Kerning and Nick Shinn's mashed potatoes preference

J_Tillman

RE:http://typedrawers.com/discussion/2198/kerning-the-space-to-glyph-and-glyph-to-spaces#latest
Nick Shinn said:
"In general, I don’t like types that are too smoothly spaced. Mashed potatoes, sure, but not creamed."

Goodness gracious! Do you mean this for text fonts? Are you saying that some kerning "mistakes" make it better or easier to read? This is new to me. If you have some more thoughts, or some examples in one direction or the other, I would appreciate hearing about it.

Even thought I put this in the technique and theory category, what I am interested in is finding text fonts good to read for extensive reading.

Hrant Հրանդ Փափազեան Papazian

Information comes only from contrast. Even texture does have a role, but is over-rated. The balance is elusive.

Wolf Böse

mhm yea, i was curious what he meant by that.

Nick Shinn

Are you saying that some kerning "mistakes" make it better or easier to read?

No. I’m saying that type needs to breathe, and fonts in which gaps are closed up—which is the majority of kerning—have the air sucked out of them.

Also, consider the default (above) of Helvetica, compared with InDesign’s “Optical” kerning (below), in which the t-o is closed up, and the r-y is opened. Such is the tacit “kerning mental algorithm” designers employ to even things out, but IMHO it contradicts the inherent balance that the original type design gives to such words. After all, these ur-grot letter forms evolved to their neutral perfection in a kern-free environment.

Bhikkhu Pesala

The Autokern feature in FontCreator will remove space between t-o, while adding space between r-y. My instinct is to kern too tightly, probably because I am looking at pairs in isolation and at a large size on screen. I did not use to add positive kerning pairs, or lowercase-lowercase kerning pairs, but at least some are needed. Lowercase ry in my Verajja bold font is a good example of a pair that benefits from more air. The pair t-o is perhaps OK as it is. 

Below are the unkerned and kerned pairs that FontCreator suggests. The tightness of auto-kerning can be adjusted, and the results can be change manually wherever necessary. 

Kent Lew

I have often contemplated this matter.

In my opinion, one of the pinnacles of text production was the era of American book publishing in the 1940s and ’50s — publishers like Alfred A. Knopf, Random House, et al., and great composing houses like Plimpton, Kingsport, and MacKenzie & Harris. These texts were nearly all composed on either Linotype or Monotype machines, both of which boasted libraries of now-classic book faces.

But neither of these machine casters employed pair kerning as we think of it today. The Linotype could not accommodate any kerns whatsoever. Although the Monotype could accommodate kerns in the traditional sense of the word (overhangs extending beyond the body), this is not the same as the kind of individual pair kerning that we think of today.

And yet, despite the occasional usual suspect that might catch the eye (To, We, Va, etc.), these unkerned settings remain a high standard of pleasant readability that I think we are still hard pressed to equal today.

Of course, one can’t completely separate the typesetting from the other variables that contribute to making these books what they are; but still, focusing on the type, it makes one think.

I can’t help feeling that, these days, especially among newbies, the role of kerning can tend to get overemphasized at the expense of other fundamentals.

To adopt a different culinary metaphor, pair kerning is like seasoning. A truly great recipe will often need only a dash of salt and maybe a pinch of pepper. And it doesn’t necessarily improve by throwing a bunch of extra herbs and spices at it.

John Savard

Nick Shinn said:

No. I’m saying that type needs to breathe, and fonts in which gaps are closed up—which is the majority of kerning—have the air sucked out of them.

Well, that does change matters. While it might not be clear if it is possible for a typeface to be kerned too well, it certainly is possible for one to be spaced too tightly.

And this is true even if the history of the hot metal era might indicate that the importance of good kerning is overrated. If people hardly ever noticed no kerning, then clearly perfect kerning, even if we don't need it, isn't going to drive people crazy: they won't notice it either.

Nick Shinn

The tightly fitted type I put in ads in the 1980s hurts my eyes now.
I don’t think it’s a result of my advanced years, though!
At the time, that was the look, and nobody—clients nor readers—objected.

Ori Ben-Dor

@Nick Shinn

I've used Word 365 on Windows 10 to set "story" in Helvetica, first with kerning turned off (above) and then with kerning turned on (below):


As far as I know, Word doesn't apply its own kerning, so it seems the default, original kerning of Helvetica agrees with the optical kerning of InDesign.

Either way, in both our examples, the looser "ry" (below) looks much better to me. I do agree, however, that "to" is too tight and would be better off without kerning.

Nick Shinn

Ori, Helvetica has no original kerning.

Ori Ben-Dor

Nick Shinn said:

Ori, Helvetica has no original kerning.

Are you sure? I mean, even back then some pairs, such as "AV", really needed kerning, and it's not as if it was impossible to implement.

Anyhow, my experiment at least shows it's not just the optical kerning mechanism of InDesign, but also whoever digitized Helvetica for Linotype.

Chris Lozos

While I agree that the metal typesetting of well printed books made for agreeable reading and it was the best that could be done with the technology of that time, I don't think it should be today's standard. Prior to printing types, texts were hand written and kerned by eye as the were written.  The more skilled the writer, the better it looked.  When metal type came along, this could not be done and I am sure there must have been some grumbling about it.  This did not stop the process of improvement of type nor should it have.  When photo typesetting came along, there were tracking sets and kerning.  This was pushed beyond the good during the bashing and crashing 80s and it was heralded by many as great stuff.  Still, there was grumbling [me among them]. Again, progress did not stop, thankfully.  Today, we can space and kern in any way we wish, both as type designers and as type users.  With this freedom, some good work comes out but so does some shit--just like always.  Fashions of type style and spacing comes and goes but we never stop evolving both what we design and what we accept as good work.  My guess is that the human mind and perception is very good at adapting to these changes and it always will be.  The tools are now there and are ever improving.  Peoples abilities to use well or poorly will always endure as well.  It is neither the current tools or fashionable preferences which are the problem as long as the skill of the user is up to the task. The "Good old days" had their shit work, too.  It just does not survive well or get retained for long.  At some point in the future, our work will be, in some cases put on a pedestal, and in others cast to the trash heap. 

Go ahead, kern away or not, just do it well.  The trouble is do it well for whom? Your current generation or those that might come later?

Hrant Հրանդ Փափազեան Papazian

I haven't set Helvetica by hand (thank the gods :-) but most of the fonts I have set by hand (including the two sizes of Pascal that I own) do feature some kerning... in the traditional sense of bits that hang over the rectangle.

Kent Lew

do feature some kerning... in the traditional sense of bits that hang over the rectangle.

Exactly — These are equivalent to negative sidebearings. But not pair kerning, in the sense of specifically targeted adjustments.

John Savard

Kent Lew said:

do feature some kerning... in the traditional sense of bits that hang over the rectangle.

Exactly — These are equivalent to negative sidebearings. But not pair kerning, in the sense of specifically targeted adjustments.

Of course, though, kerns "in the traditional sense" have one thing in common to targeted adjustments versus a negative sidebearing. If a kerned letter like an f  with an overhang is followed by a letter like another f, or an l, or an i, instead of a letter like a or o, either one has to follow it with a space so that the kern has something to rest on, rather than colliding with the succeeding letter and being damaged, or one has to use a ligature instead.

The idea came from metal type, after all. Applying kerning to letter combinations like Ya, Yo, and so on, while much more common in phototypesetting, was also on occasion done with metal and wood type, at least in display sizes, by physically cutting the type slugs.

It is precisely because kerning was considered so valuable that it was even done in metal type, despite the difficulty of doing so in metal type, that it is generally considered to be very worthwhile to do in digital type, now that there are no longer practical considerations preventing it. If kerning hadn't been invented until digital type came along, then the suggestion that it isn't really worth bothering with would perhaps meet with a warmer reception.

But, as noted, using the technique of kerning to facilitate setting type in as tight a fashion as attainable, while it may have some validity for attention-getting advertising typography in large display sizes, is indeed something in the deprecation of which for general text typesetting I am happy to join.

The fact that kerning can be abused doesn't mean that kerning isn't a good thing.

Chris Lozos

while it may have some validity for attention-getting advertising typography in large display sizes, is indeed something in the deprecation of which for general text typesetting I am happy to join.

This is a crucial point.  There is a vast difference between adverting or display  typography and running text.  The small quantities of words in advertising allow for deliberate individual kerning and tracking to directly affect that bit of text.  In long text, we cannot and need not put so much labor into it as type users. Running text should be very close to out of the box use in most cases.  Display text almost always needs more adjustment.

Kent Lew

If anybody read my comments as saying that kerning is not a good idea and shouldn’t be done, then my point was misinterpreted. (And they must not know me very well.)

I was echoing what I understood to be the original point of Nick’s mashed vs. creamed potatoes statement: that maybe everything doesn’t need to be pair-kerned with everything else, to the nth degree of “smoothness,” to achieve maximum readability (if such a thing even exists objectively) and that such a goal may even be somewhat foolhardy.

Thomas Phinney

Did FontCreator kern the "or" pair? It appeared so to me in the sample.

Bhikkhu Pesala

Thomas Phinney said:

Did FontCreator kern the "or" pair? It appeared so to me in the sample.

It would have done if I had set up a pair, but IIRC I only added pairs for to and ry. (A careful comparison of the two images confirms that o and r are not kerned). 

I used a glyph spacing factor of 23. If I use one of 25, the t-o pair has negligible positive kerning. With glyph spacing of 23:

* The o-r pair is kerned by -55
* The t-o pair is kerned by -35
* The r-y pair is kerning by +102

Adjusting the glyph spacing factor will tighten or loosen the kerning. The default is 27, but the optimal value varies for different fonts. 

Thomas Phinney

I won't make any claims for anybody else's automatic spacing, but FontCreator seems to be making the spacing less consistent.

Bhikkhu Pesala

Thomas Phinney said:

I won't make any claims for anybody else's automatic spacing, but FontCreator seems to be making the spacing less consistent.

Why do you say that?