Re: WebFonts

> From: Erik van Blokland <evb@knoware.nl>
> To: "Hannes Famira" <hannes@dataweb.nl>, "Martin Wenzel" <Wenzel@dataweb.nl>,
>         "Jonathan Hoefler" <hoefler@aol.com>, "Www" <www-font@w3.org>

> Scalability
> Having outlines available at the client makes it possible to scale the 
> text and print a page.

More importantly, you don't know the size or resolution of the screen
that the browser is using.  I have a 150dpi screen on one system, and only
a 100dpi screen on another.  Our main Macintosh actually has a 75dpi screen!!!
With such extremes, a bitmap font that can be read easily on one screen
is utterly useless on another.  And one that is large enough to be
read easily on most devices -- say, between 20 and 30 pixels per em -- is
much too large for a 640x480 PC screen, or even an 800x600 portable PC
or Mac screen.

> Some sort of anti-aliased bitmap font will do just fine.
We have users with black-and-white (not greyscale!) terminals.
They can use an outline font.  They can't easily use a greyscale bitmap.

> Printing
> Other argument: printing a webpage. First, are there figures available on 
> the amount of webpages that actually get printed?

Not that I've seen.  I can tell you that we had very high demand to add
printing to HoTMetaL and HoTMetaL PRO, the SoftQuad HTML editors.
We also had strong demand to add printing support to the free version of
SoftQuad Panorama, the SGML browser.

> I think that the ratio 
> of number of read webpages that get printed to the number of unprinted 
> ones favors the unprinted pages in such a major way that making 
> complicated and expensive provisions to make webpages printable from the 
> start is not necessary.

Nothing is more complicated or expensive than it need be.

Your argument would also apply in this scenario, I think:
    Most computer-printed documents were generated with Microsoft Word.
    Microsoft Word does not use ligatures.
    Fonts would be simpler, smaller and cheaper without ligatures.
    Therefore, fonts should not contain ligatures.

Well, Jonathan Hoefler would have saved himself a lot of work, and
if you agree that Q and K are not used very often, and drop those too,
you have saved more work.


> Secondly, typography for the screen is different from typography for the 
> printed page. Unless screen and pages actually become the same physical 
> object, a design for one medium will always look bad on the other. Not 
> just because of the fonts, the typography for each medium requires 
> different solutions: the way the text is read differs.

I think this is a good argument for separate style sheets for printing --
as supported by SoftQuad Panorama PRO, and by the current HTML stylesheet
proposals.

[...]

> Economy
> Outline font are more economic (smaller) compared to a bag of bitmaps 
> that contains all sizes one could make with the outline font. That makes 
> it a good solution for desktop publishing where you don't know what the 
> next job will bring. But a web page contains a limited number of 
> characters in a limited number of sizes. Up to a fairly large bodyheight 
> a pre-rendered bitmap font is smaller than an outline font describing the 
> same thing.

Note that the TrueDoc system at least -- I'm not sure about the others
yet, but the Agfa one can probably do this too -- download only the
outlines (and hints!) for the glyphs you actually use in the document.
A TrueDoc font is often smaller than the corresponding bitmap font.

However, for reasons outlined above, anything using a bitmapped font is
doomed to failure, even if the format you use is the Microsoft Windows
".fon" format or the Adobe/X11 .bdf format, both of which are already
widely implemented.

> The letters can be even be colored with many colors in one letter and 
> modified using normal image processing software. This is, I think, a must 
> for popular acceptance of a type on screen system, just look at what 
> people are doing with text in images on the web right now.

(does the I in this paragraph mean that only one of the people cited
at the start agrees?)

I agree that this would be a good feature.  It's a shame that Type 1 fonts
generally have to be one-bit-deep.  I've used systems like NeWS that
supported colour PostScript fonts.  But earlier you talked about not
introducing complexities, or new features, or extra cost...
What happens if there aren't enough colours to display all the colours in
a font?  How does the browser ensure that the page is still readable?

> Copyright & piracy issues
> As a typedesigner I've seen and experienced how people deal with fonts 
> and ownership, and it sucks. If a system came to be where 
> character-outlines are broadcast to every single browser in the world I 
> would stop making type, or at least, stop using anything nice and new 
> because it would become instant public domain.

If this were true, I would strongly oppose outline fonts on the web.
So would many other people on this mailing list (and elsewhere!).

One reason why it hasn't already happened is the effort involved in
making sure that the fonts can't easily be re-used.  It has to be at
least as hard as getting a font out of a PDF file, say.

Again, I'll mention the TrueDoc system because it's the one I know
most about....  the font arrives in an encrypted format.  The user can
save it on the local hard disk -- big deal.  FontMonger won't open it,
and neitehr will Suitcase on the Mac, nor anything else but a compatible
Web browser.  Some proposals allow the same screen font to be used in
multiple documents (but you won't generally have all the characters you
need!).  Personally I favour being able to use a font fot all the web
pages at your site.  But none of the proposals I've seen let you receive
a font associated with a web page's style sheet and use it yourself.
The font is never installed on the system.

Adobe and Agfa and Bitstream would not be interested in this kind of
technology if it were otherwise.  Their lawyers would be _very_ interested,
of course.

Font embedding _will_ happen, with outline fonts.

A better question, it seems to me, is
``how do we work together to make sure the result is as good as possible?''

By `good', I mean (most important first, roughly)
. Sufficiently secure
. Supports those people who need (or whish to use) Accessibility options,
  such as audio screen readers, braille conversion, `large print' editions,
  and other requirements.
. Wide range of commercial high-quality `font software' (typefaces)
  readily available
. Supported in commercial Web authoring tools and browsers, such as
  HoTMetaL, SiteMill, InternetAssistant, and NCSA Mosaic, Netscape,
  Internet Explorer, etc...
. Highest feasible rendered quality
. Fast as possible
. Internationalised -- e.g. not restricted to a `Latin 1' character set,
  but can use Unicode in conjunction with script/language tagging as per
  the HTML Internationalisation Internet Draft to access a full range of
  Hangul, Kanji, Roman, Hebrew..... glyphs
. Platform independent, so that Mac users can join in too
  (at least one proposal I have seen excludes the minority of WWW users
  who are on the Macintosh, for example... oops) (please don't ask which)
. Device independent, so that what you see on your screen is as close as
  possible to what I see on mine
. Supports printing adequately.
. Possible to use existing investment of Type 1 fonts.  (The ability
  to use TrueType fonts might also be important, but I claim that most
  people who pay for commercial professional-quality typefaces today
  buy Type 1 fonts.  One day, QuickDraw GX will be ported, and GX fonts
  will also be possible on non-Mac computers, but not yet)


I think it's really good that typeface designers participate in this process.

Lee

Liam Quin, SoftQuad Inc +1 416 239 4801 lee@sq.com
HexSweeper NeWS game;OPEN LOOK+XView+mf-fonts FAQs;lq-text unix text retrieval
<URL:http://www.sq.com/> SoftQuad Panorama, HoTMetaL, Services, SGML tools
`The future holds promise for those who have faith in it' [Inglis billboard]

Received on Wednesday, 27 March 1996 23:51:29 UTC