Font embedding for the web

Gustavo Ferreira's picture

there have been some efforts in making font embeding in html-based sites possible: weft, glyphgate, sifr. (are there others?)

the problem with weft is that it works just for ie on pc.

the problem with glyphgate is that it is expensive, and that it does not really work well - the demo pages don't work on safari, firefox crashes when i try to enter the gg site (on osx).

sifr uses flash to deliver the fonts - it seems to work well... but just for titles.

ah, there is photofont too, which is not an embedding tool but a xml-based font format for the web. but it also doesn't work on my safari or firefox either... and it seems like nobody is working on the technology anymore.

how far are we from a solution that works?
what are the technical, legal, commercial etc. obstacles?
are we doomed to use flash for delivering anything that is not verdana, arial, times or georgia...?

Nick Shinn's picture

Why is it necessary to have a new font format?
Surely browsers should be able to work with any font, and why not the OpenType format?

I just don't understand what the difficulty is.
1. Load fonts on site server with other web info.
2. Download fonts used on a web page when it is opened.
3. Delete font when page is closed.

If engineers can figure out streaming video, surely streaming fonts shouldn't be a problem.

Frank Jonen's picture

I'm not so much for streaming fonts, think about streaming CSS how maddening that would be I'm glad CSS is cached so conveniently by all web browsers and internet explorer 6. If a font can be securely cached why not do this instead?

For the photofont question. It's not dead at all, there's a lot going on behind the scenes that will solve this and other difficulties in a really slick way for the end user. Just wait and see :-)

Si_Daniels's picture

I'm feeling my age, this seems like ancient history to me now. The archives of the W3C font list should still be available and are worth looking over with regard to this. Anyway to summarize...

Plain fonts are too large to attach to Web pages, especially if you have to download them every time you view a page. Compression and subsetting are required to keep the font sizes small. There was also the issue of a page designers rights to embed a given font and the protection of the embedded font.

The web font wars, at least with respect to embedded fonts, revolved around these issues. Which compression format to use? Agfa's lossless MicroType Express or Bitstream's TrueDoc which didn

aluminum's picture

The web is a medium that is very much about accessibility. It is as much about end-user preferences and whims is it is about back end designer's preferences and whims.

Font embedding didn't take off because it's not really needed for the medium to work.

Flash methods like sIFR are actually quite useful for a bit of specific font spec'ing here and there, but I'd never dream of setting an entire page using the technique. And, honestly, I'm quite happy with the current selection of typeface for most web work I do. In the end, it doesn't matter how much I like Univers, if the end user prefers verdana, so be it.

Nick Shinn's picture

>As I said on the other thread, lack of interest, competing technologies, designer opposition and a great set of free Web fonts

aluminum's picture

"Why can't browser manufacturers agree on a compression technology?"

Why can't they agree on simple things like CSS? Politics. Firefox/Mozilla has actually done a commendable job, but they simply haven't seen any value in font embedding, so haven't focused on it.

"Users are prepared to wait for images to load, so why not fonts? "

Well, users aren't big on waiting for anything. And, as for fonts, it's rarely the top priority for your web viewer.

I agree with your last point...the 'people will copy it' argument is just silly and applies to all digital media. DRM of any sort is rather futile and more often annoying than helpful.

Gustavo Ferreira's picture

nick wrote:
> Why is it necessary to have a new font format?

i think it does make sense having a new format which adresses things unique to screen media/web

cheshiredave's picture

I'm surprised that sIFR hasn't made a bigger splash, actually. It's a little hinky here and there, but it basically works. Its biggest drawback is that it can't reliably draw with a transparent background. Also, at least with Safari, sIFR'd text, though it can be selected, can't be copied to the clipboard, which breaks one of its benefits. I'm cool with existing screen fonts for text. It's not great, but I can live with it for the moment. But I'm glad to have the option of any font I choose for headings without resorting to images.

Mark Simonson's picture

Chesh, I'm pretty sure you can copy text from sIFR by using the contextual pop-up when you right-click or control-click on the text.

Speaking of web font formats, anyone know what's happening with SVG fonts?

cheshiredave's picture

Thanks, Mark -- I stand corrected, but that's pretty lame. Hope there's

Mark Simonson's picture

Yeah. Not ideal or obvious, but at least it's possible.

Forrest L Norvell's picture

The real issue is that it's not safe to assume that everyone browsing your site has any given fonts at all installed. While Microsoft was distributing their core fonts for the web, it was safe-ish, and we can assume that most clued-up web users will have the usual suspects (Verdana, Georgia, Courier, etc.) installed, there's no real way to present a consistent typographical feel to everyone.

sIFR is an elegant hack for dealing with display type (and I appreciate the lengths its creators pursued to avoid interfering with assistive technologies), but it sucks for body type. I've read enough working group messages to think this may be infeasibly complicated in the near to medium term, especially because browser developers don't seem to see much percentage in tangling with it anymore.

Si_Daniels's picture

Well, the Web fonts are more common than ever, as the machines that didn't have them pre-installed become increasingly obsolete.

Forrest L Norvell's picture

Simon, the Mac requires you to have Microsoft Office installed. On Linux systems, you used to be able to get at the Core Fonts package via various installers, but these days even that's legally dubious, given the legal restrictions on redistributing the package. I agree, almost everybody ends up with copies of the core fonts eventually, but if you're trying to make assumptions about what users are going to have available, and catch

Si_Daniels's picture

That's not accurate, sorry.

There's certainly lots of cases of unauthorized commercial redistribution of the fonts, as well as plenty of unauthorized modified versions out there. That's why the fonts were pulled from microsoft.com.

But the fonts were distributed under a broad EULA which the Linux community continues to exploit, taking some care to observe the EULA conditions. I've not heard of anyone seriously suggesting that these actions are illegal, but if you have a different interpretation of the EULA I'd love to hear it ;-)

On the Mac the fonts were part of IE

Forrest L Norvell's picture

I'd have to search the Debian archives, but I think there was some muttering recently about not being able to use the core fonts any longer, because there was no legitimate place to host them for redistribution. Debian is unusually sensitive to licensing concerns, like some other folks I could

Si_Daniels's picture

Interesting. If they have no way of making sure the fonts don't find there way onto a retail CD or if they're only available as a value add to users of their product then they may end up breaking the no commercial use aspect of the license.

Forrest L Norvell's picture

Actually, the problem goes the exact other direction: Debian is the most exactingly anal free software distribution on the planet. They had a three-month flamewar recently on whether it was permissible to distribute software what would load binary-only firmware blobs into peripheral devices, if the license for the firmware was insufficiently free.

Anyway, this is all sort of off the point, because most users of desktop Linux are incapable of installing that package anyway (Debian's installer for it, which does some mildly bizarre things to avoid violating the MS EULA, is more straightforward than most). Surfing the web on Linux has been a pretty lame experience until pretty recently, and even now, it's kind of a pain. Free X servers only started offering decent antialising in the last few years, and until two years ago it was really difficult to configure.

My point, such as it was, is that it would be really nice to have a method for ensuring consistent type display on the web that was independent of the underlying operating system. For the reasons we've been discussing, CSS isn't up to the job (even leaving aside browser implementation issues). I understand the technological, political, and social reasons why we don't

aluminum's picture

"is that it would be really nice to have a method for ensuring consistent type display on the web"

BUT...and this is a key point...that's not how the web works. Again, the web gives a lot of control to the end-user.

Web design is much more about suggestion than dictation. Suggest a few fonts, and let the chips fall where they may...

raph's picture

Ergo, the way the web works is not really nice.

aluminum's picture

"Ergo, the way the web works is not really nice."

Well, depends on your perspective. Personally, I think it's a great thing. It's simply part of the medium.

Frank Jonen's picture

darrel: Web design is much more about suggestion than dictation. Suggest a few fonts, and let the chips fall where they may...

We've all seen where this leads, people that have no clue about design at all will use Excel to design brochures (tables for design) and use images where they should use text. The best example is the Apple website where the head menu could easily be done with text instead of images and the infamous Amazon.com Jeff Bezos greeting image, where an image is used to display a long piece of text, usually multiple paragraphs.

All because of the 'no control about fonts' issues. I strongly believe we can make the web a nicer place if we just give the designers what they need to create a great user experience without cheating. If they cheat and user image or even Flash gadgetry, people with disabilities will suffer from that.

aluminum's picture

"We've all seen where this leads"

Well, it leads to having fairly accessible content and functionality for folks.

Everyone has a 'clue' about design. It's just more refined for those of us with formal training. And if grandma likes her 18pt Times New Roman, who am I to argue with her?

The web offers the unique feature of allowing for a highly distinct separation of presentation and content. For some...typically long-time print designers, this is seen as a horrific loss of control. For others, it's seen a big benefit for a variety of reasons.

It's been said for years, but the web isn't print. Font embedding sounds great, but it's actually a technology that has come and gone with little fanfare. We do have alternatives now (as mentioned) and there are certainly times when wanting a specific typeface is called for and images or Flash take over quite nicely. SVG may come along soone or later (though I'm not holding my breath). For now, the web seems to work just fine without too much of a public outcry over a lack of font embedding. ;o)

"what they need to create a great user experience"

Having 100% control of the typeface isn't necessarily a prerequisite for creating a great user experience. Many would argue giving the end-user some control is actually what gives them a great user experience.

Forrest L Norvell's picture

darrel says: The web offers the unique feature of allowing for a highly distinct separation of presentation and content. For some...typically long-time print designers, this is seen as a horrific loss of control. For others, it's seen a big benefit for a variety of reasons.

I overstated my case, maybe. I'm all for accessible sites, assistive technologies, and my elderly relations being able to decipher my rambling screeds about Satanic heavy metal by using their own fonts and colors. Also, every time I have to wade through some schlocky overdone Flash site to get at someone's photography, type, or writing, it makes me tetchy and unpleasant to be around.

At the same time, "separation of presentation and content" is a canard that's been a canard since Marc Andreessen hacked the IMG tag into HTML, and it continues to be a canard even today in the days of XHTML 1.1 and CSS 3 (maybe, someday). Not only have the two always been hopelessly tangled on the real Web (not the theoretical, W3C Web), but I'm not sure that we, as typographers, would really want it to be otherwise. The only real justification we can provide for having hundreds of text faces and thousands of display faces is that typefaces become part of the texts they spell out. Even digital type rendered on lo-res screens has personality and character, and that character informs the text.

That's the reason I would like to see embedding. If I want absolute control over positioning and appearance, there's always PDF. What I actually want is a way to make my blog look nice for everyone by default. They can tweak it on their end all they want, but having no control at all

Frank Jonen's picture

Font embedding sounds great, but it's actually a technology that has come and gone with little fanfare. Pffrroopht... hmm ok. gone :-)

No really, if that 'embedding' needs to work it cannot be 'embedding'. Would you embed the images of an interface of a website? Obviously not. Only content images and even there, it's just linking to files.

Having 100% control of the typeface isn't necessarily a prerequisite for creating a great user experience. Many would argue giving the end-user some control is actually what gives them a great user experience.

Well actually it is. Think about it from this POV: Either designers render their headlines and subtext into images (even CSS replacements like the Fahrner / Inman replacement, which is a lot of work if done manually) or they assign a font file to their headline ID or CLASS and all content is content.

This way you have the choice of 'breaking the intended design' and using your own font to display the document or doing nothing and use it as designed. Just image Nick Shinn or HHP could disable Helvetica for websites and assign one of their fonts instead :-) *ducks*

aluminum's picture

"At the same time, "separation of presentation and content" is a canard "

This is where the debate forks. I can appreciate and understand both paths here. There's plenty to argue that it is, indeed, just a canard, while there's plenty to argue the practical beneifts of it as well.

I, myself sit on the fence on that one. ;o)

And I see your point about simply wanting the ability to suggest more options. I can't really disagree with that...it would be nice! I guess it's just not a priority for the web right now.

"This has less to do with the merits of the technology and more to do with money politics and browser wars, like much else on the Web."

Well, that's true about the software industry, and perhaps most product-consumer industries in general. ;o)

"or they assign a font file to their headline ID or CLASS and all content is content."

Well, that's what we have now, Frank. You can do this with user-style sheets.

Forrest L Norvell's picture

"At the same time, "separation of presentation and content" is a canard "

This is where the debate forks. I can appreciate and understand both paths here. There's plenty to argue that it is, indeed, just a canard, while there's plenty to argue the practical beneifts of it as well.


Well, I meant in practice. People have been preaching the separation of the two since the inception in SGML, and my point was that even in the beginning, HTML confused the two, and people using HTML have been tangling the two every day.

"or they assign a font file to their headline ID or CLASS and all content is content."

Well, that's what we have now, Frank. You can do this with user-style sheets.


You can spec a font in a style sheet, but you can't include an actual font file. This was even briefly a part of CSS 2.somethingorother, but it died with embedding.

Syndicate content Syndicate content