TrueType Typography
A History of TrueType
Back in the late 1980s, it was clear to most of the major players in the personal computer world that scalable font technology was going to be an important part of future operating systems. Adobe was trying to get Apple and Microsoft to license its PostScript code for this purpose. This solution (which became Display PostScript) had the right features, but both companies were naturally concerned about handing control over key parts of their operating systems - not to mention millions of royalty dollars - to Adobe. (DPS was too slow for the target machines anyway, ending up only on Steve Jobs' NeXT computer.) Also, Apple was irritated that Adobe licensed PostScript to printer manufacturers who undercut Apple's own LaserWriter. So, Apple and Microsoft agreed a cross-licensing and product development deal, the fruits of which would be available to both parties: Microsoft would bring a PostScript-style graphics engine to the table (TrueImage), while Apple would create a font system even better than Adobe's...

Nothing ever came of TrueImage. It was buggy when delivered, and Microsoft and Apple realized they didn't much need it anyway. (It's clear to see who got the better deal out of the arrangement!) Application developers want to do things their own ways, not get locked into writing clones of Adobe Illustrator. [Years later, Apple (whose customers are most demanding that graphics be handled effectively) released a system much better suited to interactive graphics and typography: QuickDraw GX. But application developers have been slow to use its amazing features - perhaps because if they do and their product is a success, everyone else will copy them immediately.]

Adobe took the loss very seriously. It had failed in its bid for total control over PC font technology. Its response was twofold. In mid-1989, when they learned that Apple would not be requiring its technology, they announced a program, Adobe Type Manager, before it had even been written. About a year later, you could buy ATM to display Adobe Type 1 fonts on the Macintosh, without any help from Apple. ATM was sold cheaply, or was bundled with fonts bought from Adobe. The second bold move was to publish the Type 1 font specification. (Previously font foundries had to pay Adobe royalties to create Type 1 fonts. The font data was encrypted, and it was not possible to retrieve the control points - except deep inside an Adobe PostScript engine. But now anyone could write a Type 1 font editor.) In fact Adobe had this move forced upon them, since the TrueType specification had been made public, and because Bitstream cracked the Type 1 format anyway. Bitstream soon released hundreds of Type 1 fonts and a fast ATM clone, FaceLift.

Apple had been developing what was to become TrueType from late 1987. At that time there were many competing font scaling technologies, and several would have been suitable for the Macintosh. It was by no means certain, according to lead engineer Sampo Kaasila, that Apple would adopt TrueType. In the end though, it proved itself on performance and rendering quality (at high and low resolution) against the others. Kaasila completed his work on TrueType, though it didn't yet have that name, in August 1989. The following month Apple and Microsoft announced their strategic alliance against Adobe, where Apple would do the font system, Microsoft the printing engine. Apple released TrueType to the world in March 1991 - the core engine in much the same form that Kaasila left it back in 1989. This first customer version was an 80K add-on to System 6.0, available until recently on the Apple website! The system needed fonts of course, and the first TrueType fonts - Times Roman, Helvetica and Courier - were great examples of what could be done with the technology. TrueType has been built into the Mac operating system ever since.

Microsoft introduced TrueType into Windows with version 3.1 in early 1992. Working with Monotype, they had created the superb core set of fonts - TrueType versions of Times New Roman, Arial and Courier. These fonts showed, just as Apple's TrueTypes had, that scalable fonts could generate bitmaps virtually as though each size had been designed by hand.

With non-fancy fonts, the system generally worked well. However, since Windows 3.1 had to run on machines with slow 16-bit 286 processors, the TrueType system had to be reconfigured as a 16-bit implementation of Kaasila's fundamentally 32-bit architecture. (While a 32-bit simulation would have worked, some thought that would have been too slow on these base machines.) Memory allocation worked only "most of the time". At large sizes, the whole system would became less precise, although that wasn't the worst thing. In some fonts that were made faithfully to the specification, complex characters would sometimes fail to display at all, or they'd appear on screen but not on the printer, mystifying and infuriating font users the world over.

Things were bad for the font developer too. Complex glyphs had to be simplified. Hinting code had to work around the 16-bit limitations. Fonts that worked fine on the Mac, developed with the TrueType hinting tools (all Mac programs, mostly written by Kaasila), would fail in Windows. TrueType hinting was hard enough already without this to contend with, and several font foundries abandoned earlier commitments to release their type libraries in TrueType format. (Even now, type foundries have many typefaces just waiting for a big customer to say "I'll have 10,000 licenses please" to justify the man-years of TrueType engineering.) So it was that the main type foundries left the huge market for TrueType fonts on Windows wide open. The market was soon flooded with cheap fonts scanned or stealthily converted from other people's work - mainly bug-ridden fonts of dubious ethical quality, with wobbly outlines and useless hinting. The perceived quality of TrueType as a whole suffered from these abominations, at least on the Windows platform, and for years TrueType itself was much maligned by type professionals.

Only in August 1995, with the release of Windows 95, did Microsoft's TrueType engine become 32-bit, complete and reliable. Indeed, it now features greyscale rasterization (anti-aliasing), enhancing on-screen text substantially. Microsoft have commissioned some superb new TrueType fonts that they give out free on their website. The rehabilitation of TrueType is well underway.

by Laurence Penney

Some comments from Greg Hitchcock

Greg Hitchcock is one of the engineers at Microsoft who adapted the Apple code to Windows, and has continued updating the Microsoft system ever since. He had a couple of comments on my article above:

I disagree with your assessment of the 16 bit TrueType rasterizer in Windows 3.1. Technically you are correct. It was a terrible mistake to move the code to 16 bits, and I totally opposed the move. Emulating 32 bits in the 16 bit world would not have had that serious a performance impact. Most of the problems we ran into with the 16 bit rasterizer were scan conversion problems, and this was solved by actually having two scan converters in Windows 3.1, a 16 and 32 bit version. The 32 bit one kicked in when the 16 bit would have problems. Most of the problems that font developers encountered with 16 bit were not "show stoppers," almost all were able to come up with workarounds. The rasterizer that shipped with Windows NT 3.1, which followed later, was a 32 bit rasterizer. And the rasterizer that shipped with Windows 95 was a major rewrite, that was not too visible on the outside (except for font smoothing) but had major architectural changes internally. (As an aside, these architectural changes have allowed all the additions we made to the rasterizer a breeze, as well as porting to other processors.)

The technical issues aside, the user impact of the TrueType rasterizer in Windows 3.1 was in my opinion, one of the biggest impacts on the most number of people that type has ever had. I'm not one for hyperbole, and only make this assessment in hindsight. At the time we just thought this was some cool stuff, we didn't realize it would have any impact. Here are my reasonings for this statement.

  1. We went to a lot of effort in the development of the core fonts, as Chuck mentioned in the article in your page. We spent just under 2 years tuning and re-tuning the quality of the core fonts. When we made the first public showing of the core fonts, at the Monotype conference of 1991 in NYC, there was consensus of the type industry that these were the best screen fonts ever made. That is obviously a matter for debate, I've worked so closely with them over the years that I see flaws all over with them, and I'm disappointed with our internal hinting code - not the results, but the way we got the results.

  2. There was an explosion of font development for Windows. All the major foundries put out font packs. The first font pack that Microsoft produced did unbelievably well and shocked all the executives at Microsoft.

  3. Windows became a viable platform for publishing. In the Windows 3.0 and earlier days, any printer font you wanted got mapped to Helv or Tms Rmn on the screen - courtesy of the "Windows 3.0 Font Mismapper."

  4. Finally, the press picked up on this. About 6 months before Windows 3.1 shipped, our group made a serious effort to educate the press on the power of TrueType and the quality of fonts that can be produced. We had many demos that compared our core fonts with the ATM core fonts. This really helped to stir up the "font wars." We initially didn't see much response to our effort - but once Windows 3.1 shipped, there were tons of articles praising TrueType in Windows 3.1. Interesting enough, and somewhat unfortunately, the press gave Microsoft credit for TrueType, instead of Apple. This had nothing to due with us, except for the fact that we were actively evangelizing TrueType, and Apple said almost nothing.

All of this hype for TrueType that Windows 3.1 generated did have a big downside. TrueType fonts are very difficult to hint properly and get good results, but many people wanted to get into the TrueType game. There was a great reliance on autohinters built into font tools. This in my opinion led to the great disparity of quality seen in TrueType fonts.

The TrueType Development Team at Apple

Kathryn Weisberg: project lead, instruction editing.
Sampo Kaasila: coding - interpreter, instructions, scan conversion, glue code.
Mike Reed: coding font editor, glue code.
Charlton Lui: coding font manager, QuickDraw mods, glue code and hash buffer.
Richard Becker: coding translator from URW VS format, outline font data protection scheme.
Lee Collins & Dave Opstad: coding international data structures.
Jim Gable: hardware product management.
John Harvey: developer technical support.
Dianne Patterson: Inside Macintosh and format documentation development.
Andy Yarborough & Pam Martin: system and printer software quality and test.

Further Reading

A Brief History of TrueType
Microsoft's article on TrueType history.

A talk with Sampo Kaasila
A chat with the guy who, from 1987 to 1989, created TrueType.

What is the history of TrueType?
An article by David K. Every.

TrueType vs. Type 1: What's the difference?
Thomas W. Phinney's article discusses the reasons that Apple developed TrueType, and how it compares with its oldest adversary.

Charles Bigelow writes... Times (New) Roman was used as the litmus test by the competing font technologies (from Apple, Adobe, Imagen and Sun) in the late 1980's. "It was perhaps the supreme era of the Digital Fontologist".

Font Wars
Chapter 11 in Accidental Empires, the history of the microcomputer by Robert X. Cringely of InfoWorld magazine (Addison-Wesley 1992).

TYPE*chimérique | TrueType Typography | TYPE*links