"best" Spacing- and Kerning-Methods?

Sebastian Nagel's picture

Dear Members,

I am currently working on my third typeface Terra Nova, but it's the first one to leave the ambitious draft phase and enter the serious precision work phase.

Until now, when testing the face, i relied on Indesign's "Optical Kerning-Feature", that did the job passable for now. The further i get with the font, the more necessary it gets to think about spacing and kerning, but actually I have no experience yet, so i just don't know how to do it right, and fiddling without a plan does not lead to anything...

Is there a well-tried and documented method to do the font-metrics for a serifed font (in fontlab 4.6; targeted font format is opentype)? What about kerning, where do you start, where does it end? What is it about class-based kerning? You see: I am an absolute noob on this...

Thanks for any direct answers, links, etc.
sebilar

Sebastian Nagel's picture

of course, the link does not work... :(
next try: Terra Nova

Miguel Sousa's picture

Sebilar, have you tried to read this?
http://typophile.com/node/15794
It might help with the spacing.

ebensorkin's picture

I have not used this tool in a delibertae way yet but when I get to the point where I am making a font like your I expect I will.

http://www.typotheque.com/type_utilities/lettermeter/

" Inspired by Ligature Counter, LetterMeter is a text analysis tool, used in the Type&Media classes (postgraduate course of type design at the Royal Academy of Art in The Hague). LetterMeter is designed for comparing multilingual texts and measuring the frequency of appearance of glyphs."

It also show glyph pairs.

cool no?

Even cooler that you can have it today & for free. Hooray for Peter Bilak & Just van Rossum!

Sebastian Nagel's picture

I was actually "remembered" of my problem by reading your thread, mikesousa, but i couldn't find any other methods yet and I wanted to do some recherche about common techniques before starting (this doesn't imply that your method is not good). By the way: congratulations for your Calouste-typeface and especially to your presentation-pdf, which is very inspiring to me.

I will also have a look at Peter Bilaks lettermeter, which seems to be quite useful even on first sight. Edit: argh... osX only :(

Thanks for your tips!

Randy's picture

walter tracy's book: letters of credit, may be helpful. He breaks spacing down into a step by step.

sim's picture

I've read this book, and it's really helpfull especially for the spacing.

Sebastian Nagel's picture

Thanks for this proposition, the book is already ordered :)
Are there any other books about typedesign (spacing or letterforms) that can be recommended?

I am waiting for Karen Cheng's Designing Type, but it won't be out here in Germany until June 2006, and I don't even know what's written in it, but it's a must-have for me already because of the publishing house that is doing the german edition produces only (expensive...) high-quality-books.

I already have:
- Albert Kapr, Schriftkunst
- Max Caflisch, Schriftanalysen
- Gerrit Noordzij, Letterletter

Thanks
Sebilar

hrant's picture

http://typophile.com/wiki/triumvirate

BTW, who knows what about that Cheng book?
A translation of that German page, somebody?

hhp

Sebastian Nagel's picture

Sorry about posting a german link here without translation.

There is much "bla" in the description. Short summary:
Karen Cheng has/is writing this book, looking at each letter and its specifics and showing examples. Not only the main characters a-z A-Z are described, but also numerals and special characters. Then it says you can lern to create your own type, or just train the eye to recognize and select high quality typefaces.

As far as I have found out, the book is translated from english, as Karen Cheng is Professor at an university I can't remember (her Homepage isn't working at the moment, but it did a few days ago, so it's a temporary problem i guess).

You can also find an english version of the book (with different cover-design) at amazon.co.uk.
It's not available here, too (may 06), and I am wating for the german edition for the reason of mother tongue and for knowing that the edition made by Hermann Schmidt (publisher) will be high quality (I love books :)

hrant's picture

I just emailed her for some insight.

hhp

hrant's picture

I got a reply from Karen Cheng. She was kind enough to
send me a summary as well as a PDF of sample spreads.

An excerpt from the summary:
"This book provides a detailed guide to developing a typeface and is suitable for readers at all levels, from the novice to the serious professional. Issues of structure, optical compensation and legibility are clearly illustrated with numerous diagrams and type specimens. The book describes the process for developing both serif and sans-serif typefaces with precise detail. Special emphasis is given to the systematic relationships between letters and shapes in a font."

Just based on that paragraph, one might be tempted to think this book is the missing Holy Grail. But since Cheng is not an accomplished type designer* it can't be that. However, what she does seem to be doing is going quite deep into the time-honored practice of analysing and comparing existing fonts, and this is both rare and valuable**. The spreads reveal some very well thought out comparisons that discuss many things of concern, including: relative widths; vertical proportions; serif structures; the B/P/R relationships; modularity; spacing; and even diacritic placement. And at over 200 pages, that makes the whole pretty promising I think.

* As far as I can tell. Unless she does it for the CIA or something.

** The mention of Young's "Fonts & Logos" in the summary is telling, since that book also enagages in a fair amount of comparison.

Since looks can be decieving, I might regret saying the following, but: this book is a Strong Buy recommendation, especially for beginners (if also probably interesting to most of the more advanced designers). In fact -once I get my hands on a copy- I think I might recommend it as a textbook for my type design class at ArtCenter.

hhp

dezcom's picture

Hrant,
Is it available now?

ChrisL

david h's picture

April 06 - Yale Press; May 06 - Amazon

> But since Cheng is not an accomplished type designer* it can’t be that

why not? but she's graphic designer

hrant's picture

For a painful illustration of what happens when a graphic designer (who is
not also a type designer) tries to make an actual "finished" font, check out
the fiasco-landscape in this recent book: http://www.rotovision.com/description.asp?bookid=2487#

Fortuntely for her book, Cheng seems to be leveraging her perceptive and analytical abilities without trying to detail the design and delivery of a fully prime-time type design. But again, I have yet to really check it out.

hhp

david h's picture

So not to buy the book?

Sebastian Nagel's picture

Although the thread is off-topic now, I think it was worth sacrifying it. Sounds promising. Thanks for asking her about details, hrant

antiphrasis's picture

Hrant,

How'd you like the Type Specific book? Worth buying?

hrant's picture

I thought some of it (like the opener: the killer Houston Chronicle face) was good, but most of it was just depressing: design firms resisting the common sense of hiring a type designer, ending up either with geometric constructed suicide pills or decon hooligan backwash (often not even a font but an illustrator file with a bunch of outlines), but not even seeing a problem, and on top of it getting published as real, good work. But I guess there's a silver lining on the cloud: the demand is there, and there's a lot of room for education, hence potential.

So the book is totally worth looking at, as a reality check
for us; but the money could certainly be better spent.

hhp

antiphrasis's picture

Hrant,

Thanks for the mini-review! So the book might then be a bad buy if you want to learn type design...you'd pick up bad habits. I've got my Cabarga's, Bringhurst, and some other books on typography (even an old Aaron Burns book).

dezcom's picture

I just finished "Type Specific" as well. Mostly, it is a show book and very little of a tell book. The "show" part is in the sense of typical award annuals, you get a quick flash of something but not much meat. There is minimal description of the problem, solution and method.
The main thing that bothered me was the presentation. The design/layout of the book overpowers the typefaces they are showing. This is one of those "it's all about me" designs. The type designs themselves are nondescript and pretty uninventive--they look like off-the-shelf faces. I wonder why someone would pay for a custom font when it was not customized for them?

ChrisL

david h's picture

> but the money could certainly be better spent

yeh....like stuff from Plantin-Moretus Museum

Sebastian Nagel's picture

Now I've had the time to read "letters of credit" and do some spacing-experiments on my font, and I've come to a problem where I need some advice:

I have spaced uppercase (HOHOHO etc.) and lowercase letters (nonono etc.), respective to the inner room of H, O, n and o. When setting uppercase, everything is fine, same is when setting lowercase.

The problem occurs when mixing them: The spacing of the uppercase obviously is wider, so words like "Terra Nova" look more like "T erra N ova". This happens with most uppercase letters, not only with T, V, W etc.
Is this an error by design (H too wide --> spacing too loose)?
Is it an error of spacing (uppercase spacing chosen too loose)?
Or does it need adjustment between upper- and lowercase via kerning?

Thanks for your help
sebilar

Mark Simonson's picture

Caps usually do look better set looser, but fonts are not usually spaced that way for the reason you discovered.

If you want to retain that feature, I can think of a couple of options.

1. Shift all the caps to the right, or shift all the lowercase to the left, or a little of both. The downside of this is when you have a lowercase letter followed by an uppercase letter the spacing will be even looser. Fortunately, this doesn't come up in normal words, mainly just hi-tech brand names (iPod, FontLab, etc.). People who need to set stuff like that could simply do manual kerning in such cases.

2. If you are building an OpenType font, you could space the caps tighter for use with the lowercase and use the 'cpsp' feature to specify looser spacing for all-caps settings. This would only work in OT-aware apps, but is probably the smarter way to do it.

dezcom's picture

"1. Shift all the caps to the right,"

That is my solution as well. Determine your sidebearings for cap to cap spacing, then shift the caps to the right using the same width (meaning the left sidebearing will increase and the right sidebearing will decrease) so that the right sidebearing of caps are now correct for positioning next to lower case.

ChrisL

Nick Shinn's picture

If typographers want all-cap spacing, they should do it themselves, with tracking.

Related:
Having an alternate version of a font with different sidebearings and kerning, for use at different letterspacing (tracking), is an idea which I put into Nicholas, the tightly-fitted display version of Goodchild.

Sebastian Nagel's picture

Thanks for your comments. So this is a general problem, not one specific to my font. I was just wondering about, that Walter Tracy doesn't mention this problem (and my results are the consequence of his method, i did nothing else yet than follow his step-by-step instructions).

Terra Nova is a opentype-Font, so cpsp is defintly an option and I think I will narrow caps-spacing for standard-settings so it looks good in normal text as well as in words like iPod.

Just thinking about the other method: When shifting the Caps to the right, isn't there the problem that there will be holes in the left side of text-blocks, like:

 This ist just an
example what happens to
 Terra Nova when done
this way.

so caps will look indented?

Mark Simonson's picture

When shifting the Caps to the right, isn’t there the problem that there will be holes in the left side of text-blocks....

Yes, if the caps are generously spaced.

Most type designers do as Nick suggested and leave it to the user to space the caps correctly, trusting they do such things. The cpsp feature is the best general solution because it lets you control the caps spacing yourself without doing weird things with the sidebearings which, I admit, is sort of a hack.

hrant's picture

> If typographers want all-cap spacing, they
> should do it themselves, with tracking.

?
Are you sure you mean it's OK to produce fonts with lousy all-caps setting?

Yes, making all-caps setting water-tight is a bear, but just dumping the work on the user is no good: it makes them pay less for your work; and it can make your fonts look bad (since many users either don't care or don't know).

hhp

crossgrove's picture

It's now possible with OpenType to provide features to make all-cap settings look good without wrecking normal text. I imagine a domino effect if the caps are shifted; then what happens to lines of text with lining figs, or spacing on punctuation, caps at beginnings of lines, or..... ?

A book face will look tight when set all-caps, because the primary concern in spacing is correct fit in text. If the typeface is intended for display, maybe a huge amount of kerning is necessary to get the desired syrupy effect. But for a book face, don't throw everything else out of kilter just to get loose all-cap settings. How often do you need them when setting text?

Again, OpenType obviates this. You can have it all. Don't use hacks, use cpsp.

Nick Shinn's picture

>Are you sure you mean it’s OK to produce fonts with lousy all-caps setting?

I would have thought you'd have figured, from my mention of tracking, and the direction of the thread, that I was referring to an optimal all-cap setting that is slightly more tracked out, or letterspaced, than what happens when caps are optimized for use with lower case.

Another problem with adding extra sidebearing space to the left is that it will throw off left column alignment.

Actually, a legitimate plan to space out a default all caps setting would be to add extra kerning to every cap-cap kern pair, including figures. That wouldn't affect any other typographic parameters.

However, I don't think it's the type designer's job to make this kind of decision for the end user. Therefore I'm not a fan of the OpenType "cpsp".

hrant's picture

> a legitimate plan to space out a default all caps setting
> would be to add extra kerning to every cap-cap kern pair

An idea I've presented a few times before. Or you could couple such an approach to the offseted-caps idea: make comprehensive space-UC negative pairs to "put things back" so to speak.

> I don’t think it’s the type designer’s job to
> make this kind of decision for the end user.

The thing is, quite often a "decision" is never made!
And it's simply beneficial to make the font work well
as often as possible (although admittedly never always)
sometimes in spite of the user!

That said, any time saved on a given design does
mean there's more time to work on other stuff...

hhp

dezcom's picture

I think what Nick means is that he is leaving the option to the end user, not making a timesaving for himself. Some users might like their caps tight and some even looser than the type designer might. Creating a style sheet in InD for all caps settings is a piece of cake. I always create one for small caps as well because I usually like it looser than it comes out of the box.

ChrisL

hrant's picture

> he is leaving the option to the end user

Yes, but not of tight/loose, but of good/bad.

hhp

dezcom's picture

"Yes, but not of tight/loose, but of good/bad."

I think you missunderstood him. The only thing tracking can do is adjust tight/loose.

ChrisL

hrant's picture

It's possible I misundertood. But then so did Mark; except Nick didn't correct him.

hhp

Thomas Phinney's picture

I should add that Adobe puts the "cap spacing" feature into all our OpenType fonts. However, the cap spacing is pretty mild.

You can test this easily enough in InDesign: take some text, typed in all caps. Dupe the line. *Format* the second line in all caps. You'll see the spacing get looser, dependent on the setting in the font.

T

dezcom's picture

"...*Format* the second line in all caps. You’ll see the spacing get looser, dependent on the setting in the font."

Thomas, are you saying that if the original typing is done in all caps, the spacing will be tighter than if you apply the "all caps" formatting command in InD? If so, why?

ChrisL

John Hudson's picture

Tracey was talking about how to space letters, not how to space a font. What he has to say about the spacing of uppercase-to-uppercase is primarily of interest to typesetters, rather than font developers, because the latter have to space the uppercase letters primarily to work with lowercase.

I strongly advise against the idea of spacing the caps relative to each other and then offsetting them to the right to space appropriately with the lowercase. The will produce THIS sort of mess, where the space to the left of caps is larger than the space to the right. Of course, this also results in giant gaps at the beginning of words.

Proper cap-to-cap spacing is primarily the duty of the typesetter, hence the existence of tracking. Caps in fonts should be optically evenly spaced and fitted to the lowercase. The feature can be used to increase the spacing between caps, but I generally find it best to use this only to uniformly loosen the setting slightly -- i.e. a kind of in-built tracking --: it is too blunt an instrument to produce optimal spacing of all-caps.

John Hudson's picture

Thomas, are you saying that if the original typing is done in all caps, the spacing will be tighter than if you apply the “all caps” formatting command in InD? If so, why?

Because the all caps formatting function calls the layout feature, but simply typing in all caps does not. The layout feature description notes that the feature could be called heuristically, though, so it is possible that an application could invoke the feature contextually whenever two uppercase characters occur next to each other.

hrant's picture

> The will produce THIS sort of mess

Missing link? Or did I miss something?

> this also results in giant gaps at the beginning of words.

(You must mean capitalized words.)
Unless you add comprehensive space-UC negative kerning pairs...

Which isn't to say that I feel comfortable pushing caps to the right.

hhp

John Hudson's picture

Sorry, Hrant, Typophile is being clever and removing double spaces. Also, I meant to write 'this also results in giant gaps at the beginning of sentences. Here is an image that shows the kind of word spacing problems I was trying to describe:

Note that among the other problems, you also throw off margin alignment, and no kerning is going to solve that for you.

k.l.'s picture

Capital spacing is something which troubles me since some time, and I am happy to find many of my nightmares reflected in this thread! Some thoughts on it.

(a) I do not trust designers -- any typographers left? -- and so think that the job needs to be done by type designers.

(b) cpsp indeed requires that the "All Caps" option is selected from the menu. Yet there are people who TYPE uppercase letters. (I do not want to discuss whether they should or should not do so. They do.)

(c) cpsp, as used now, merely increases the overall spacing of capitals. (Whether you add more space to one side of glyphs, or distribute additional space equally to left and right sides of glyphs, is a matter of taste. Each with its own consequences.)
In my opinion, uppercase letters need to be spaced a bit differently for uppercase context and for lowercase context.* Compared to increasing sidebearings in a linear way merely (which is what the cpsp feature normally does), A L T V W Y may require less flesh leftside/rightside for UC context.
I still like David Kindersley's model for UC context. As a side effect, UCs spaced that way don't need much kerning at all.

In Litteratra, I included a second set of uppercase letters which are spaced not only more openly but a bit differently. Normal UCs are spaced for lowercase context, a special set is spaced for uppercase context. A lookup in calt then replaces normal UCs by special UCs in uppercase context.
However, this blows up the character set unnecessarily, and requires kerning for even more glyphs -- with punctuation marks &c at least. I will not repeat this.

My hope is a bit in contextual positioning. Then one set of UCs would do, and it would be possible to add spaces INSIDE spaced words only. So, the effect which Mr Hudson illustrated would not occur.**
In a second step, the space character's advance width could be increased if UC-only or SC-only words precede and/or follow it.

Karsten

* Depends on typeface design. Rather narrow sanserif UCs do not require different spacing. Increasing spaces the linear way works fine.
** InDesign's optical margin alignment may compensate for it. But this does not help poor Word users ...

Mark Simonson's picture

Typophile is being clever and removing double spaces.

This is normal web browser behavior, nothing to do with Typophile. (If you do "view source" on this page, you will see all the spaces you typed are still there.) In HTML, if one or more spaces are typed, only one is displayed; none if it's the beginning of a line.

But, here's a little trick that works on Typophile: Type non-breaking spaces instead when you want multiple spaces or spaces at the beginning of a line to be displayed. On the Mac, use option-space. Control-shift-space on Windows (I think).

hrant's picture

> this also results in giant gaps at the beginning of sentences.

Although that's not as bad as extra gaps before capitalized words mid-sentence.

> you also throw off margin alignment

Good point.

> there are people who TYPE uppercase letters

In fact I'd say that when a person wants all-cap text he's going to bloody type all-cap text (especially with that handy caps-lock key) so it's a problem if an app requires the user to do something counter-intuitive to trigger good cap spacing.

--

BTW, I'd spell the name "Terranova". I've always thought
that Tarragona (in Catalunya) has a really great ring to it.

hhp

Sebastian Nagel's picture

> BTW, I’d spell the name “Terranova”. I’ve always thought
> that Tarragona (in Catalunya) has a really great ring to it.

This sounds interesting, but I don't get it yet (a language problem maybe; I am german-speaking). Could you please explain?

Ontopic: I've now done narrow spacing for the Caps, so they fit to lowercase. I am also trying to balance it a bit more, so it will fit smallcaps as well (it does already, except for some combinations).
And next thing is that I will have a deeper look into cpsp (this talking about linear increase of spacing does not sound good for my constellation, but I will have to look at it first.

hrant's picture

Let me try this: ¡Tarrrrrrragona!
(Maybe you had to be there.)

hhp

John Hudson's picture

cpsp support doesn't need to be linear, that's just the easiest way of approaching it if all you want to do is loosen the spacing a bit (effectively, the same as applying tracking).

The cpsp feature uses GPOS lookup type 1, which is a single glyph adjustments. The quick and easy way to implement the feature, as described above, is simply to define a glyph group (or class in FontLab parlance) and apply the same GPOS adjustment to the whole group. But there is no reason at all why you could not take the time to make individual and varied adjustments to each glyph, or to smaller groups of like glyphs.

The relationship of the cpsp and kern features is an interesting question. Do you kern caps-to-caps base on their default spacing, or based on their cpsp spacing?

k.l.'s picture

"cpsp support doesn’t need to be linear [...]"
"But there is no reason at all why you could not take the time to make individual and varied adjustments to each glyph, or to smaller groups of like glyphs."

I know. ;-)

"The relationship of the cpsp and kern features is an interesting question. Do you kern caps-to-caps base on their default spacing, or based on their cpsp spacing?"

I do not & will not use cpsp because it requires manual activation of the All Caps option on the user's side.
Still it is open which feature is the best candidate, but it should be one which uses contextual positioning and is on by default, so users get a good result without thinking about the matter. I opt for calt. In this scenario -- contextual positioning, feature on by default --, kerning should be based on the spaced version.
It is a very tricky issue.
(However, as long as cpsp is an option which requires manual selection, kerning better be based on default spacing -- which gives acceptable results even if All Caps is not activated. In this context, using cpsp the linear way indeed makes sense!)

Hello Mr Papazian. I use caps lock myself much too often. :) My point is: It is better not to expect that users activate All Caps. But then it's up to the type designer to make sure that in all-caps-setting spacing is right -- nevertheless.

Karsten

hrant's picture

Just call me hhp. Vowels: who needs 'em? They are, after all, anti-Semitic. ;-)

hhp

John Hudson's picture

I do not & will not use cpsp because it requires manual activation of the All Caps option on the user’s side.

That is true of one company's current implementation in one set of applications. I don't think you should presume that other companies will implement the feature in the same way, or even that Adobe's implementation will remain this way.

I opt for calt. In this scenario — contextual positioning, feature on by default —, kerning should be based on the spaced version.

calt is not a positioning feature, it is a substitution feature.

Syndicate content Syndicate content