How do I start working on italics?

Ori Ben-Dor

I'd like to start working on italics for my font (Portfolio Mono, see a separate thread).
The problem is I've never done italics before, so I'm not sure how to approach this task.

It's going to be true italics, but related to the roman enough to consider the roman as a starting point.

I'm considering starting from:
1. scratch.
2. the roman slanted.
3. the roman manipulated by some fancy tool that is supposed to give better results than simple slanting.
(Am I missing any option?)

How would you approach this?
Any tips?
What tools do you use for better slanting?

Jasper de Waard

I would go for 2. Yes, there are tools, but since this is your first italic I think editing the slanted glyphs would be a useful lesson, and allow you to judge the quality of existing tools for yourself.

Tools I know of: RMX slanter, and Glyphs' slant function.

John Hudson

Start with slanted roman for the uppercase, and spend a lot of time figuring out what you need to do to the round letters and bowls to make them work. That's the best way to understand the difference between slanted forms and forms written on a slant.

A 'true italic' lowercase, i.e. one with a cursive ductus construction, is a completely different beast from a roman, so there's no point in starting with a slanted roman. Better to start from scratch.

Tip: because a diagonal line between two heights is longer than a vertical line between the same heights, the x-height of an italic tends to appear optically larger than that of a corresponding roman, so you should compensate for this by making the italic x-height slightly shorter. You don't need to do the same compensation for cap height or extenders, because these don't have the same impact on perceived text size.

Johannes Neumeier

One thing to keep in mind when using a slanted roman (and generally when having a first go at italics) is to compare darkness between roman and the italic you are developing. Simply slanting them will likely not result in an even grey value between the two.

Hrant Հրանդ Փափազեան Papazian

Ori Ben-Dor said:

It's going to be true italics

There's nothing "true" about a subordinate style distracting from the tone of the Roman.

John Hudson

I avoid the term 'true italic', except when quoting someone else's use of it, even though I know what it conventionally means in these discussions, and it does make sense in terms of the original model of italic derived from Italian renaissance cursive. There are so many ways to design an italic, and so many ways to balance or subordinate, harmonise or distinguish in combination with a roman.

Hrant Հրանդ Փափազեան Papazian

John Hudson said:

There are so many ways to design an italic, and so many ways to balance or subordinate, harmonise or distinguish in combination with a roman.

True. It remains that the formulaïc approach we usually see yields an Italic that might be great on its own, but is too informal to properly fulfill its typical role. I call it the Aldine shotgun wedding.

Slant being the cornerstone of what readers expect out of an emphasis style, without any further context a (non-mechanical) slanted-Roman is a much more sober starting point than what we type designers are conditioned to value.

[Deleted User]

The user and all related content has been deleted.

Hrant Հրանդ Փափազեան Papazian

James Montalbano said:

Ed Benguiat's advice on making Italics from uprights was to slant the round forms half the amount of the desired angle and then rotate them again by half.

I think I'd heard that from Briem too. Maybe he got it from Benguiat...

Thomas Phinney

It is also Briem’s advice, yes.

I came out to the same place, after getting less-satisfactory results from both slanting and rotation on their own. Just as a starting point, for round-ish shapes.

Chris Lozos

I have spent quite some time attempting to marry the partial rotation partial slant technique.  I get what it is going after but feel it is more time consuming than just redrawing the curves as I see them.  This goes back to the very long process of learning to see and evaluate what you see.  Whatever methods or experiments I have tried, including mathematical variances, have helped me learn to see but you still have to do the work.  First you have to face the facts that having done the upright does not mean you are even half way to the italic.  Hopefully, you have come to enjoy drawing letters by know since that is the most rewarding part of the process. I can state emphatically that the italic is much more work than the roman to me.

Ori Ben-Dor

Thank you all for the valuable input, I really appreciate this!
I leaned toward starting from scratch before and I think I'm ready to take this approach now.

Rafael Cases

James Montalbano said:

Ed Benguiat's advice on making Italics from uprights was to slant the round forms half the amount of the desired angle and then rotate them again by half. So a 12 degree italic angle would be slanted 6 degrees and rotated 6 degrees. I have mixed results using this technique. I think it probably worked better in film.

Here's a typeface that applies Benguiat's technique:

https://www.myfonts.com/fonts/nootype/solanel/

Georg Seifert

What "fancy tool" did you try?

There is a "cursivy" option in Glyphs.app that does some of the compensation that is needed when slanting.

Daniel Calders

Charles Nix has an excellent online tutorial on making an italic, starting from an upright roman, which is worth checking out, especially as it is your first italic: https://www.linkedin.com/learning/design-an-italic-typeface

Rafael Cases

This one is a fundamental for designing Venetian-Garalde hybrids (for LGC) based on seeing numerous Latin, Greek and Cyrillic typefaces:

Latin:

Uppercase fundamentals: (all have a contrast axis of 30 degrees, except for diagonal contrast and rhythm proportions)

M - basic serifs, square and diagonal contrasts, rhythm and proportions
O - basic round shapes
V - basic triangular and diagonal shapes
------
H - basic hairlines (prerequisite: M)
L - basic horizontal serifs (prerequisite: M)
------
C - basic open curves, basic colour balancing (prerequisites: O)
S - basic spine construction (prerequisites: C)
U - basic curved bowls and contrast alterations (prerequisites: V, O)
------
N - basic rectangular proportions (prerequisites: H, M)
Y - basic junctions (prerequisites: V, H, M)

From there, all other Latin glyphs can be designed.

Lowercase fundamentals: (axes are variable, though default is at 30 degrees)

h - for double-vertical elements and serifs (default axis)
l - vertical elements (default axis) (prerequisite: h)
o - round elements (start at default axis, but end the axis of contrast at around 0 degrees, but at 120 degrees in italics)
v - basic diagonal elements (same as uppercase v in upright, but in italic, if the v has a bowl-type counter (e.g. Caslon and Stempel Garamond), axis is at 60 degrees)
----------------
z - for reverse-contrast elements and diagonals (axis is at 60 degrees)
----------------
c - basic open curves, teardrop and ball terminals, basic colour balancing (default axis) (prerequisites: o)
e - basic hairlines and diagonals (default axis) (prerequisite: c, H)
b - basic joints for curved and straight elements (default axis in upright, 150 degrees in italic, except for the vertical element) (prerequisites: l, o)
m - basic joints for curved tops and vertical elements (default axis) (prerequisites: h)
----------------
s - for lowercase spine and colour (default axis) (prerequisites: o, S)
----------------
a - for ball terminals, exit strokes, double-storey and single-storey studies in upright and italic (default axis in upright, 150 degrees in italic, except for the element with the attached finial) (prerequisites: s, c, b)
g - for ball terminals, exit strokes, double-storey and single-storey studies in upright and italic (0 degrees in upright, 150 degrees in italic, except for the descending stroke element) (prerequisites: a, o, s)
i - for diacritics
x - for the basic x-height and diagonal crossing elements (prerequisite: v)

Once these are done, all other Latin lowercase glyphs can be designed.

Greek:

Uppercase fundamentals: (all have an axis of 30 degrees, except for diagonal contrast and rhythm proportions)

Α - basic hairlines, triangulars and diagonal shapes (prerequisites: H, V)
Μ - (pull from Latin M)
Ο - (pull from Latin O)
Ξ - basic hairlines, visual centres, proportioning (prerequisites: H, L, N)
------
N - (pull form Latin N)
Δ - triangulars and heavy horizontal contrast placement (prerequisites: V, A)
Σ - triangulars and proportions (prerequisites: Ξ, M)
------
Ω - basic heavy horizontal and curve joints (prerequisites: Δ, O)

From the Latin and the Greek basics, one can create the rest of the uppercase Greek glyphs.

Lowercase fundamentals: (axes are variable, though default is at 120 degrees)

ο - round elements (start at default axis, but end the axis of contrast at around 0 degrees, italics in default axis)
η - for double vertical elements, entry strokes (or semi-serifs) and exit strokes below the baseline (default axis)
ν - for diagonal elements; take care not to conflate with Latin v (default axis)
ι - for exit strokes and / or semi-serifs (for best overall grey colour, the semi-serif version for uprights must be at 60 degrees, italics with exit strokes must be at 120 degrees)
----------------
ζ - for mirrored contrasts, half-descenders and basic vertical curved elements (for best overall grey colour, axis in both upright and italic must be at 60 degrees) (prerequisites: Latin z, S, s, g, c)
----------------
π - for perpendicular junctions and top-heavy glyphs (for best results, uprights msut be at 60 degrees, and if the italics will take the pomega form, italics must be at 30 degrees)
----------------
λ - for diagonal elements and curved top ascenders (default axis) (prerequisites: v)
κ - for the basic kappa height (in line with Latin x) (prerequisites: ν, ι, Σ)
χ - for diagonal crossing elements and diagonal / curved descenders (default axis) (prerequisites: λ)
----------------
ε - for open bowls, contrast pattern changes, colour for Greek glyphs, hairlines, lunate glyphs and possible change of form in italic (default axis) (prerequisites: c, o, e)
θ - for items within a counter and for a single-stroke version in italic (default axis) (prerequisites: o, λ)
----------------
ρ - for round items joining vertical elements (for best overall grey colour, uprights and italics must have an axis of 150 degrees) (prerequisites: η, ο)
σ - for round items joining horizontal elements, must be narrower than ο (for best overall grey colour, uprights and italics must have an axis of 150 degrees) (prerequisites: Ω, o, z, g)
α - a fundamental glyph showing the most Hellenic character; take care not to mistake for italic a, and it must be wider than italic a (default axis) (prerequisites: ο, λ, χ)

Once these lowercase glyphs are done, these can be used to add to IPA and to design the rest of the Greek lowercase, and a few Cyrillic glyphs.

The system may look complicated, but once the basics and the most difficult to design glyphs are done, all the rest will be easy.
In defense of Eric Gill, he was attempting to form a system to transpose Latin elements to Greek, but he did not take into consideration Greek legibility, geometry and sensibilities, which are now more widely available thanks to the Internet.  My gratitude goes to Andreas Stötzner for showing us one of the more playful ways to greatly improve on Eric Gill's system.
All axes have been determined by testing with a broadnib tool.  These have also been determined by my understanding of the historical broadnib pens, as well as by other designs.  I know my guidelines and definitions won't sit well with other people, but this is the best compromise that I know that unifies numerous conflicting axes to fit Latin, Greek and Cyrillic in one superfamily.

jeremy tribby

After some light Saturday reading (James Victor Gaultney's thesis on italics) I decided to give all the methods mentioned on pages 182-185 a try. I knew about Briem's method, but not Lucas de Groot's method (move points first, slant after). I didn't find any of them satisfactory on a sans. (Tangentially, I'd be curious if Mario Feliciano is using the same method or what he is hinting at in that Instagram post is substantially different, because his results are quite good IMO).

I experimented with a few more methods, and I think, when put together, the result isn't bad. So I scripted it for Glyphs, and published it here. It's not complete — like all automatic methods for this, it requires manual intervention. But I think it gives a decent starting point. I like it better than Glyphs' built in "Cursify" in any case.

Jacques Le Bailly's method for round letters, as implemented by Alexei Vanyashin, is quite good, so I brought that in. For less round letters, like /n, I found a method that is also passable: rotate by italic angle, skew by italic angle, rotate by negative italic angle. Charles Nix describes this in one of his videos, but I think I first heard this one elsewhere.

So my script does both, depending on the letter. And it simply slants the rest. It does re-adjust the spacing based on some math from Lukas Schneider and Frederik Berlaen.

Obviously there are problem with my script — I need to fix the way anchors behave, and components get messed up. But, it's just a start.

Thanks to Thomas Phinney, too, for providing a list of which glyphs not to slant on this forum many years ago.

Has anyone come across other automated/semi-automated methods I should explore for scripting?

Cheers

Jeremy

Craig Eliason

Thanks for sharing @jeremy tribby. I have a current sans project that I've Glyphs-cursified and then adjusted, but I might try this script out as well as RMX Tuner and compare results.

jeremy tribby

Would love to know how it works for you. I am still tinkering. Does RMX Tuner do something different than Glyphs' built-in slant? I hadn't realized if so!

Craig Eliason

I know RMX results in extrema nodes so it must be at least a bit different. 

Andreas Stötzner

jeremy tribby said:

…

Has anyone come across other automated/semi-automated methods I should explore for scripting?

…

For a sans the generic slanting of the upright may be a coarse starting point, for the capitals more than for the lc perhaps. For some glyphs (e.g. a, e, o/O, s/S) a slanted upright is just useless.
For all other sorts of type style I recommend this method: envision the glyph before your inner eye first, then draw from scratch. Guaranteed automatism-free!

James Puckett

For a sans or capitals I slant everything and fix it manually. I don’t think this takes any more time than using tricks and cleaning up all the details that get fucked up.

Chris Lozos

Diagonal letters like X,V,W,K R get the slanted strokes more distorted than the rectilinear ones. check stem weights of your uprights for reference.

Christian Thalmann

James Puckett said:

For a sans or capitals I slant everything and fix it manually.

My go-to routine for fixing slanted rounds (like /O/) is to power-nudge the top vertices left, the left vertices up, the right vertices down, and the bottom vertices right by about 20 units each (for a 1000 UPM), and then run RMX Harmonizer over it. That counters the over-shearing that the rounds tend to suffer.

jeremy tribby

James Puckett said:

For a sans or capitals I slant everything and fix it manually. I don’t think this takes any more time than using tricks and cleaning up all the details that get fucked up.

I don't necessarily disagree with you, but it's been fun to try the different tricks nonetheless.

I do think jacques le bailley's method for the very-rounds requires a bit less manual intervention, though you may disagree

and good point, chris — I hadn't really considered angled letters in this script, but now you've got me thinking — cheers!