Phasing out “PostScript” Type 1 fonts
Recently, I’ve spoken at a couple of conferences, and made presentations to several key customers, about Adobe phasing out sales of Type 1 (“PostScript”) fonts. This posting is adapted from my talks and presentations on the subject.
[Update 14 Oct 2007: We still haven’t stopped selling Type 1 fonts yet, although probbaly 85-90% of our sales are in OpenType. But pretty much everything else I’ve written below remains true. I sure wouldn’t buy a font in Mac or Windows Type 1 or Mac TrueType format today unless I had a very specific reason for it. – T]
Background:
Here’s a brief introduction to set the scene. This is all discussed at much greater length in my article on article on font formats.
Type 1 is the original multi-platform scalable outline font format, introduced by Adobe in the mid-80s, with the format made fully public around 1991. Because Type 1 was initially supported only by PostScript devices, the format is sometimes itself called “PostScript,” which ignores the fact that there are other font formats supported by PostScript (including TrueType), and that since 1991 it has been possible to print Type 1 fonts to pretty much any printer as well as scale them on-screen.
TrueType was developed by Apple and licensed to Microsoft, and introduced around 1991 as well. Apple and Microsoft wanted to have an outline font format that they owned, which would be WYSIWYG (what you see is what you get) on screen, and was specially optimized for the low-resolution monitors and black-and-white text imaging of the time. Plus they didn’t have to pay Adobe royalties on the technology.
After a long period of conflict between the two font formats, often referred to as the “Font wars” (cue heroic music here), Adobe and Microsoft decided to bury the hatchet. In 1996 we announced OpenType, which was intended to supercede the older formats. Taking the table-based structure of TrueType, OpenType offers several key features:
- cross-platform, single-file fonts
- up to 64K glyphs per font
- potential for extensive multi-lingual support in a single font, using a Unicode encoding
- options for advanced typographic features, whether needed for linguistic reasons or just for fine typography
As part of the same transition to OpenType, Adobe licensed its ATM rasterizer code to Apple and Microsoft at no charge, giving them the ability to offer system-level support for Type 1 and OpenType CFF (“PostScript flavored OpenType”) fonts. They started doing this in Mac OS X and Windows 2000.
We have various information about mapping from Adobe Type 1 to OpenType fonts, both for multiple master Type 1 fonts and for regular Type 1 fonts.
What’s wrong with Type 1?
Well, Type 1 does not allow for more than 256 glyphs to be encoded in a single font. The special CID flavor of Type 1 gets around this, but only by defining rigid “character collections.” Type 1 also doesn’t use a Unicode encoding. It doesn’t have typographic extensions to deal with various languages that require them and to support advanced typography. The font files are not cross-platform, and there are multiple files required to represent a single font. OpenType addresses all of these problems. So, from a technical perspective, Type 1 is a thoroughly obsolete format. However, that doesn’t mean we’ll stop supporting it tomorrow or anything (see below).
Phasing out Type 1
Adobe stopped developing new Type 1 fonts in 1999, and introduced its first OpenType fonts in 2000. As we converted our type library to OpenType, we made the corresponding Type 1 fonts much less prominent on our web site. Now that we have out whole library in OpenType, people really have to go out of their way to get Type 1 fonts. This has been true for several years now.
As a side note, we stopped selling the multiple master flavor of Type 1 fonts several years ago, and stopped active tech support for multiple master fonts at the end of 2005. However, MM fonts still work in Adobe applications, though I wouldn’t want to guess on the likelihood of bug fixes in this area going forwards.
In any case, the overwhelming majority of Adobe font sales today are in the OpenType format, and our Type 1 sales continue to decrease. Nonetheless, we are well aware that there is still a very large installed base of Type 1 fonts.
Editorial break:
People expect their fonts to continue to work forever. But when thinking about Type 1 eventually going away, it’s worth keeping in mind the value that customers have gotten from their Type 1 fonts over the years. What other software do you have that you bought in the late 1980s that still works today? It’s amazing that these things have had such a long lifespan. We still get occasional tech support calls for an issue that we fixed in two Type 1 font families in 1993!
Of course, a lot of people don’t think of fonts as software, but that’s really what they are: little plug-ins to your system software. The operating system vendors now have the Type 1 rasterizer code in their OSes, and it’s up to them whether they support it moving forwards.
THE FUTURE
Avalon:
What is Avalon? Avalon, the code name for what is now called Windows Presentation Foundation, is Microsoft’s next-generation graphics and text system, which will ship as part of Longhorn (Windows Vista), and at the same time as Windows Vista. But it will also be made available for Windows XP. Most news reports suggest Avalon and Vista will ship in the second half of 2006.
Technical specs and a public beta version of Avalon are available now, and it does not support Type 1. It will, however, have a native rasterizer for OpenType CFF fonts – that is, PostScript flavored OpenType fonts.
So, what does it mean that Avalon won’t support Type 1? It means that if you have an application that is written to take advantage of Avalon – on either XP or Longhorn – installed Type 1 fonts will not show up in that application’s font menus. Note that no applications today use Avalon; an application has to be written specifically to take advantage of Avalon.
Now, GDI (Graphics Device Interface) is the main predecessor to Avalon, and it’s through GDI that Type 1 is supported in Windows today, and will continue to be under Windows Vista. Existing and future GDI applications installed on XP or Vista will continue to use Type 1 fonts as they always have. Also, Adobe applications that use our own shared font engine can continue to support Type 1 regardless of Avalon or other OS support.
So, when we talk about Type 1 support going away, it’s more accurate to talk about Type 1 not being supported in Avalon. After all, Avalon can be installed on XP, and Vista continues to support Type 1 in GDI, so saying that Windows Vista won’t support Type 1 is not exactly true. Now that you understand all this, you can have the fun of explaining it all to other folks, as I do. 🙂
Adobe & Our Customers:
So, looking at this situation, the main thing Adobe sees as a type foundry is if we keep on selling Type 1 fonts, starting in a year or two there will be a bunch of applications that won’t support them, at least on Windows. At the same time, we’ve been moving away from Type 1 sales for many years.
So our current plan is that no later than when Windows Vista ships (late 2006?), Adobe will stop retail licensing of Type 1 fonts for Mac and Windows. This is subject to review and change based on market conditions, but it’s our current best estimate.
Adobe recognizes the huge investment our end users have made in Type 1 fonts over the last 20 years. We want to help our customers as much as possible. This is why we have been taking the rather unusual step of getting this information out sooner rather than later. We believe that for customers licensing fonts today, unless they have a very specific reason for getting Type 1 fonts, they should license OpenType fonts as their best long-term investment for the next two decades.
We’ve also made extensive information available online, notably matching old Adobe Type 1 fonts to our corresponding OpenType fonts. The intention is to help our users with migration from Type 1 to OpenType.
Adobe expects to continue to offer technical support for our existing Type 1 fonts well after the end of sales. We also expect to continue to support Type 1 fonts in our major publishing applications for years beyond that. We currently expect that Type 1 fonts as embedded in PDFs will be supported as long as those versions of PDF are supported – which right now looks like “indefinitely.”
Because I like to keep our lawyers happy, I offer the following caveat to my comments above:
This posting includes “forward-looking statements” within the meaning of the safe harbor provisions of the United States Private Securities Litigation Reform Act of 1995. Words such as “expect,” “estimate,” “project,” “budget,” “forecast,” “anticipate,” “intend,” “plan,” “may,” “will,” “could,” “should,” “believes,” “predicts,” “potential,” “continue,” and similar expressions are intended to identify such forward-looking statements. Forward-looking statements in this presentation include, without limitation, Adobe’s future actions with regards to the Type 1 and OpenType font formats, and other matters that involve known and unknown risks, uncertainties and other factors that may cause actual results, levels of activity, performance or achievements to differ materially from results expressed or implied by this presentation. Actual results may differ materially from those contained in the forward-looking statements in this presentation. Additional information concerning risk factors is contained in Adobe’s most recently filed Forms 10-K and 10-Q.
Adobe undertakes no obligation and does not intend to update these forward-looking statements to reflect events or circumstances occurring after this presentation. You are cautioned not to place undue reliance on these forward-looking statements, which speak only as of the date of this posting. All forward-looking statements are qualified in their entirety by this cautionary statement. fnord.
22 Responses
Comments are closed.
I applaud you, Tom, for passing out this information early and often. All technologies come to the end of their lifespan sooner or later, and Type 1 has had a good run.Another factor in how long Type 1 fonts remain usable is how long related tools remain usable on current OS’s. I know Fontographer remains a common tool in Japan for authoring Type 1 gaiji fonts. Other tools like FontLab could step up to fill that niche, but when will it happen in Japan, and how long will it take?I guess Type 1 font downloaders have already passed into history, so they are no longer an issue.
Through Avalon, will Windows antialias OTF-CFF or even TTF fonts? OSX introduced this to the Mac crowd — html text in any browser looks as good as a Photoshop generated GIF.Is Microsoft stepping up the system-wide graphics quality in Longhorn/Vista, or are they leaving non-hinted fonts in pixelated limbo?By the way, Tom… are you a Mac or PC guy? (You can plead the 5th on that, if you have to.)
Thank you for this information as an Instructor this has made it easier for me to explain the difference to students
This mostly continues the corporate FUD we have been hearing for years now. It mixes misinformation – a Type 1 font file can of course contain more than 256 glyphs, while you have the same encoding difficulties with large OpenType fonts when programming PostScript – with clumsy attempts to boost sales – it is not necessary to buy an OpenType font if you have licensed its Type 1 variant as Adobe’s standard agreement allows you to convert the font. Indeed, OpenType does have some fine additional features, but that does not justify spreading FUD.
Jim,Good to “see” you again. With regards to tools, the FontLab folks now have a license to revise and distribute Fontographer, and apparently expect to make their first update around the end of the year.I agree with you that font downloaders are not overly important these days. We get few calls for them any more.Silas,Windows already offers anti-aliasing for all major outline font formats. However, in Avalon, ClearType will be on by default and will apply to both TrueType and OpenType CFF fonts.I have no idea what will happen to unhinted fonts in Avalon. I do have the impression that MS is doing a number of things for system-wide graphics and UI quality in Vista, though.As for my own OS preferences, I have switched back and forth between Mac and Windows about five times in my career so far. Currently, my main box is a Windows laptop, for two main reasons: FontLab comes out with new beta versions on Windows a year or more before they do on the Mac, and I can get 1600×1200 screen res on a Windows laptop, which allows me to do font development with a 1000-unit em square at 1-to-1 res. I admit to a great fondness for the elegance of the classic Mac OS (not so much for OS X), and I think that in general I am pretty non-partisan in the “platform wars.”Cheers,T
Anonymous,Actually, one of the major purposes of this article was to dispel FUD, not create it. I’m trying to be very clear about what it means that Avalon doesn’t support Type 1 (not as bad as it might sound, thanks to GDI still being in there), and also to be very clear about our own plans, which include supporting Type 1 for a long time to come. Font sales are not a major part of Adobe’s total revenue. The transition to OpenType is much more about enhancing our users’ workflows and typography and working well with our applications, than it is about directly making money.Yes, a Type 1 font can contain more than 256 glyphs. I didn’t say otherwise. But if they aren’t encoded, that’s a big problem for character access.As for being “a clumsy attempt to boost sales,” I wasn’t trying to boost sales at all, so I suppose if you thought I was, it would seem clumsy. I didn’t really get into the details of upgrades. It’s true that Adobe’s licensing terms allow users to convert their own Type 1 fonts to OpenType. However, such conversions will not be identical with Adobe’s own OpenType fonts, and the end results may also differ slightly depending on whether one started with Mac or Windows Type 1 fonts.However, if one needs cross-platform fonts or compatibility with Avalon, and one is on a budget, converting (on one’s own) existing Adobe Type 1 format fonts might make sense. I’d suggest TransType from the FontLab folks (http://www.fontlab.com) as probably the best tool for the purpose Note also that such converted fonts will not be eligible for technical support.Cheers,T
Well, to migrate from Type 1 to OpenType as mentioned in your article very much implied to buy OpenType equivalents of Type 1 fonts. Anyway, the fact that you posted my comment and how you replied proves your sincerity. It remains to hope that the companies involved describe the cheap way to migrate clearly in their transition notes; it could make a really convincing statement that OpenType is primarily about typography and not about making money by licensing fonts twice.
It’s an interesting question about migration paths.Unlike Adobe, most foundries’ licenses (EULAs) do not allow their end users to modify or convert the licensed fonts. So in most non Adobe cases, end-user conversion won’t even be an option.I can’t promise that we’ll promote other routes to get OpenType versions of Adobe fonts. This is for two reasons: (1) although one ought to be able to get working fonts this way, they won’t overall be as good as the in-house conversions we’ve done; (2) even if money isn’t our prime motivation in doing OpenType, it’s not like we want to actively discourage users from spending money with us.Still, it’s an interesting idea. I’ve been writing up the relevant documentation for end users about Type 1 compatibility with OpenType equivalents. I think it’s worth mentioning in that doc end-user-conversion and the differences between those and Adobe’s own OpenType fonts.Cheers,T
Perhaps you could at least supply added value with more of the OpenType versions. For example, the glyph complement PDF for Garamond 3 shows that there are an Euro sign, oldstyle figures, and some small caps, nothing which had not been in the Type 1 version. Spending $99 just for the same content in another packaging does not seem fair. If at least complete proportional figures, superscripts/subscripts, and more ligatures were added, there would be some plausible reason to buy the OpenType variant.Otherwise if making money is not the prime motivation in this migration process, tell us what else is? If a software supports OpenType/CFF/Type 2, also supporting Type 1 is possible in at most a few thousand lines of program source code; parsers for two simple file formats, some character code tables, and a converter from Type 1 to Type 2 charstrings is all what is needed. This amounts to approximately nothing in a large software project such as Avalon. So if Microsoft omits Type 1 support, it is not for technical difficulties. At the technical core, Type 1 and Type 2 charstrings are nearly equivalent. Type 1 name to Unicode mappings exist. Telling the people that something is technically obsolete or hinders their workflow simply because it uses two files instead of one cannot be it. Converting between Mac and PFB is easy. Ultimately, we cannot set type anything better with Garamond 3 OpenType than with Garamond 3 Type 1, same for many (most?) other fonts. So why are the corporations really doing this?
I’d like to respond to some of Anonymous’ questions & comments. First, the problem Microsoft faced is harder than s/he seems to think. One of the huge limitations in Type 1 is that the lack of Unicode support led to massive numbers of glyphs which claim to represent characters they actually have nothing to do with. Character code tables can’t help when the input is bogus. Glyph names are a patial solution, but again heavily abused.A second problem Microsoft faced is OS security. (Yes, they care a lot about this, whatever you think of their track record). Fonts play at a pretty low level in the OS, and although I’m unaware of any attempts to abuse this, the fact remains that fonts can easily crash the system. I know of several that have achieved this accidentally. A font file format that be verified and certified gets pretty attractive in this context.One of the big advantages offered by OpenType is an effective mechanism for correlating text (i.e. Unicode characters – nothing else is reliable in enough contexts) with image (the glyphs we read). For storage, search, transmission and repurposing – all of which are important to modern publishing – this is a crucial function.Regarding the functionality of the OpenType fonts in Adobe’s library, it’s important to notice the difference between the fonts Adobe owns, and those it licenses. A large and growing percentage of the fonts Adobe owns have significantly extended character sets, and rich sets of alternate glyphs. You can expect this trend to continue. If anyone wants to see this kind of extension in Garamond 3, s/he should take that up with Linotype. In the mean time, I’d recommend the recently-released Garamond Premier…
David,”lack of Unicode support”: This is an argument for creating new fonts in OpenType format, but it is absolutely no issue for existing fonts in Type 1 that use Adobe standard or expert encoding as these have well-defined names that even match those in CFF. So while lack of Unicode support in Type 1 is indeed a good reason to introduce OpenType, it is not a good reason to drop existing Type 1 fonts, and the latter is what I am talking about. “bogus” – no problem to create bogus glyphs in an OpenType container with bogus Unicode tables, so this is a non-argument.”OS security” – OpenType fonts did already crash Windows as well (use Google). Furthermore, the security community very much agrees that verifying the origin of programs by certificates – the mechanism used by OpenType and ActiveX – is ineffective for this purpose.Your last point is valid but since it is Adobe that wants to sell these fonts, it is their concern how they manage to add sufficient value in cooperation with their licensors that it makes sense to buy OpenType variants.
Anonymous,”lack of Unicode support”: Sure, it’s possible to create OpenType fonts with bogus Unicode encoding information. But bogus glyph names were commonplace in Type 1, to a far greater degree than bogus Unicode encodings have been so far in OpenType. It’s not that it’s impossible to screw up a Unicode encoding (of course it’s easy), but the available track record of the first 10,000 or so OpenType fonts suggests it is fairly rare.With regards to OS security, of course, one can make a bad font in any format. I for one do not regard digital signatures backed by certificates as the primary route to better system security with regards to fonts. Th4e primary route is instead to have a more robust and secure rasterizer that will protect the OS from bad fonts. We have put considerable effort into improving our existing OpenType CFF rasterizer for Avalon. Porting the code, making it more robust, and tying it in to ClearType was far from a trivial task, and in fact we have used all our qualified people full-time to do this. Even if Microsoft had been willing to put a Type 1 rasterizer in Avalon, I do not believe that we could have rewritten both rasterizers at the same time and still done it in time for the Avalon schedule. (And no, hiring contractors would not have been a viable option: for this critical work we are using only highly senior programmers, who were already expert on rasterizers in general, this rasterizer code in particular, and system security issues including those associated with rasterizers.)Cheers,T
On bogus glyph names: I am tempted to assume that this is not much of a problem with the high-quality Type 1 fonts in the Adobe Type Library or other professional fonts with Western characters, and these are the ones for which migration to OpenType will be expensive. It is hardly an issue when low-quality fonts with bogus glyph names do not render properly, or when problems with fonts for non-Western languages for which Type 1 did indeed not provide appropriate support persist.I agree that making rasterizers robust is a key step regarding security. Even more important is to move font support from the OS layer to the application layer completely. Did Microsoft do this in Avalon? If they did not, there is no point in talking about OS security anymore here as they had fundamentally violated the principle of least privilege. If they did, fonts do not challenge OS security at all.But why are we talking about rasterizers at all? A Type 1 rasterizer is technically equivalent to a CFF/Type 2 rasterizer. Now if you have a CFF/Type 2 rasterizer, all you need to rasterize Type 1 is code that converts a Type 1 font program to CFF/Type 2. This is not wizardry but a standard programming task which requires to read two specifications and to rewrite data linearly. It is also perfectly possible to achieve this with portable code. You also had about ten years to do this so no Microsoft-imposed schedule can serve as an excuse. Actually it is obvious that you already have such code in the AFDKO, and the release notes do not indicate serious problems with it within the last years.Really, the technical issues concerning Type 1 support have been solved so many times in the last fifteen years that any attempt to take them as an explanation here must fail. What remains is that Microsoft is not willing to support Type 1 in Avalon, and that you at Adobe do not “want to actively discourage users from spending money with us” in consequence of it. Honi soit qui mal y pense.NB: I repeat that regardless of the points above, it is great that you and Adobe have the honesty to publish these statements.
How will phasing out multiple master affect the use of multiple master fonts for substitution in Adobe PDFs, as used by Acrobat?
Cari’s question is a good one. Adobe stopped selling multiple master fonts some time ago, and replaced them with OpenType versions. Over time, it’s likely that some applications will stop worrying about any bugs related to handling multiple master fonts.However, the technology is still very useful in limited circumstances. For example, we use multiple master tools as part of our type design process. And inside the PDF code, the notion of multiple master substitution fonts is likely to remain a key component for a very long time to come.- thanks,David Lemon
David Lemon wrote :”NB: I repeat that regardless of the points above, it is great that you and Adobe have the honesty to publish these statements.”Henrik Holmegaard
Henrik,It wasn’t David Lemon who wrote that, but “Anonymous Coward.”Cheers,T
I wonder how one would convert their type one fonts into Open Type Fonts?Also, i wonder if there is a way for a Lowly Graphic Designer to create Open Type Fonts of their own?
Phil,You’d use a program such as FontLab to make your own OpenType fonts, or convert Type 1 fonts to OpenType.Regards,T
How are OpenType fonts accessed in PostScript? I have my own hand-coded PostScript job and an OpenType font file. How can I embed and use that Open Type font file in my hand-coded PostScript job?[Although some other reader may comment, this question might best be asked on the BlueWorld OpenType mailing list or somewhere like that. I don’t know all the details. The short answer with what I do know is, if it’s an OpenType font with TrueType outlines it needs a PostScript wrapper to be sent as “Type 42,” while if it’s got CFF outlines, you need to extract the CFF table and probably decompress that to send it as Type 1 (although PostScript devices can handle Type 2 fonts on the device, not sure you can download them that way). Anyway, you’ll want more details than that…. – T]
Thanks for the reply. I forgot to mention this in my previous post, but I searched the Adobe web site for this information. Doesn’t Adobe provide documention on how to do this? But maybe it doesn’t even make sense to ask the question if the answer is to convert back to a Type 1 font. Might as well start with a Type 1 font in that case…[I’ll see if I can get somebody who knows about this print stream stuff to comment…. – T]
Reality check, Sept 17, 2009. I bought a PostScript Type 1 font from the Adobe Type Store today. While Open Type fonts are featured, it is now easier than in the past to find and buy Type 1 from Adobe. Vista handles Type 1 fonts just fine. Accessing pi characters using Open Type’s Unicode scheme continues to be a royal pain as Unicode continues to be a step backwards in terms of usability.[We continue to sell Type 1 fonts, but these are intended only for people who are using older software that cannot support modern font formats, such as OpenType. It’s been ten years since we crafted our last new Type 1 font, and the updates and enhancements we’re making on a regular basis apply only to the current (OpenType) fonts.Also, while older (GDI) environments on Vista still support Type 1 fonts, GDI+ and WPF do not. In addition, it seems that Apple’s latest iteration of Mac OS X (10.6) has some issues with Type 1 fonts, sometimes causing reflow for documents authored in earlier versions of the OS. We expect a growing number of applications to move to environments that don’t work well for Type 1 fonts.With regard to accessing pi glyphs via OpenType’s Unicode mapping, which I assume refers to the use of PUA code points, which are used because many pi fonts have glyphs that have no appropriate homes among Unicode’s standard (non-PUA) code points. Type 1 fonts were severely limited in what code points could be used, specifically a subset of 0x00 through 0xFF (256 code points). A technique sometimes referred to as “code point poaching” was necessary, and it had the clear disadvantage in that it falsely assigned attributes to pi glyphs. And, I should point out that pi fonts are not the only culprits of this method of poaching.The intent of using PUA code points for encoding pi glyphs is to ensure that each glyph can be encoded in the most appropriate way possible, and for those not mappable from Unicode, PUA is the only choice. This allows such fonts to be used with the maximum number of applications, including those that have no GSUB feature support or otherwise require glyphs to be directly encoded. We appreciate you raising the pi font issue, and we’d like you to know that we are working on developing keyboard maps so that OpenType pi fonts can more easily be entered via a keyboard. – KL]