path: root/usergrps/uktug/baskervi/4_3/truetype.tex

\hyphenation{Micro-soft}
\title{A tutorial on TrueType}
\author[Ian D Chivers]{Ian D Chivers\\\texttt{udaa260@bay.cc.kcl.ac.uk}}
\begin{Article}
\section{Background}

For a better understanding of what True Type has to offer it is useful
to look a little closer at fonts and typography.

\subsection{Type}

It is important to realise that the design of fonts is part science, 
part art and part magic. When we look at the highest quality fonts we
find they are designed to look good at a particular point size. 
A particularly good recent example is \emph{Telephone Book}, the font used
in your BT telephone directories, and much credit must be given to 
John Miles of Banks and Miles for what has been achieved. The font
is small, yet very legible when printed on relatively poor quality
paper, and the directory is now 4 column rather than 3 column. This
represents a considerable saving on trees!

\subsection{Scalable fonts and digital type}

Let us now look at scalable fonts and digital type.
What the computer offers with the possibility of scalable 
fonts seems to go against accepted practice. The process is
\begin{enumerate}
\item     take the font outline and scale to the correct size;
\item     fill the outline with pixels, \ie generate a raster bitmap;
\item     transfer to the display device
\end{enumerate}

There are two main problems here. Firstly a scaled font does not look
good at all points sizes --- remember that fonts are designed to
look good at specific points sizes. Secondly rasterisation has inherent
problems that are particularly acute with low resolution devices where
for example we have to map a character onto a small grid of dots or 
points. For more information on the general
problems of rasterisation one should look at a book on 
computer graphics, and there is one given in the bibliography.

The following illustrates one of the problems,

{\footnotesize
\begin{verbatim}
    .    .     .     .     .     .     .     .     .
              +-------+            +-------+
    .    .    |.     .|    .     . |   .   | .     .
              |       |            |       |
    .    .    |.     .|    .     . |   .   | .     .
              |       |            |       |
    .    .    |.     .|    .     . |   .   | .     .
              |       +------------+       |
    .    .    |.     .     .     .     .   | .     .
              |                            |
    .    .    |.     .     .     .     .   | .     .
              |       +------------+       |
    .    .    |.     .|    .     . |   .   | .     .
              |       |            |       |
    .    .    |.     .|    .     . |   .   | .     .
              |       |            |       |
    .    .    |.     .|    .     . |   .   | .     .
              +-------+            +-------+
    .    .     .     .     .     .     .     .     .
\end{verbatim}
}

\noindent where we are mapping an upper case h onto a low resolution
device.  The right hand upright is only one pixel wide, compared to
two for the left. The human eye is surprisingly good at noticing
things like this. To see this more realistically Paintbrush offers the
possibility of zooming in on some text, and one can see the actual
pixels and how the characters are made up. It is quite illuminating.

Adobe were the first company to achieve widespread success with 
scalable fonts with \PS. They managed to do this by 
firstly adding special rasterisation algorithms to the 
interpreter to overcome the problems of small point sizes 
and low resolution devices, and  secondly adding hints 
to type 1 fonts to further massage the  characters to 
enhance their appearance.

\section{True Type Fonts}

One of the heavily marketed features of Windows 3.1 was the addition
of True Type fonts to Windows. As someone who has been involved in
electronic publishing for over ten years I had a high degree of
scepticism about Windows and True Type.

There are a number of advantages that True Type should bring and
some of these are:
\begin{enumerate}
\item For all Windows applications the screen display should match the printer
display, if the printer supports bit mapped mode, \PS\ or TrueType.

\item The disappearance of jagged screen fonts, with no need to build 
each font at each point size that you wanted, as the fonts are scalable.
This was the case with earlier versions of Windows, and one that will
be familiar to most people involved in the dtp field, not just under Windows.

\item In a university environment with staff and students using public access
systems with Windows installed it means that they can take their document
from one system to another and always have access to the following fonts
\begin{quote}
        Times New Roman\\
        Arial\\
        Courier New\\
        Symbol\\
        Wingdings\\
\end{quote}
and print them on a range of printers and get the same results, as the 
rasterisation takes place in the pc, rather than the printer.

If Word for Windows is installed then the following additional fonts
are also available
\begin{quote}
        American-Uncial-Normal\\
        Fences\\
        MT Extra\\
        News Gothic\\
\end{quote}     
\item The ability to send documents to bureaus without breaking the
copyright on the font. With True Type you can embed a font in
the document. The fonts are encrypted, but are easily decrypted
for screen display and printing. 
\end{enumerate}

Well, does it all work?
In the main, yes. Though as with most things that Microsoft are
involved with there are some minor problems.

The statements below are based on a system based on a 33 MHz 80486 DX
PC, with 8 Mb memory and a 1024 $\times$ 768 colour display.

Firstly let us look at the rescaling of text as we alter the window 
size or actually alter the point size of the text.
The on screen display is pretty good in most applications though
I was surprised at the variation between them. Discussions with
someone involved in Windows developments indicates that there is
more than one way to achieve what should be the same end result.

I had Adobe Type Manager installed and found little noticeable
difference between them. ATM handles the \PS\ fonts, and 
Windows handles the TT fonts.

There were differences in what packages running under Windows
offered:

\begin{quote}\footnotesize
\begin{tabular}{lP{.3\textwidth}}
Word For Windows 2.1&           4 to 127 points, 0.5 point increments\\
Excel 4.0            &          1 to 409, 1 point intervals\\
Write                 &         4 to 127, 1 point intervals\\
Designer 3.1           &        1 to 144, 1 point increments\\
Ventura 4.1.1           &       1 to 254 in 0.5 points \\
Corel Draw               &      0.7 to 2160.0, 0.1 point intervals\\
\end{tabular}
\end{quote}

The sizes of the output files when printing surprised me initially.
The following figures are for one A4 page that contains a table
with each character in the Windows character set. 


\begin{quote}\footnotesize
\begin{tabular}{P{.3\textwidth}r}
Times Roman, \PS\ printer, using internal printer font & 173,958b\\ 
TT Times New Roman, \PS\ printer, download bitmap,
  1270\,dpi & 5,648,944b\\ 
TT Times New Roman, \PS\ printer,
  download bitmap, 300\,dpi & 509,841b\\ 
TT Times New Roman, Epson printer, download bitmap, 180$\times$360\,dpi & 599,905b\\
\end{tabular}
\end{quote}


If we look at one of the graphing packages or spreadsheets then we would
have figures similar to those below:
\begin{quote}\footnotesize
\begin{tabular}{P{.3\textwidth}r}
\PS\ printer,
     A4 page, 11 $\times$ 8 inch image area, 300 $\times$ 300\,dpi        &    990000b\\
Epson 360 $\times$ 180\,dpi,     
     A4 page, 11 $\times$ 8 inch image area, 360 $\times$ 180\,dpi        &    712800b\\
\end{tabular}
\end{quote}
For a 20 page document involving text and graphics therefore we could easily
have a file size of 20 $\times$ 600000b, or about 11.5 Mb.

Printing this kind of document is therefore slow on dot matrix printers, and
not wonderfully fast on \PS\ printers.

This obviously has some fairly important ramifications in a university
environment where increasingly people are moving to Windows and Windows
based applications. Good business for the printer manufacturers!

TT works well with non-latin fonts, which did again surprise me.
With Arabic, for example, there is a separate screen font for
menus, dialog boxes etc, as the subtle angles in Arabic writing 
cannot be scaled cleanly at normal screen resolutions (72\,dpi --- ega
or 96\,dpi --- vga). However, when printed it looks simplistic. The second
Arabic font does look quite good printed and does a reasonable job
of the cursive Arabic style.

\section{So who will use TT?}

For people with dot matrix printers, and early HP Laser Jets (and there are 
a lot about) they have for the first time access to a print quality normally 
associated with \PS\ devices.

For people involved in document interchange the fact that you can legally
move documents and the fonts around is a strong plus. How many people reading
this article will have been given a \PS\ file to print and had
problems with missing fonts?

For people with access to local \PS\ facilities and typesetters
at ULCC and Oxford then I can't easily see TT displacing \PS.

An examination of the major players in the typesetting industry shows
that \PS\ is the de facto industry standard. Xerox offer for
example a complete publishing solution with their DocuTech aimed at
the professional printer, and that is based on 600\,dpi scanners and \PS.

\section{Further reading}
For those with Internet access there are a number of sources of
information on fonts and TT in particular. Microsoft make available a
mass of material at \texttt{ftp.uu.net} in the
\texttt{/vendor/microsoft/TrueType-Info} directory. Some of the files
held with descriptions are:
\begin{itemize}
\item \texttt{ttspec1.zip}, \texttt{ttspec2.zip},
  \texttt{ttspec3.zip}: The complete TT specification.  About 400
  pages if my memory serves correctly.
\item \texttt{tt-talk.zip}: Contains three papers, one on digital
  type, one on linear versus non linear scaling and the third on the
  Lucida family of fonts.
\end{itemize}
There is also a very good FAQ on fonts that looks more generally on
fonts with a good coverage of a variety of hardware and software
platforms. Good start for a general overview of what is available.

\end{Article}
\endinput
Bibliography

Typefaces for desktop publishing --- a user guide, Black, Architecture
Design and Technology Press.

Good introduction with very useful bibliography. Scholarly coverage
rather than the shallow approach found in the computer comics.

Computer Graphics, Principles and Practice, Foley, Van Dam, Feiner, Hughes.
Addison Wesley.

Very good coverage of most aspects of computer graphics. The chapters
on rasterisation are worth a read to get a wider perspective on the
problems inherent in scalable fonts and 
display on low resolution devices.

Visual and Technical Aspects of Type, Roger D. Hersch
(Ed) CUP.
Collection of papers by professionals from a range of disciplines.
The 3 main sections are

Letterforms: The Basics
Digital Standards and Algorithms
Typographic Research

Based on lectures held at the first European summer school
in Digital Typography held at Lausanne, 1991.

Designing Instructional Text, James Hartley, Kogan Page.

Good practical guidelines with coverage of text, illustrations,
tables, graphs, diagrams, charts, symbols, forms, lists
etc. 

Windows 3.1 Secrets, Livingston, Info World. 

Actually contains some information about True Type unlike the
documentation that Microsoft provide. Effectively the technical 
manual that Microsoft should have provided with Windows.

Raster Imaging and Digital Typography II, Morris and Andre
(Eds), CUP.

Collection of papers from Raster Imaging and Digital
Typography 1991, held in Boston MA.
Coverage of
Outlines
Greyscale
Halftones
Non-European type
Rendering

Publishing from the Desktop, Seybold and Dressler, Bantam.

One of the better background books on dtp. Provides a good foundation 
to dtp, without the normal marketing hype found in the computer
comics.