Elemaints - A Serif Family with Optical Sizes

Florian Pircher

Minion Math does something interesting in this case: the Latin /v and /w get substituted by rounded versions while the Greek /nu is kept pointy. That gives the reader a chance to distinguish /v and /nu. See the presentation at 19:19:

https://youtu.be/Xw7oBf8EmoM?t=1159

Your design, while it uses different paths for the two letters, is still too similar to be useful in formulas where both symbols appear. Maybe exaggerating the change you just made would suffice?

Or maybe, authors just should not use /v and /nu in the same formula, similarly to how /upsilon and /omicron are rarely used in mathematics.

Daniel Benjamin Miller

Florian Pircher said:

Minion Math does something interesting in this case: the Latin /v and /w get substituted by rounded versions while the Greek /nu is kept pointy. That gives the reader a chance to distinguish /v and /nu. See the presentation at 19:19:

https://youtu.be/Xw7oBf8EmoM?t=1159

Your design, while it uses different paths for the two letters, is still too similar to be useful in formulas where both symbols appear. Maybe exaggerating the change you just made would suffice?

Or maybe, authors just should not use /v and /nu in the same formula, similarly to how /upsilon and /omicron are rarely used in mathematics.

Indeed, and you will find this done for other fonts as well, for math (e.g., STIX). For text, meanwhile, there is no need to distinguish v and ν that much, because I'll be able to tell a running Greek passage from a Latin-alphabet one. I personally have no problem with the rounded v and w for math.

I'll also note that TeX, if memory serves, does not have a separate omicron for the math font. Lower-case upsilon is there, though, if I recall…

Linus Romer

I am already aware of the practice of rounded math italic /v for maths (e.g. Barbara Beeton confirms this practice in this stackexchange thread) and personally I do not like this convention too much. But the comment of @Florian Pircher probably points in the right direction. I should really consider a rounded math italic /v (and /w as well) and still use the pointy /v for text...

By the way: STIX2 has two variants vor math italic /v,/w and even /u (probably to distinguish from /upsilon). But I design Elemaints mainly for LaTeX, so I will choose only one version per math italic glyph.

@Daniel Benjamin Miller Yes, you are right about standard TeX not having a separate omicron. The OML-encoding does not include it whereas it includes almost every other lowercase Greek glyph.

Anyway, I want to keep the new /nu (please excuse this bad phonetic joke) for text. In the following you will see the combination "/nu /v" for different faces.

Florian Pircher

The text /nu looks good to me. As @Daniel Benjamin Miller pointed out: “there is no need to distinguish v and ν that much”. For mathematical typesetting you could also consider using vertical stress for /nu to further differentiate it from /v. Minion 3 conveniently offers both forms of stress, which makes for an easy comparison (Latin-style stress on top, vertical stress on bottom):

Linus Romer

@Florian Pircher Interesting idea but I doubt whether this makes /nu and /v easier distinguishable. For now I have concentrated on your first suggestion because there seem to be many people that expect to have rounded math italic /v and /w. The following picture shows /v, /w, /v.var and /w.var:

Linus Romer

What I have done since the last post:

1. I have written a patch for fontforge that approximates merged paths better. FontForge has merged my pull request.

Before (left) and after (right):

2. The patch helped me to improve the Italic Caption face. I have completed the Greek glyphs (without spacing/kerning optimization):

3. I am planning to add Cyrillic to all faces. The sketches for the Caption face are still quite rough. I have used "modern" straight/triangular shapes:

Christian Thalmann

Wow, nice bit of coding!

Greek looks good to me, apart from the missing kerning (/eta/mu/ will need some positive kerning). Perhaps /varrho/ could be more slanted? It currently looks rather upright. I'm also wondering whether the «foot» of /sigmafinal/ isn't a bit too steep (or low?); it might benefit from a bit more «sitting» on the baseline, if that makes sense. The /zeta/ seems to work fine though.

I suspect the Cyrillic will need a bit more work (note that I'm neither a native reader of Greek or Cyrillic, though!).

• /De-cy/ will need some wider shoulders around the triangle. I know it makes kerning a headache, but it is what it is.
• Cyrillic descenders are letter parts, not serifs. Yours should probably be larger, and I suspect they would look better angling inward than outward.
• The triple letters (/Zhe-cy/, /Sha-cy/, /Shcha-cy/) strike me as too narrow. 
• The top of the bowl of /Be-cy/ and related letters looks a bit sloping to me; I suspect it would profit from some bulge (rising slope where it leaves the stem).
• The /Ia-cy/ looks crowded and is probably too narrow.
• The breve in /Iishort-cy/ feels small and floats significantly higher than the dieresis in /Io-cy/. (Possibly the former should be wider and the latter higher?)
• The stem of /Ef-cy/ should overshoot a bit, and looks small. 
• As for /Dje-cy/, check out this link.

Cheers!

Craig Eliason

Is /zeta a little too light above the baseline?

Linus Romer

@Christian Thalmann & @Craig Eliason Thank you very much for your helpful hints!

I have made the thigh of /zeta above the baseline thicker. The /varrho should now look more slanted (mainly achieved my adjusting the descending tail). I have changed the descender of /sigmafinal and I think it looks better but I still fear I might not have understood the hint of Christian completely...

/De-cy has now more prominent shoulders and the descenders are darker and leaning inside:

The triple letters /Zhe-cy, /Sha-cy, /Shcha-cy are now a bit wider:

I have bulged /Be-cy and related letters. At first glance the bulge might seem to be too strong but for small sizes (as captions are) it seems appropriate to me:

Regarding the /Ia-cy I restarted by extending the bowl instead of shrinking the leg and I think it looks less crowded (and surely is wider):

The accents were wrong due to old scripts that did not include cyrillic. Here is what it should look like:

The overshoot of /Ef-cy is now implemented. (Unfortunately, my main Cyrillic reference typeface PT Serif does not overshoot the /Ef-cy.)

New /Dje-cy and /Tshe-cy:

Linus Romer

The cyrillic glyphs have been completed to cover the T2A encoding and the W1G subset. According to http://latex3.github.io/help/documentation/cyrguide.pdf, T2A supports the following languages:

Abaza, Avar, Agul, Adyghei, Azerbaijani, Altai, Balkar, Bashkir,Bulgarian, Buryat, Byelorussian, Gagauz, Dargin, Dungan, Ingush, Kabardino-Cherkess, Kazakh, Kalmyk, Karakalpak, Karachaevskii, Karelian, Kirghiz, Komi-Zyrian, Komi-Permyak, Kumyk, Lak, Lezghin, Macedonian, Mari-Mountain, Mari-Valley, Moldavian, Mongolian, Mordvin-Moksha, Mordvin-Erzya, Nogai, Oroch, Osetin, Russian, Rutul,Serbian, Tabasaran, Tadzhik, Tatar, Tati, Teleut, Tofalar, Tuva,Turkmen, Udmurt, Uzbek, Ukrainian, Hanty-Obskii, Hanty-Surgut, Gipsi, Chechen, Chuvash, Crimean-Tatar.

Additionally, I have replaced the standard breve by a cyrillic variant.

The spacing is not yet optimized and there is no kerning at the moment:

Christian Thalmann

Very pretty overall.

The /el-cy/ looks very small to me. I'd try widening its stance.

The triple-stems /Zhe-cy/ and /Sha-cy/ still strike me as too compressed. And /E-cy/ is perhaps a bit too wide? But it does match your /C/... hm.

Linus Romer

The stance of the /el-cy is wider now. The /Sha-cy is wider as well. /Zhe-cy and /zhe-cy are wider and a bit more open at its joints. /E-cy is quite a bit narrower now - I think it matches the /C still well:

Vasil Stanev

(picking up from when I last saw this thread last year, two pages ago)
The Cyrillic is solid, it is reminiscent of Times New, so you can't go wrong. I would only push the roof of the Д to the right-hand side, because now it looks too much to the left - due to the two diagonals being of different weight.
Considering your earlier efforts of the mathematical symbols, I noticed a tiny thing you could improve:

In the picture the bowl and the serif look glued together, while they should flow more naturaly. Feel free to read on the so-called "bone effect", an often overlooked but important part of oval-to-straight relations.

Doing this, of course you must consider the stroke thickness of the outline, but this can be done by an experinced designer, eg you can see it in fonts like this one by a Bulgarian  foundry:



 https://www.myfonts.com/fonts/typedepot/centrale-sans-inline/


Please notice the elegant solutions that give it it's character and feel free to use them in your design.

Linus Romer

@Vasil Stanev I have now shifted the roof of the Д and the д slightly to the right and I think it looks better, indeed:

About the ℝ: Don't you think the bad impression is due to the joints that are not thinned out? Here is a comparison with the old ℝ (left) and the new ℝ (right) with thinned joints:

Vasil Stanev

Vasil Stanev

But this is minutia. If the sign is viewed at small sizes, this doesn't matter so much.
Д is fine. maybe a biiiiiiit more to the right. Care to not clog the foot of the right diagonal when it is near the right descender. Descenders of Д may sometimes be slightly assymetrical and of slightly different weight if this looks better. You can make the left one slightly heavier and the right one slightly lighter to rhyme with their slanted strokes respectively. Think of the serif of G that has to compensate the lower right part of G. Balance.

Linus Romer

@Vasil Stanev Regarding the ℝ: I restarted the process starting from an outline (image left) to its mathematical stroked version and then finetuned (image right):

Note that the starting outline really pays attention to a smooth change of curvature from straight paths to curved paths: The second handle lies on the Tunni point and therefore the curvature is 0 on both sides of the points that join straight and curved paths.

I have already tried merging the two counters to the very right to one counter; but this is difficult without changing the shape of the ℝ heavily.

Linus Romer

And here are the Д (with even more shifted top) and the д with asymmetrical dark descenders:

Linus Romer

What I did not yet : spacing and kerning

What I did:

1. Added Cyrillic smallcaps for the Caption face:

2. Added Vietnamese for the Caption face:

3. Made accents less rounded but more edged (in the course of adding Vietnamese accents).

Any remarks regarding this typeface are welcome!

Drawcard

Linus Romer said:

Any remarks regarding this typeface are welcome!

I'm lovin' it! That is all.

ducxcao

Linus Romer said:

What I did not yet : spacing and kerning

What I did:

1. Added Cyrillic smallcaps for the Caption face:

2. Added Vietnamese for the Caption face:

3. Made accents less rounded but more edged (in the course of adding Vietnamese accents).

Any remarks regarding this typeface are welcome!

Hello Linus, a few comments on the Vietnamese diacritics:

1. Overall, Vietnamese does requires a more generous space when typesetting, meaning that some diacritic combination will be higher than the ascender height, which is why I suggest you designing the double-stacked ones for the uppercase letters first — especially the combination that requires the most generous space found in Ẩ and Ổ or Ẫ Ỗ (/circumflex + /hookabove or /tilde) — that way when you switch to the single one, the design doesn't change (as can be seen now in ấ or ẫ, ă and ằ). It'd be great if we can avoid squeezing them altogether.

2. Tilde feels thinner than the rest.

3. All the diacritics position right now is a bit off to the right. This also happens in the /tilde + /breve combination.

4. /ư & /ơ definitely needs their own spacing and kerning values, not just following the /u and /o. If you look at the word "nước" in the first line, for example, ơ and c are too close to one another. Watch out for this. 

5. Watch out for the diacritic clashing with the uppercase letter in some of the common letter combination e.g. Tề, Về, Tằm, Vằn.

Hope these help!

Linus Romer

@ducxcao I appreciate your advice!

1. & 2.  I have tried to make the accents more consistent and to avoid squeezes:

The accent circumflex+tilde itches me a bit: I feel the urge to move the tilde to the right in order to reach an optical balance.

3. This is a problem of the base anchor: So, you think I should move the base anchor of /A and /a to the left?

4. & 5. I will try to keep this in mind when I optimize the spacing and kerning in the end.

Linus Romer

So here is what the Vietnamese looks in the different optical sizes:

I have the impression, that the accents of the lower optical sizes tend to be a bit too dark at the moment. And I should probably shift the tilde for capitals like /Ohorntilde a bit higher.

Here is the cyrillic in the different optical sizes:

Any hints on Vietnamese and Cyrillic are welcome.

Christian Thalmann

I still think a lot of your Cyrillic caps are off in width. In particular, many of the display caps look condensed. Best talk to a native type designer. (I might be wrong.)

As for Vietnamese, perhaps the /circumflex_acute/ and /circumflex_grave/ look a bit too similar at small optical size? The secondary accents are very point-like there.

Linus Romer

@Christian Thalmann About the widths of Cyrillic caps: Luckily, I have a native reader at hand, but unfortunately no native type designer. One model typeface that I used when forming the shapes of the Cyrillic glyphs was PTSerif. When I compare the display caps of Elemaints (blue) with PTSerif (red), I get the following picture:

With respect with PTSerif I would consider "Ef", "Che", "Dje", "Tshe", "Ka with vertical stroke" and "Shha" as too condensed. Are these the glyphs you had in mind as well?

Christian Thalmann

I think it might have been in particular /Sha-cy/ and /Zhe-cy/, and to a lesser degree /Pe-cy/, /Tse-cy/, and /Che-cy/ in the Display cut. The lowercase versions are also affected (/tse-cy/ is positively compressed). But maybe that's just part of the design of the Display cut, and the Latin looks compressed too?

As I said, better ask a native type designer. I do tend to get the feedback that my triple-stem letters are too wide, so maybe my eye is at fault here.

Jānis Kalaus

Hello!

Look for the book “Book about letters from Aa to Яя”. Currently the latest edition (https://store.artlebedev.com/books/) is sold out. Maybe you can find it on Amazon or elsewhere. It is in Russian. But it clearly shows the structure of the Cyrillic letters. The scanned book can be downloaded here >> https://vk.com/doc89880629_438607155?hash=df4607c107a11af588

Jānis Kalaus

"Книга про буквы от Аа до Яя", Юрий Гордон
Or try asking the author of the book >> https://yurigordon.com/

Jānis Kalaus

The third supplemented book has now been published. Some excerpts:

Jānis Kalaus

The Taste of Text Like the Taste of Honey with Yuri Gordon

ENG tranlation >>