Elstob: a Fell-based font

Peter Baker

I think you're right, Thomas. When I actually measured the contrast in the model, I could see I was way off. Does this work better?

I found the n very hard to do. I ended up widening the leftmost vertical somewhat.

Yes, there is an optical size axis, and there will be less contrast at smaller sizes.

Thomas Phinney

That is certainly an improvement as far as contrast, for a “display” set of masters.

What size is the display intended for? Is this aimed more at print, or on screen, or...?

Peter Baker

I'm thinking of it as primarily (but not exclusively) a variable webfont with weight, optical size and grade axes. The opsz axis runs from 6 to 18 right now, but maybe 8 to 24 would be better, since 6 isn't all that useful on a screen, and, as you say, Double Pica is approximately 24 points.

Dave Crossland

Will you develop the typenetwork parametric axes?

Peter Baker

@Dave Crossland: I like the idea of parametric axes, but have been hesitant about them for this project for a couple of reasons.

One is that my audience is mostly academics in literature, history and linguistics--not the most technologically sophisticated people out there. My experience trying to introduce users of my earlier fonts to OpenType features (they rarely rtfm) suggests that it's going to be hard enough to get them to understand my three axes, weight, optical size and grade: the parametric axes are harder to understand and perhaps less likely to be used.

My second concern is simply the additional labor of producing that many more masters, plus the complication of having to devise a new workflow that includes fontmake as well as Glyphs (which I believe still limits fonts to six axes).

Perhaps I'm underestimating the sophistication of my audience and overestimating the difficulty of implementing parametric axes. I'm certainly open to argument.

Peter Baker

Still working on italics. After the last few posts I realized that I had somehow made them boring and returned to the specimen for closer study. I'm more satisfied now:

But I'd mainly like to solicit advice about the caps, for which I have (so far) stuck pretty close to the specimen, including serif length (different for every letter), and A, V, W, which appear to have a much more extreme angle than the other caps. These things together are making work on the metrics something of a nightmare—I suspect that the italic was never meant to be set in all caps. You can see here that I'm still some distance from achieving good spacing for the caps:

And here that spacing was also a problem in the 1925 specimen:

Should I normalize the lengths of the serifs? And should I change the angle of A, V, W? And of course any other suggestions are very welcome.

LeMo aka PatternMan aka Frank E Blokland

For your information: a first series of overviews of the ‘Text’ weights/styles of DTL Fell was published last week on DTL’s vintage website (since 1998). Images of the special ‘Display’ editions, alternates, discretionary ligatures, and ornaments will follow soon. An overview of the glyphs can  be found here (PDF).

DTL Fell’s development began in 1996 when my wife, who is an art historian, and I visited the Bodleian Library to study the related type specimens and to order photographs. We then visited the Oxford University Press to examine the matrices of the English Roman and Italick.

Peter Baker

I was aware of DTL's Fell project, though I hadn't seen a sample until now. It's an interesting article, and beautiful work! The English is I think the most beautiful of the Fell types: I chose the very different (and not so distinguished) Double Pica as a model because it seemed to me that it was likely to lend itself better to development as a variable font.

LeMo aka PatternMan aka Frank E Blokland

Thank you very much for your compliment. DTL Fell had some time to mature: its development began in 1996, as mentioned. In the course of time, a handful of designers worked on the typeface. So far, the first (British and American) customers have been very positive about the result. We will pay some extra attention to DTL Fell in 2020 when we celebrate the 30th anniversary of the Dutch Type Library.

Adam Jagosz

There’s something odd going on with the hinting. Looking at the github pages site on Windows, 118ppi, at medium sizes, the /i is much taller than the (prevalent) x-height.

Christian Thalmann

The /l/ in Elſtob also looks to me like it's hovering.

Peter Baker

For right now I'm just running the generated fonts through the beta variable-font-enabled version of ttfautohint (is this Adam Twardoch's project?) and accepting the output without alteration. When the font is in a more stable place I'll look harder at hinting, perhaps with VTT.

Adam Jagosz

Oh, but you did correct this issue, either voluntarily or by accident Though now /i is a bit short in some sizes. The black magic of hinting, I will never get it. And no one cares, since everyone who could fix this globally already uses hi-res screens.

Peter Baker

I've been experimenting with VTT this morning, autohinting and then editing selected glyphs (including i). I uploaded one version to github. That v is pretty awful--the autohinter leaves the right serif stranded.

Peter Baker

Manual hints don't seem to coexist very well with the autohinting. I suppose people who use VTT seriously develop their own control programs and hint everything manually.

Peter Baker

Recently I've been adding Greek to Elstob. I'm no great expert on Greek type, though I've gone through whatever resources I can find online (especially Gerry Leonidas's stuff). Some years ago I added Greek to Junicode, but the stress of those letters was Latin-like, and I would like Elstob's stress to be more traditional—so I am in relatively unfamiliar territory. I could go on about my doubts and queries, but maybe it's better to just post what I've got so far (with Latin for comparison) and let people react. Thanks to anyone willing to weigh in!