The Future of Type Design

thomascbryan's picture

My biggest question at this point in time is that of the future of type design. Since type has been around for so long, how does one avoid overlap with designs that are already out there. What is the best way to not end up in the situation where you finish designing a typeface and discover that you just re-invented the wheel? Is there an easy way (other than flipping through the FontBook / font catalogs for hours in order to avoid re-creating an already released font?

In short, I feel like as time progresses, the chances of reinventing the wheel increase, how does one avoid this conundrum?

Thanks in advance for your input,
Tom

oldnick's picture

The short answer is: you can't avoid it. Check out the Type ID board: no single individual has all the answers all the time.

Si_Daniels's picture

>Is there an easy way (other than flipping through the FontBook / font catalogs for hours in order to avoid re-creating an already released font?

Upload the key letters into Myfonts WTF and see what quality of hits you get.

thomascbryan's picture

Good point, oldnick. I guess I was just looking for that answer to life, the universe, and everything...

sii, I didn't think of that for some reason, and it would definitely work once the face was complete (or near complete). Was just hoping to see the 'dead end' sign at the top of the road rather than at the bottom.

Many thanks to both of you for your input.

Thomas Phinney's picture

There's no way to avoid the risk. The better part of a century ago, Frederic Goudy was saying of type design that "The old fellows stole all our best ideas!" Meaning that many "new" ideas people came up with had already been thought of. Even landmark type designs ofttimes turn out to have been preceded by lesser known designs that feature similar innovations. For example, look at Stellar before Optima.

Just be creative, and learn lots about what's been done already. Equally importantly, learn the *craft* of type design. There are a bunch of optical principles for strokes, there's the basics of spacing and kerning... most of which is not entirely obvious.

Cheers,

T

jshen's picture

I think you need to have an internal program for your font, i.e., a purpose in mind, a reason for a new font serving that purpose. If your program is specific, then you have criteria for your design decisions as you create the typeface. It's unlikely that you will repeat what someone else has done exactly.

Nick Shinn's picture

It's possible for two people to have the same concept independently, but the execution will come out differently.

There's no danger of accidentally reproducing someone else's typeface, if you start from scratch and pay scant attention to reference material.
After all, nobody could recreate, say, even well-known faces like Helvetica or Times from memory; there is something too personal and physical involved in the act of drawing, of judging, weighing and adjusting curve, proportion and space over an entire alphabet--and the original glyph shapes are extremely subtle.

It's the subtleties that make or break a typeface.

Nick Cooke's picture

It's like that old saying of giving a monkey a typewriter and it will eventually type the Bible (or something - I'm paraphrasing here).
That may well be but it would take a trillion years.

kentlew's picture

But give a monkey a Sharpie and it will pretty quickly come up with Comic Sans, right? ;-)

Rob O. Font's picture

>In short, I feel like as time progresses, the chances of reinventing the wheel increase, how does one avoid this conundrum?

First of all, if one closely examines the invention and reinventions of the wheel, you find literally millions of variations on the wheel that we're all glad for, many of which don't even roll. So, it's not like the wheel was, once invented, not in need of reinvention for new technologies or the desires of fashion.

Cheers!

Ray Larabie's picture

Once we can break free from the rigid technical limitations of current font formats, we'll have more room to maneuver.

Thomas Phinney's picture

What limitations in particular are you thinking of, Ray? I don't find current font formats all that limiting, in terms of design constraints.

T

dezcom's picture

I think Nick Shinn is right. If you don't set out to copy something existing, you won't. You may see it as a distant cousin but never an identical twin. There may be a few exceptions when it comes to some very rigidly geometric forms.
The big thing is, why would you want to go through the prolonged task of designing a typeface unless it had some parts of your soul in it?

russellm's picture

Theere is a finite number of places to put nodes in a 1000×1000 M quad, so sooner or later, all possible fonts will have been made. Then our work will be done and we can all relax.

dezcom's picture

"...and we can all relax."

I guess that means Russel is buying :-)

Mark Simonson's picture

...a finite number of places to put nodes in a 1000×1000 M quad...

True, but the number of possible combinations of nodes, control handles, their order, quantity, and position is very, very high. You could simplify it to a bitmap of 1000 x 1000. That would give you 2 to the millionth power* of possible bit patterns. Even if you only include distinct and desirable forms, for practical purposes, it may as well be infinite.

* 9.9006562292958982506979236163019032507336242417875... x 10^301029

russellm's picture

… So, Mark, what you seem to be saying is that there is still a lot of work to be done here.

:o)

Sorry Chris, I'd be happy too, but it looks like may have to wait a while.

Mark Simonson's picture

Especially if you work in TT format (2048 unit em).

dezcom's picture

OK, Russel, if you make it across the big pond by then, I'll buy you a tinner :-)

Ray Larabie's picture

@Thomas:

  • no alpha channels: currently, all semi-transparency must be done as dots etc.
  • no gradients: pointless without alpha channels I guess
  • no blur channel: to simulate liquid, gels. Less blue near the edges, more blur on the thick parts
  • no specular (variable glossiness)
  • no assignable color groups: if you want to do multicolor effects, you currently need to make separate layers. I'd like to, at least define a base color and a highlight color.
  • no stroke/plotter fonts: like the old days. Let the user define a stroke thickness.
  • no stroke/plotter scripts with real joins: the renderer can simulate a continuous stroke through an entire word, perhaps changing quality as it goes along. The renderer can animate the stroke so handwriting could appear to be written. By "real joins" I mean the entry and exit stroke is defined so the pen moves in the proper direction and order.
  • no 3D: if you think there's no point having 3D lettering, walk though a cemetery.
  • no bones: like a skeleton to a 3D modelling skeleton system. An example of how this would be used would be to make a swash g at the end of the word, loop under to the beginning of the word. No matter the length of word, the loop's skeleton would would deform the loop's mesh.
  • no smart way of dealing with diacriticals: we should be able to define rules on how accents should be placed.
  • no tiling alpha channel textures: this would be a great way to simulate textures like "pencil" while still having a scalable vector fonts. Using a detailed textured would more be memory affordable than actually rending it on every glyph like Photofonts. They would also tile properly across diacritics. I can explain if more detail if you don't know what I'm getting at.

These features should degrade properly so they still look readable without all the extra features.

At the very least, I want something PostScript has supported since biblical times: gradients & alpha channels. Trying to simulate pencil, marker or watercolor with dots in 2010 is ridiculous. Yeah, yeah Photofonts, whatever. I want a real format with vectors, not pixels. Font rasterizers already render gray levels for smoothing so it might not to be as big a technological leap as some of the other things I've mentioned.

Since the beginning of digital type, font formats have offered no new visual capabilities: decades of dumb solid shape after dumb solid shape.

Rob O. Font's picture

>...decades of dumb solid shape after dumb solid shape.

Centuries, actually... and for very good reasons.

Cheers!

Mark Simonson's picture

Ray, if PostScript Type 3 font support were revived, most if not all of that stuff should be possible.

Thomas Phinney's picture

Mark is correct that PostScript Type 3 fonts could do those things. However, they were only viewable with a full-blown PostScript environment... in other words, they only output to PostScript devices, and to get live viewing on screen, you needed Display PostScript. That hasn't been standard on any computer since NeXT went away, AFAIK.

Although there would be some advantages, I am willing to bet we won't see either PostScript Type 3 or Display PostScript making a resurgence in my lifetime. Or ever, in all likelihood.

I would not be shocked to see support for some of the things Ray wants in a future font format or extension to OpenType, though. Color, transparency / alpha channels, and gradients are the most likely advances, IMO.

However, balanced against all of the above are the twin issues of complexity and performance. The OS vendors in particular, who need to implement and support any new font format, care a great deal about text rendering performance. I mean, good grief, Microsoft still hasn't turned kerning on by default in Word because of performance concerns. Do you really think they'll embrace these sorts of features in fonts? I think the perception will be that this is a major escalation in cost for a really minor benefit.

The one small caveat has to do with accent positioning. Perhaps Ray can explain more of what he's looking for there. Seems to me like that may already be supported well at the format level, perhaps just not well enough in the tools he's using?

Regards,

T

Mark Simonson's picture

It seems like it would be possible to do a sort of meta-format, based on sets of current-format fonts that are designed to be used together for "chromatic" effects, such as Rosewood/Rosewood Fill. You could then have an app that would let you build layer/color recipes that could be used like normal fonts via a plug-in for graphics apps.

Rob O. Font's picture

>Seems to me like that may already be supported well at the format level,

Actually perhaps not, and the inventor of TT once agreed that by following PS's lead in both T1 and T3, he may possibly have erred. The two parts only have an anchor to share knowledge between them in tt: they scale and render separately and the bits are pinned together by this anchor. When glyphs need more information between them to get along better, across the range of scripts and styles, tools must be used to manage these more complex glyph compositions, before dumping the dumbed down composites into a puny machine format for output, which usually makes them bigger. TrueType provides a complete language for managing all the information of more intelligent compositing, if only the language could be applied to the recipe for positioning the parts, and subsequent parts on the stack could reference points of previously placed parts.

The swash g dream is near. That's GX variations with a backward-looking OT feature. No bones required, just the addition of 128 frames of change, and composition software that can count backwards and tell the glyph how far to stretch. Milo did this I'm guessing for Arabic.

Most of the rest of the list falls into the area of treatments, better left to the user at output, for size considerations. I understand the desire of the type designer to wish for more control, but the user has so many more tools, that if they want an effect they can get it with any outline under clever hands.

I miss stroke fonts, and I wish I had variable width stroke fonts, so I'm doing something about it. Otherwise I think the dumb shape format is good for what it does, but we need more tables, particularly for more complex authoring of color, texture and time-based media. But all we ever seem to get is less use of the formats we have now, and dumber formats being generated from smarter ones for dummies with programs.

Cheers!

Ray Larabie's picture

>>...decades of dumb solid shape after dumb solid shape.

>Centuries, actually... and for very good reasons.

Not for good reasons. It's not like metal type ever had an alpha channel gradients option available. It's not a choice that ever had to be made. The reason was that it couldn't be done. Over centuries, type designers have developed techniques to adjust weight. A bowl's strokes get thinner when they meet the stem. I want them to get thicker when the meet the stem. Thicker yet more transparent. I want my W to use variable transparency to reduce weight, not just thinner lines. I want my O to have an edgeless counter. To do this now. I have to use shitty dots and lines which render poorly on-screen.

> Most of the rest of the list falls into the area of treatments, better left to the user at output, for size considerations. I understand the desire of the type designer to wish for more control, but the user has so many more tools, that if they want an effect they can get it with any outline under clever hands.

Please forget the idea of "effects". These are not just effects. These attributes are integral to a font's design. It's not just an effect on a font, it is the font.
Current font formats were designed for text fonts, not for display fonts. Hence the technical limitation of dumb black shapes.

> clever hands

I want these to render automatically and dynamically on the web. How can a web designer apply a realistic looking pencil font to blog headlines with current font formats? It's impossible.

I want customers to be able to use these on the web. How can I define which parts of a glyph have higher specularity with current formats? It may seem pointless now, but when the web is 3D and surfaces can reflect light, we need systems in place to define these properties. When you tilt your iPad or move your head, parts of letters will shine in a pleasing way. Current print magazines have this feature. Perhaps it could be used to simulate hand painted lettering where the paint is partially dry. Where strokes overlap the paint is still shiny. Dots and dashes have heavier ink, take longer to dry and are very shiny. Perhaps what looked at first like a plain font actually has an intricate specular pattern when page angle is shifted. Unnecessary to readability? Yes. Awesome? Yes. It's not just a matter of how shiny a glyph can be, but which parts are shinier. It's not just "treatment". That kind of thought is the reason font designers are stuck with these bullshit, ancient formats. Having to use separate fonts with dots or lines to create gradients to define as attributes is stupid and looks like shit on the web. It's a clear sign that font formats were developed by people with no imagination. They must have looked at the fonts of the past and imagined that fonts in the future would be exactly the same. We've all seem beautiful calligraphy using semi-transparent inks; overlapping strokes which have have real depth. You can sense the speed of the brush by the variation of transparency. That's not just treatment. I want people to be able to set their blog headlines with fonts that can perform that way. No matter how clever a designer is, they can't get these effects using current formats without actually using a tablet and painting the letters "by hand". Even if they can, this doesn't address, what I see as the one of the main uses for display fonts in the near future: dynamically generated web headlines. Really look at pencil lettering and you'll realize that recreating this with OTF or TTF is impossible. It can't be done by clever designers with filters and effects. You need variable transparency. It's not just about simulating analog effect but discovering new ways of designing the alphabet.

3D: this is not just treatment. I want to be able to define a real 3D mesh right in the font format. Yes, a designer can use current flat font formats to create their own 3D models of glyphs but I want to define those shapes myself. I want them to be rendered exactly in the shapes I've defined, not a generic 3D extrusion. Yes, I can sell 3DS models of letters but I want these things to be used as blog headlines, not manually set up and rendered. You've seen those wonderful; gold leaf. deep engraved letters that sign painters make? It's not the same effect as using some lousy "extrude" filter. The font designers need to be able to define these shapes precisely. Why can't I provide that to a customer? Because current font formats don't allow it. Not because I don't know how to 3D model. It's a format limitation.

A blur channel isn't just treatment. How can I define what's transparent & blurry and what's just transparent? I want to define that if I want to. If a new font format had channels which could be defined, we could have as many definitions as we like. How else can I make a font out of alien ghost orb gel?

Accent positioning: a smart font format should define how diacritics link to glyphs. A dumb, Reagan-era PostScript based font format (like OTF) actually contains every variation that the font designer could be bothered with. At least TTF has components, that's a little smarter. If I make a font with hundreds of letter variants, am I supposed to create hundreds or thousands more accented variations? If I'm supporting Vietnamese glyphs, am I supposed to throw in an another extra thousand variations to support every possible combination? The more languages are added the bigger the Unicode Private Use Area trash heap gets. I've made some textures fonts where I had to eliminate Vietnamese glyphs just to make the font load properly in Windows. It's hard enough dealing with complex textures but when you need to support more and more languages, the files can get really large and the time delay when Windows rasterizes it gets longer and longer.

russellm's picture

Serious question: Who needs fonts with gradient fills etc? It is so easy to ad your own effects, why would a designer want to be tied down to the ones that the type designer liked.

I can just see the tidal wave of cheesiness that would follow.

The fact a thing can be made doesn't mean it should be made. :o)

Nick Shinn's picture

Right, just like all typographers do their own kerning, in Quark, because they can, to avoid being tied down to the font designer's idea of spacing.

And like nobody uses Photoshop filters, but makes their effects with the basic tools, just like in version 2.0.

If anybody is going to make animated crimson dimensional dripping blood typography, why not a type designer integrating the effect into the typeface?

John Hudson's picture

Accent positioning: a smart font format should define how diacritics link to glyphs.

Er, you mean like OpenType GPOS anchor attachment points for mark positioning?

The more languages are added the bigger the Unicode Private Use Area trash heap gets.

Don't use PUA for language support.

russellm's picture

Well Nick, because if animated crimson dimensional dripping blood typography is really needed (cool as that sounds), it will not make the world a better place, and probably will have to be customized to the design it's need for in any event, and that customization is not difficult fo the graphic designer to do.

Kerning s different. Its a basic question of functionality.

Ray Larabie's picture

Not every font is a text font. There are other fonts called display fonts. What about using gradient alpha channels to adjust weight as opposed to thinner lines, dot and line patterns? Do you think I shouldn't have the tools to create that? What about realistic pencil lines? Look at real pencil writing? Achieving that with pure black and white is a farce. What about watercolor script? Real marker? Fake letterpress with varying ink transparency. It seems like you want to limit the technical capabilities of fonts as a matter of taste.

As for better dripping blood fonts, that's a good idea. Currently it's all the same on/off consistency. If if could be smeared with transparency, that would look more like real blood.

> Er, you mean like OpenType GPOS anchor attachment points for mark positioning?

Yeah, like that except that it would be supported to the point where I wouldn't have to include my own accented versions.

Nick Shinn's picture

How about flourishes, Russel?
Isn't it better to have them integrated into the font design, rather than either "snap on", to be assembled by the typographer, or achieved by drawing them in Illustrator?
How are other potential effects generically different?
This is the age of smart fonts, since OpenType features.
Either an end-user works on the basic typesetting that the font provides, or it is packaged as a labour-saving feature.

russellm's picture

All good points, Nick and Ray.

I was not asking to be contrary. I was just having a hard time getting my head around the point of some of those things.

eliason's picture

This kind of transparency could potentially be useful on text settings too. For example, if it's true that serifs help serve readability by strengthening terminals, perhaps that strengthening could be accomplished in a sans face by relative opacity.

Vladimir Tamari's picture

A very interesting thread. There are now so many fonts available, and more being churned out as young people discover the joys of making their own designs. Fonts reflect the personality of the designer and user, and hence their popularity. We are all the fools Eric Gill referred to to explain the variety of types. One can take the simile further. There is an infinite variety of human faces based on a handful of common features (eyes, nose, mouth, etc.) yet we can recognize our loved ones, friends and enemies using subtle differences. Same with typefaces. And as rare as it is to find two identical human faces, fonts may look a lot alike yet have subtle differences that make one fine or noble, funny, leering, ugly, friendly, etc. etc.

I second the need for one of typodermic's items in his wish-list: skeletal linear fonts that can be defined by a few points. In use, thickness, stroke angle and other features allowed by such a font are applied. This would make Arabic type for example behave almost like the calligraphy it grew from. The software will have to be programmable to avoid clashes and other effects deemed ugly or otherwise unacceptable in the given script. A friend once explained to me in detail why such a system would not work, yet I hope it may one day.

Dare one mention animated effects and fonts that blush when the text gets too daring? Will the end of civilization as we know it be displayed in headlines made up of pulsating toothpaste-like florescent green 3D letters that emit little sounds as they read themselves aloud?

John Hudson's picture

Ray: Yeah, like that except that it would be supported to the point where I wouldn't have to include my own accented versions.

Well that's not a font format issue, and a new format would put us back to where we were with OpenType in 1997.

John Hudson's picture

Craig: This kind of transparency could potentially be useful on text settings too. For example, if it's true that serifs help serve readability by strengthening terminals, perhaps that strengthening could be accomplished in a sans face by relative opacity.

Bzzzzzt. Using relative opacity to strengthen some feature implies reducing opacity, and hence stroke density, in other features. One of the few really solid pieces of empirical evidence we have about legibility at typical text sizes is that strong, consistent stroke density is crucial, which is why antialiasing isn't preferable at all sizes and resolutions to b/w aliased bitmaps. So anything that involves reducing the opacity or otherwise diminishing stroke density is a bad idea for text readability.

I think Ray's desiderata are almost entirely limited to display type. Which is not to say that they are ignorable. Ray, have you looked at Photofont?

John Hudson's picture

Vladimir: Will the end of civilization as we know it be displayed in headlines made up of pulsating toothpaste-like florescent green 3D letters that emit little sounds as they read themselves aloud?

Probably, and I'm pretty sure that MS Word will support such features before it supports GPOS kerning.

Si_Daniels's picture

Nick >After all, nobody could recreate, say, even well-known faces like Helvetica or Times from memory;

Sorry to veer back on topic, but I think John Downer has proved otherwise...
http://www.flickr.com/photos/nicksherman/1063935926/

eliason's picture

Bzzzzzt

Okay then, forget opacity, how about just having blinking lights at the terminals?
;-)

Nick Shinn's picture

John Downer has proved otherwise...

He painted a similar-looking grotesque.
Look at the two "S"s on the right of his demo, they are different from one another, and in both the top counter is almost the same size as the bottom counter; in Helvetica, the top counter is smaller than the bottom one, the diagonal stroke slightly higher.

Si_Daniels's picture

Maybe "proved" is too strong. But John's point is that he believes it's possible to clone a font from memory, and that he's not the only person with such an ability. Although i don’t agree with John on everything, I think that a sign painters consistency demonstrates that this might be possible.

John Hudson's picture

John Downer wasn't just recreating Helvetica from memory, he was also painting it freehand with a brush, with only horizontal guidelines. So I'm not surprised there is some variance in the letter shapes. The point is that he knows Helvetica well enough to hold the letters in his head, and presumably in another medium could reproduce it with even greater fidelity, e.g. as a digital font.

Ray Larabie's picture

Photofonts: I never got into it because it didn't seem to be something that would catch on for general use. It was more of a gut feeling I had that it wasn't going to be the format that people were going to go for. If Photofonts had caught on, I'd have less to complain about. What I like about Photofonts is that I would be able to buy lots of tax deductible items to photograph. Sausages, carrots, licorice . . . I could use most of my groceries to generate fonts.

Another use for varied transparency could be to increase the visibility of certain aspect of extralights and ultralights. If you've designed an ultralight, one of the problems that always comes up is disappearing dots. A tittle can vanish if it's not lengthened or thickened (moan). Commas need exaggeration. Periods and diacriticals can be easily lost. Rather than fattening them up, you could increase the transparency of the rest of the font, leaving the little parts more opaque.

Another one: use transparency gradients to simulate capillary action of letterpress. Or perhaps better, do a similar effect in a non historical context. The edges of the shapes could remain sharp, while the middle of the shapes could be subtly transparent. The degree to which this is done could be used to fine tune overall color. It's one more tool apart from line width that we can work with. Craig mentioned beefing up the terminals by varying transparency: I've been itching to do that for years. I've seem it done in calligraphy. You can put more emphasis on the parts of a letterform that you think are important.

Stroke fonts: I'd love to see stroke fonts make a comeback. I'd like web designers to be able to declare stroke widths in their style sheets. Neato.

Ray Larabie's picture

Back to the original direction of this thread. Everyone has a different idea of what's a copy, what's close or what's nothing like the original. The more modular a font is, the greater the chance of coming to the same conclusion someone else has. When I made GGX88, I made no visual references to Helvetica. It was commissioned as an set-top-box text companion for Avant Garde headings. I built a generic sans heavy weight. Balanced the widths and blended it to a mechanical ultralight. No Helvetica was used to make it. The more you balance a modular, sans-serif, expressionless grotesque, the more it looks like Helvetica. While there are many differences between GGX88 and Helvetica, you can see a resemblance, right?

Avant Garde is another font that's easy to inadvertently resemble. Make a geometric sans using circles straight lines and make rational optical adjustments. Rimouski & Meloriac.

When I find myself in that kind of situation, I have to make a decision. Should I take it in another direction? Abort? What's more important? Making a useful font or coming across as original? As the century progresses, font designers will have to become more comfortable with how close they can get to existing designs, no?

Nick Shinn's picture

The point is that he knows Helvetica well enough to hold the letters in his head, and presumably in another medium could reproduce it with even greater fidelity, e.g. as a digital font.

I'm sure he could create a very convincing Helvetica from memory, but it would nonetheless be an interpretation.

Here is the same character from three different Helvetica Regular fonts. (Helvetica, Helvetica Neue, and Helvetica BQ.) The cap height, sidebearings, stem width, proportions and curve shapes are different in each.

As the century progresses, font designers will have to become more comfortable with how close they can get to existing designs, no?

That's not my intention.

Ray Larabie's picture

>> As the century progresses, font designers will have to become more comfortable with how close they can get to existing designs, no?

>> That's not my intention.

Perhaps now it's not an issue but add another half million fonts to the mix and all font designers will be treading close to existing designs no matter what they make. Maybe. My point is that the gaps between designs are getting filled in.

Nick Shinn's picture

Relax Ray, history has no end.
Sure, with proliferation there is always an increase in "ordinary" work, we've already seen that.
But there is always plenty of opportunity for new stuff, as Mark proved mathematically.

The problem is not that there is nothing new, but that most people prefer something more familiar.
So strike the balance.

John Hudson's picture

Ray: What I like about Photofonts is that I would be able to buy lots of tax deductible items to photograph. Sausages, carrots, licorice . . . I could use most of my groceries to generate fonts.

You're thinking way too small. Let's say I'm going to make a font out of carrots. First I need to grow the carrots, so I deduct the cost of the seeds. Then I deduct the cost of the lumber and power tools that I used to build the raise beds in which I plant the seeds. Then I deduct the cost of the fence I build around my property to prevent the deer from getting in to eat the carrot heads, not to mention the cost of the electronic opening gate.

Vladimir Tamari's picture

John Downer wasn't just recreating Helvetica from memory, he was also painting it freehand with a brush, with only horizontal guidelines. So I'm not surprised there is some variance in the letter shapes.

John and John may have discovered the basic mechanism for font development - Darwinian evolution. Each new font varies slightly from the others, and the ones fittest for use survive.

Syndicate content Syndicate content