[Openfontlibrary] Re: Free Design Software Community Roadmap

Raph Levien raph.levien at gmail.com
Mon Apr 17 12:01:20 PDT 2006


> On 15/04/06, Raph Levien <raph.levien at gmail.com> wrote:
> >
> > In this specific case of websites for free font design, there
> > are many problems to be overcome:
>
> > 1. Lack of type skill in the community, especially the intersection
> > with the free software community. Almost all of the good free fonts I
> > see are either license transfers from established businesses (the URW
> > set, Vera), or from people like Victor who has serious traditional
> > training.
>
> The data/information levels of that 'serious traditional training'
> will end up on the web one way or another, and the knowledge level is
> already available with community review sites like typophile. This
> will in improve things a lot.

Indeed. I pretty much feel that a website for free font developers
will stand or fall based on how appealing it is to people learning to
make fonts. In particular, howto guides (adapted for free tools),
critique from more experienced designers, and meaningful praise for a
job well done.

> > I fear that if you just throw up a site that says "post
> > creative commons (or whatever) licensed fonts here), you'll end up
> > primarily with stuff like http://www.copacetix.com/fonts.php
>
> There are two issues here.
>
> The first is managing a cornocopia of smalltime display fonts. Tagging
> and trust^H reputation metrics means that its possible to automate
> 'best of' data mining, and the effect can be social'd too with myspace
> style uid homepages with 'my favourite free fonts' polls and such.
> OpenClipArt and OpenPhoto.net have been dealing with a similar issue
> for a while, and moving to a common platform like ccHost was suggested
> by Jon to leverage their technology for this.

I don't think the technical problems are that hard to solve. The
number of fonts posted will be small enough, I think, that low-tech
solutions like asking an expert review board for ratings is practical.
More important is the _emphasis_ of the site on quality, something
that I pretty much see missing from most of the sites for free fonts
out there.

> The second is that bigtime text fonts are hard and complex work.
> However, this is solvable. TypeForge emphasises collaborative font
> development, and CVS/SVN has been discussed earlier on this list for
> OpenFontLibrary just for tracking all the changes. However, IMO hosted
> CVSish infrastructure means the development of more serious typefaces
> is more probable. Distributed, version controlled font development is
> the future of type design in general, too.

Yes, and here's an area where there's both a lot of potential and the
need for farly extensive development work. A font-in-development
_should_ be a web-accessible repository, and some things like posting
of comments for individual glyphs should  be very open, in wiki style.
The repo should also easily handle multiple variants of the same
glyph, so the "best" can be chosen through either voting or individual
choice when "cutting" a .ttf from the repo.

Most of this work would be server-side, but support for font-as-repo
can be added client-side as well. Instead of "save," font editing
tools might commit to the repo.

> This will also discourage the Vera problem of 50 different versions
> scattered to the web, all with slightly differently extended glyph
> sets. Instead we'll have everything in one source tree and get bonus
> network effects - lower overheads encourage people to contribute just
> one or two changes. This also relates to (1) - I quote Sascha
> Brossmann's post from December 6th 2005:

Right. I think one of the strengths of font-as-repo is to make
extended glyph sets more manageable. I wrote about this fairly
extensively in the typophile post that got the ball rolling on this
discussion.

> -- 8<  --
> I would especially like to encourage to
> start with tutorials on slightly easier topics than designing and
> implementing a whole font from scratch, e.g.
>
> (1) how to add glyphs to an existing font, so that they properly fit
>     in (this includes non-trivial type design basics - trivia should
>     be just linked, mainly in respect to micro typography, kerning
>     etc., and technical issues like optimal curve construction,
>     encoding, import/export, etc.).
>
> (2) how to optimise fonts for low-resolution rendering (e.g. screen
>     and standard desktop printers): hinting (truetype and ps),
>     letterspacing/kerning, etc. (-> how to improve technical quality)
>
> (3) all things unicode, open type features and the like
>
> (4) how to design a new [bold|italic|smallcaps|...] face to be added
>     to an already existing font family.
>
> -- 8< --
>
> In the long term I suspect Letterror's UFO is the perfect format for
> this, and read/write support in FontForge is implicit. (I don't mean
> to even imply any pressure on George whatsoever, I'm talking /long
> term/ here :-)

I'm somewhat familiar with UFO, but don't have at the tip of my brain
an answer to this question: is it specific to Beziers, or can it
easily accommodate other curve representations, such as (for example)
a refined implementation of the Ikarus ideas?

> > 2. Overlap with existing efforts. After posting the typophile thread,
> > I became aware of TypeForge and the others posted by Dave above. Does it
> > still make sense to build something new? Is there "low hanging fruit"
> > that might be added to one of the existing projects to make it truly
> > useful?
>
> I didn't say anything about building something new.
>
> I think its quite important for everyone to keep chipping away at
> their own projects, going with the grain of the decentralised
> net.culture, a bit of healthy evolutionary diversity if you will, or
> 'small pieces loosely joined' as they say.
>
> Whats lacking is the 'loosely joined' part, which is where everyone
> getting a shared vision of the future comes in.
>
> This might mean, for example, discussing how TypeForge and
> OpenFontLibrary can complement each other :-)

Yes. In practice, this question will boil down to choices of file
formats, protocols, and extension languages/API's.

In the file format world, we have at least FontForge sfd, UFO, and of
course "legacy" formats such as TTF, complete with various textual
representations for the OpenType extensions (Rogier van Dalen's
otcomp, the comparable tool in Adobe's FDK, whatever ships with
Fontlab, etc). I'm also developing a file format for my curve work,
for a variety of reasons. That's a pretty fragmentary set.

For protocols, I'm primarily thinking about network access to the font
repository. This basically boils down to which version control tool is
chosen, and how it's set up. Should each font be one big text file, or
perhaps a directory with a separate file for each glyph?

Extension languages and API's for accessing font data can also be
important for tying all this stuff together. Python is becoming
something of a de facto standard in the (commercial) font world, and I
also use it extensively for converting between formats and other
similar tasks.

> > 3. Lack of resources. I have lots of ideas and programming ability but
> > relatively little time to sit down and do the webmonkey work to make
> > something like this happen. Other people I've talked to are
> > enthusiastic but lack the requisite 2d graphics/font programming
> > expertise, and may not have all that many hours to devote either. The
> > "free font design" community is small to the point where starting an
> > ambitious project may be marginal.
>
> Lets just start and let things manifest themselves :-)
>
> > I don't mean to throw cold water on the idea. I wouldn't have posted
> > my thread on typophile, etc., if I didn't think there was some real
> > potential here. But I think the above questions need to be answered
> > first, before we all start writing ambitious-sounding roadmaps and the
> > like.
>
> You're exactely right about real potential: SIL has got good links
> with the FSF, OSI, debian-legal, and Ubuntu with the OFL. OpenClipArt
> has great links with Creative Commons and Inkscape. FontForge is on
> board. As coalitions of the willing go, this is serious stuff :-)

Ok, that sounds reasonable. But I'd still like to hear more detail on
how much time individuals are going to be able to contribute.

Raph


More information about the Openfontlibrary mailing list