Kerning in Kepler

ncaleffi's picture

Hi there, I've just started using the wonderful Kepler by Slimbach. The font is great, of course, but there's a strange kerning value set between the lower case f and the open parentheses and brackets. The default kerning here is 126, which looks a lot. It can be manually adjusted, of course, but it would be interesting to know the reasons behind such a choice.

Dan Gayle's picture

I like Kepler. Especially the display weights.

Thomas Phinney's picture

I think that looks like a bug. Perhaps Miguel or Paul can check with Robert and comment....

T

Florian Hardwig's picture

My guess: +126 is the amount that’s necessary in the Italics to keep the bracket and the f descender from clashing?

kentlew's picture

I only have the original Kepler MM fonts to look at, but I think Florian is on to something.

Checking the AFMs for the Light Display masters, it looks like in Roman the parenleft-f pair would have a kern on the order of -20. But in the Italic, the parenleft-f pair gets a kern more like +110 to +130. Factoring in some interpolation, I imagine that's where the +126 value comes from and was just accidentally picked up from the Italic and applied to the Roman.

So, Thomas is right: that would be a bug.

-- K.

Miguel Sousa's picture

Thanks for bringing this to our attention.
I looked at this issue with Robert and he first thought it could be caused by a bad kern value in one of the masters. But then we looked at all the Roman fonts in the family and the kern value between the left parenthesis/bracket/brace and the 'f' is pretty consistent. So, yes it's a bug, but we no longer have enough information to know exactly how it happened.

eliason's picture

Don't think of it as a bug, think of it as a subtle homage to Linotype machines. :-)

johnnydib's picture

Wow interesting!
I can imagine how Slimbach would react to this!

ncaleffi's picture

I'm going back to this thread because I don't want to create a new one about the same typeface.

I've noticed that in Kepler the "Th" is set as a standard ligature.

(above: ligatures turned on by default; below: Th ligature manually desabled)

I respect the designer's choice, but from my point of view it would have been better to have it as a "discretionary" ligature. When set at very small sizes, say 6 or 8 points, the "Th" ligature can be a little confusing to the reader's eye, with the two different glyphs resembling one single letter. Who types the text can correct them one by one, since you can't turn off a single ligature without affecting the others - a hard work if you're setting a book, for example. So I believe that having it as discretionary, as someone else suggested (in general term, not specifically to Kepler: http://lamb.cc/blog/2008/07/14/css-ligature-support/), should help. Just my two cents!

Thomas Phinney's picture

since you can’t turn off a single ligature without affecting the others

You can use search-and-replace to do exactly that. At least, in InDesign it works... haven't tried it in QXP. Admittedly, this needs to be done after setting all the text, but it's not difficult.

I'm not saying this to argue against having some way to turn off specific ligatures as a feature: I think it would be useful. But at least there is a work-around.

Also FWIW you'll find this as a standard ligature in a whole bunch of Adobe typefaces, including mine.

Cheers,

T

Florian Hardwig's picture

Thomas said “Admittedly, this needs to be done after setting all the text

In InDesign CS4, this actually can be done live, on-the-fly, thanks to GREP styles: Create a character style that disallows ligatures. In your paragraph style options, set up a new GREP style and let it automatically apply the character style to every instance of ‘Th'.
F

ncaleffi's picture

"Also FWIW you’ll find this as a standard ligature in a whole bunch of Adobe typefaces, including mine".

Yep - and here's how some Adobe fonts look, all of them set at 6 points:

Ok, the quality of the image is bad - but I still think that the text reads better with this ligature disabled, as in the Hoefler Text, Cycles and Whitman examples. Though correcting the "Th" ligature is possibile with the search function, it would be better to have it as a separate choice. I'm on Indesgin CS3, by the way.

On the same level, this leads me to think that it would be useful to have fonts with particular glyph options; I know some typefaces have it. For example, after setting a book in Hoefler Text, the author asked me if the Q cap italic could be changed with a more conventional one - an alternate glyph could have helped, in that situation.

John Hudson's picture

Thanks for the comparison image, Nicola. It is interesting that the Th ligature seems to cause less of a legibility problem in the types with smaller x-heights.

Thomas Phinney's picture

Two comments:

1) Most of those Adobe fonts have optical size variants, so you should be using the appropriate one if you're setting text at 6-8 point size. I question the validity of critiquing their legibility at such small sizes, when variants optimized for those sizes are also available.

2) Aside from that, I don't see a legibility problem specific to the Th ligature in the samples shown. Indeed, I prefer it to the spacing created by the Th combination in the non-ligated examples. I think this is a matter of taste, though.

Cheers,

T

kentlew's picture

I suppose it is a matter of taste. I find the Th ligature still unfamiliar and distracting. I, too, wish that it were activated in a set separate from the standard ligatures -- individually "turnoffable."

For whatever reason, I see all of the examples above as starting with some sort of backward Gamma.

charles ellertson's picture

One thing to consider - I don't kern letter pairs when I'm going to replace them with a ligature that will be included in the liga feature. Maybe Adobe does, I don't know. But if you don't use the ligature, Th is a pair that frequently needs kerning.

Florian Hardwig's picture

Charles: They do (I checked Arno Pro Caption).

Thomas Phinney's picture

Charles wrote: I don’t kern letter pairs when I’m going to replace them with a ligature that will be included in the liga feature.

Ah, but Adobe doesn't kern pairs at all any more, except when they need special treatment. They kern classes instead. So in a typical typeface the bhkl will all be in the same class for kerning, and will get the same kerning against the T (if any). To un-kern the Th combo would actually *increase* the size of the font for no gain. Thus Florian's observation.

Cheers,

T

charles ellertson's picture

Ah, but I ignore that class kerning -- If I'm reworking an Adobe font, I either strip out the class kerning entirely if I have to, or use old-fashion pairs kerning as "exceptions." At least for the base characters.

b h k l may all have equal sidebearings -- or equal optical side spacing, but I don't think they should necessarily be kerned the same. I remember when I squawked about the Adobe using kerning values of "2". I think there is more than that needed with kerning between letters involving b, h, k, l etc.

YMMV

Thomas Phinney's picture

I remember when I squawked about the Adobe using kerning values of “2”

That would be an artifact of interpolating kerning values between different masters in MM space. Probably at least one of those masters had a significant kern, but some of the middle values might not. IIRC, Adobe filters values less than 5 from the exceptions, but it doesn't save any space to filter them from the class kerns, so I'm not sure if they do.

T

Christopher Slye's picture

IIRC, Adobe filters values less than 5 from the exceptions, but it doesn’t save any space to filter them from the class kerns, so I’m not sure if they do.

Our kern filtering policies have varied quite a lot over the (OpenType) years. Probably some (or most) early OT fonts were not filtered at all, but eventually we did apply some post-processing to remove kerns below some value like +/-2. (The kerns would be commented-out of the input feature code.) Probably in some cases that value got up to 5, but I think 2 was regarded as pretty safe. I'm actually not sure what has been done in recent fonts like Arno. Anyway, I recall that the filter was applied to all kerns -- class pairs and otherwise -- for quite a while, whether or not it actually resulted in size reduction. It was just a standard production step.

Renaissance Man's picture

Although this thread never did go far enough for me, "[Bad] Kerning in Kepler" is yet another example of Unprofessional Fonts from Professional Designers. Fonts should work out of the box, and not have to be modified by end users.

While I'm at it, Kent Lew's Whitman is the most comprehensive, flaw-free text font I own (out of about 1900 fonts). It almost pisses me off, because it's my go-to font when I want legibility and readability. Which is to say that I wish I had more choices as good as Whitman.

Syndicate content Syndicate content