Idea: Crowd-made Unicode font

[Accident]

This topic was imported from the Typophile platform

Hello there, everyone.

I am interested in Fonts for quite some time now. A few days ago I was searching on the Internet for a good(i.e., nearly complete), free Unicode font. I was amazed how hard it was to find fonts which had a big amount of glyphs and were free(and most of the 'free' ones had strings attached like being only free for private use on one computer, etc).

Now I had an idea. What if a website would be made, where people can pick one of the 100.000+ glyphs that Unicode has nowadays, and draw it, with their mouse, on a simple white with black canvas, with an undo and a clear button. The style will of course be a bit hand-drawn, shoddy-ish. That's a consequence of this technique but I don't think that it's a bad thing, because it will be uniform through the whole font:

 

Multiple people can draw the same glyph, and then votes will decide which one looks best.

I believe that in this way, with a large group of people, it might be possible to create a big unicode font which covers one or multiple of the unicode planes.

What do you, experts of typefaces, think? Is this idea viable?

[Karl Stange]

Have you seen this page on Wikipedia?

http://en.wikipedia.org/wiki/Unicode_fonts

Probably far from exhaustive but a good starting point.

What are your requirements for a 'free' font? If you do not want to be bound to the terms and conditions of a license agreement then creating your own font would definitely be the best way to go, but guaranteeing any kind of consistency and quality across such a vast undertaking would be very difficult.

For what purpose do you require this kind of font? It will almost certainly fall down in terms of engineering if not design. If you need a font which meets the requirements of coverage for testing then you may have more luck with SIL's Unicode BMP Fallback Font or Dr. Ken Lunde's "Extreme" OpenType font for stress-testing which is also documented in his Adobe blog.

[Karl Stange]

As an exercise I think the proposed project sounds like fun and could generate some very interesting results. However, whether you would ever get to the end of the project is another matter...

[Luma Vine]

There have been some similar projects out there, though nothing quite like what you are proposing. Remember that a font is a lot more than a collection of glyphs.

[quadibloc]

Frankly, I've been astonished by how many free Chinese fonts, and how many free Unicode fonts, are available out there. (There is, however, only one free font I know of, Hanazono, that provides all the Unicode Chinese characters, and thus there is no free font that provides all of Unicode. However, Hanazono has a BSD license, not a GPL license, so it may be possible to fix that with a font editor).

[hrant]

It would be interesting as an experiment, but couldn't produce a really usable font, because:

I was amazed how hard it was to find fonts which had a big amount of glyphs and were free

You shouldn't be amazed. Good type design is a lot of work.

And voting? Hopeless.

BTW "crowd type design" has been attempted before, but it was bitmap-based. And it still didn't get anywhere functional.

hhp

[John Hudson]

there is no free font that provides all of Unicode

The Unicode character set now greatly exceeds the 65,536 glyph limit that is built into the sfnt font format (TrueType and OpenType). Considering also that a fully functional font will need more than one glyph per character for many scripts, it is clearly impossible to create a single font that supports all of Unicode in today's principal font formats.

One could, however, use a composite font format to bundle multiple fonts into a single pseudo-font.

[quadibloc]

I thought that OpenType was a composite font format, in the sense I think you're using that term, so clearly I have much to learn.

[John Hudson]

No, OpenType is not a composite font format; it's a discreet font format. I'm talking about things like the 'composite font representation' in ISO/IEC 14496-28:2012, i.e. a wrapper around multiple fonts that present them as if they were a single font, with metadata indicating which cmap ranges of each component font take precedence.

Each sfnt font can only contain a maximum of 65,536 glyphs (with functional limits on other tables in the font based on the same architectural restrictions). But multiple sfnt fonts could be packaged within a composite font representation to cover all of Unicode.

[Accident]

You shouldn't be amazed. Good type design is a lot of work.

True. I think 'amazed' was a bad choice of words. Rather, I am astonished how many of the 'better' unicode-fonts have all kinds of restrictions placed on them.

And voting? Hopeless.

Could you elaborate this? Certainly, the font won't win a beautyprize, but that's not the purpose of the font.

Basically, my goals of the project are the following:
-An interesting experiment, to see how well it works to create a font together.
-Creating a font that covers all(or as much as possible) of the unicode planes, to create one font that is able to display all characters when you don't have another font that has them installed.
-Creating a simple unicode-font that can be used by anyone for any project without being restricted by any weird licensing.

I realize that one TTF/OTF has a 65.535 cap, but indeed as pointed out, using multiple fonts is a solution. Composite fonts are something I didn't know of yet, interesting.

@hrant: Do you have any source about earlier crowd type design projects? I'd love to read about them. The only thing I know of right now is the http://czyborra.com/unifont/, which in fact is one of the most complete, free unicode-typefaces right now. The big problem of it is that each of its characters consists of blocks in a 16x16 grid and glyphs were made with editing the ascii representation of a glyph. This means that the font looks 'pixelated' and doesn't looks nice on any size bigger than 16px.

Have a nice day,

~Accident

[hrant]

Voting can't work because too many ideas of what's good ruin a font.

You will end with a very large number of ugly -and often incorrect- glyphs that are roaming around looking for trouble, individually or in lynch mobs.

I'll try to dig up that crowd bitmapfont project.

In the czyborra thing restricting it to a 16×16 grid is actually a pretty smart way of limiting the ugliness.

hhp

[hrant]

OK, I found it. It was actually an official Typophile project which ran ten years ago, called "Smaller Picture".
http://www.howardesign.com/4.0/meta/smallerpicture.html
http://www.imarlin.com/sandbox/smaller/

And here's some partly decent commentary:
http://www.kuro5hin.org/?op=displaystory;sid=2002/9/21/13257/5648
http://www.underconsideration.com/speakup/archives/001220.html

hhp

[Nick Cooke]

Design by committee always results in a (usually bad) compromise.

[Nick Shinn]

A student I was at art school with made films by picking up scraps from the editing room floor and splicing them together randomly. You could get a similar effect by tipping many type trays (same size, different typefaces) on the composing room floor and then sorting them all into one tray; similar to the OP’s suggestion, then taking one glyph from each section of the tray, to represent the composite font. Floor-sourced.

With or without voting, the effect is random, because the process does not address the fact that type is a system where the parts are supposed to work together.

It would be a Frankenfont.

[hrant]

tipping many type trays (same size, different typefaces) on the composing room floor

But that damages the sorts (furthermore causing them not to fit together any more). I know because I once pied a galley of a full specimen. :-/

hhp

[John Hudson]

I think the criticisms of the initial proposal are sound, but the recent assertions that 'design by committee' necessarily results in compromise or that the results of voting are 'random' seem to be to overstate the case and ignore the possibilities of a well designed process.

If there are established design principles and directions, agreed upon by consensus of participants at the beginning of the project, then this kind of collaborative design approach doesn't need to end up in compromise or a Frankenfont. 'Crowd-sourcing' is probably the wrong term to use, in that it implies little or no active coordination, but in the same way as an individual designer decides upon a set of criteria by which to judge the success or failure of individual glyphs -- within what Nick rightly identifies as a system within which the parts must work together --, a group of people can come to the same kinds of criteria and make collective judgements through discussion, consensus and, if necessary, voting. In such a situation, I would consider voting to be something of a last resort, when there are two or more proposed designs that appear to equally meet the criteria, in which case majority preference is a legitimate basis for deciding which to adopt. I would not set up the process to be based on voting as a default or initial decision making process: a discussion/consensus approach, while time consuming, will produce much better results.

I did a workshop with the MA students at Reading earlier this year in which they designed and built an 8-bit display font in two days. I provided them with five key glyphs by way of establishing the characteristics of the design (obviously in a longer time frame this too could be arrived at by group decision): H O i o n. Responsibility for individual glyphs was determined by drawing cards, and this happened three times during the two days, each set of cards building on what had been accomplished in the previous round (e.g. there was no point in having two people independently designing left and right quote marks whole someone else designed the comma, since these are all related marks, so the first round included the comma, the second the left quote, and the third the right quote). Key roles in the project were those of Curator, also chosen by card draw and responsible for collating the glyphs and updating the master FontLab source file, and, in the later rounds, Consistency Checker, who would check alignments, overshoots, stem widths, etc..

Ben Mitchell wrote blog post about the workshop.

The most important aspect of the project was the group crit halfway through in which the design was discussed and, for instance, the b d p q glyphs compared and decision made, by the group, as to which bowl shape worked best and what approach to take to harmonising them.

[hrant]

The key thing there though is the conversation between the participants, which stems from dedication. It seems very unlikely that the project described by Wiebe would enjoy that, at least not in proportion to the size of the project.

hhp

[Nick Shinn]

…recent assertions that 'design by committee' necessarily results in compromise or that the results of voting are 'random' seem to be to overstate the case…

My comment that the OP’s process was random stemmed from the statement “Multiple people can draw the same glyph, and then votes will decide which one looks best.” I assumed that voters would be choosing glyphs by comparing each glyph of a Unicode character with all other glyphs of the same character, irrespective of what the other characters looked like. Hence the selections would be random in relation to one another.

However, on further consideration, it would be possible to create a live, interactive decision process, in which a block of multi-character text is displayed online, using all the currrent selections, continually updated. Then the voter could insert different glyphs for a particular character in the text, and see how well they worked in context. Rather like the Metrics window in FontLab.

Mind you, it would be a lot of work.

[russellm]

With respect, sounds like having your teeth cleaned when you have 65,536 teeth.

[dberlowgone]

"...sounds like having your teeth cleaned when you have 65,536 teeth."

...and when that 65,536th dentists arrives and asks how you're feeling today, on this, the 435th day of cleaning...

[Bendy]

I'm not sure this could work. What use is a font where every glyph is drawn by someone else? The whole point of a useable typeface is that the characters work together. And a hand-drawn 'shoddyish' font? I don't understand the need or use for that. The demands of even one complex script are a difficult thing for even a professional designer, so I wouldn't have faith in a crowd-sourced version. Key dimensions and positioning have to be shared among the designers, just to make the orthography correct, never mind about aesthetics. And someone would have to do the VOLT production, just to make the thing work.

John's right, the most important thing is the curating, or regulating, of the design. In the Reading project this wasn't too difficult, as we were dealing with basically an extended Latin character set, but a lot more questions would arise in the scope of a full Unicode project. (It would be interesting to see how far a group of Reading students could go with broader script support!)

[russellm]

What Bendy said.

But, I think that with a bit of persistence and a well defined process and methodology it could be interesting... if not very useful in the commercial sense.

I mean... I do like having clean teeth and all, even if I don't look forward to sitting in the chair looking at the ceiling for 40 minutes, listening to the dental assistant going on and on about her boyfriend who can't seem to be able to commit... or is too eager to do so... I can't remember which way that was swinging the last time I was there, but she does tell a good story.