Lowercasey uppercase, or how to extend a unicase typeface

Adam Jagosz

I’m extending my existing typeface which was released a while ago as a unicase design with mostly lowercase appearance. I drew some inspiration from uncial, which might explain this decision a bit.

A shorter while ago I spotted this design used with modifications to add capitals. That was a clear sign that caps are a “useful feature” that most fonts should have. So now I took to adding them, which for this design turned out to be pretty fun and easy (or so I am mistakenly believing) using FontLab (it would take twice the time in FontForge, I guess).

• As usual, weird and useless ideas pop in my head, but this time even I don’t need being talked out of it. There’s definitely no point in making the capital /B /H (possibly /K) enlarged versions of their lowercase counterparts with reduced serifs, right? I trust there is no precedent for this (?) (except for, to the modern eyes, Blackletter H and K) and neither should I create one?

Ugh.

• But perhaps there is also no merit in keeping the other typical lowercase forms in the UC? /a/g/e
  
• Is the height difference too small? I’m uneager to make it more pronounced because I’d have to make the caps really wide to preserve the proportions.
• Should I differentiate stems at this weight? Currently there is no difference, but now that I look at it, it could use one or two units.

K Pease

All of this looks great. The big-lowercase form /B/H/D will be unwanted by most, but a few uncial purists will be interested in them as options. To make them clearer, both as caps and as the letter they are, you could extend their ascenders a bit more into the diacritic zone and then continue them with a horizontal bar (similar to the blackletters you mentioned and echoing /G). I wouldn't change anything else, it's very attractive.

Christian Thalmann

Yes, it’s pretty!
Maybe bend the ascender of /D/ diagonally to the left, as in blackletter?
The /C/ is perhaps a bit more closed than it needs to be. 

And if you want a LC flavor, maybe try a narrower /L/?

André G. Isaak

I like this in the all-caps setting, but I find the capital E and A a bit disconcerting in the mixed-case setting.

Adam Jagosz

@André G. Isaak I think I can see what you’re getting at with /A: its structure (the stem only as tall as the x-height) is problematic and might cause a momentary lapse before it is parsed as capital, or might even be parsed as a weirdly big minuscule.

As for /E, I’m not sure if the problem boils down to a single thing other than the form not being typical? I think providing three cuts, “pretty normal”, “unicase but still normal”, and “pretty weird”, could satisfy most people (except for all those who’d rather have the choice made for them, of course). I’ll think about it later on.

Adam Jagosz

In the meantime, what about these ogoneks?

Consistency of the diacritic shape or orthodox positioning? The top version is the original one (with slight correction), and I still feel it’s more at home here.

Adam Jagosz

Btw. I feel it could all be spaced a bit looser, mostly the verticals. Hardly anyone will use it at sizes that this was optimally spaced for (rookie mistakes).

Thomas Phinney

This is a fun design. I like it a lot.

I think the shape of the ogoneks is better in the top version — a bit deeper curve — but the joins are better in the bottom.

Could use more optical compensation in the joins for /n /a /d, etc.

Straight-to-curve transition is a bit abrupt in /n — too much like a straight line and a curve glued together.

The overall spacing is very tight, OK for large sizes. But that makes the spacing more tricky. Might need a bit more space around the straight elements, such as the lowercase /i /l. It is problematic in words like “grail” and “scrolls”.

André G. Isaak

Adam Jagosz said:

I think I can see what you’re getting at with /A: its structure (the stem only as tall as the x-height) is problematic and might cause a momentary lapse before it is parsed as capital, or might even be parsed as a weirdly big minuscule.

As for /E, I’m not sure if the problem boils down to a single thing other than the form not being typical? I think providing three cuts, “pretty normal”, “unicase but still normal”, and “pretty weird”, could satisfy most people (except for all those who’d rather have the choice made for them, of course). I’ll think about it later on.

I think the problem is a combination of the form being atypical and the very generous x-height — it takes one a moment to realize that these are majuscules rather than oddly-sized minuscules.

Vasil Stanev

If you're going for a more medieval look, maybe you should try a monocle g instead of a school room globe one.

Edit: also, you can include alternatives of the g and other glyphs so the font gets more bloated. 

Adam Jagosz

@Vasil Stanev I’m actually not going for a medieval vibe, maybe even the opposite. Actually I was considering removing the crossbar extension in /e since it’s jarring to some.

School room globe, huh, I’ll disregard this veiled insult Good idea, definitely a monocular /g would fit and be at least a useful alternate, even though I think the current one is as good as it gets.

Light, Medium, Bold, Black. (Yes, I know, my Bold is barely Regular, thanks.)

I’m not too happy with the Black yet, it’s a bit jumpy and random glyphs look too black/too light depending on point size. I’ll think about the serif treatment later as well, to maybe make them more bulky.

@Thomas Phinney What do you think of this weight progression? I’ll bet you don’t like how the advance widths barely budge. But unlike in that other project, here inner light is already abundant in glyphs with lower stroke density and I’ve tried stretching them wider but it only made them even more unbalanced. I increased overshoots to counteract it, but I don’t really have much ascender space left to increase the x-height altogether.

André G. Isaak

I like the g. If you opt instead for a monocular g, I'd at least keep the existing one as an alternate.

Vasil Stanev

I'm in love with this bold Maybe add some spacing in the heavier weights

Christian Thalmann

I like the globe /g/. I'm not so sure about the /e/'s nose or the very low stem in /A/a/.

I'd also consider moving the apices in the arches of /n/ etc. a bit away from the stem, as most typefaces do (even the geometric ones).

Adam Jagosz

Loosened spacing in Bold & Black, sanitized /e, + a preview of a less lazy approach to serifs (bottom part). Which do you like better? (Or maybe something in between?)

Adam Jagosz

Hmm. Torn between wanting to preserve proportions and to condense /ae into a single visual letter. Um, should the /e component of /oe and /ae be exactly the same?

Vasil Stanev

It gets better and better. In my mind, if the overall appearence is roundish, so should be the serifs. I would try a version with completely semicircle serifs just to see how it would look. You can add it to the family so the buyer has more options. This definitely has potential.

Adam Jagosz

Some ideas for the Greek: a naïve, Latinized approach, followed by a take on achieving a more native feel (open image in new tab for detail).

Both pitches contain some unicase + uncial extremes that I am afraid might be rejected by many readers.

I am obsessively tempted to depart from the most basic solutions, because I feel like those are already repeated ad nauseam. I’m not sure whether that is because Greek types are also for Western people, and so they must be super-legible to the point of being blandly obvious; or is that because the Greek themselves hold the script venerable and disapprove of any deviation?

Also, first micro-draft of an italic (I think it might need a bit of condensing):

I applied some easing on the serifs, which I consequently also applied to the (bottom) Greek. But then what will I do for the Greek italic?

Christian Thalmann

I find the /ae/ jarring; the free-style form of the /a/ part clashes with the geometricity of the rest of the typeface...

Love the top Greek! The bottom less so... I've been criticized for using such a loopy lambda in a geometric typeface once, so I wouldn't recommend it. (Then again, I believe geometrization and Latinization are different things, and this is the former, so...)

Adam Jagosz

@Christian Thalmann Apart from /lambda, do you like the top /kappa /pi /nu /upsilon /sigma /theta /xi better as well? In general, the top version has serif treatment copied directly from Latin, while the bottom one has them eased out sometimes, and is more varied in this regard.

I feel neither /delta is there yet, right?

Christian Thalmann

I generally prefer the top version. A lot of these letters look the same to me in both versions. The /delta/ is a bit aggressive, but I guess it fits the Latin globe /g/.

I don't think /s/ works as a final sigma; it needs a descender. Not sure about zeta either, but at least there the capital looks the same.

Adam Jagosz

A lot of the letters are exactly the same. A closer look then (I put a checkmark next to variants I think are better, and pointed arrows towards features I'm not sure about):

I eased out the /delta a bit.

As for /sigma1, I’ve seen some examples of this in Greece, in relatively plain sans serif fonts. (Those fonts tended to have the loopy lambda as well, though...) It’s also common in handwriting. I’m having an admittedly hard time finding a font like that right now, though.

/zeta (and the connected /xi) are directly modelled on Greek uncial, so...

Also, it’s a bit perverse to leave /eta with no descender while introducing one into /gamma and /pi...?

Adam Jagosz

A bit more streamlined /AE /ae. Not completely mutually cohesive, but I'm not sure which direction to push yet.

Christian Thalmann

I don’t mind this eta and love this pi, so I wouldn’t change the descenders. Eta and upsilon are perhaps a bit too wide, though. Eta also suffers from the apex problem of /n/. 

The omega looks sterile to me, as if from a sans. Maybe give it some organic stroke ends?

Andreas Stötzner

e and c are too helvetica, æ doesn’t work, a doesn’t work, bow-into-straight of h and n needs more care …

overall distribution of serifed vs. non-serifed endings is still unsolved. e.g. watch s, t and c: three glyphs from three different typefaces.

Adam Jagosz

@Christian Thalmann Thanks. I guess /n might be too wide as well then. As for the apex, I already tweaked it a bit, but going too far destroys the stem, so I'd have to give it a different serif altogether (lowered maybe).

@Andreas Stötzner Thanks for all the catches.

> e and c are too helvetica

I might open up the counters a bit more (I already did for /c, /e probably needs to follow suit (right now it has identical terminal without compensation). Thanks for pointing this out! But in general... what’s wrong with Helvetica? People “like” it.

After opening up (following a mask) I feel that /e is unbalanced, it falls over its nose. If I tweak it, it will no longer be geometrical (unless I make it follow /s somehow...). I don’t like this.

> æ doesn’t work

Too wide? Too narrow? Too dark joins? Too assymmetrical? Too... helvetica? (Just kidding, working on it). Just not right, right? Does this help? (New version at the bottom.)

Or did you mean a deep structural change? (Terminal borrowed from /s, so not a completely alien solution.)

> a doesn’t work

He is on vacation.

Oh, you mean the curvature? I agree.

> bow-into-straight of h and n needs more care

Woah, it seems it's heavier than both the vertical stem and the round curves. Must've come out like this when I played with the ease of transition. Not sure whether it's an issue on its own, or if it can be justified by curve smoothness... I tweaked the curvatures in /a /n /h (old top, new bottom), /n /h are narrower as well. I adjusted the botom terminal of /s too.

> overall distribution of serifed vs. non-serifed endings is still unsolved

But it is, straights get serifs (with minor exceptions, if anything, your examples should point there), curves don’t.

Christian Thalmann

That last /ae/ works best.

Why shouldn't curves get serifs? Some letters end up looking plain/sans without them. Perhaps try some serifs on /c/ and /s/, at least at the top?

I don't think keeping the current serifs on /n/ and smoothing out the arch is contradictory. Don't worry about geometricity; you have quite some leeway left before it starts to look ungeometric.

Have you tried an /a/ with a higher stem perhaps even up to x-height?

The serifs on the /t/ crossbar seem the wrong way around...

Adam Jagosz

I never claimed is was a serif face. Actually I want it to look sansey. Looking at /n, it's more of a semi-serif. I think of the existing serifs more as tapering or instroke/outstroke. I also want to be truthful to the initial design (as I mentioned, it was first released in 2017, and was even featured in a movie, hah, so it must have been at least a bit good the way it was). Serifs on /s/c would no doubt look fine, but a different typeface, and also a bit crowded.

/t follows /f /g which had to have an up serif to avoid crowding the counter / the join. I can see how it looks 'flipped' to you — you are imagining a classical roman T with two slightly diagonal, vertical serifs. But this is a bit different here.

On the other hand, /t could also follow /z. Then again /z is that way to follow uncial script's downstroke, despite sufficient inner space.

There is a similar issue in /Xi vs. /Theta.

If anything will make this look plain and sans-serif, it would be /a with full stem. But it consistently continues to go up and up, in Black it will be just slightly short of the x-height. (In “snap nine ache” above it went up by some 10 units). Unlike some features, this actually avoids one problem: a join. In addition, it repeats the shape of the serif. It serves a purpose.

Previously you referred to the thinning of the join in /n so I thought you were still onto that. by apex I thought you meant the serifed top of the stem. Regarding the arch... what more can be done?

Christian Thalmann

I never claimed is was a serif face. Actually I want it to look sansey.

Call it what you want — you do have serifs on many (most?) letters, so that sets up certain expectations for the others.  I do agree that it's not a fully serifed face, given that the feet of /n/ don't have any, for instance. I think «glyphic» is probably the best drawer to put it in.

Sure, if the unconventional /a/ is important to you, keep it!

As for the apex, I mean the top vertices of the arch. I suggest power-nudging them a bit to the right (and power-nudging the top vertices of the right stem up a bit in response).

Adam Jagosz

Then she’s a transsans 🌈

Will see about that /n, but to be in the clear: do I worry about getting increased weight in the top-right of the shoulder? Should it be between stem and bowl width? At bowl width? Can it be heavier? Or do I just trust my eye?

I think I will draw these /s /c /a after all, just for the fun of it.

Here's an /n drawn from scratch, with no optical correction beyond the minimum, created from narrowed down /o, so that we can start over instead of correcting over corrections:

What I did before was to power nudge the CPs down to ease out the transition. So from here, y=300 (x-height at y=600), do you still suggest I go up?

Christian Thalmann

I think the previous version was a better starting point than the redo from scratch.

Yes, I suggest to move the lateral points up (and the top points right).

This is what it looks like in my own geometric sans, Quinoa:

This is most likely too asymmetric for your case, but note that a lot of the asymmetry is seemingly lost when you look at it in context. In fact, the stem introduces an irreducible amount of asymmetric to begin with, and the arch shape is just chosen to balance the stem and the arch out in a harmonic way.

Remember, a geometric typeface should look geometric rather than be geometric, and it definitely cannot do both.