Hmm. Good question. I used to be vehemently anti-TTF, but that is the "pre-press in the 90s, TTF fonts not downloading to the RIPs" talking.

I would recommend keeping the TT versions (but obviously back up the PS versions someplace). Microsoft have already announced that Type 1 fonts will not be supported by their new Avalon graphics system, and it is only a matter of time before support for the format becomes an issue.

There are still some PS RIPs out there that have problems with TTFs, but if you encounter one it is usually a sign that the pre-press bureau or print house hasn't updated their systems for some time, which I would take as a warning about using their services.

For a more complete explanation on the phase-out of PostScript Type 1 fonts see - http://blogs.adobe.com/typblography/2005/10/phasing_out_typ.html

Si

Seems from Thomas's blog that CFF-flavoured OpenType WILL still be supported.

My initial reaction to phasing out of Type 1 was, "Dammit! Does that mean that my CFF-flavoured OpenType fonts should actually be TrueType-flavoured OpenType fonts?" The answer seems to be "no", but I would love to hear from Si and Thomas - and anyone at Apple who might be lurking on this site - what the best policy is for small foundries. CFF or TTF?

From an operating system perspective, supporting one rasteriser is always going to be more attractive than supporting two rasterisers, and having one of those rasterisers provided as a black box by another company (Adobe) probably makes a lot of MS and Apple engineers quite nervous. So the internal push for OS developers would favour a single flavour of OpenType, namely TrueType. On the other hand, there does seems to be acknowledgement of the reality of font distribution and the inherited preference for PS fonts in the professional graphics market, so I think these companies understand that supporting CFF as well as TTF is important.

But my uncerstanding is that there was a period when it wasn't certain that there was going to be any PS rasteriser in Avalon, which might give you pause for thought. About 95% of the fonts I've made in the past ten years are TTFs; in fact, Adobe are currently the only people for whom we make CFF fonts. But then I'm making custom fonts, not retail fonts for the professional graphics market.

If I had to make a recommendation on purely technical grounds -- based on robustness of the format, availability of thorough QA tools, ability to get at the internals of the tables, and superior system and application support --, I would have no hesitation in encouraging foundries to favour TrueType over CFF. The only downside to TrueType is the relatively complicated hinting model and the additional work this requires to get decent results.

I think, the original question is from a users perspective, "Which format should I collect?"

I think: "It doesn't make any difference." Long after the T1 format is gone-for-good, if such eclipse does in fact occur, OS and devices are still going to handle it. You don't need to have two rasterizers to deal with both, or at least that's what MS and Apple have been saying for some time, (auto-hinting converters with dual overhead OT generators, you know). And if anyone's nervous about the black-box Adobe rasterizer, it must be a slow-burning nerve...

I think from a founders perspective, it's best to design and produce in the format you expect the market to use the font in most readily and easily. Most print clients want PS first, and everything else at the end. Most online, or on screen font clients, want TT and some get everything else too.

Will TTFs made for future Windows versions require hinting at all?

I remember Mr Connare's Magpie sample, accompanied by the comment: "Below is a link to view an image of Magpie (UNHINTED) using ClearType in the Microsoft Reader for Windows."

To my eyes this looks quite good.

The importance of hinting is steadily decreasing, and will eventually approach zero. Aside from hinting, the technical differences between OpenType TT and OpenType CFF are also not that significant - the encoding and fancy contextual features are the same across the two, the only real difference is the representation of the outlines.

IMNSHO, a reasonable way to deliver fonts for Vista and future systems is to forego hints altogether, and tune the "gasp" table to enable y-direction grayscaling. See this thread for discussion and examples of the latter, a new feature for ClearType. Btw, I wasn't able to find the documentation for the new gasp flags for controlling y-direction grayscaling. Maybe one of the ClearType experts here can point me in the right direction.

Basically, the effect of this approach is that contrast will be slightly softened compared to well-hinted fonts, but you're pretty much guaranteed no distortion or artifacts, and increasing resolution will lessen the importance of contrast over time. Of course, all this depends on the nature of the font. If your goal is good screen rendering of large blocks of text, as it is for MS's new ClearType font collection, then you probably do want to pay attention to the hinting. For display fonts, it shouldn't matter much at all.

As far as the expectation for future support, I think both TT and CFF are going to be around a long time. All the new Microsoft stuff (XAML, XPS, WPF, if you can keep track of all the alphabet stew) supports both TT and CFF, and of course anything that deals with PDF has to as well. The code for unhinted TT and CFF rendering is pretty simple.

So I would say that the choice between TT and CFF boils down to which tools you're most comfortable using.

As a graphic designer or end-user, I actually prefer OpenType PS (.otf) fonts over OpenType TT (.ttf) fonts.

From the production perspective, both formats have their problems.

The B-spline outline representation is just stupid. Everyone, including John Hudson, designs using cubic Beziers, because they are the standard design tool. The subsequent conversion to quadratic B-splines is a tedious operation that you don't have complete control over (various tools will convert the same Bezier outlines into various B-spline representations). So there is a certain level of uncertainty and increased QA that needs to go into a production of a OT TT font.

On the other hand, the internal structure of an OT PS font is just absurd. The "CFF " table that holds the Bezier outlines of the OT PS font has a structure that is totally alien to the rest of the OT tables. In OT fonts (both PS and TT), the glyph names are stored in the "post" table, the encoding in the "cmap" table and the font names and copyright strings in the "name" table. In OT TT, the "glyf" table only holds the glyph descriptions. But in OT PS, the "CFF " table, holds the glyph descriptions as well as... glyph names, encoding and a set of font names. But the glyph names, the encoding and the font names need also to be stored in "post", "cmap" and "name", where they belong.

The OpenType specification advises software developers that only the information from the latter tables is relevant and should be relied upon but since there is no guarantee that some piece of software does not accidentally rely on the "other" information, the font developer (or the software he uses) needs to make sure that the redundant information is kept in sync. Although that information shouldn't actually be there in the first place.

The reason for the whole thing is that CFF -- not the entire OpenType PS font, but *the "CFF" chunk itself -- was originally conceived by Adobe as a self-contained font format. It was a replacement for Type 1 that was used in PDF files -- instead of Type 1. So whatever was in Type 1 (glyph definitions, font names, glyph names, encodings), ended up in CFF. At some point, Microsoft and Adobe started talking and agreed to "merge" the TrueType and PostScript font formats. Since Adobe already had a spec for this new compact chunk of data called CFF, they decided, let's put the CFF chunk directly into the OpenType file structure. So what you end up with is actually a complete font inside of a font. In fact, when an OpenType PS font is being printed to a PostScript printer, the entire OpenType "wrapper" gets thrown away and only the CFF chunk remains. It gets converted to Type 1 and put into the .ps data stream. So yes, Type 1 lives *and will live*. Because PostScript doesn't know anything about OpenType PS: any OpenType PS font travels into a PostScript device as a plain Type 1.

Type 1 fonts had also their problems, the biggest being that the kerning information was not part of the actual file but had to be stored externally (.afm, .pfm, FOND, you name it). Also, it was not really set up well for complex scripts and generally, non-Western typography. On the other hand, its beauty is in simplicity. Type 1 is very simple to create, edit and debug. In fact, one can write Type 1 source code in a text editor and the compile it using very simple code.

But since Type 1 lacks advanced typographic features that would make it usable for writing systems such as Arabic or Devanagari, OpenType is a more mature solution. When it comes to the difference between OT PS and OT TT, I can tell you that creating and debugging OT TT is much easier. Several different rasterizers and toolkits exist for TrueType / OpenType TT while for OpenType PS, one is practically limited to just Adobe code.

Whenever a problem occurs (e.g. you print a text set in a font and you get .notdef boxes on the printer), with OpenType TT, it's relatively easy to track it down. You build the font using different tools and check the font on different rasterizers, and you'll be able to quickly nail down whether the problem is in the font, the software used to create it or perhaps in the rasterizer. With OpenType PS, it's more problematic. I spent nights and nights trying to solve several particular problems once, and it turned out that the fonts were fine but the code used to create it (the Adobe FDK for OpenType) wasn't.

Fortunately, the Adobe type people are very helpful and responsive when it comes to such problems, and are usually able to help one in debugging. Also, the quality of the code gets better and better. Since the Adobe FDK for OpenType is used as a library in FontLab products, many people have built OpenType PS fonts over the last years, so the "infant illnesses" of the code have already been fixed.

So these days, when I have a choice which font format to pick, I usually pick OpenType. If the font has been hand-hinted, I pick OpenType TT, otherwise I prefer OpenType PS.

A.

BTW, Leslie Cabarga's book "Learn FontLab Fast" includes a contribution of mine: a handy comprehensive pro-and-contra comparison between all major font formats from the user's perspective. You may find it useful when picking a particular font format for a particular project. It's on pages 81-82 of the book.

A.

I would go with John Hudson's recommendation also, for several very specific reasons.

Consider that the most recent version of FontLab makes no distinction between TTF-flavored Opentype fonts and Truetype fonts: both have the .ttf extension. Second, forget the myth that .ttf fonts cause problems printing. I have been involved with the printing trades for almost forty years and with the desktop version of same for fifteen, and I can tell you categorically that the anti-TTF propaganda is a load-o-cr*p, not unlike the Macs-are-inherently-better-than-PCs hooey that is pitched in the Mac community.

I have used both Postscript and Trueteype fonts to output to imagesetters. It is true that there were some problems with .ttf fonts and level 1 RIPs, but we are way beyond that point now. I am also not entirely convinced that the perpetuation of that myth wasn't fueled by Adobe (following the principle of NIH -- not invented here). It is curious that, NOW, with the advent of OpenType, developed jointly by Adobe and Microsoft (versus Truetype, which was developed jointly by Apple and Microsoft), Adobe is OK with .ttf. And I have used both Macs and PCs in a serious production environment, and neither machine has an advantage over the other (other than the fact that PCs are a LOT less expensive to purchase or lease).

So, bottom line: keep the Truetype, sell the Postscript to unsuspecting suckers on eBay.

> Consider that the most recent version of FontLab
> makes no distinction between TTF-flavored Opentype
> fonts and Truetype fonts: both have the .ttf extension.

The Adobe-Microsoft-owned OpenType specification replaced the Microsoft-owned TrueType v1.66 specification. OpenType is a Microsoft trademark and TrueType is an Apple trademark. This means that essentially, from the point of view of Microsoft or Adobe, TrueType doesn't exist anymore.

Microsoft Windows displays a green OpenType icon for .ttf files that include a DSIG (digital signature) table. This is only a cosmetic distinction from the fonts with a blue TrueType icon. You can have .ttf files with a blue TrueType icon that include OpenType Layout tables and .ttf files with a green OpenType icon that include none. OpenType Layout tables are not a mandatory element of OpenType fonts.

In short, one could write OpenType >= TrueType.

A.

"The importance of hinting is steadily decreasing, and will eventually approach zero."

I have heard other people say this too, including some Microsofties. As far as I know, this is bull, or it's a round-about way of saying, "you'll just have to edit bitmaps for low resolution, we ain't spendin' the dime or the time on either fonts or technology to make this work". The reason I know I'm telling the truth is because a. I've seen the effects of uncontrolled multibit scan conversion on small sizes and it looks like mud. b. there are no magical resolution increases that will transport all of the 10-12 point screen fonts in the world to a 27 ppm minimum. and c. because I know the people who are saying this have abandoned the original screen font quality of the mid-80' to mid-90's in favor of a lower priced lower cost alternative, look Ma no hints!

"This means that essentially, from the point of view of Microsoft or Adobe, TrueType doesn’t exist anymore."

Really? or is this just the view of the techocrats and typocrats, while the lawyers have other opinions?

"The B-spline outline representation is just stupid."

Wow, Adam, you've designed fonts with a B-spline outline representation?

it was not this Microsoftie. I agree with David, hinting is still very important for screen rendering, and also about the “mud” comment. I may have talked about a lighter approach to hinting, i.e., controlling the main y direction features in the font, x-height, cap height, etc as well as hints to control serifs and to keep certain glyphs from closing up at smaller sizes. This is still very much needed, and without too much effort can result in very sharp type on the screen that faithfully represents the original outline.

> “The B-spline outline representation is just stupid.”
> Wow, Adam, you’ve designed fonts with a B-spline
> outline representation?

I tried but failed. I know some (very few) people who do. I'm not saying that it's conceptually inferior -- there are even other curve representations that people use, such as the on-curve-point-only Ikarus geometry. In fact, I must say that I'm more comfortable with the on-curve-points-only Ikarus system than the TrueType B-splines.

My point is that practically, it's stupid because the PostScript drawing model is standard in the publishing industry. Most people draw in PostScript curves. Even that FontLab has allowed drawing in native TrueType B-splines since version 3.0, I've yet to see the big run for it. In the last two years, I don't remember the TrueType drawing mode of FontLab/FontLab Studio being raised even once on forums or tech support submissions.

So the reality is that most people draw in PostScript curves. Then they have a choice of either sticking with their carefully drawn curves by going into Type 1 or OpenType PS, or to rely on an approximative conversion by going to OpenType TT / TrueType.

To sum up: I don't see a clear favorite. Each of the formats has its drawbacks. Type 1 does not support multilingual typography properly, OpenType TT / TrueType uses an unpractical curve system that is not native to most designers, and OpenType PS is still a bit of a double-shrink-wrapped black box, which, if something goes wrong, nobody really knows what and why. Of course, each format has its advantages as well: Type 1 being rock-solid, simple and splendidly debugged over the last decades, OpenType TT / TrueType being cleanly designed so to large extent, it's backwards-compatible to Windows 3.1, and OpenType PS being a kind of a blend of both worlds.
A.

To take this a bit OT: who (or what) is going to rasterize my library of PS fonts in future?

It's still impossible to buy many fonts as Opentype, of any flavour; I'm reluctant to invest in TT fonts when buying type for professional use, and it's often not an option anyway as many foundries only offer PS. Do Adobe, Apple or Microsoft intend to offer a tool to wrap PS fonts as PS-flavoured OT?

R

: ) Glad to hear Mike still believes. Thing is, if we HAD higher resloutions on screens, peopled use smaller sizes, and we'd still need to hint...or should, unless bitmaps are on the upswing again... Small sizes need ALL the hints, especially side bearing and counter controls, (which would only leave stems loose, which of course is impossible)...so...

"To sum up: I don’t see a clear favorite. " But doesn't TT, with it's "lossless" conversion to T1 and the reverse being "lossy" make the answer obvious?

All of this techie stuff makes my head spin, but I do know that PS is now and will be used in the future. If you don't like it, convert it! Use TransType Pro, or the other PC thing, whatever it is called. Seems like everyone here must have either worked at or owned an output service shop, as they used to be called. I remember in the dim past trouble with TT fonts in rasterizers not rendering correctly or going to Courier too, but don't hear much about that these days. I take Adam's points about the differences between the formats for the practical side of things, My head hurts even more now. Phase out of Type 1 - now that is scary.
Jordan

I was hoping that someone would step up and state unequivocally that either TTF or CFF is the "right" format for all future type development and publication. I should have known that there is no right answer, only informed opinions. In the meantime I will continue to develop CFF-flavoured fonts.

I would be curious to know what my colleagues are doing in their development process; whether they plan to offer TTF- or CFF-flavoured OpenType fonts.

Here's what I see so far in this thread, and by visiting the sites of foundries which offer OpenType fonts The first list is clear and marked with "v" prefix when verified:

v Adobe ___ majority: CFF / minority: TTF
v Dalton Maag ___ TTF
_ Emigre ___ CFF
v Linotype ___ CFF
v Karsten Luecke Type Foundry ___ CFF
v Monotype ___ CFF
v Orange Italic ___ CFF
_ Porchez Typofonderie ___ CFF
v Shinntype ___ CFF
v Mark Simonson ___ CFF
v Jeremy Tankard ___ CFF
v Terminal Design ___ CFF (Rawlinson, Giacomo) / TTF (ClearviewHwy)
v Thirstype ___ CFF
v Tiro ___ retail (minority): CFF / custom (majority): TTF
v Type Initiative ___ CFF
v Village ___ CFF
v Underware ___ TTF (Auto) / CFF (Bello)

The second list is foundries which do offer OTF fonts, but whose flavouring is unclear or unstated:

_ Font Bureau (David Berlow) ___ TTF (?)
_ House Industries ___ ?
_ Lucas Fonts ___ ?
_ OurType ___ ?
_ P22 ___ ?
_ Process ___ CFF (?)

I invite everyone who is releasing OpenType format fonts to throw in to this list; I will update this post as information comes in.

Thanks, c

UPDATED 2006.01.10 / 13.07 EDT

On most of the issues being discussed here, it seems the consensus has settled on The Truth, so I'm not going to re-hash those things. I have just a few niggly comments:

Adam wrote any OpenType PS font travels into a PostScript device as a plain Type 1.

This is technically true as long as one keeps that word "travels" in there. However, it is interesting to note that fonts resident on PostScript RIPs may be CFF!

Old Nick wrote It is curious that, NOW, with the advent of OpenType, developed jointly by Adobe and Microsoft (versus Truetype, which was developed jointly by Apple and Microsoft), Adobe is OK with .ttf.

It's true that during the font wars of the early 90s, both sides denigrated the other's format. On the other hand, it's also true that "now" = 1996-97. So the period in which "Adobe is OK with .ttf" has now been longer than the period when Adobe was more begrudging in its support of TrueType. But even in the early 1990s Adobe was doing a lot of work to make TrueType function well with Adobe applications and PostScript (including licensing the TrueType rasterizer and building it into PostScript).

rjohnston asked: Do Adobe, Apple or Microsoft intend to offer a tool to wrap PS fonts as PS-flavoured OT?

Adobe has made its OpenType font development tools available at no charge for quite a few years now. However, most folks will be best off using other tools - some of which incorporate Adobe source code, such as FontLab and DTL FontMaster. These can all take in Type 1 fonts and generate OpenType CFF (PS-flavored), but even they require a fair bit of setup and understanding.

Chester's chart confused me a bit. Adobe offers 12 OpenType TT fonts and about 3000 OpenType CFF fonts. Currently, we continue to develop and release OT CFF.

For what it's worth, I gather that Linotype and Monotype are doing primarily OT CFF. I wouldn't necessarily take that as an endorsement of technical superiority, but likely a combination of where they have the best quality outlines and what their markets prefer. The professional publishing market seems to have a strong preference for CFF, even if there is no technical reality behind that.

Cheers,

T

Thomas, thank you for the post, and for clarifying the situation for me. I do feel stupid, but when it comes to this subject, I would rather feel stupid and ask than NOT feel stupid and change course based upon a misapprehension. There is a lot of confusion in the marketplace about OpenType, which most people believe to be a unique and monolithic format, not two different formats wrapped in the same paper. I hope that our industry can find ways to inform and educate our customers in a "top-level" way, instead of burying this important information.

Chester, I think Tiro's entry on your chart should read majority TTF, minority CFF. What we make is very much in response to what our clients request. The only vaguely retail family on development at the moment is Ross' Plantagenet Novus, which I believe will be CFF. We've made CFF fonts for Adobe recently, as noted above. All other clients want TTF.

I am canvassing some foundries via email and am updating the table of who offers what. I hope that this table will be more and more complete as we go along, and that other people will start to fill in any holes. Perhaps when it's more complete, it can go into the WIKI. In the meantime, I'm posting this note in order to keep this node somewhere nearer the top of the tracker.

I have released one OT font, in CFF format, which is what I will continue to use.

I am reluctant to start releasing in two OT formats, as I had hoped that OT would make things simpler. At the moment, though, due to market dictates, I am contmplating releasing my next typeface in FOUR formats, two versions of OT, plus "old" .ttf and type 1.

There would be TWO versions of .ttf, one for OpenType where small caps are included in the base font, and another where there is a separate "Expert" SC&OTF font.

This appears to make no sense, and I can see why it would cause Adobe to go all-OT.

But when I look at my font sales, I see that I *might* be foolish to forgo releasing a new typeface in a "legacy" format that is still alive and kicking for a significant market segment.

And then, there might be even more versions of the same typeface, "Standard" and "Pro" versions of OT.

I am also pondering what to do with Cyrillic and Greek support. It's all very well for non-foundries (yeah, browse their websites and see*) like Microsoft and Adobe to cram everything into massive Pro fonts and, if not distributing by bundling, flog them for peanuts, but that may not represent a very good ROI for the indie guy.

But I suspect that it IS actually the way to go, and that there are massive multi-user corporate licensing deals waiting for that kind of monster font.

I also suspect that many indie foundries are developing such product right now, so my prediction is -- 2006: Year of the monster fonts.

*I'm not undervaluing their contribution to typography, and its importance, but they do: wouldn't it be something to see MS and Adobe flaunting their typographic prowess on their home pages?!

Nick, I totally agree with you. With a monster font containing small caps, several numeral sets, and supporting several ISO standard character sets, a 4-font OpenType family could easily be generated in hundreds of instances. Who has the time to manage all of that stuff? and why do it? Why step back in time? You might as well release your type in cold metal and wood and dry transfer sheets too. I say: make the legacy formats for those who request them, otherwise encourage users to get on board with this great new format. Users don't have to access the small caps or the alternate numeral sets or the additional accents and glyphs, but pretty soon the old formats will be like leaded petrol; only needed by those whose engines are antiquated.
PS; I have added you to the table. Thanks.

>You might as well release your type in cold metal and wood and dry transfer sheets too.

LOL!

I can verify that my OT fonts are all CFF and I intend to continue releasing in that format for the most part.

It’s true that during the font wars of the early 90s, both sides denigrated the other’s format. On the other hand, it’s also true that “now” = 1996-97. So the period in which “Adobe is OK with .ttf” has now been longer than the period when Adobe was more begrudging in its support of TrueType.

With all due respect, Thomas, mud has a way of sticking long after it has ceased to be thrown. If the "fact" of Adobe's (albeit grudging, perhaps?) acceptance of Truetype were universally known, we wouldn't be having this present discussion, would we? Six of one, half a dozen of the other; half full, half empty; the check's in the mail; and some of my best friends use Truetype...

As far as I could ever tell, most of the mud being slung was not from Adobe, but from end users, who were (and are) far more conservative than software companies. That was certainly my impression at the time, long before I came to Adobe in 1997.

T

BTW, all the Emigre OT fonts I have seen to date were CFF.

T

Dalton Maag release all their OTs in TTF format. Like John H, most of our work is custom related, with many corporate clients still in the world of NT4, and even OS/2. Also, all our OTs support a Latin A Extended character set, and from what I understand, there are (or were) some issues with CFF fonts in that respect. Furthermore, supporting TTF will give you backward compatibility on larger charactersets with Windows systems and Office apps because it's all proper Unicode.
And we have the option of applying high quality hints if we want to. As far as I am concerned I sell OT, not ttf or cff. It's OpenType, who cares what outlines are behind it as long as it works.

Bruno Maag
Dalton Maag Ltd

I am about to purchase type for use in Windows. It is for a book being developed with Adobe InDesign and Photoshop. My choices are TrueType or Postscript Type 1. The type may be released as OpenType in the future, but there is nothing definite about this.

My sense, reading this thread, is that I should purchase the TrueType version. Does anyone disagree with this?

Thanks.

Well, it depends. If it's for print, your chances are actually better with PS, because
all too often TT fonts are converted "blindly "from PS, and can have problems.

Go with TT if you have faith in the foundry's abilities to
output good TT (and especially if screen rendering matters).

hhp

Thanks. Given that the price is well in excess of US$500, I would like to think that their TrueType and PostScript are comparable, but perhaps this is something that I should discuss with them. To the extent that we are designing on screen, screen rendering matters, but not otherwise. My main concern about PostScript Type 1 is that it is being phased out.

“I have used both Macs and PCs in a serious production environment, and neither machine has an advantage over the other” except for the near-impossibility of typing any characters outside the US-ASCII symbols on your U.S. keyboard. On which platform? Take a wild guess.

If you're talking about Windows: wow, I
didn't realize Armenian was in US-ASCII...

hhp

Joe,

both Mac OS X and Windows ship with a large number of keyboard layouts which you can use to type in pretty much any language. The Mac keyboard layouts are often a bit more interesting (they are more "complete", i.e. map pretty much every Option/Alt keystrokes to more or less useful characters, while the Windows keyboard layouts are far more conservative and only use a small number of Alt keystrokes just for the very essential characters). On the other hand, Windows has a far more superior support for various writing systems of the world, and ship with more fonts for non-European languages.