That's not a thread, that's a book! :-)
But here's something small: http://www.themicrofoundry.com/ss_trapping1.html

hhp

That’s not a thread, that’s a book! :-)

he he.
\\thanks for the link

http://briem.ismennt.is/2/2.3.2a/2.3.2.06.illusion.htm

http://acg.media.mit.edu/people/fry/alphabetbook

Hrant — Are you writing that book? Because if you don't, I will! :)

As for your Trapping Flower. To what extent is the model supported by real observations on the behaviour of type when printed? I mean, did you run tests, is the model useable without further parameters such as type size, paper stock, pressure, ink quality, machine speed etc?

I inquired about this some time ago on the PPLetterpress list (in particular my posts from July 18 and 24, 2005). You seem not to be on that list (anymore)?

I am very very much interested in this stuff, as I am looking to develop letterpress optimised type. Of course, the technology could also be applied to offset printing, flexo, intaglio and even to digital laser printing at, say, 600 dpi.

The producers of dfRialto (which is probably one of the most chirographic faces designed in recent times ;) claim to have looked at your Trapping Flower, although it is not clear if they actually used it when applying the traps to the Pressa edition of their font.

[ What the heck is going on with the font here? Georgia is just fine in regular setting for on screen use, but I don't like to read the full body in italics. ]

Alessandro, thanks for these great links. really helpful.

Ludwig, the Flower is based on the revolutionary scientific observation that ink bleeds outward. :-> The parameters control it quite finely, but are not tied in any quantified way to things like point size, pressure, etc.

That could be done of course, but my point was mainly to provide a tool that a type designer could use to quickly and consistently implement trapping in a font once he knows what kind/size/frequency traps he wants. This latter depends on various attributes of the font on hand and its intended usage, and is done with a mixture of experience and iteration - something I suspect no tool could do for you! :-)

The Flower has a lot of power*, but it's still just a tool - you have to use it well. And first of all you have to use it; virtually all trapping we see out there is pretty shoddy: uneven, inconstant, malformed, you name it. Even something like DTL Fleischmann, as wonderful a face as it is overall, has pretty laughable trapping. If they had just used the Flower, it would've been smooth and silky.

* For example, it can accomodate asymmetrical bleeding (where -usually- the vertical will gain more than the horizontal, mostly due to usage of a cylinder press and/or unevenly-bleeding paper) with a simple modification. Can you guess what modification? :-)

The Rialto boys (the younger of which I met in Rome in 2002 - where I remember we had to communicate in Spanish! :-) made their -amazing- Pressa cut before learning about the Flower. They subsequently expressed interest in its usefulness, and were enthusiastic about using their letterpress experience to some day test a Python script of it. (Don't ask...)

As for the book: write away, my friend. Sadly at this point I
couldn't even promise to review the book you'll hopefully write... :-/

BTW, Satya, the "or" in the subject line of this thread makes me want to point out that sometimes visual corrections can become part of the style! Jurgen Weltin's Yellow font used by the British phone books is such a case. Also, sometimes visual corrections, such as traps, can migrate to the sphere of "style", pretty much abandoning their original intent; FB Amplitude being the best example of that (and FF Bradlo a much earlier such effort).

hhp

> the revolutionary scientific observation that ink bleeds outward. :->

So why the need for spikes? Ink does "bleed" (shrink) inward as well. That's precisely why Bell Centennial has those bulky stems.

> The parameters control it quite finely, but are not tied in any quantified way

This is somewhat contradicting, isn't it? Parameters are of course variable values, that are expressed by measurable quantifiers. We only need the right equation, measure the variables and there we are. Sure point size is of extreme importance: ink bleed is an absolute given, so it will be proportionally bigger in smaller sizes. Pressure, stock and machine speed (all measurable) determine the amount of bleeding.

> depends on various attributes of the font

I don't think so. The ink doesn't know whether it is squashing a serif, a high contrasted transitional face or one with a slant of 6 degrees. It's just outlines and angle width.

> something I suspect no tool could do for you! :-)

Sure. Everything a man can do, a machine can do. If it is understandable, and as long as understanding guides a man's doings, it can be programmed. And if it is programmable, it can be executed by a machine.

> DTL Fleischmann, as wonderful a face as it is overall, has pretty laughable trapping

Gerhard Kaiser was trained by Typo Art in Berlin, which was renowned for it's build-in ink bleed compensation in fonts for photo composition. Do you have showings of the fallacies? (E.g. an overlay of printed specimens over the digital outlines.)

> asymmetrical bleeding

This is a hard one indeed. I was reading in the Fleuron yesterday; the press-work is undoubtedly very good, but I noticed that the lines were thicker at their endings. Could it be that this phenomenon caused the invention of those eighteenth/nineteenth century tapered lines, that might compensate for this, which only later on became a fashionable feature as such?

> to some day test a Python script of it. (Don’t ask…)

Errr… Okay, I won't. Just, my two cents. I am convinced that building ink traps (etcetera etcetera) is not the task of the type designer. It shouldn't even be done in the outlines of the font, but at the stage of the RIP. The RIP can be fed with all the given variables (such as stock, machine type and speed, pressure, ink, etc.), which would allow it to calculate the ink bleed compensation on the fly. There is simply no other way, because of the huge amount of parameters. Compare it to dot growth compensation in screened images.

> As for the book: write away, my friend.

I will! Unfortunately I have no time for it either. (I am already engaged in writing a PhD with a completely different subject matter. Oh man, if only typography had been a prime academic research field!)

Spikes? If you mean thorns (the opposite of traps), the Flower does not do those. Although they're also useful, they tend to be less so, as well as much more intrusive. In the photo days of massive definition loss thorns were critical in preserving corners, but offset or even letterpress aren't nearly that bad (except when the printer is a dork).

Bell Centennial has traps to counter bleeding (and optical correction) but the stems are thick not because they get thinner during printing! They're thick because: you're probably looking at the Bold Listing weight; and small type needs to be heavier.

Ink bleeds in?! If you mean the "halo" that forms around the outlines of letterforms: I don't see how a font could help that; when they're too obvious isn't that simply really bad printing?

> This is somewhat contradicting, isn’t it? Parameters are of course
> variable values, that are expressed by measurable quantifiers.

What I meant was that there is not yet a formulaic relationship defined between the Flower parameters and the factors causing ink bleed. There can be, and that would be amazing, but the Flower is still very useful without such equations because it's a tool for efficiency and consistency. A kitchen knife won't make a salad for you, but without one you're a caveman.

> Do you have showings of the fallacies?

All you have to do is look closely at their PDF.
The traps are often missing, crooked and of random size.

>> depends on various attributes of the font

> I don’t think so.

You don't think that traps need to harmonize with things
like the vertical proportions, color, etc. of the font?

In fact even the style of the font matters. For example pronounced, angular traps would look totally out of place in something like Garamond #3, but totally harmonious in the work of John Downer*. Why? Because in digital type traps can end up being visible. This is in fact why some designers avoid traps entirely. But when you consider that a font is ideally designed for a not-too-large range of sizes, and when you consider that type is all about the details, traps make sense - often if not always.

* Unfortunately Downer is (or at least used to be) a member of
the Slimbach/Porchez school of "don't ever need 'em", IIRC.

> Everything a man can do, a machine can do.

Well then I'm glad I'm not married to you.
I don't believe that humans are algorithmic, but heuristic.
A System Cannot Understand Itself.

> This is a hard one indeed.

Not for the Flower. All you need to do is use ovals instead of circles! :-)

> I am convinced that building ink traps (etcetera etcetera) is not the task of the type designer.

Theoretically, I agree. But as long as that's not a reality on the ground a craftsman can rightly worry about traps in his fonts.

Also, something you're ignoring: traps aren't just about bleed, they fight optical aberrations too. And that can depend on things beyond the printer's control like the presumed age of the reader of the font (like in a newspaper), the lighting, etc.

> There is simply no other way

Only if you think Perfection can exist, and that it's good. Like if you think your camera needs to be in Perfect Focus before you hit the button. I think Perfection is a succubus. The bottom like is that, in practice, humans can benefit from traps in fonts. Anything else is too Modernist, too romantic and naive.

hhp

Okay, this is bad printing (letterpress). But it shows the more obvious what happens in _all_ cases, albeit less extreme. Ink gain deforms the glyph shapes in a typical way, patternlike: corners get smoothed, the centre of stems gets thicker etc. This you can analyse and the behaviour can be programmed/simulated.

> Only if you think Perfection can exist, and that it’s good.

Perfection, by its very name, is good per se. I don't know (yet :) if it can exist, but as imperfect beings we should, and by our inept nature, naturally tend to strive for its realisation. You imply a relationship between modernism and romanticism? I like that. And take the naivity for granted.

Satya, if you mean by 'visual correction' a deviation from simple geometrical shapes and alignments, then practically every letter has 'visual correction'.

Briem's wonderful web site indicates some of these corrections: the overshoot of the o and O, the top half of double story letters being smaller, compensating for optical illusions such as the 'bone effect' and 'broken straw' in the X, the round vertical arches thicker than the vertical stems etc., etc. But beyond this is the adjustment of shapes for uniform color. You could almost say that what is specifically typographic in type design is a matter of optical adjustment.

Ink trapping, which can be helpful in small type, I would not call optical adjustment, but rather design to counteract ink spread and get a different shape in print than what you design on screen.

What I would call optical adjustment is a matter of designing those final shapes for the eye so that printed letters in mass knit together in words and read easily. And that's at least half the battle, the other half being aesthetics.

> This you can analyse and the behaviour can be programmed/simulated.

Yes. On the other hand:
1) It's much much harder than using the Flower, and more significantly hard enough that I think it is unlikely to ever be implemented, basically because there's not enough money in it, in fact not enough money in type as a whole.
2) Randomness (evident in your sample - and every other sample) cannot.

Perfection: I believe that striving for it is a waste of Life.
Imperfection is where the taste is.

hhp

> If you mean thorns (the opposite of traps), the Flower does not do those.

On second thoughts, yes it does. Just in the same way as you use the Flower for building ink traps, but only the other way around :)

Wow, I should have known! Thanks much.

hhp

You're welcome. BTW, do you have a scalable file of the Flower (.eps, .pdf)? I'd like to try out.

I have it as a glyph within FontLab, but growing one from scratch is easy.
Let me know if/how you like it!

BTW, if you do use it to make thorns, use smaller numbers than you would for traps.

hhp

I guess an eps file will do. But if it's too much trouble, I can get away with almost all formats. I turned on my contact setting, so you can mail me.

I'll try the thorns too. And I am planning to run tests (laser printer, letterpress and offset), so that I can approximate the exact amount of protruding for each output device.

OK, I've prepared a new Flower for you, with a somewhat finer grain (and in EPS). But I'll need your email address since Typophile messages can't carry attachments (except -inline- images). Mine is: hpapazian-at-gmail-dot-com

> I am planning to run tests

Awesome! Thank you.
Please let me know the results.

Try starting with an Aperture of 80 ems and a Multiplier of 3, and then vary to taste. You can also vary the Minimum ring (by selecting a larger ring, or by adding smaller rings by scaling down the largest ring in steps of 5%) to control the trapping cutoff.

BTW, as you might already be aware, the historic pronounced (like in mid-grade newspapers) gain amount, which surprisingly still seems to hold to this day, is one mil (1/1000-th of an inch). I suspect a laser printer and letterpress will generally come close to that, while (good) offset will be much lower.

hhp

Hrant — Thanks a lot for the file. It will take some time, though, before I will actually use the Trapping Flower. I am still busy designing my typeface, and since I am a novice in type design, this takes time. Lots of time. I only started last month, but hope to have a useable version till the end of this year, that is one without optical and ink squeeze corrections.

> and then vary to taste

This is precisely what I don't want to do. I aim to apply the corrections based on empirical data, rather than taste. For the tests I have in mind, I will take printed specimens (letterpress to begin with), scan them and overlay the original digital font. This will show both the overall deviation and the trouble spots. From there I will apply ink traps and thorns, using your Flower, along with a general slimming of the outlines, using Gerald Lange's sequences.

Good luck with finishing your typeface! But you can still
test things before you're finished (or even very far ahead).

> I aim to apply the corrections based on empirical data

1) This will be very hard to pin down properly - I mean for more than a single configuration of paper, press, ink, etc.
2) Although I wasn't using "taste" in that way, you can't remove taste from something made by a human, for humans, can you? I wouldn't even want to.

But still, any quantitative progress you make would be highly valued.
I would certainly incorporate your findings into the Flower's "readme".

hhp