Regis, a new serif font

Daniel Benjamin Miller

Hello everybody.

First, a little about me: I’m an undergraduate student at McGill University. I have been interested in type design for quite a while, but this is my first serious endeavor towards actually creating anything which would seriously be worth using, which I’m calling Regis (in reference to Romain du roi).

I was reading some books printed by the Cambridge University Press set in Monotype 178 “Barbou,” which drove me to search for a digital font which was reminiscent of this face. I found a few which similarly follow, like Barbou, in the footsteps of Pierre-Simon Fournier, but none which replicated the warm, soft impression I felt when reading these books (I would post a scan, but unfortunately the text is under copyright and I don’t wish to run afoul of any authorities). Furthermore, being an avid user of free and open-source software, I wanted to design a font which would be able to be used by anybody under a free license. (In terms of Fournier-inspired free fonts, Bitstream Charter and Adobe Source Serif Pro both exist, but their designs go in a very different direction from my intentions and I preferred beginning from scratch to making modifications.)

I’ll note that Stephen Coles’ specimen of this face has a number of details which are not present in my face (such as the gap in the lowercase w). As the linked sample and the books I have are both set at 10 point, I find it likely that these variations are the product of the printing process. My desire is to replicate the feel of these books to some extent, not whatever Monotype happened to have been doing. In any case, I've also taken some liberties with my design.

So far, after a few weeks, I have a lowercase alphabet (minus a z), an uppercase M and a period/full stop. The kerning I have yet to seriously work on, so forgive me if it is off-kilter. This is my first design, so please let me know what you think, though please keep in mind that this is an extremely early draft. A few of the letters are wonkier than others at the moment. The line spacing is also a bit too cramped in this sample.

Thanks are due to Luc Devroye, who happens to be here at McGill, with whom I’ve discussed this design over the past few weeks and who has provided some input.

The source (FontForge format) is available on my GitHub.

Thank you!

Adam Ladd

Subtle, but the tittles of /i and /j feel a touch dark (large) and high/floaty to me (particularly when set in a word with no other ascenders, like "sit" above... not as pronounced when near ascenders though). Also, the /t feels a touch too narrow and dark... maybe the cross stroke and tail extend slightly to the right, or the cross stroke is a little thinner?

Christian Thalmann

The bowls of /b/q/ look jarring — did you mirror /d/p/? The heel of /b/ is also very unusual.

The /a/ is lower-contrast than the rest, and there's something gooey about its shape. It feel to me as if it were part of another typeface.

The /g/ is significantly too wide for my taste.

Daniel Benjamin Miller

Adam Ladd said:

Subtle, but the tittles of /i and /j feel a touch dark (large) and high/floaty to me (particularly when set in a word with no other ascenders, like "sit" above... not as pronounced when near ascenders though). Also, the /t feels a touch too narrow and dark... maybe the cross stroke and tail extend slightly to the right, or the cross stroke is a little thinner?

Thanks, I think I will probably shrink them a little bit and see how they see. The cross stroke of the /t is this way in order to line up better with the /u. Not sure how to rectify it, then.

Christian Thalmann said:

The bowls of /b/q/ look jarring — did you mirror /d/p/? The heel of /b/ is also very unusual.

The /a/ is lower-contrast than the rest, and there's something gooey about its shape. It feel to me as if it were part of another typeface.

The /g/ is significantly too wide for my taste.

They’re not simple mirrors, but that was the initial model. They still need some significant adjustments. The heel of the /b, I think, reflects Barbou (as seen in the linked specimen). I also think this is generally the case for the width of the /g, though it's possible it could be thinned a little bit. The /a is fairly new, and I'll try to get it a bit “sharper,” so to speak; it’s posed a few more challenges than some other characters.

Christian Thalmann

It's true that the /b/ in the sample shows the same heel as yours. My other concerns seem to be well-addressed in the sample, though. The /g/ has more balanced counter sizes and looks more centered horizontally, the /a/ has pleasingly clear-cut curves and contrast, and the bowls of /b/q/ agree with the stress direction of the typeface.

Kent Lew

The heel of b is a distinctly Fournier characteristic. It never really caught on with his contemporaries.

The wide-ish g is found in Monotype’s Fournier and Barbou, but was not necessarily always the case across Fournier’s œuvre. To make it work better, you should balance the two bowls a bit more. If you look again at your exemplar, the eye is a bit larger and the bottom bowl a bit smaller than you have.

Here’s a snippet of Barbou from Stanley Morison’s Tally of Types.

P.S. Don’t even think about kerning until you’ve got the design and fitting well in hand. You should be able to make it look decent without any at all, and then the kern pairs are only for fine-tuning.

Daniel Benjamin Miller

Christian Thalmann said:

The bowls of /b/q/ look jarring — did you mirror /d/p/? The heel of /b/ is also very unusual.

The /a/ is lower-contrast than the rest, and there's something gooey about its shape. It feel to me as if it were part of another typeface.

The /g/ is significantly too wide for my taste.

How does this look, after a little bit of adjustment?

Henning von Vogelsang

I think it’s the stroke angle that could help you improve the font. It’s somewhat variable with serif fonts and depends on the era and the style. Caslon fonts, or Times New Roman, have a rather straight angle, whereas Rennaissance or Venetian serifs have a rather tilted angle.

You see the angle in the thickness of the stroke, inherited from drawing letter shapes with calligraphic pens and brushes. The angle is more of visual guidance for and rarely applied strictly across the entire font, but if you compare your letters /b, /d, and /q, /p with Adobe Garamond, or with ITC Caslon, for an example, you see how they applied the stroke angle.

One trick I use is drawing the letter /d and then turning it by 90º. It gives me an idea of how the bowl of the /p should look.

Craig Eliason

180º works even better!   

Henning von Vogelsang

That’s actually what I meant, @Craig Eliason. Thanks for correcting me!

Daniel Benjamin Miller

Still trying to make the whole b-d-q-p set work. In the meanwhile, I did do a new /g, which I think responds at least in part to some of the concerns voiced by @Christian Thalmann .

Thomas Phinney

The g is looking just a *little* light, right now. And the top part of the bottom loop feels a bit stiff and odd to me. I think it may be because the top of the stroke starts changing well before the bottom... and the right side of that bottom loop is thinner than the lower right—that is just off considering the stress of the rest of the font. The whole bottom loop feels a bit like a curling stone, somehow.

Daniel Benjamin Miller

Thomas Phinney said:

The g is looking just a *little* light, right now. And the top part of the bottom loop feels a bit stiff and odd to me. I think it may be because the top of the stroke starts changing well before the bottom... and the right side of that bottom loop is thinner than the lower right, which is just off. The whole bottom loop feels a bit like a curling stone, somehow.

Considering all the characters, it may be less that the g is light, and more that the direct-vertical strokes are a bit heavy in the i, l and derived characters. so that might be adjusted soon. (ed. Or, maybe, on second thought, I'll darken the g.)

As for the bottom loop of the g, it's been pretty tricky to get just right. So it's still very much a work in progress.

Henning von Vogelsang

With the example of these two serif fonts, Sabon and Spectral, you see the stress corresponds to the angle (more or less) of the stroke drawing down.

Kent Lew

And yet you see with later examples, not in the Garalde vein, such as Fournier/Barbou, which had been identified as a primary source of inspiration for this effort, such a stress pattern is not really the case at all.

Daniel Benjamin Miller

To add to the samples of my inspirations, check out the following links to some scans I've put on my website: https://fr.dbmiller.org/178/grayscale.pdf https://fr.dbmiller.org/178/bw.pdf

Henning von Vogelsang

It is not strong, sure. The original specimen is a letterpress print version in small size. I think you can emulate the bleeding effect in thin stroke lines with digital fonts if you want to achieve the same kind of washed out look. It’s a stylistic decision.

The example I gave was meant as a guide to look at common patterns, it was not meant as a set of instructions.

Daniel Benjamin Miller

Some updates. (See also my GitHub, linked above --- you can download this and compile to OTF in FontForge if interested.)

Jasper de Waard

d and q still look awkward. Have you tried vertically flipping their bowls? (but keeping the bowls of b and p the same!)

Also, g is falling backwards (s do c and e, actually), and the diagonal thins seem a bit too thin to my eye. 

Daniel Benjamin Miller

Jasper de Waard said:

d and q still look awkward. Have you tried vertically flipping their bowls? (but keeping the bowls of b and p the same!)

Also, g is falling backwards (s do c and e, actually), and the diagonal thins seem a bit too thin to my eye. 

That would look like this:

The g, c and e I might adjust as well. The diagonal thins are thin on purpose, as they are also a bit thin in the source materials (see the links above). I might try to thicken them very slightly, though.

(I also shrank the tittles in i and j, by the way.)

Daniel Benjamin Miller

The lowercase, updated. The tittles have been shrunken; the t is a bit adjusted.

And a specimen (missing characters and all from Lorem ipsum, plus alphabets as they stand):

Craig Eliason

I'd suggest thickening the end of the loop of /g/ (i.e. the lower left). /f/ and /t/ crossbars read as thick. Maybe reconsider the treatment of bottom left of /b/ and top right of /q/. I find the connected serifs of /w/ distracting. The balance of counters in /a/ seems out of whack with the imbalance in /e/. /m/ may be a bit wide but the proportions overall are sound. You can cheat on the /r/ by extending the right baseline serif to help fill the void and keep it from tipping over. 

Daniel Benjamin Miller

Craig Eliason said:

Maybe reconsider the treatment of bottom left of /b/ and top right of /q/

What do you mean by this exactly? To consider getting rid of the serifs?

Craig Eliason

Yes. 

Daniel Benjamin Miller

Not a consideration for me, then, sorry—it's a key distinguishing element of Fournier's design.

Craig Eliason

Maybe cheat those serifs a little shorter then. And the notch gets a bit lost in the /b/.

Daniel Benjamin Miller

I have not tried tweaking the b and q yet per those suggestions, but I did deal with the diagonal thins a bit @Jasper de Waard , and also tried to balance a few of the letters which were said to be falling over.

Michel Kern

Hello @Daniel, seems for me that the dot on the /i is too high like about 20%

Daniel Benjamin Miller

How's this? @Michel Kern

Michel Kern

@Daniel Seems better, you shrinked like 15% no ? Maybe a little more still
Theres's also the Serifs on /p /q whih seem too larg IMHO

Daniel Benjamin Miller

Yes, @Michel Kern , I shrunk them around ~15% if I recall.