I don�t know if your time is of any value, if so, it is probably cheaper to just buy OpenType fonts, if they are available, and if they have all the glyphs you need. (Likely they won�t have all the glyphs, though. In the rush to put out OpenType fonts, most foundries aren�t including things.)

Here�s what I mean. For years now, we have made a database of all our Type 1 fonts. When we bought a set of fonts, using Fontographer, we would make up a font file where all the glyphs were included, small caps, old style numbers, etc. We�d also add whatever glyphs we needed, such as tabular oldstyle figures, tabular lining figures, extra diacritics, and so on. (To USE this font, we�d write an encoding vector selecting the glyphs we wanted, but that�s a different story.) In short, our Type 1 fonts had most of what you�d want to include in an OpenType font.

Starting from this database, it takes me several hours to convert the font to the OpenType format.

One reason is that I don�t like using the Unicode private use area for small caps, oldstyle figures, etc, & so rename them. See

http://www.typophile.com/forums/messages/4101/25695.html?1076260212

If the glyphs are *correctly* named (proper AGL name) in the Type 1 font, FontLab will preserve the correct name them & assign the Unicode number automatically. But if you are using the Type 1 fonts as purchased, the numbers and small caps will not be properly named, they will be in a separate font (a small-cap, old-style figure font), with the small cap glyph named �a� �b� etc, instead of �Asmall�, �Bsmall� etc. The os-numbers will be named �one� �two� instead of �oneoldstyle� �twooldstyle� etc. This means you will have to do the renaming yourself. Moreover, you�ll have to open up the original Type 1 font in FontLab and rename there in it in order to have the program rename the kerning & create an AFM file when you generate a Type 1 font. When you then paste the glyphs into your full font (OpenType), you�ll have to type in the names again. You can at least just read in the AFM in an �add only� mode to get the kerning. I�d actually do it differently, creating a dummy AFM file that you�ve gone over first to make sure nothing is in there twice with a different value. We don�t have this problem with our Type 1 database, because all the kerning is in the one AFM.

I�m making this sound harder than it is; once you figure it out, it isn�t hard, but it is time consuming.

A second amount of time is spent writing the �features� that give you small caps, ligatures, old-style numbers, etc. InDesign only supports a few of the OpenType features, so you don�t have that much to write, but you do have to do it.

There is a learning curve � if it takes you all day to do the first font, it will take considerably less after you get the process down. And if you use the same *features* for all your fonts, you can create them once, write them off to a file, and then read that file in for the next set of fonts.

One plus to all this is that you can create your own glyphs if/when you need them, augmenting what the foundry supplied. I don�t worry about modifying a foundry font, if the G@%D!%& foundry had given me what was needed the first time, I wouldn�t have to do it myself.

A second plus is that by the time you are done, your knowledge of both type and its use in InDesign will have gone way up. But on a strict dollar basis, it is likely cheaper just to re-buy them.

Good luck,

Charles

Charles--

Good luck indeed! Thanks for your response. Let me see if I can distill your answer in order to make sure I grasp what's going on:

First, I would have to open in FontLab all the individual Type I fonts whose glyphs (or some subset thereof) I want to combine under an OTF wrapper--so, for example, Scala LF Regular, Scala Regular, Scala Expert, Scala SC, Scala SC Expert.

Second, using the AGL naming conventions, I would need to rename all the glyphs in these five Type I fonts I want to use in the final OTF.

Third, I would need to save from these fonts five new Type I fonts in order to generate kerning tables with the right glyph names in new AFM files.

Fourth, I would need to open all five of these new Type I fonts and combine all the glyphs from them I want to use into a sixth, new OTF. (Which names, exactly, would I need to "type in again"? Doesn't FontLab have drag and drop glyph insertion? My own copy of TypeTool does.)

Fifth, I would have to read all the old AFM kerning data for all the glyphs and their combinations into the new OTF (making sure there are no duplicate kerning calls with different values in the file).

Sixth, and finally, I would have FontLab generate the new (CFF-flavored) OTF.

I see what you mean by things being cheaper when you just buy the OTF versions. Of course, that's not yet possible for FontShop's foundry, and in any case, it would/will be pretty pricey when FSI does start offering OT versions of their fonts to get versions of everything I have licensed (small as my collection is).

Not to mention, wouldn't renaming glyphs and combining kerning tables start getting me into trouble with FSI's (or any foundry's) EULA? I know you wrote that you wouldn't worry too much, and your reasoning is on the surface sound, but is that a consensus opinion? My guess is that at least some of this is ethical, else why would Adobe offer software that allows people to do this?--unless, of course, they meant only for people to edit and combine their own work.

Lots for a neophyte to chew on. Other thoughts and advice are welcome.

Maurice,

With regards to our own FDK, I recommend its proofing/testing tools but not its building tools. You have to be a real gearhead to get very far with those (no offense to my fellow geeks intended, here). Also, you'd still need something else to combine original fonts and so forth, so you might as well use FontLab for the entire process.

Mind you, that assumes that you're going to actually do this. I can't see how this would not constitute modifying/editing the fonts. If the FSI license does not allow such, then I would think you'd be out of luck as far as being able to do this legally. Not that I'm a lawyer, and of course this is a legal question.

T

Thomas--

I had another look at FSI's EULA. Given Charles's account of the process with FL and your hints about FDK's steep learning curve requirements, I'm inclined to read that EULA very conservatively. Less facetiously, I agree with you: I can't see how combining kerning tables and renaming glyphs doesn't mean you are changing fonts.

<sigh>

If wisdom consists in knowing that you don't know, then I come away feeling pretty wise.

Oh, well--it's not as if I can't just sub in the right ligatures after finishing the rest of the setting for those FS faces, and at least InDesign comes with some pretty sweet fonts that are already OT. --MM.

> since I'm not editing the font
> (changing the metrics, hinting, or kerning)

Any conversion between formats implies editing. Converting from Type 1 or OpenType PS to OpenType TT / TrueType involves coversion of outline types and of hinting, hence editing. Converting from Type 1 to OpenType PS (.otf) involves changing the font's encoding and other modifications.

Your steps 1-6 are correct, with a few remarks to be made:

1. Do not use glyph names such as "oneoldstyle" or "Asmall". They are not Adobe/FontLab-recommended glyph names. Glyph naming recommendations from Fontlab Ltd. and Adobe can be found at:
http://groups.msn.com/fontlab/tipsandtricks.msnw?action=get_message&mview=0&ID_Message=3065

2. Combining glyph sets of fonts into a single font in FontLab 4.6 can be comfortably done using the "Append glyphs" feature in the Fonts panel. In the upcoming FontLab Studio 5, this will be even easier -- we will have a new "Merge fonts" feature which will automatically merge kerning as well.

Regards,
Adam Twardoch
Fontlab Ltd.

WARNING: Long, and maybe boring.

This whole matter of the licensing agreement is pretty charged. My own practical position has just two elements:

1. Nothing I do should cost the foundry (or designer or whoever) any money.

2. Nothing I do should cause the foundry (designer, whoever) any embarrassment.

Everything else in the licensing agreement is lawyer talk. To take an example outside our industry, every time you rent a car, when you sign your name, you not only agree to the conditions of the rental agreement, but certify that you have READ and UNDERSTOOD the rental agreement. Now that agreement is about three or four pages long (8.5 x 11 pages!), set in 6-point type. If you were to read it, everybody in line behind you would be screaming for your head. Moreover, it was written by lawyers (who typically can�t agree among themselves about what something means), so your chances of understanding it is pretty slim, even if you read it. But you need the car, so you sign, and I suppose you�ve just perjured yourself.

Now this next bit maybe wrong � I�m not a lawyer � but this is my understanding. In the U.S., you can�t copyright the particular form of letters. That is, what makes Scala different from Joanna can�t be protected by a copyright. What can be protected is a particular piece of software that �draws� the letterforms. To me, this is silly. Martin Majoor should be able to protect his art, not simply a piece of software. I believe the copyright law has lead to some lawsuits, by the way. If you were to scan in a printed font � using ScanFont, for example, then use, say, FontLab to generate a new Type 1 font, and use a new name for your font, you will have broken no law. It is a lot of work, but it can be done. At least one major foundry did just that, was sued, and was found to be perfectly within their rights. While this may be legal in the States, I think it absurd, and breaks my rule number 1.

My Rule 2 is obviously a little subjective. But let�s start here: Kerning values are to some extent determined by type size. The right kern between two letter pairs for 10-point type will not be right for 40-point type. Now programmatic kerning (as found in the AFM) doesn�t allow for this. Since I set books, I�m going to adjust the kerning values for a range of 8-12 point type, and use hand-kerning when the same font is used in display. And I�ll make a categorical statement: I have NEVER found a �foundry release� font where the kerning values were right.

How about glyphs? One of the designers we work with is a Brit � a brilliant designer, recently selected to the RDI. His background and eye use a word-spaced en-dash when a dash is needed. American editors usually won�t agree to this though, and want an em-dash. We came up with the idea of using a 3/4-em dash on an em set, and this satisfied the aesthetic demands of both the American editors and the designers from the U.K. Now I�m going to put this glyph in the em-dash position, so that it can be used automatically in setting text.

Does this break the licensing agreement? When you think about it, the main reason this hadn�t been done is historical. The type case was small, room for only a limited number of characters. In hot-metal or early photocomp, the case was still small (Teletypewriter requirements, for just one example). There wasn�t room for a two-em or three-em dash as separate characters in the �case,� so the em-dash had to set full width in order to make up two- and three-em dashes. Now, the �case� has room for a lot more characters, so we can have a two-em dash and a three-em dash (full width to make up even longer dashes)as well as a stand-along em-dash.

Or consider the �Bembo mistake� where the wrong capital �R� was sent from the committee down to the room where the characters were cut. We were stuck with that long-tailed �R� for quite a while. Eventually, Monotype put the regular �R� in the PostScript font as an alternate character. But the regular �R� was what you wanted for running text; the long-tailed �R� was for display. We simply re-named them to make composition easier. Did we break the law?

When Martin Majoor cut Scala, most production followed the following path: repro was run out & sent to the printer, who shot negatives, stripped them up on a flat, then burned a printing plate. The type picked up a bit of weight at each stage. The printed books looked quite nice, Scala was a good font. Just a few years later, the compositor sent a PDF file to the printer, who, likely as not, burned the plate directly from the computer file (CTP). Moreover, there was apt to be live art in the PDF, so the printer had to ensure that the dotgain of the art wasn�t increased beyond necessity. All of a sudden, Scala appeared too light, when printed offset at 2,400 dpi using a direct-to-plate process. What we did was to increase the weight of the characters 2/1000 of an em for composition where CTP printing was to be used. All this did was restore the weight of the type to what it originally looked like with the old printing process. I have scans of the type I used in a lecture which show this, but the file is over 1 meg, so I can�t really post it. As far as I�m concerned, all we did was to follow Rule 2 above.

Consider too that this morass of Type 1 fonts (Regular, Expert, Small caps, OS figures) came about because QuarkMaker had hard-wired encoding. If everyone could have written their own encoding vectors, the fonts likely would have been made with all glyphs in a font. You would have been limited to 256 characters at run-time for a font, but you cold select the ones to include.

Finally, the current licensing agreements (if I understand them) are going to marginalize the small foundries out of existence. Today, when you send a job to the printer, you had better send it in a PDF file. I suppose you can subset the characters, most printers prefer full embedding. Moreover, most academic publishers have contracts with NetLibrary. When they decide to publish a book, they are contractually bound to make the book available to NetLibrary, who requires a PDF file with the fonts fully embedded. But consider the publication process: The book may be sent to a freelance designer, who has a legal copy of the font. Next, it goes to a compositor, who also has a legal copy of the font. They�re going to make a PDF file to send only to the printer, so the book can be printed. The printer returns this PDF file to the publisher, who, as far as I can tell, doesn�t need a legal copy of the font at all. After a bit of work on the images (if any) the PDF file is given to NetLibrary. Who is to police all this? Worse, the first lawsuit that arises is going to have one outcome: Press directors are going to instruct their designers to use only fonts where embedding in a PDF is permitted. Since most of the larger foundries permit embedding the font in a PDF, any protests will fall on deaf ears.

I don�t know where all this is going, or what the solutions are. But if you are breaking the law by converting your Type 1 fonts to OpenType, by changing the kerning values, or by tailoring the glyphs to your needs, the law simply won�t be obeyed. Type designers are NOT type users. The few I�ve worked with (Matthew Carter, Sumner Stone) are quite willing to listen to the users of type. What they are NOT about to do is to re-issue a font whenever a small change is needed.

Charles Ellertson

Charles, thank you for very interesting contribution. I agree with you. I face very similar problem, let me introduce you to it.

Many foreign companies are starting their businesses in the Czech Republic where I live. These companies also have corporate identities defining fonts. And that's the problem � majority of foreign fonts doesn't contain characters necessary for setting type in Czech language.

I offer a localization service for these companies or advertising agencies working for them. They send me files and I create CE fonts for them. I am happy � I have an interesting job. They are satisfied � they have functional fonts with correct Central European characters. According to Czech law, I am allowed to change a piece of software to work seamlessly with other software or to modify the software to work with Czech characters without modifying the functionality of the software.

I think there are two problems of the localization:

1) Collision with EULA. I had some discussions with lawyers what is more important: Czech law (Author�s Law, Nr. 121) or EULA and the results of their interpretation of the law are not explicit at all

2) Original diacritics in the font are often unusable or even totally wrong. In such cases I have to redesign the diacritics. Doing so I alter the original design, but the resulting font is usable for type setting Czech language

What is your opinion on font localization?

> If you were to scan in a printed font � using ScanFont, for example,
> then use, say, FontLab to generate a new Type 1 font, and use
> a new name for your font, you will have broken no law.

No U.S. law. However, if you sell your products worldwide, you can be sued in any country, including any of the E.U. countries that offer copyright protection to typeface designs. If you lose, you risk the ban for the sales of your products from the entire European Union.

> And I�ll make a categorical statement: I have NEVER found
> a �foundry release� font where the kerning values were right.

Properly made typefaces include "optical sizes", i.e. different masters for small and large type size (at least 2, sometimes 3 or 4 different masters). These masters have not only different kerning values but also different glyph outlines -- the small type sizes have often less stroke contrast, wider proportions etc. Adobe has made such masters for Brioso Pro, Chaparral Pro, Cronos Pro, Jenson Pro, Kepler Std, Minion Pro, Sanvito Pro, Utopia Std, and Warnock Pro. The newly released Scangraphic Digital Type Collection (http://www.myfonts.com/foundry/efscangraphic/ ) has two masters for most of their typefaces; sometimes they differ in metrics & kerning, sometimes also in the outlines. There are quite a few other foundries that have done just that. For example, Carter & Cone and The Font Bureau offer optical masters for many of their families.

Filip makes a good point that European copyright (the Czech legislation is similar here to German, Polish, and most likely other continental E.U. laws) explicitly allows modification of software that is made to obtain interoperability with other systems. If applied, this legislation overrules any contractual limitations. However, it is still up to interpretation whether these particular regulations can be applied to the question of font localization. A person who localizes fonts would most likely want to use a liberal interpretation of these regulations, the original font manufacturer prefers to apply a narrow interpretation. This is truly such a "border case" that, when a conflict occurs, a court would need to make a specific decision. So far, I know of no cases in the E.U. where this question would be answered.

In the U.S., contracts have a very strong binding, i.e. when an EULA prohibits a certain action, the user is most likely bound with it. In the E.U., legislation often can overrule contracts (this is the case for copyright, for customer protection legislation and many others). In general, U.S. legislation is more pro-vendor while the E.U. legislation is more end-user-friendly, although the copyright protection of typeface designs is very much the other way around.

Still, there are two aspects of the situation: the legal aspect and the ethical aspect. And there are two possible scenarios: modification purely for your own use, and modifications that are being distributed.

Some people may disagree, but I think since type is not really an artistic work intended for final use, but much more a creative resource that is being used to create further works, the "consumer" of type is always a creator of other works. This is similar to an actor's performance of a stage play. The text of the stage play is written down by the author, but the stage performance uses it only as a resource, and creates a visual, theatratical performance based on that. In many cases, the theatre company may alter or shorten the original text and do artistic adaptation if necessary.

My own view is that font localizations and modifications done purely for your own use are O.K. You are a consumer of the type, but also are creating a work of your own. There may be situations where the original type does not 100% fit into your concept and you may need to alter it. This starts with trivial changes such as setting the type on different spacing, or stretching the letters, and ends up with redrawing portions of characters or adding other characters.

Font localizations and modifications done for other people -- either under contract, as a 3rd party "localizer" who works for a particular client, or completely for distribution -- are much more tricky. My view is that you really need the original vendor's consent to do it. Contractual trickery such as: a client buys a license for a font, transfers the license to you, you make the modifications and transfer the license back to the client -- are just very disputable. I think it is not all-right from the ethical point of view. Because in this case, you make your business as a "localizer", meaning, you are in a way competing with the original font creator.

I should add that all above is just my private opinion, not associated anyhow with companies that I do work for.

Best,
Adam

If you purchased these fonts from FSI, have you tried writing them to see if they'd give you a discount on the Opentype versions?

It's worth a shot.

Again about localization.

Adam wrote: My view is that you really need the original vendor's consent to do it.

Adam, I agree with you. I usualy ask the client to contact the vendor of the fonts first and ask for the localization. Most of the clients is not satisfied with the answer of the vendor and come back to me. Maybe the reason is I can do the task very quickly and this is the most important thing for them. They don't want to wait weeks or even months.

Another problem is correcting diacritics. There are fonts distributed by large companies containing wrong diacritics and the vendor is satisfied with its product. Some clients want to modify the diacritics to respect typographic tradition and use for example correct carons (instead of breves) or commaaccents (instead of cedillas) etc. In such cases, me (and other localizators) offer the only solution.

I think *localization* must be a term that has come to have a meaning I�m not familiar with.

And it doesn�t matter what my opinion is when it comes to legal matters, I have no legal training � when it comes to legal matters, I�m with Shakespeare.

My opinion on the ethics of the matter also probably doesn�t count for much. Since I don�t make any money by selling fonts, I could take the high road & it wouldn�t affect my pocketbook. Still, I have my two rules.

I�m a typesetter, not a type designer, which means I�m what Mr. Twardoch calls a �consumer� of type. I don�t offer my modified fonts for sale. I�ve given a couple of my modified fonts to two people; my wife and a close friend, both book designers. Even here, I made sure they had already purchased the foundry version of the font before I would give them the modified version. In fairness to them, both are honest, and would not in any case use fonts they haven�t paid for.

* * *

My categorical statement that I�ve never found a foundry release font to have the right kerning values wasn�t based solely on optical size. Just a few examples:

How about Yg? (One spelling of King Arthur�s wife, Ygraine)

How about the period that follows capital *U*? In the abbreviation of U.S.A., the period following the U is usually much closer to the �S� than the �U�. Same with the period following the �S�, its closer to the �A�.

I spend a lot of time kerning punctuation, by the way.

Or the lower-case �f� � some fonts come with the quoteright & quotedblright kerned, some do not. But there is other punctuation, like the question mark & exclamation point, parenleft, bracketleft, asterisk, etc. Or, if the designer/foundry has used Fontographer�s version of class kerning and the pair �fa� has a kern, so too is �fagrave�, but that will give a serious error.

And on & on

Charles

All--

This is what happens when you ask a simple question on this site . . . you learn a lot, more even than you thought you would.

Since I'm what most on this site would call a hobbyist, with no other reason to contemplate converting/rewrapping T1 fonts than pure convenience, there's no reason for me to stretch the bounds of ethical practice, especially considering all the work that would be involved (think of it as a happy convergence of practicality and integrity). But the discussion (technical and ethical) has been enlightening.

FYI: FSI has not (yet) packaged their material in OTF. I suspect that's coming, but considering all the steps involved and the size of their catalog, it will take them a while, I guess. --MM.

P.S.: Charles, Ygraine was Arthur's mom. His wife was Guinevere.

>> Charles, Ygraine was Arthur's mom. His wife was Guinevere. >>

Ah, you're right. I've gotten too old to rely on memory for important things like legends and proverbs.

There is a moral here: vant your galleys while you can.

Charles

Charles, right on - great stuff.

> My reading of FSI's EULA suggests there are no ethical problems with this

An EULA does not cover ethics...

> Properly made typefaces include "optical sizes"

But Adam, do you realize what you're excluding by saying that wonderful thing? :->

> How about Yg?

It's interesting -to a geek at least- to note that Minion contains a "Yq" kern (like in "Yquem") but not a "Yg" (like in "Ygrec"). Tsk tsk tsk...

But what's "vant"?

hhp

Rather than endlessly hypothesize and argue about how FSI might feel about this, the best people to ask about this are FSI. You never know, they might be working on their own OT versions and need an existing FSI customer to do QA testing for them. If you can demonstrate to them that you have paid for their product, value their rights and control, and will help them in ways that save more time than waste it, who knows what they might agree to. Try writing them a thoughtful email explaining what you're doing in detail.