What is good kerning?

[Ngoc Ngo]

My two cents...I consider "the activity that a type
designer does in a font" as Hrant has put it, as adjusting the sidebearings since this is performed by the type designer.

[hrant]

But we also do kerning after/during -what I call- "spacing",
which is not just adjusting, but determining, sidebearings.

hhp

[imageWIS]

Is the 'Siena Italy' picture from the new James Bond movie...?

Jon.

[david h]

> I just came out of the new Bond movie

How did you take the screenshots -- a digital camera? video?

[John Hudson]

Hrant: But we also do kerning after/during -what I call- “spacing”, which is not just adjusting, but determining, sidebearings.

Spacing, following the typical font model that we've inherited from previous technologies, is a two-stage process involving, first, the default spacing of individual glyphs for arbitrary glyph sequences (sidebearing setting) and, second, the spacing of pairs of letters for specific glyph sequences* (kerning).

* OpenType extends this model by permitting contexts spacing that extend beyond the pair, but few non- 'complex script' fonts take advantage of this and, if I recall, the Adobe FDK and hence FontLab do not support this yet. I use it in Arabic fonts.

[beejay]

the location titles for Quantum of Solace were done by Tomato.

you can see them posted on Tomato's site

but I think these images came from the Tomato flickr.

[asen Tsvyatkov]

if those were done by tomato, i wouldn't doubt for a minute that the decision is deliberate and aesthetically considered.

[redpushbuttons]

The more I read into this the more uncertain I get.
Why do I keep thinking there must be a science to this?

I get fixated on numbers and have a compulsive urge to calculate stuff knowing this won't actually result in decent kerning.

I seem to trust number more than my eyes.

[charles_e]

Why do I keep thinking there must be a science to this?

It's been 45 years now, but you remind me of Rudolf Carnap. There was a branch of logical positivism that believed, essentially, all science could be reduced to physics. Psychology, for example, could be expressed by laws of physics if only we knew enough (statistical "laws" allowed).

It's the same sort of thing. Maybe kerning could be reduced to programmable numbers, if we only knew enough.

My thinking about Carnap & his followers was always, "So What?" Carpenters who are not physicists will still build houses. The theory is interesting, just not to those who have to do the work; the underpinnings of theory are not always useful.

[John Hudson]

Maybe kerning could be reduced to programmable numbers, if we only knew enough.

The tricky aspect of automating kerning is determining how much knowledge is enough, given that the amount of knowledge required is likely to vary from design to design. A few years ago, at dinner with some of the clever folk from URW, I outlined how I thought an automatic kerning tool should work: it would accept iterative 'knowledge' from the designer, starting from a set of spacing for key glyphs that are determined by the designer not to need kerning. Then the program runs and presents the designer with a proposed set of kerning, at which point the designer can look at individual pairs and say 'No, this kind of combination should be more like this...'; then that adjustment becomes additional knowledge in the next pass of the program. In other words, the program doesn't seek to replace the informed eye of the designer, but to capture what the designer thinks is appropriate to the design and apply it across the glyph set. To my knowledge, no one has built this yet.

[charles_e]

John, the other part of programmatic kerning is it is between pairs, and the notion that one value between a pair is correct won't fly. As mentioned above -- probably by you -- programmatic contextual kerning is what's really needed. As one move from preventing disasters to aesthetics, ever more contextual kerning is needed...

And the right value changes, depending on the size, esp. for fonts where there is only one master size.

Etc.

Or, just space the font widely. Seems to be a modern trend. That covers a multitude of small sins.

[Nick Shinn]

You might just as well say there is an ideal way to play a classical music composition.

[charles_e]

Very well put, Nick.

[hrant]

Ideal, shmideal - perfect, shmerfect. You do it because it helps.

hhp

[John Hudson]

The fact that there are conditions in which the desired spacing between a pair of letters might differ from that in other conditions is beside the point. That is already the case, and yet we still find it worthwhile to put kerning in fonts. What I described is what seems to me the best approach to automating aspects of what currently requires considerable man-hours while not obviating all the experience and knowledge that a good kerner brings to the job. I know, when I am kerning, that the decisions I make for some pairs have direct implications for other pairs; ergo, it makes sense that if a program can capture those implications and apply them, it can save a lot of time.

Of course, all this still presumes the current, inherited sidebearing/kerning model.

[dberlowgone]

"...there is an ideal way to play a classical music composition."

For a given performance, per listener, there is. The issue besides taste with kerning, is that glyphs and spacing don't scale linearly. "Modern" font formats hypocritically offer typefaces as scaleable when all are not, and the situation, once thought to be temporary, is now permanent enough to regularly circulate thru here like bad air.

The first example above goes to my first point, I'm fine! said one viewer, with "LONDON". The other example, in the direction of my second point, is so hideously selected and composed, you'd want to be fixing the way these characters were selected, and then how each scaled, before fixing the kerning... I think.

[Renaissance Man]

No mention of Igino Marini's iKern?
http://ikern.com/k1/

[hrant]

Some people use iKern because they hate spacing (and don't want to spend money outsourcing it). But I'm surprised that there are some big, talented names among iKern users too. I guess it saves times when you have very large character sets, since the kerning grows exponentially. But personally I would never trust it for "base spacing" (sidebearings).

hhp

[Renaissance Man]

It has been suggested that if you really hate someone, teach them to recognize bad kerning. That way, they'll be forever doomed to recognize examples of typographic ignorance and carelessness that mere mortals cannot see.... In no time at all, your friend/victim will be so well-versed in kerning that they may join the ranks of people who cannot order lunch from a poorly kerned menu.

But how do you ensnare an unsuspecting victim? One way would be to show them KernType, a kerning game. The site offers interactive kerning practice where you drag letters within words to distribute the given amount of space.
http://www.creativepro.com/article/learn-kern-kern-type

TEST HERE: http://type.method.ac/#

[Nick Shinn]

https://typography.guru/forums/topic/96189-forwarding

[quadibloc]

@Nick Shinn:
You might just as well say there is an ideal way to play a classical music composition.

While this is an important point, and it is applicable to font design in general and kerning in particular too, I'm not so sure how applicable it is in the current situation.

There is too much bad kerning out there, and there are hardly any unique creative approaches to kerning. Type design and typography are arts as well as crafts, but they do lie on a different point in this continuum than musical performance.

[Nick Shinn]

All analogies break down under close comparison, because they’re analogies.
I wasn’t suggesting a close correspondence across the board, merely one common principle!

The classical score is fixed, just as letter shapes are.
Notes are like letters, spaced in time.
In that respect, performance corresponds to setting type.
That’s all.

One doesn’t expect the music maestro to imitate a player piano, so why should the type maestro use default “Metrics” or InDesign’s “Optical” spacing for display when manual kerning is available?

There is even the option of custom kerning a font in Quark XPress.

…there are hardly any unique creative approaches to kerning.

Manual kerning is always unique to the individual doing it.
It doesn’t have to follow a defined methodology, it’s more a matter of individual taste and expression.

(However, as an exception to the rule, I have managed to employ one completely brain-dead method, in Softmachine, in which the closest proximity between every glyph combination is equalized.)

[PabloImpallari]

The tricky aspect of automating kerning is determining how much knowledge is enough, given that the amount of knowledge required is likely to vary from design to design. A few years ago, at dinner with some of the clever folk from URW, I outlined how I thought an automatic kerning tool should work: it would accept iterative 'knowledge' from the designer, starting from a set of spacing for key glyphs that are determined by the designer not to need kerning. Then the program runs and presents the designer with a proposed set of kerning, at which point the designer can look at individual pairs and say 'No, this kind of combination should be more like this...'; then that adjustment becomes additional knowledge in the next pass of the program. In other words, the program doesn't seek to replace the informed eye of the designer, but to capture what the designer thinks is appropriate to the design and apply it across the glyph set.

To my knowledge, no one has built this yet.

I have played around with a similar idea ( knowledge → suggestion → refinements ) for a Spacing Macro.
It's not a finished product yet... it's more like a "proof of concept" test (It works only on the lowercase glyphs so far)... but the results are quite decent, and far better than the FL autospacing.
I think it has potential if someone else what to improve it... as it is 100% customizable.

Screencast on how it is supposed to be used: https://vimeo.com/45648496
Download Macro and play around: http://www.impallari.com/projects/overview/spacing-macro