Tesserae — an experiment in modular variable fonts

Christian Thalmann

With work on Quinoa drawing to a close, I've allowed myself to indulge in a little side project that's been bumbling around in the back of my head for a while. It's a typeface made only from five elements: a square and four corners, used as components to build all glyphs. One could then use stylistic sets to swap out these elements to radically change the flavor of the typeface, or build in a number of variation axes along which one could hollow out, expand and contract, rotate and jiggle the elements for live visual effects (not implemented yet).

Here's what the first draft of the basic alphabet looks like:


And a few variations:





Actually, now that I see those last two examples, I'm tempted to make a spin-off typeface based on this piping principle that connects the free horizontal lines between letters wherever possible...  hm...

Christian Thalmann

A little bit more spacing definitely improved the box version. And the component-wise rotalic also works better than I expected:

Arne Freytag

Very nice, especially the last one with these connected lines. Remembers me at Kombinat by Josef Albers and the ideas of modular alphabets and "Universalraster" in the 1920s. It is still fascinating me.

Christian Thalmann

I figured I'd better complete a basic character set before goofing around with features that would require divergent Glyphs files. For now, I've added the basic Mac and Window characters and the Pan-European lists in Glyphs.

Some characters were surprisingly easy to translate into this scheme (the tiny figures in particular!) whereas others remain a headache.

For instance, is the top /ø legible? The only two alternatives I came up with are the ones below, but they seem much more contrived.

Similarly, I'm having trouble with /ł. Does this work?

Ray Larabie

I don't think you need to squish the Œ œ Æ æ.

Christian Thalmann

Good idea, if a bit disconnected. How about this?

As for the conjoined letters, I agree that they are currently too narrow, but just placing the two full-sized letters next to each other defies the purpose of the ligatures. Of the following, I think I prefer the one with full-sized /o but reduced /e (second line).

I'm wondering whether that also works with Nordic expectations of /æ?

Christian Thalmann

Mixing up the tile central block with the pipe corner blocks also produces an interesting result:

EDIT: It works even better if the pipes are extended so as to bridge the gaps to the blocks:

EDIT: Or with circular holes, maybe? 

Christian Thalmann

Ooh, or this one!

Craig Eliason

Ray's /oslash works better, as yours is too p-like. 
I like the circular holes variant which looks plaid!

Hrant Հրանդ Փափազեան Papazian

Cool stuff. Especially the rotalic one!

BTW you might like to check out the work of Paul McNeil, and recently Just van Rossum.

Craig Eliason

Christian Thalmann said:

A little bit more spacing definitely improved the box version. And the component-wise rotalic also works better than I expected:

License this one to 3M to sell Post-It notes!

Ori Ben-Dor

The 4 bothers me.

Maurice Meilleur

Albers being the obvious point of reference, the value of pursuing this project beyond your personal interest (which isn't nothing!) lies in surpassing what he made, which is why it's especially good to see you're playing with the shapes and thinking about OTVar axes to modulate them. I'd also look at numbers—how could you move from 3x and 4x arrays to 2x and 3x (like Albers's original Kombinations-schrift) and to even wider arrays? How about height? Ascender and descender extent? Changing the quarter circles to right triangles?, etc.

Maurice Meilleur

Maybe also get text set in the different variations into large fields and think about how it looks, behaves, signifies at that scale and in those amounts …

Christian Thalmann

Today's morning exercise.

Christian Thalmann

It looks a bit better with convex cuts:

Christian Thalmann

Concerning the ø, I agree that my latest proposal has some dangerously p-like qualities. However, the hovering angles feel very out of character for this typeface, considering the slash is supposed to be intimately built into the glyph rather than added as an afterthought. In the pipe versions of the typeface, many of these approaches look particularly weird:

This feels a bit like Conway's Life.

I actually still prefer the original, minimalist approach on the left. I'd be interested to hear the opinions of actual readers of ø on whether it parses (also in the block version).

Christian Thalmann

@Maurice Meilleur: I just had a look at Albers' Kombinationsschrift, and I am pleasantly surprised at how different it is. I think there's no danger of confusion or conceptual overlap. I also must congratulate Mr. Albers at the elegance of his approach. In general, though, I prefer not to study existing typefaces too much to avoid contamination.

Yes, the variable fonts aspect is supposed to be the core feature of this project, though if the idea of connecting letters in the pipe design works out as well as I hope, it could be a worthwhile spin-off on its own.

Christian Thalmann

I've started working on the connected pipe version to see what it would look like. Not incredibly readable, but it could still be fun.

I'm getting the impression that maximizing the number of connections is not a desirable thing, though. The combination /u/n with three links becomes too monolithic. I'm thinking of connecting at the baseline and x-height only in those cases (middle row) or only in the bottom two levels (bottom row).

Adam Twardoch

Christian, your Łł works very well in the solid fill versions, though I'd be curious to see how it works in the "lighter" variants! 

Adam Twardoch

Overall, this is brilliant. Last year, I co-designed Milka with my friend Botio Nikoltchev: 

https://www.myfonts.com/fonts/lettersoup/milka/
https://www.myfonts.com/fonts/lettersoup/milka/regular/glyphs.html

We had tremendous fun doing the Greek and the Cyrillic together  

Christian Thalmann

Hi Adam — here's how the slashed ells look in the pipe style. If anything, they look more convincing here than in the tile style.

Milka looks like lots of fun indeed! And yes, I've had a thought or two about Cyrillic and Greek, but I had my doubts over whether the conceptual space would be flexible enough for those scripts, especially the many small-scale graphemic differences in the Cyrillic (e.g., between и and н). Not impossible though!

Hrant Հրանդ Փափազեան Papazian

Christian Thalmann said:

I've started working on the connected pipe version

Run with it!

Christian Thalmann

Huh, it turns out Cyrillic wasn't all that hard after all.

Kent Lew

I’ll be curious to hear what native users think of Cyrillic Лл. To my eye, they are not terribly convincing. Дд perhaps only slightly more so. I think capital Ъ probably needs to be one column wider. But again, a native user would have to confirm.

(I love the lepidoteran Ж, however! ;-)

Either Ελληνικά is spelled wrong, or I’m just not seeing some of those characters.

Christian Thalmann

I never liked my solution for /x, and the problem gets worse in Cyrillic where it competes with /ж. I'm thinking of going extra-wide instead:

Christian Thalmann

Kent, I see what you mean with л. Are any of these better?

Yes, Ελληνικά was misspelt. I made some more Greek in the meantime:

Christian Thalmann

How about these for capital Л?

Hrant Հրանդ Փափազեան Papazian

!!

https://twitter.com/michaelbierut/status/894212868635230209

Marc Oxborrow

Surprised there's been no mention of Fontstruct, the the long-running site/tool for modular type-building.
https://fontstruct.com

Kent Lew

Christian — I’m not really qualified to say. But, from the capital selection I’d lean toward the second or fourth. Not sure what that means for the lowercase.

Again, we should hope for a native speaker to weigh in.