x-Height Variance Only

hrant's picture

Are there any typeface systems that feature large and small
x-height cuts that don't vary in other way? I mean pairs of
fonts where the ascender, descender, cap, stroke thickness,
etc. are all the same and only the x-height is different.

hhp

blank's picture

I'm actually tempted to do this with a fat face family I'm working on, so that users can choose between classic and contemporary feels.

clauses's picture

Unger's 'Vesta' and 'Big Vesta'.

clauses's picture

Licko's 'Mrs Eaves' and 'Mrs Eaves XL'.

hrant's picture

Maybe the MyFonts renderer isn't reliable enough, but it
seems that Mrs Eaves XL has longer extenders and more
weight in the stems than the original. The caps seem to
be the same though.

Let me try to check Vesta.

hhp

hrant's picture

Just checked Vesta (comparing screengrabs of 1600% zooms of the PDFs)
and it looks like everything else -except the stroke weight*- is being varied
slightly too. Which is normally a good thing really, but not what I'm after.

* Where BTW there's an offset of one weight step between the two.

hhp

William Berkson's picture

Frutiger says that Glypha was made out of Serifa by increasing the x-height. You can read it starting on p268 of Adrian Frutiger Typefaces: The Complete Works.

Jack B. Nimblest Jr.'s picture

Hrant, be reasonable, how can xht alone change, practically in a useful design?

hrant's picture

Normally it shouldn't (as I implied by "normally a good thing" above) since most obviously the relationship between the ascenders and descenders goes out of whack (since the baseline has to stay put).

However I can think of two cases where such a setup would be a good design compromise, in effect a lesser evil than varying all the proportions:
1) Using a larger-looking cut of a typeface for emphasis, instead of Italics. You don't want to do this by increasing the point size; and since it's an embedded snippet you want to maintain as much of the horizontal alignment as possible.
2) If you're mixing writing systems (or even languages*) inline and you want to respect each script's apparent size and Cartesian-space-usage needs without making a mess of horizontal alignment. For example Armenian has the same vertical "bands" as Latin, but -for text- the "x-height" needs to be much smaller. Ideally I would do what I've done in Nour&Patria: vary everything (only maintaining the "style" and as much of the mood as possible). But it gets complicated. So I'm trying to think of a simpler method that can work 90% most of the time and can be applied more efficiently. What I'm thinking is two cuts per typeface, one with a normal x-height/"x-height" and one with an x-height/"x-height" that's too small/big. Everything else would align. What you'd be sacrificing a bit is the descenders (not a huge deal in Latin), but you'd be gaining alignment.

* Think of the differences in density/size among
Latinate languages, like Latin versus Polish.

hhp

1996type's picture

Lexicon and Trinité

Nick Shinn's picture

Palatino and Aldus.

I used to use Palatino Bold as the bold for Aldus, which has none.
But never thought to use Palatino Regular for emphasis in Aldus text!

I have also mixed ITC Garamond with other Garamonds, especially for its Light and Condensed styles, which are not available in the traditional Garamonds.

hrant's picture

Bill, it looks like more than the x-height varies there. But again, the MyFonts renderer might be too wonky to tell.

Jasper, I'd forgotten about Lexicon's (and Trinité's) cuts that vary proportions! Thanks - and shame on me. Looking closely again, it seems the x-height stays the same and the extenders change; which could work if point size changes to equalize the vertical span would make a pair of nearby weights click... The Lexicon* PDFs are 600dpi, but do include some sufficiently large settings**, so I checked: it gets very close, maybe even close enough (although one would have to mess with point size differences).

* The Trinité specimens aren't comprehensive enough to tell.

** Although no descenders or caps.

Nick, although (once again as per MyFonts) those two don't click exactly the way I'd like, using Palatino for emphasis in a body of Aldus (or maybe even ITC Garamond within some normal Jannon) is a great idea - it would serve as a proof of concept - let me try it...

BTW, for some reason I hadn't realized Aldus doesn't have a Bold!
Somebody fix that, please!

hhp

miha's picture

I don't know, but there might be some old Multiple Master fonts where you can change x-height yourself.

hrant's picture

True! Good thinking. A combination of Optical Size and Weight axes might allow things to click... Although, as above, the ascenders and descenders are likely to change in proportion (as opposed to the descenders becoming relatively more cramped).

hhp

cuttlefish's picture

This talk has got me editing one of my fonts to make a higher x-height version. Early attempts show the LC coming out looking condensed to maintain the same width, and the hook features of the f and t going all wonky. I'll post some pics later.

William Berkson's picture

Hrant, what the Frutiger book shows is that he stretched the letters vertically in the x-height range. That also moved the ascenders higher. He did some other stuff, but the basic thing is stretching in the x height region. So that gives you a concept anyway of what happens. Personally I don't like either Serifa or Glypha.

hrant's picture

Jason, cool, but just know, as discussed you'll
be sacrificing the ascender/descender balance.

BTW, I'm assuming the widths will be let loose, not
maintained. Otherwise besides looking compressed
as you say, the apparent size difference wouldn't be
as dramatic.

> Personally I don't like either Serifa or Glypha.

Me neither.

hhp

cuttlefish's picture

Well, I used the Change X-height feature in Fontforge, which worked latgely as you describe, though some glyphs had unexpected distortions that had to be corrected by hand, and since I misunderstood you I also applied the Condense feature to bring the width back down to that of the original. It wasn't possible to match the widths exactly, but they came close.

I think I understand you now. I recall reading that versions of Avant Garde from different vendors have different sized lower case. That obviously couldn't have been planned as the variation you're looking for, but it looks enough like it.

Jack B. Nimblest Jr.'s picture

HrantI >1) Using a larger-looking cut of a typeface for emphasis, instead of Italics.

but...if the l.c. xht alone grows, and the ascenders do not absorb this growth and grow as well... if there's uppercase in this scheme, despite the possible overshoot of ascenders above caps in some designs, just changing the xht is impossible without raising the caps. Either way, (overshooting ascenders alone, or caps growing with ascenders), you'd need to have a style that originates with lots of extra vertical space.

Hrant II> 2) If you're mixing writing systems (or even languages*) inline and you want to respect each script's apparent size and Cartesian-space-usage needs without making a mess of horizontal alignment.

Nobel effort. One of the truly divisive underlying issues in all of font tech, is the distinct differences in the way scripts address the em. Listening to much older hands than myself, I learned that the least evil solution is always to favor the script of the "client".

The web, of course, 'hopelessly' complicates modern use of the term "client". (So please support the OFF mvar and uvar tables if they ever come up ;)

Thomas Phinney's picture

Find somebody with Ares FontChameleon and a computer still capable of running it, and you could generate varying-x-height pairs of fonts all day.

JoergGustafs's picture

Hrant:
BTW, for some reason I hadn't realized Aldus doesn't have a Bold!
Somebody fix that, please!

Check out Aldus Nova, by Zapf himself, together with Kobayashi. It features two bolds and matching italics!

hrant's picture

Oh. (Thanks.)

hhp

Nick Shinn's picture

Oh dear.

hrant's picture

> if the l.c. xht alone grows, and the ascenders
> do not absorb this growth and grow as well

I actually mean that they do "absorb" it - so the overall height of the ascenders does not change. The point is to keep as much of the alignment as possible - not least for reasons of consistent line-height (although I do realize that a larger x-height ideally requires more leading).

> just changing the xht is impossible without raising the caps.

Why?
Certainly if the x-height gets too big the caps (especially the ones that are structurally similar to their lc-s) can get drowned out; but the good news is the caps can actually be allowed to get wider along with the x-height increase (since that doesn't ruin any alignment). There's a bigger problem though*: you can't emphasize something in all-caps (like "I"). Not sure what to do about that.

* Similar to the case of so-called upright-Italics.

> you'd need to have a style that originates with lots of extra vertical space.

Sure. Text fonts. :-) I'm assuming you mean extra space within the vertical span (as opposed to extra talus - AKA "internal leading").

> Find somebody with Ares FontChameleon and ...

Or this exquisitely-timed contribution...
http://typophile.com/node/73827

> Oh dear.

I just took a closer look at Aldus Nova to figure out Nick's complaint, and I'm not sure this is what he meant, but: what's with the Bold looking just like Palatino Nova Bold? (Also, I can't see any difference between the two Bolds.)

hhp

hrant's picture

As we say in Armenian, I "forgot the big donkey":

> One of the truly divisive underlying issues in all of font tech,
> is the distinct differences in the way scripts address the em.

Indeed this is something I've been campaigning on for over a decade,
writing articles, giving talks, etc. Some scripts (like Arabic) suffer
very obviously when their proportions are imposed from Latin so most
people oppose it; some other scripts, like Armenian, don't suffer overtly
so everything is made to align with Latin - but it does suffer when it
comes to readability. And I think this sensibility should be applied even
to combinations of Latin-using languages that use space very differently.
And certainly Cyrillic -when its "x-height" is simply taken from a Latin
master- has suffered too much for too long - it's time for Cyrillic type
design to make such alignment the exception, not the rule.

There is merit to alignment - it does have a place in any design
compromise (which to me is a redundant phrase actually) but
it's certainly given too much weight almost all the time.

> the least evil solution is always to favor the script of the "client".

But some clients are -or at least want to be, or appear- multi-script. :-)

> OFF mvar and uvar tables

What are those exactly?

hhp

Jack B. Nimblest Jr.'s picture

>Indeed this is something I've been campaigning on for over a decade...

Well, what do you need in fonts to offer a solution to the issue(s)?

hrant's picture

At the top end is what I call a "multilateral" system where each script* has a master design (that fully respects the script's needs, Cartesian and otherwise) and subordinate designs in each of the other scripts in the system (where each subordinate adapts -although not to the point of "slavery"- to the master design). And the master designs try to belong together stylistically, as equals. This gets very hairy very fast, and AFAIK has only been done once, yours truly's Nour&Patria system. :-)
http://themicrofoundry.com/image/s_rome2-1.gif **

* Or technically each language even...

In the Ernestine&Vem system (the Dublin thing) Nina actually led the way and figured out a valid simplification of the system: since Armenian prefers to be slanted -at least when serifs are involved- she thought to use the Latin's Italic as a "pivot". This worked great, but is pretty limited in versatility; the full multilateral system can accommodate pretty much any typesetting task, with any combination of hierarchic and parallel settings. This thread is basically a search for another valid simplification.

--

In terms of breaking the vertical proportions across scripts, one has to decide where to end up! To this end I've used a quasi-numerical method. I render a line for each script where each letter's width comes from its frequency in text:
http://themicrofoundry.com/image/s_rome1-4.gif **
In this way I can quickly visualize how a given script uses the Cartesian space, in particular in comparison to another script. In the above example one can see: how little variety there is in the Armenian "x-height"; how much more Armenian relies on extenders. Based on this I arrived at ballparks of how to vary the proportions between the scripts, but then had to do some iteration/tweaks to make things just right. It would be nice to have a more robustly numerical system that can spit out the "final answer" based on: letter frequency; and the vertical "weight distribution" of each letter.

The results have so far been applied to Nour&Patria and Ernestine&Vem; the former system has been used in a handful of projects, and we have high hopes that the latter will see glory once it's finished and FF releases it.

--

I think the first thing to do, after accepting that vertical proportion imperialism :-) is a bad thing, is to get Cyrillic and Greek to break free.

--

** Images from: http://themicrofoundry.com/ss_rome1.html

hhp

Nick Shinn's picture

To address Hrant's issue:

I don't see this as primarily a font problem, but one for typographers and graphic designers; ultimately a question of aesthetics and branding.

There is a school of thought that would like to have uniformity in type spec between different languages, alphabets, and scripts (in designs ranging from page layouts to corporate identities) — but this desire for economy of means and simplicity of appearance may not be entirely virtuous, it could also be seen as a reluctance to allow typography to cater to the complexity of what it represents.

If typographers are able to manage the intricacies of mixing sans and serif type, of mixing bold and light weights, of mixing lower case and small caps, oldstyle and lining figures, of applying optically adjusted sizes, then why can't they mix different fonts for different languages and scripts?

But perhaps type designers can help by providing more typefaces with x-height alternates, which would seem to be an opportunity in the marketplace.

hrant's picture

> ultimately a question of aesthetics and branding.

There's more to typography than that.

> why can't they mix different fonts for different languages and scripts?

They can, and sometimes it makes great sense. But if they don't want to produce stuff that's too decorative, if they worry about visual distractions taking away from the content, then their options are extremely limited, because most multi-script type systems are very poorly balanced, exactly because they ignore what I've described. They're basically a lot like forced marriages.

It's not a matter of making one thing that will cover all styles. It's a matter of making tools that are good enough for typographers who care about functionality. Such tools remain too rare in the non-Latin world.

hhp

Jack B. Nimblest Jr.'s picture

>At the top end is what I call a "multilateral" system where each script* has a master design...

each with its own baseline?

hrant's picture

No, baseline fixed (for scripts that have one).

hhp

Nick Shinn's picture

There's more to typography than that.

That's what I said.

Jack B. Nimblest Jr.'s picture

Start making sense.

Nick Shinn's picture

Let me try again: aesthetics and branding are the elephant in the room of multi-lingual/multi-script typography.

Their call for uniformity generally overides the subtleties that might do greater justice to regional typography.
This is why they are the "ultimate question".

How does one counteract the global force of homogenization, as represented by reductive (modernist?) design, and strict corporate identity manuals?

Having local type producers is a start, and we have certainly seen that in the last two or three years, with a swath of "Cyrillic" type designers now marketing globally through MyFonts and Paratype, and the emergence of linguistically-positioned foundries such as Typotheque, Rosetta, Cannibal and Parachute.

hrant's picture

I for one don't ignore aesthetics (and I'm having trouble thinking of anybody who does). "Branding" is better termed "identity" here, and one might say it's the raison d'être of my type work (which is frankly why I don't make enough money in it). As for uniformity, it's exactly what I'm fighting; my disdain of Modernism is no secret. The difference is my view of subtlety is much more... subtle :-) than expecting people to simply use different fonts for different scripts (or languages), which I find highly exhibitionist and anti-reading. Breaking "vertical proportions imperialism" is the sort of subtlety I'm promulgating (and implementing); this is part of the subtleties of text typography (as opposed to the subtleties of display typography).

> Having local type producers is a start, and we have
> certainly seen that in the last two or three years

Which is great, but notable is the paradox that John Hudson once made me painfully aware of: historically "natives" have often been the ones doing the most damage to the authenticity of their script, because they try to emulate too closely the admittedly more refined world of Latin type. It's difficult to filter the formalities of Latin type to extract only what really works for a given non-Latin script.

hhp

Jack B. Nimblest Jr.'s picture

>historically "natives" have often been the ones doing the most damage to the authenticity of their script

Now that's something I'd like to see some data on. Or perhaps it was true up until... I mean, the technology of type is foreign to most scripts, and that i think, is where the most damage comes from.

riccard0's picture

It seems (http://twitter.com/#!/iEmmanuelRey/status/74143508478427136) that the upcoming New Fournier BP (http://swisstypefaces.com/blog/2011/5/14/new-fournier-bp-teaser/) will have two versions with different x-heights. No word about ascenders/descenders, though.

John Hudson's picture

David: Well, what do you need in fonts to offer a solution to the issue(s)?

Not speaking for Hrant, but as someone who has to deal regularly with the issue of different scripts within a single set of vertical metrics:

What I think is needed is script- and language-specific vertical metrics, i.e. applying the OpenType Layout script and language system tagging system to vertical metrics sets. So, for instance, a font that contains both Latin and Arabic designs would have deeper descender metrics for the Arabic script tag, and if its Latin support includes Vietnamese this might have taller ascender metrics for the appropriate language system tag. Applications using vertical metrics for linespacing would then apply the tallest ascender and deepest descender of languages/scripts present in the text, either at the line, paragraph, story or document level depending on user preference.

Jack B. Nimblest Jr.'s picture

JH>as someone who has to deal regularly with the issue of different scripts within a single set of vertical metrics:

Yes, thanks.

JH> What I think is needed is script- and language-specific vertical metrics, i.e. applying the OpenType Layout script and language system tagging system to vertical metrics sets.

So, the sfnt's table(s) include VM 'per script' in general (?) and then, per language (?)...

JH> Applications using vertical metrics for linespacing would then apply the tallest ascender and deepest descender of languages/scripts present in the text, either at the line, paragraph, story or document level depending on user preference.

And the other way around(?) is that if a script is being used alone, from a multi-script font, where said script had been "VM merged" to work with the other scripts in the font, perhaps somewhat uncomfortably, app/user can find specific metrics more comfortable to that script/language in use alone? (complete with any damage that might have been inflicted on a given script's contours from above merge, which can't be fixed this way), some way other than trial and error.

And then backward extension of this, into the font tool development arena, type design could be perfect decisions for each script, including the contours, with an intelligent set of transformations available to "can" combinations for particular pairs, trios or quads, with optmal VMs for each and all combinations.

And then the forward extension back into user space would be to have this mechanism in client-side font software, or serve the results, (making it so that per-script damage is minimized and could be truly at the user's discretion).

To do this a tool'd have to change it's attitude to the em and VM, thanks.

nina's picture

Hrant, getting back to something from before, here's something kinda-related that may be interesting (disclaimer: I've only skimmed [most of] the thread, sorry):

> 1) Using a larger-looking cut of a typeface for emphasis, instead of Italics

I think it's an interesting idea. If I remember correctly, Erik van Blokland briefly showed a mouseover effect like that at Webfontday in Munich last year – the idea being to have a website navigation rendered in a given webfont where on mouseover, as a way of highlighting (!) the selected item, instead of boldening or something the x-height would be raised. I don't quite remember how he proposed to do this technically (I think it wasn't about swapping fonts, more like some sort of MM-type OT/CSS magic) but I did find it a striking effect. BTW, in that case the widths stayed the same (which I think is good for this purpose, since having stuff wiggle around in width on mouseover can be distracting).

The bad news is I can't for the life of me find any reference to this now. It might even have been part of that wildly animated/scripted Safari 5 (?) / WOFF demo page he made at one point, but I can't find that either. Some help, eh?

Karl Stange's picture

Nina, is this what you are referencing? According to the page it will only work with Firefox 4, at least in the context of that example. I found the link through this page.

nina's picture

Oh, yes that was the test page I was thinking of (thanks!), but it seems the x-height thing isn't actually part of it.

Jack B. Nimblest Jr.'s picture

It's actually pretty amazing that you would confuse a coooooool FF toy with something globally useful ;)

I'm still waiting for Hrant to add the obvious to his excellent idea of larger-looking cut of a typeface for emphasis; how the rest of the planet should handle strong, and emphatic strength , and do so throughout a "normal" glyph repertoire, where all four states of emphasis & strength are broadly required. And to do so over a spectrum of designs, a range of sizes and a battery of rasterizers might only be achievable beyond the gravity of typographic reality.

hrant's picture

I think Strong (The Style Formerly Known As Bold) can be applied in parallel to... what to call it?... Large? So no problem there as far as I can see.

But yes, there's is the problem of sometimes having no way to exhibit an increase in size, for example for a character that already fills the Em (mentioned in my second post of May 4). In fact some writing systems (those that don't have extenders) probably can't use the idea at all (at least not without ruining [their version of] leading).

I guess what I'd say is that no solution works well for everything; you use what works best, and even if by some strange cosmic shift the idea of Large ends up becoming the Standard, there will be times when Italics will work better.

--

Something else though:
What's sacrosanct about The Four Styles?

hhp

Jack B. Nimblest Jr.'s picture

HP>What's sacrosanct about The Four Styles?

Not to me, but to some they represent the four winds.

To others, and here I am included, the two tranformations, bold and italic, are arguably the easiest to apply to any regular width and weight font of any script with the four resulting options readers are familiar with, brought to them over time by a wide variety of design and composition customs that use these four winds to blow the user smoothly along through documents.

hrant's picture

I think contrast in weight (tellingly dubbed "color") is indeed natural. But I also think size is much more natural than skew*; even rotation** might be more natural than skew. And at every given instant a responsible designer needs to strike some moral balance between serving the status quo versus trying to improve it. As Robert Plant (?) said: "Revolution, go slow."

* And certainly not cursiveness - that's bogus.

** http://typophile.com/node/82853

hhp

William Berkson's picture

>not cursiveness - that's bogus.

How do you explain that no serif font has a successful slanted roman?

hrant's picture

Let's define:
1) Slanted Roman.
2) Success.
3) The relevance of said success to cultural progress.

hhp

Jack B. Nimblest Jr.'s picture

Let's define:
A. Think
B. Natural
C. Plant and
D. Improve.

Syndicate content Syndicate content