Italics: Optical Correction; Punctuation

SebastianK's picture

Good evening/morning,

I'm drawing an italic to accompany an old-style face (drafts here and here), and slanting + cursify-ing the Roman just doesn't do. My o looks like an easter egg.

Apart from this wonderful comment, I couldn't find any specific tips on how to do optical correction; neither here nor via Google (Books). So here's the newbie asking for your help:

1. How do you "push out" and "pull in" curves, e.g. in o, c, and e, without ruining the slant, to make round curves round again? Do you, to be precise, just shift control points back and forth til' you're happy? Is there a mathematical function of angle, thickness and contrast factor? — Or could I establish such a thing and become rich?

2. What are good guidelines for slant variation? My f seems like it's tipping over. Again, I can't find numbers, and I don't completely trust my eyes (yet). Then again, I personally dislike the un-slanted capitals in Arrighi- and Granjon-inspired fonts.

Unfortunately – I'm an engineering student, not a designer – I don't own any literature; and the few fonts I've looked at failed at teaching me an efficient way of consistent correction.*

(Recommendations are more than welcome though, there are libraries after all.)

---

One other, unrelated question to the book designers here:

3. Bringhurst §5.3.2 recommends, among other things, upright parentheses in italic text. While I don't disagree with the look of it, I was surprised at the very least. Is this normal practice?

Thanks!

Sebastian

* I'd open the Roman, slant it, paste it onto the italics and look at the differences.

kentlew's picture

Is this normal practice?

No.

eliason's picture

I started a thread on that Bringhurst comment here.

Thomas Phinney's picture

Besides Berlow's comment, see this very helpful tutorial from Briem:
http://briem.net/2/2.3.4a/2.3.4.31.italic.htm

Especially (though not only) this page:
http://briem.net/2/2.3.4a/2.3.4.34.curves.htm

Cheers,

T

oldnick's picture

re #1: sometimes you can obtain a pleasing appearance by ROTATING characters such as o and s clockwise, the same number of degrees as the italic angle slant. In such circumstances, you would of course have to relocate the extrema. This little trick doesn't always work, but it's worth a try...

Nick Shinn's picture

Is this normal practice?

It is the best practice when an italic word/phrase within a paragraph of roman text is parenthesized.

Nick Shinn's picture

could I establish such a thing and become rich?

Ask Igino Marini how wealthy ikern has made him.
He is also an engineer.

The short answer is no, because it's one thing to invent something impossibly marvellous (even if you can get general agreement on that), quite another to make any money from it. That's been the type case since Gutenburg.

SebastianK's picture

Thank you! Briem's tutorial looks helpful indeed.

Rotating sounds good too, Nick ... I'll try that.

(Nick Shinn: I was joking. No ambitions, really :-)

kentlew's picture

It is the best practice when an italic word/phrase within a paragraph of roman text is parenthesized.

Most publishing houses have style guides regarding this. There are a handful of schemes, usually referencing either the surrounding text or the enclosed text, then sometimes adding specific guidance regarding mixed styles.

Nick describes a circumstance where the style of parentheses follows the surrounding text regardless of the nature of the enclosed text. This approach certainly has many adherents. I believe this is more or less Chicago Manual of Style’s current practice.

But Bringhurst’s rule is advocating something slightly different — always upright, regardless of context (surrounding or contained matter), even if all italic, like so.

Bringhurst’s own sidenote explains “This rule has been broken more often than followed since the end of the 16th century.” It is worth noting that the period prior, when this ‘rule’ was more often followed, is also a period wherein the capitals used with italic are still largely upright as well.

I’m not arguing that one should or should not employ this approach. The point of my terse reply to the third point in the initial post is simply that this strict approach is not what most compositors would consider “normal” practice today.

blank's picture

Is there a mathematical function of angle, thickness and contrast factor? — Or could I establish such a thing and become rich?

I think this is where you would likely run into the problem Metafont has. Sure you can do it all with math, but then the font designers also need to be mathematicians, and people who can be both are rare.

Thomas Phinney's picture

One trick I've tried for designing italics of letters like O and S, is after getting the initial weight and condensation correct, get the slant by doing half of it via obliquing and the other half by rotation. One catch on the rotation is that it affects the height, so you still can't get away without doing at least a few manual corrections.

Cheers,

T

Bert Vanderveen's picture

This may be of interest: RMX Slanter. To be found here:
http://remix-tools.com/

SebastianK's picture

That's extremely helpful insight, guys. Thank you so much. I'll consider updating the Wiki once I'm comfortable with the method (would that be appreciated?)

Bert: Some day, yes, I wish ... right now it's a low-budget project (Ubuntu + FontForge) :)

Thomas Phinney's picture

RMX Slanter totally rocks. It doesn't do everything, but it is pretty helpful as a starting point on a lot of glyphs....

Of course, I am a big fan of the Font Remix Tools in general:
http://www.thomasphinney.com/2010/03/font-remix-tools-rmx-and-multiple-m...

Cheers,

T

oldnick's picture

I am a big fan of the Font Remix Tools in general

I would certainly like to give them a try, but every time I have attempted to install the demo, I get an error message saying that the module os.path is missing...

blank's picture

RMX Slanter totally rocks. It doesn't do everything, but it is pretty helpful as a starting point on a lot of glyphs....

It does until a design requires non-extrema control points on curves :(
But that’s why I also licensed UFOStretch!

Tim Ahrens's picture

Nick,
this error message is puzzling. The module os.path should be available on any system, it's a standard Python module. Cann you tell me which OS and Python version you are using? Any further error messages? We might be able to track down what causes the problem.

oldnick's picture

Tim,

I'm running Vista Home Premium SP2. The only instance of os.path I can locate on my hard drive is within a text file which explains what's new with Python v2.5.2, so I am presuming that's the version I have installed.

When I drag the RMX installer into FontLab, I get this...

Traceback (most recent call last):
File "", line 4, in ?
ImportError: No module named copy_reg
Traceback (most recent call last):
File "", line 4, in ?
ImportError: No module named os.path

The "No module named copy_reg" comment seems to come up ANY TIME I get an error message, so I don't know how relevant it is...

Tim Ahrens's picture

Nick,

as mentioned on the Remix Tools website, RMX requires Python 2.3 or 2.4.

This is because FontLab in general seems to have problems with the newer versions of Python. See the official FontLab Python info on FontLab Studio 5 for Windows:

FontLab Studio 5 and AsiaFont Studio 5 for Windows work with Python 2.4. You need to download and install it on Windows. Python 2.2 and 2.3 will also work. Python 2.5 will not work.

I suggest you install Python 2.4, then RMX should work.

Tim Ahrens's picture

Btw, the older installers are a bit difficult to find on the Python website.

This page has the download link:
http://www.python.org/download/releases/2.4.4/

oldnick's picture

I suggest you install Python 2.4, then RMX should work.

I downloaded and installed 2.4.4 and RMX installed just fine. However, I'm not seeing some of the features "doing their thing." I tried Harmonizer on an uppercase and lowercase s, and nothing noticeable changed. I tried Slanter, and it slanted alright, but didn't maintain orthogonal extrema. Curious...

Tim Ahrens's picture

Nick,

I tried Harmonizer on an uppercase and lowercase s, and nothing noticeable changed.

I'll simply quote the RMX manual:

Note that the Harmonizer does not modify the shape as a matter of principle. How much a glyph gets modified glyph depends entirely on the deviation from the required conditions. Consequently, a shape that is already harmonious will not be changed at all if the Harmonizer is applied to it a second time.

Did you see the Harmonizer video? It also demonstrates why this is the case.

I tried Slanter, and it slanted alright, but didn't maintain orthogonal extrema. Curious...

With extreme slanting, or if there are too many additional (diagonal) control points then some orthogonal handles are not retained. If you want you can also send me a (one-glyph) FontLab vfb file, then I can have a look.

oldnick's picture

Tim,

I watched the Slanter and Harmonizer videos. Perhaps it's reassuring that my Ses can't be improved upon ;)

I don't think that 12° isn't an extreme angle, nor is the glyph over complicated...

Tim Ahrens's picture

Nick, the RMX Slanter and Tuner only correct nodes that connect two curves, not connections of straight line and curve like in your example. In that case, would you simply add nodes at the extremes? That can be done with FontLab's built-in tools. Or, would you maintain the orthogonal extrema by converting the straight to a curve and then adjusting everything? This would be a bit complicated, I am afraid, and it involves design decisions. The Slanter, like all RMX tools at the moment, does not insert nodes or change the node structure. Not treating the outline is the only sensible thing I can think of in the above example - do you have an idea for a better behaviour of the tool?

oldnick's picture

Tim,

Changing the lines to curves before applying Slanter didn't change anything in the noted case; I presume that this means Slanter works with smooth curves, but not tangents, since I applied it to an all-curves S and it performed as advertised. Thanks for the personal attention in this matter.

Syndicate content Syndicate content