that are meant to reproduce the look of damaged lead type.

      1.  Presumably because the wear process for lead type is essentially the physical analogue of the “averaging” process undertaken here. That’s actually a kind of neat result, and makes the meaning of the Spanish word even more appropriate.

        1. Kinda. This font has averaged a jillion different faces – presumably including italic, bold and regular – and looks correspondingly amorphous or “blobby”. Whereas worn lead type (which I had the pleasure to use to set an entire book years ago) is nowhere near as ‘mushy’-looking. It just looks like itself, but it has cool little nicks out of stems, funny bent serifs and things like that which give it a personality that perfect, computer-generated fonts will never reproduce – the human touch, in fact.
          Hand-composed text, skillfully executed by a trained, experienced typesetter, is truly a joy to read. A dying art these days, I’m afraid – although I hear you can still pick up small letterpresses with a couple of sets of fonts on eBay and the like – if you could transport them!

          1. oh I love it how you use the word never, but unfortunately, I dont think that word means what you think it means.

            With enough time and effort, it is quite feasible to create a computer simulation that can simulate real-world wear on a mechanical object, and produce those “human touches” to a point where you cannot tell the difference. It would be extremely complex, time consuming and expensive to do, but shows that you just cannot throw around that ‘never’ word and not expect to be wrong.

          2. @ gekko

            Yes, it theoretically could be possible to create similar wear patterns by some analytical jiggery-pokery. But if it’s so complex, time-consuming and expensive that no-one would bother (because it’s just easier to tumble, etch or otherwise distress something), isn’t that essentially the same as never – I mean, to all intents and purposes?

      2. Type designer is a designer of typefaces. Typographer is not. There is a clear difference there!

        1. Yes – that’s why I said “or even”. As in, use of that term would be a stretch, but at least is more correct than ‘fontographer’. (!)

    2. I can’t help but like suberbanhick. He’s just so damn serious about typography and as someone who was trained as a graphic designer but never really worked as one, I know the world of graphic design needs more people who are passionate and totally serious about type because most graphic designers know squat about it.

      1. Thx for the shout out! ;-)

        Seriously, though: I was thinking further about this last night; and gekko is right, of course – in a strictly literal/interpretive way.

        But it set me to thinking: can you imagine the computational horsepower it would take to actually model/produce even just one full case of a font. That’s one of those cool lead font drawers you see photos of: usually a complete set of one face including italics & bold, plus all the wingdings and numerals, lines, etc. That’s all upper and lower case too, of course. All in multiple point sizes – usually from 8pt up to about 60pt. And many copies of each individual letter too.

        That’s a s***load of individual pieces of type.

        And every individual piece has its own unique complement of dings, nicks, scratches, pits, dents and other blemishes caused by handling and usage, which is constantly being updated in real time.

        You’d need to:

        Have a finite amount of actual individual letter forms, periodically refreshable with a few new ones – which only seems fair, seeing as printers would periodically get new type to replace worn/damaged letters.
        Do a ‘separate character’ count (that’s counting how many times every character gets used) for every thing you set in that face, and average that out with the finite number of letterforms you have.

        Make it real time updatable, changing over time with the age/usage of the font.

        ….and thats just getting started with the variables.

        I suppose you could theoretically model that, but man – it’d be like computing the human genome.

        Time and money indeed, but it would be cool as hell!

      2. Me too, for the same reasons. I have similar passions that aren’t nearly as practical or useful.

        Now I am compelled to render type with dynamically changing, procedurally generated flaws. I expect I’ll need to tolerate some flak for practicing fontography.

    1. You could call it “Yes It Is Your Average Schro’s”  .
      (No chance of trademark infringement there ;-) )

  1. Very cool!  The fully aliased edges in that graphic hurt my brain, did the transparent png from the site not work in the BB layout? Maybe jpeg works better:

  2. Ooo, I just noticed that the italicised and unitalicised “a” glyphs are two different basic forms. *UNF*

    1. I’d assume that’s because changing the “a” glyph between italic and plain is pretty common. Times New Roman and Garamond both do it.

  3. Yclept!!!  One of my favorite words! 

    adj having the name of; called:  http://www.thefreedictionary.com/yclept

  4. This looks pretty much like what a worn serif font goes to, when the type metal wears down.

    Which, in context, kind-of makes sense.

  5. Looks entirely like a ReCaptcha to me. Which would make it the perfect “control group” font for building Captchas. Nice.

  6. Hmmm.  I like the type, but mono-spacing is not exactly what you would call average.  I don’t think I would like to read a volume set in it.

    1. It’s not monospaced. It’s proportional, with the kerning being an average (who’d have thought?) of all the fonts that provide kerning data according to the creator.

    2. I put Avería into my latest newsletter and it looked quite pleasant. A bit like printing on cheap paper where the ink has spread a little. A soft and friendly font. My only fear is that these qualities make it the serif version of Comic Sans but for now I might publish the newsletter with Avería Serif and see if anyone notices.
      This doesn’t apply to Avería Sans which looks quite dreadful.

  7. Nothing new. Type designers have been messing around with this since the 1980s (if not earlier – 1580s?). The question for this one is: What typefaces were used? Almost always, they are horrible choices based on some unknown personal picks instead of some more important factors like if they are popular or not. What used for, etc.
    But unlike this one, type designers only do it to prove a point about the essential visual elements that make a letter form most readable. Then they make actual typefaces from that research!
    I’d like to see this done but with the negative space between letters. The negative space is *more* critical than the letter in reading. So, can examining negative space averages make a better letterform for design? Unlike with a typeface, there many more negative spaces to average! A challenging computing problem that should only be done with an expert type designer to guide which typefaces to compare.

  8. “Avería is an incredibly beautiful word regardless of its meaning”
    Speaking of euphonious words regardless of their meaning, it’s vaguely fitting that the alleged most beautiful word in the Spanish language, “Almorrana”, means “Hemorrhoid”.

  9. Should I just give up on correcting people who don’t know the different between a typeface and a font? If I hear or see, “I don’t like the font” one more time, I’m going to cry.

Comments are closed.