Elemaints - A Serif Family with Optical Sizes

Linus Romer

As a heuristic to get from the upright font style to the italics, I condensed the upright characters to 94% of their width and then slanted them 11° to the right. There are various problems with this: Diagonal widths change to varying degrees, arches appear distorted and serifs have to be partly shortened and partly lengthened. This technique is only suitable for cursive characters that are similar to their corresponding upright characters. The left \B is the result of the heuristic, while the right \B has already undergone some further optimization:

If the manual corrections were not successful, I sometimes resorted to expanding a curve with an elliptical pen:

The Cyrillic \Б (Be) is shown here in three stages of development: On the far left, with a bow horizontally adjoining the shaft at the top. This gives the optical impression that the bow on the right is hanging down slightly. In the middle is a first correction - the droop has disappeared, but the bow is somewhat dented. On the far right is the successful arch, which strives gladly towards a possible following glyph.

For visual reasons, arches are often less inclined than stems when slanted. Following the Goldilocks principle, I finally chose the middle variant for the Cyrillic/Greek \Φ (Ef):

The italic \Я (Ya) on the left is a condensed and slanted version of the upright \Я. On the right, the extremes of the Bézier curves are set correctly and the arch is bulging. The lower arc connection was later changed to an overlap in order to avoid interpolation errors:

Often only subtle changes are necessary. The middle part of the \S is inclined too steeply in the left partial illustration and is also set a little too low. The upper and lower counter appear rounder in the right partial illustration:

In cursive Cyrillic, some characters raise the question of how much handwriting should be incorporated into the form. The cursive \я has become increasingly handwritten from left to right:

A comparison with the basic characters \O, \H, \n, \o provides information about suitable variants. The more right-handed \δ fits better with \o and appears somewhat smoother, especially in the lowest rounding. It is also slightly condensed.

The left \θ is too narrow and not inclined enough compared to the \o. The right \θ fits better:

The left-hand \λ is the result of purely technical condensation and slanting. Compared to the \n, the right-hand \λ has more appropriate serifs and corrected stroke widths. Because the joint of the \λ has a smaller acute angle than the \n joint, the hairline of the \λ joint must be somewhat thinner to achieve an optical balance:

Linus Romer

The left \2 is the result of a purely technical condensation and slant. Compared to the \n, the right \2 has more fitting serifs and rounder arches:

I initially designed the italic \E too narrow. I also had to visually adjust the middle serif:

The Serbian ligature \љ (lye) on the left disregards the fact that the \л part should be more handwritten. The correction on the right also required that the upper arch connection be steeper:

The Serbian ligature \њ (nye) also disregards the fact that the \н part should be more handwritten. Such errors often only become apparent when the different characters are compared with each other:

The left \M looked too dark. The right \M is lighter because the left and right shafts have been widened more:

I have made the drop endings of \f and \j more similar to each other and also to the drop ending of \r (on the far left is the old \j). The drop ending of \r is naturally somewhat darker:

The drop ending adjustments made me wonder whether I should make the upper drop endings a little more open for the figure \3. I have therefore chosen a more open variant (third from the left) in the upright sections on a trial basis, which fits better with the \f drop ending:

I then tried out different variants in the italic versions with the new and old drop ending. In the end, I liked the old upper drop ending better and the rightmost variant is the one I have now chosen:

For round shapes, it is often better as a first step not to simply tilt the upright letter by 11°, but to stretch it slightly vertically, slant it by 8° and then rotate it by 3°. Both are linear transformations that have the same effect in the middle height, but produce a different effect at the top and bottom. The variant on the right is less distorted and closer to an ellipse:

The \A and the \V are both based on the \Λ. The adjacent illustration shows the expansion of the italic display section:

Based on feedback from a typedrawers.com user, I added a drop ending to the \J. As a basis, I slightly enlarged the drop ending of the \ɲ's and after several iterations finally arrived at the rightmost variant:

The \S from the cursive display cut was basically created from the cursive tiny cut: After slanting to the left by 8°, the character has been thinned out along the contours with an eraser pen of width 30 and rescaled evenly to the correct size:

The inclination of the stem of the left \p is exactly 11° and therefore the same as the inclination of \I. However, the left upper serif gives the impression that the \p is inclined too little compared to \I. The inclination of the right \p is 12°, but the optical slanting effect is the same as that of \I:

Linus Romer

I consider making all lining figures align with the capital glyphs and also widening them. The "too wide" line of the following picture shows an early attempt that I considered as too wide in a second thought.

I think that the new figures are more legible and not only fit the capital glpyhs better but also the lowercase glyphs. What do you think of the new figures?

Andreas Stötzner

Linus Romer said:

I consider making all lining figures align with the capital glyphs …

Why?

Craig Eliason

In addition to all the time you're putting into this design, it must have taken a lot of time to assemble these well-explained and well-illustrated posts. I appreciate that you have, an interesting read on the subtleties of type design and a useful resource!

Linus Romer

@Andreas Stötzner It was pointed out to me that in Elemaints Tiny the figures generally look too narrow and additionally too small compared to capital letters in all-caps. It was suggested to enlarge them at least for a .case OpenType feature. I then immersed myself in the subject and found that capital height figures are not that rare as I thought they were. Since Elemaints as a mathematically affine typeface uses quite large mathematical symbols (+, =, <) compared to usual text typefaces, I toyed with the idea of fitting the figures according to the capital height. And at this point I thought I might ask on this forum for feedback.

Your question reminded me that I have forgot to add also an alternative for the figures that is only wider but not taller. So here are the three alternatives (the alternative "too wide" from above is discarded):

Christian Thalmann

Great stuff!

Just a quick note: You regularized the spacing of /n/n/ by fitting rounded rectangles into the inside and outside spaces. Visually, though, the space between letters is larger because it is open to the sky and the preceding /n/ is curving away from it. Thus, I would expect the visually balanced solution to yield somewhat narrower spacing than your purely mathematical solution. In particular, your new italic /n/n/ look too wide to me compared to the original next to it. (Then again, one would need a running text to judge that properly.)

Typofactory

It is funny that Googling “Elemaints” into image search yields pictures of a typeface that you cannot download, just because of how many samples Linus has given us.

Linus Romer

@Christian Thalmann The «mathematical pure» spacing rhythm is only applied for the tiny faces since their spacing has to be a bit more generous. Going from tiny to display faces the visual effect that you mention is respected more and more. Here are text samples for Tiny, Regular and Display:

Christian Thalmann

Good to know! The Tiny Italic does look (to me) significantly looser than the Tiny Roman, though, whereas the two match well at larger sizes. (Looking at it again, the Regular Italic zwitschernde is still pretty loose...)

konrad ritter

While I second everybody above who's praised your work on this face, I'd like to suggest a humble correction that would increase legibility, or at least make that particular glyph less distracting. The top serif on your lowercase /z/ is a butterfly serif, as it were -- it spreads out on both sides. I would change it to be a mirror reflection of the bottom serif. 

Anything that helps this beautiful face look more transparent, and also unlike Minion, which some people have had enough of. No disrespect to Slimbach, of course. 

Linus Romer

@konrad ritter I see your point for the lowercase \z and find that the single-armed serif works too for \z. However, for consistency reasons the double-armed serifs at \Z and \T would also have to change. Here is an example of how it would look like:

But I do not like the \T without double-armed serifs. So I will probably stick with the double-armed serifs for \T, \Z and \z.

Ruixi Zhang

Linus Romer said:

However, for consistency reasons the double-armed serifs at \Z and \T would also have to change.

You can keep the top serifs of /T, though. Shown below is DJR’s Fern:

Linus Romer

@Ruixi Zhang But as far as I know a single-armed \Z and double-armed \T in the same typeface is rare (also in historical context). Furthermore, "TZ" and "TZT" do then look strange (and "TZ" and "TZT" are much more frequent in German than in English), don't they?

Linus Romer

I think, I will keep the old figure height. But I have widened the figures of the tiny face. The figure widths of the display face are unchanged. The regular face is the interpolation inbetween. 

The following image shows Tiny, Regular and Display:

Linus Romer

@Christian Thalmann The spacing of the italic faces is still in the early stages, so I particularly appreciate feedback on this. I have tried to optimize the spacing of all italic faces (most side bearings were reduced):

Christian Thalmann

I like the figures in running text, but next to capitals (as in H0) I find the height different irritating. What's the purpose of it? If you want the figures to blend into text as smoothly as possible, wouldn't you use old-style figures instead?

The new Italic spacing looks good to me at first sight. Maybe a hair tight in the Display? I'm not sure.

Linus Romer

@Christian Thalmann

If you want the figures to blend into text as smoothly as possible, wouldn't you use old-style figures instead?

Yes. Currently, my font files support lining-tabular, oldstyle-tabular, lining-proportional and oldstyle-proportional figures. Lining-tabular figures are the default figures. These are the figures that are the most frequent in scientific papers.

Here is a compilation of frequently used typefaces in scientific papers that shall demonstrate that figures with a height below capital height are very common:
The lower row has figures with capital height (Palatino's figures are slightly below capital height).

Christian Thalmann

Ah, interesting! I had no idea that was a thing. Don't mind me, then.

Linus Romer

Over the past year, I have unfortunately not been able to spend as much time on Elemaints as I would have liked. However, I am still working on this project and I intend to publish it as soon as I have finalised everything. Nontheless, I am still a long way from publication. The basic characters of the italic faces have already undergone several revisions and are approaching their final form. Afterwards I will draw the remaining characters. Here is a snapshot of the current status of the project (the spacing is still suboptimal):

At the moment I am also frequently testing the effect of the tiny faces printed at 5 pt on paper. For example, the «w» at the bottom left was too strong:

michele casanova

Very interesting work! Are there plans to distribute it as a variable font, or will it only be available as separate interpolated fonts?