Sounds like a job for... Metafont!

You might ask Linus:
http://typophile.com/node/73827

hhp

What's Briem's method, please?

It's on Briem's site:

http://66.147.242.192/~operinan/2/2.3.4a/2.3.4.34.curves.htm

(Sorry for the weird URL; the site is built with frames, and this was the only way to get a URL to the pertinent page. The main URL is http://briem.net/ There's lots of other interesting notes about type design there as well.)

The transformation tool in Glyphs can do some optical correction. See page 5 of the manual.

Thanks for your comments, I'll check Glyphs out.

What about the manual methods, aside from Briem's? How do you do the obliques/italics?

Another question, the Briem's method is so schematic that it got me thinking that it must be possible to be translated into a script.

Unfortunately not. Briem does not give any information on the x-coordinate of the centre of the rotation, which is not obvious for asymmetrical glyphs.

I have thought long and hard about a universal, robust mathematical answer to the problem of slanting compensation but haven't found any yet.

Unfortunately not. Briem does not give any information on the x-coordinate of the centre of the rotation, which is not obvious for asymmetrical glyphs.

Oh, I see now.

I have thought long and hard about a universal, robust mathematical answer to the problem of slanting compensation but haven't found any yet.

Tim, I hear you. Maybe fully mathematical isn't possible, maybe slanting two uprights as multiple master could be the way to go? Could you elaborate the problems you faced?

I have thought long and hard about a universal, robust mathematical answer to the problem of slanting compensation but haven't found any yet.

Even a solution that is not universal would still save designers plenty of time. And that leaves more time for the tough glyphs like S and 3.

Before I found Briem's technique, I used a different method. I don't think it's as reliable, but the results weren't terrible. Also, I don't think it would lend itself to some sort of script. That said, here's what I did:

1. Slant the glyphs twice the target angle. For example, if the italic slant is to be 11 degrees, slant the glyphs 22 degrees.
2. Visually correct the glyphs until they don't look distorted. (This is the tricky part since it depends on your visual judgment, which may not be as good as you think it is.)
3. Interpolate with the original upright glyphs.

The idea is that in the twice-slanted glyphs the distortions will be easier to see and fix, and, if your corrections are not perfect, the imperfections will be reduced in the interpolated glyphs. I prefer Briem's method mainly because it's more mechanical and less dependent on subjective judgments, plus I'm happier with the results.

Here's another trick: don't slant too much. :->
Seriously, gentle slants are generally better anyway.

hhp

Seriously, gentle slants are generally better anyway.

I'd be interested in hearing a rational for that. Particularly with sans faces, I find that the less the angle on the oblique, the harder it is to distinguish the oblique from the upright; there is insufficient contrast. Although I like setting text in sans serif (although I'm seldom allowed the opportunity in book design) I dislike the use of obliques instead of a built italic just for that reason. Italics are often used to provide emphasis, and I think you need to be able to readily distinguish the italic from the roman for that to work.

Well, there's gentle, and then there's gentle. I agree with
your concern, and that's why I'm no fan of "upright italics".
I also think you have a point about the sans issue, although
when an Italic has too fewer serifs than the Roman that's a
problem too (in terms of "skewing" the voice of the Roman
too much - not being a good subordinate style).

The slant certainly has to be sufficient. And that depends
on a number of factors, for example the repro quality, which
can be hard to predict. However one thing it also depends on
is other attributes of the design - the Italic can deviate from
the Roman in various ways beside slant, such that the slant
can be in the single digits and still work. Look at the Italic in
FF-Ernestine*; it's only ~3.6 degrees, but works like a charm.

* http://ernestinefont.com/

But why avoid steep slants anyway? Readability. The patterns
established by the more significant Roman (and Romans) are
increasingly violated with increasing slant. This BTW is also
why a descending "f" in an Italic is a bad idea.

hhp

Slant is not the only delimiter of italic.

Thanks Mark! That's actually very clever. And very good for that "training your eyes" part.

One other thing about my old method: It might not work so well in font editors that use only integer coordinates. When I was using that method, I was still drawing my typefaces in Illustrator. I don't draw typefaces in Illustrator anymore and I don't recommend it. However, some font editors can work with fractional coordinates, including Fontographer and Robofont. (Edit: Robofont seems to allow use of fractional coordinates, but they don't seem to "stick", rounding to an integer next time you click on the point.)

Thanks, Mark :-)

Thanks, Mark :-)

In all this figuring out a workable process, don't forget the Slantarant about FontLab's angle reporting...