Well, for the sake of discussion I'll ignore the ethical issues around releasing a bunch of clones of original designs - that happened long before they were donated to the GhostScript community.

The quality is wildly variable. The "original" [sic] URW fonts are really of very good quality, and that continues in the outlines that are directly sourced from the URW originals. The added Greek, Cyrillic, small caps and oldstyle figures are generally somewhere in a range from mediocre to poor to largely useless. Some are better than others. On the down side, look at the oldstyle two in the Palatino clone.

I think this really shows both the strengths and weaknesses of open source fonts (maybe open source software in general?). There's some really first-rate material, and some utter junk, and it's mixed into a single font (or piece of software). But heck, it's free! :/

Cheers,

T

As a TeX user, I have always appreciated the times that someone very knowledgeable about type has provided insight into the fonts that are readily available in the TeX world. Usually the response has been similar to Thomas Phinney's: mixed quality. I like the fact that Knuth's Computer Modern is given an "honorable mention" on the Typophile libre fonts wiki page. That's always been where I've placed it.

Although I have purchased and installed for use with TeX many professional typefaces (thanks Adobe for allowing modification of the font files), I still have a keen interest in the freely-available font packages. It's not because of cost, but because I hope I'll find a package that contains a good typeface _and_ all the integration that's necessary to pair it with harmonizing mathematical symbols. I also hope to find packages that I can recommend to students that will allow them to improve the quality of their output.

Here's a general question: The TeX Gyre web page provides information about the automated method used to generate small caps, which involves unequal scaling in the x and y directions and application of a brush to trace the outline and thicken it further in the x direction--just enough to match the upper case. While I know that any automated process isn't ideal, could such a process produce results "good enough" not to be embarrassing to someone choosing to use those glyphs? I once did that as a personal project with Palatino (URW Palladio) and felt that the letters came out OK, but my attempts at old-style numerals definitely delimited the extent of my skills.

ST

That's how they did the small caps? I never got that far in reading the site. Automation without human correction is bad for the soul. And that explains why the small caps R of their Helvetica looks slightly off. No one corrected it.

Thomas,

Compare TeX software and your widely available commercial office suite: both have made contributions ranging from excellent to largely useless. But one is free.

I personally find the commercial suite useless period. In contrast, I thank Adobe and the LaTeX community (especially those who gave the world the MinionPro package) for the ability to prepare good looking documents at no charge.

Note that:
- they are not aware of all the bugs & problems
- they are open to suggestions
- they respond to emails
- they add or improve glyphs based on user requests

If you think that small caps R looks horrible, drop them an email. They sometimes need some time to answer, but they almost always do. And they keep improving glyphs based on user requests/comments.

On the contrary, I would not dare asking Adobe to add a new glyph to their fonts.

Right now, they're trying to find funding (even if these are small user contributions) to design math glyphs for all of these fonts.

They do a marvellous job. I'm esp. a big fan of their Antykwa Toruńska:
http://nowacki.strefa.pl/torunska-e.html

Mojca

@EK
TeX, although I've never used it, certainly fills a need for many. And the price is right. But it's hard to paint a broad stroke saying "I personally find the commercial suite useless period.," when in fact, if you ARE a commercial graphic designer with any hopes of file compatibility with other graphic designers, you'll need to use commercial software. I know of no design agencies, publishers, print shops, etc., that don't use commercial software, i.e., Adobe's Creative Suite or equivalent.

They're nothing more than tools of the trade. If you're a construction worker framing a house, you need a hammer, a belt, and a truck, right? No different.

@Mojca
I'm not complaining about the free/commercial aspect. I'm not even complaining about the glyphs being off a little. (You get what you pay for...) I'm just curious to see what kind of results have been made by a small group of noble-minded volunteers. If I see any problems, I'll certainly feel free to either contribute what I can or to notify them of my concerns.

Dan,

The reference was to a commercial office suite that I dislike, not an Adobe product. I meant to show that "some really first-rate material, and some utter junk" characterizes commercial software as much as open source.

And thank you for bringing up the compatibility argument. It's hard to find any argument to prefer that famous commercial suite over the alternatives other than compatibility. That's why its maker has worked hard to reduce interoperability.

@Mojca: On the contrary, I would not dare asking Adobe to add a new glyph to their fonts.

Why ever not? In fact, we take such suggestions quite seriously. They influence both our decisions about what to put in new fonts, and what we do when retrofitting existing fonts. I've even blogged about it....

I'm kind of tied up right now with some personal medical issues (writing this from the hospital, in fact), but I'd be happy to give a more detailed commentary on what I noticed in reviewing the TeX Gyre fonts at some later date. But the main thing the fonts need is not expert critiques, but a larger donation of time by one (or preferably more) people who really know their stuff to actually fix the fonts.

Cheers,

T

DanGayle wrote:That’s how they did the small caps? I never got that far in reading the site. Automation without human correction is bad for the soul.

I should offer the correction that I read far enough to see how they automated creation of small caps, but I didn't read far enough to see if they stated at a later time that they corrected the glyphs by hand. Having read changelogs, etc. for the Latin Modern fonts, it really does seem like they're putting serious attention into these projects.

Thomas: as a frequent lurker here, I've always appreciated your comments, and I hope you get well soon.

ST

I can say based on some recent suggestions which I've made that Adobe does listen to the users.

I took a look at the Century Schoolbook a few weeks back, and thought some of the added characters rendered the font (well, the new characters) unusable. The old-style figures were particularly bad.

I'd say Thomas has it about right.

When you think on it, typographic niceties are not required for communication -- lining figures and full caps communicate information adequately. The reason for their use is aesthetic, and if they are rendered poorly, they don't fulfill that function.

OpenSource fonts are not about letting people get something for free. The licenses for most fonts, if adhered to, prohibit anyone but a large, successful commercial corporation from fully using them. For example, consider scholarly journals, which ideally are (1) printed, (2) put on the web, (3) available for downloading, and (4) etc. Sometimes the total subscriber base is as small as 1,000, though the total audience, esp. for single articles, may be significantly larger.

What are they to do? Site licenses are prohibitively expensive. How many good OSF are there, maybe 1 or 2?

Again, Thomas is correct -- what is needed is a few good folk to donate their time & talent.

There is still plenty of business left for type designers. Just look at the design forum "I need a font that looks like (whatever), but is new!!!" IMSLTHO, most of the suggestions are inferior to (whatever). But they are new, & apparently sell.

FWIW

Charles, if your business sector can't afford to pay for good new fonts, you don't deserve them.
So quit whining and don't expect a handout!
You've been spoiled by digital publishing, paperless publishing on the internet, and faux features for small caps, fractions, and superior figures, which make things seem a little too easy. And the glyph- and feature-loaded fonts that Adobe bundles with InDesign.

Since FontLab 4.5 enabled Mac-based type designers--people like myself with independent foundries--to produce typographically-rich OpenType fonts three years ago, we have been working hard constructing these beasts, and they are starting to be published. Concurrently, Reading University has been seeding the industry with a new breed of broadly-educated (language, culture, coding, history) type designers who have a quite different foundation than the previous new generation of 200-character-per font, DIY deconstructionists. And also, an internet community of diverse stakeholders has coalesced and nurtured the development of mega-fonts*. All this is in play, and the scholarly publishing sector is beginning to reap the dividends.

Having spent thousands of hours in developing scholar-worthy fonts that are still not published, I already feel like I have donated enough time and talent, without having to put out freebies. I don't know exactly where I will recoup my investment, but I expect it will more likely be from the corporate and periodical sectors. Scholarly publishing will be a beneficiary, as the pool of suitable fonts available from independent foundries grows larger. So get out your cheque-book and build more fonts into your budget, because the new ideas that will stick will be those that are well thought out, well written, well edited, and well set--in well-designed new fonts.

*An example: in his Typblography blog, Thomas Phinney started a thread on Cyrillic character support, in which he outlined his, and Adobe's philosophy. That really put things in perspective, and was an inspiration to me by providing a logical framework for increased character support (with examples). He also solicited feedback. So right there, the dynamic intersection of the corporate, the academic, and the independent.

Thomas, here's wishing you back on your feet soon.

Nick, in general, I have a lot of respect for you. But not tonight. Scholarly publishing has gotten along only by more or less skirting the issue of embedded fonts appearing on the internet.

As far as whining goes, I have given of my talents, such as they are, to scholarly publishing. When needed, and when within the license, I have cut needed characters such as phonetic symbols, the Latin characters from Latin Extended A, B, and Additional. I've done what I can for the aesthetics of some fonts too, cutting small caps -- especially italic small caps -- sometimes roman as well. Occasionally, heart in hand, I even cut old-style numbers. I sweat a lot over the old-style 2. And other work.

How much do we charge for this? Not one damn cent.

But I do not have the skill to complete full alphabets, or some of the more intricate phonetic symbols, and the licenses forbid my releasing the characters I can do to the scholarly community generally. For the Adobe fonts, I'd give them to Adobe for re-release, but I imagine they wouldn't want them. They're OK, better that the fonts under discussion, but not perfect. They work for text size settings. Sorry I lack the skill but not the desire.

I don't care if you don't participate. Your choices are yours to make. But why council others not to contribute a little? If you think university presses have the money & are just being tight, you are wrong. From my side of the blanket, I see the whining coming from a different spot.

But why council others not to contribute a little?

I wasn't. In fact, might I recommend that you pass the hat round amongst the scholarly community and have everyone chip in a few bucks? Then hire a commercial foundry to produce the fonts you need, and avoid being caught between the rock of commercial fonts that are deemed too expensive, and the hard place of "first-rate mixed with junk" that occurs with open source.

hire a commercial foundry to produce the fonts you need

This has been done before. Those familiar with LaTeX will know that Hermann Zapf was commissioned to create the Euler math fonts by the American Mathematical Society. They were originally created to complement Knuth's Concrete text fonts, but they are also a good complement to Palatino and other similar text fonts. I'd love to see this happen a few more times, perhaps creating several high-quality font packages to complement a wide range of text fonts.

And while I'm giving my wish list, I'd love to see someone (pay to?) create a multiple-master font consisting of just those mathematical symbols that aren't letters and don't need as much design, but whose weight should be adjusted to match the text face. Many newer fonts are now shipping with Greek and other glyphs that would be useful for typesetting math, but they need matching symbols if everything is to piece together well. Computer modern is very complete and is often used to fill in missing glyphs in other typefaces, but its symbols are too light to match many of them. [The mathkit package attempted to do this using metafont in 1998, but I would guess that it produces bitmapped fonts as the final product.]

I thought of giving this a try just before my son was born... Afterwards, I didn't think about it as much.

ST

The University of Toronto Press spent several thousand dollars in the late 80s to create PostScript outlines of Knuth's math fonts. However, management at that time decided not to share those fonts with the community. There is competition even within the scholarly marketplace.

(In the end it was not a big problem. Adobe came out with the Lucida Math set, based on Knuth's mappings, within a year.)

 The TeX Gyre project does pay for professional designers to work on the fonts, though the initial work was done by programmers. They are currently soliciting for donations for Stage 2 of the project (Unicode math coverage; Stage 3 in the original project plan) which is expected to cost €40K, some of which should be covered by the TeX users' groups. Stage 3, "a cleanup and redesign of current shapes that are not okay", is supposed to follow somewhat later.
—Joel

It seems to be the consensus, at least of those here at Typophile, that Stage 3 ought to be moved forward in the grand scheme of things.

I agree, but Stage 3 is locking the barn door after the horse has bolted.
This project would have benefited from involving people who know how to draw, from the outset.

Where it has been possible to cannibalize the original roman characters to make new characters, the resulting glyphs are ok if not too much modification is required -- many Cyrillic characters are fine. But where even a little re-drawing is required, the quality suffers. Where completely original letterforms are required--the Greek lower case of "Pagella" for instance--the results are ghastly. Really, those should be re-done from scratch.

As I see it, this is the goal of this project: Create open licensed fonts with reasonably extended character sets and added typographic niceties like small caps and old-style figures, all wrapped up in a well-coded cross-platform Opentype font.

It's not to create open licensed typefaces with character sets that cover more than the basic Western Latin. That's covered elsewhere, by other typefaces such as Gentium and the Deja Vu fonts.

Is it to create math fonts for educational texts? Not really, because that has already been done.

So if your purpose is in fact to make nice, typographically rich fonts, why even release them unless the rich typographic parts of the font are well-done?

This is not to belittle the contributions made so far, but I think Nick is right. Well-intentioned does not equate to well-drawn, and rich type design is by definition well-drawn.

No, I don't mean to belittle the project. It's great to see people and organizations developing fonts with this scope, and I don't consider it competition for new commercial type designs -- if anything, I would expect that these fonts, with their introduction of typographic niceties prompted by OpenType, will stimulate interest in commercial fonts amongst people who might not otherwise consider them; as long as the commercial fonts have the character support. So the challenge for foundries is to develop fonts which have both the character complement and the typographic features. It's a huge amount of work.

Thomas,
 Could you please post your review of TeX Gyre, or has everything been said in this thread?
—Joel

Sorry, I am back at work now, but also trying to catch up from my outage. More on that elsewhere, I think. Suffice it to say that everyone is fine in my family, and we're all home. :)

I think that the comments in this thread give a fair bit of guidance. I could do a more detailed critique, but it would take a lot of time, and anybody capable of fixing things based on that would also be capable of reviewing the fonts and finding and fixing problems along the way. A *really* detailed critique could be done, that might guide a less experienced type designer, but that would be at least an order of magnitude more work. 40-80 hrs instead of 4-8, kind of thing.

Cheers,

T

After reading this thread I was encouraged to examine the fonts in question more closely.

One interesting difference I noted was the redrawing of the ogoneks (the Polish connection, I assume). Viz:

HELVETICA clones:

Nimbus, Helvetica Neue and URW Gothic (see below) appear to be missing the Ǫ (Oogonek). And note the placement of the tittle in the lowercase i in TeX Gyre Heros.

Avant Garde clones:

At least they're consistent. Look at how widely Nimbus varies in their additions. (Unless variation is a good thing. I've ever designed an Oogonek)

I had recently visited the web site for this project.

On the general subject of which typefaces are most appropriate to mathematical typography, I would like to make some comments.

Times Roman is a very popular typeface these days. Like Caslon, it has sharp points; like Baskerville and Corona it is a Transitional design, and it happens to be beautiful. However, quite aside from Donald Knuth's preferred style of italic x, its italics needed to be altered for mathematical composition, specifically to change the italic v so that it would not be confused with the Greek nu. (The reduction in slant was done for technical reasons, and the slant of the italics for Modern No. 7 is the same as that for the regular Times Roman italic, so that is not required by the aesthetics of mathematical typography.)

Needless to say, if you are using a word processing program on a computer to set mathematics in Times Roman, the italics you are likely to get are going to be based on those of Times 327, not Times 569, even if that font has been digitized (even in Times 569, the modified g, v, and w were alternate forms, however).

Computer Modern, while a contribution we should be grateful for, still "looks funny" compared to conventional fonts; there is an effort to design an improved version of it.

While the font Modern No. 7 is reasonably attractive, it still looks somewhat dull and severe from the standpoint of current fashions. Partly, it suffers from being associated with the Scotch Roman styles used in 19th-century typography which was unattractive.

If one isn't bold enough to try AMS Euler or even Cambria Math, is there an obvious alternative? I think there is, and I'm surprised it seems to be neglected.

I remember reading somewhere - and finding it confirmed by my experience - that if there are three competing magazines, each seeking its own typographic identity, a very likely result is that one will be set in Times Roman, another will be set in Caledonia, and then the third will be set in Century Expanded.

Century Expanded, of course, belongs to the same general category of typeface as Modern No. 7, but unlike both it and Computer Modern, it is fully acceptable to current tastes. So, to my mind, it is the "obvious" choice, but I don't see it being used much for mathematical publication.

You can see samples of all Gyre fonts (and some others) at http://www.nowacki.strefa.pl/wzornik/ , "Napisz i zobacz" is the place for your sample.

It seems to me that the review, if published, would be 10x more valuable than directly improving the fonts.

If it were as impossible to extract a font from a .PDF as it was from a printed book on paper, so that embedding wasn't an issue, then people wouldn't need "free" fonts to publish PDF documents - that was the issue that was raised here, which has nothing to do with not paying to license a font. Although I thought the licenses for most commercial fonts did include embedding.