The alluring trap of font editors in typeface design

The Bézier curve is not only the core tool of vector drawing applications like Adobe Illustrator, but is also the default manner of making and working with digital typeforms. This typically happens in font editors, applications that have evolved greatly over the years. They are sophisticated tools that empower us to create functionally and aesthetically refined typefaces more quickly and consistently than ever before. They influence our execution in other ways, however, and a uniquely digital workflow involving Bézier curves affects the way we create and think about typeform quality. For the non-specialist or student in particular, who may not have experience working with pre-digital tools, the alluring features of font editors and Bézier curves in an all-digital workflow can subtly oversimplify or obstruct design approaches and decisions that form the basis of successful typefaces. In becoming aware of these effects, however, designers can work to minimize their influence and retain greater control over their results.

The allure

Font editing software has come a long way since their first iterations. Originally as separate tools for recreating on the computer tasks done by machine, these tools were steadily unified into single applications. This consolidation developed alongside the ubiquity of the computer and the availability of other design software, propagating a higher public awareness of both typography and typeface design. Indeed, the newest generation of font editors are easily accessible – and marketed – to an audience much wider than type designers. FontLab, for example, has created TypeTool, a limited version of their flagship product, for a more generalist population,[1] while Glyphs is marketed with ‘Create – Produce – Release; Font editing for everyone’.[2]

These tools offer many advantages to the designer. They are relatively inexpensive and easy to install, uncomplicated for learning their main tools, and capable of producing functional fonts quickly. They contain refinements for error correction and for following current technology standards, and mimic the predictive behavior of mobile phones, such as suggestions for manipulating outline control points, template OpenType feature code, and automatic assignment of Unicode values, diacritic marks, and font names. Scripts and modules developed and shared by the user community extend the utility of font editors for batch processing, improving consistency, and streamlining workflows.

But perhaps the most significant development is that modern font editors can handle character definition, spacing, kerning, developing, hinting, exporting, and on some level testing – all of the major aspects of typeface design. The team-based design and engineering required with previous technology, in which each member contributes their specialized roles and skills, is in principle no longer necessary. A typeface can be made with one application on one machine by one person.

This can seem empowering and liberating: a freedom from previous constraints through automation, a reduction in the necessary materials and expertise, and quicker results on the road from idea to fully-functioning font. Yet behind the appealing features and possibilities of font editors, limitations removed by adopting new paradigms are replaced by new kinds of limitations. Furthermore, an important concept to letterforms is removed and marginalized in the process, and what replaces it is derivative.

The trap

Let us take a moment to describe the traditional component to letterforms and typography, the stroke. The stroke’s quality is dependent on designers and their materials, and in the context of letters, they can be used in different ways. Letterforms can be created in three manners: as handwriting, lettering, or typography. Handwriting is the creation of main parts of letters with single strokes. Lettering, synonymous with drawn letters, is the making of letters with many strokes. Typography is the creation and control of text and its arrangement with identical letters by machine.[3]

Typefaces typically consist of letterforms based on adjusted handwritten forms. As type technology evolved, the typemaker’s tools to make strokes changed, which in turn changed the role of the stroke.

Figure 1: Square capital S as (a) form, (b) pen outline, (c) scanned outline description, (d) recreated outline description
Figure 1: Square capital S as (a) form, (b) pen outline, (c) scanned outline description, (d) recreated outline description

The example in (a) is the result of three movements with the hand and broad-edged pen. The moment the pen touches the paper and starts to move, both mass and edge of the form are simultaneously created, and the variability of this form is limited to the range of interaction between hand movements, the pen, the ink, and the paper. Working with form directly is the essence of handwriting as well as manual punchcutting, in which form is sculpted out of steel with strokes of the burin.

The edges of (a) are described in (b), the paradigm of mechanical punchcutting and some types of lettering. This outline, too, involves strokes – nine movements of the pen. But the strokes’ individual quality is fixed and reduced in importance; their thickness is conceptually zero because the space within the outline will be filled in later, turning the outline into an edge. No notion of a stroke centerline is used in this approach. However, if I am required to accurately recreate (a) using drawn outlines, I either trace (a) or, without an image of (a) available to work from, I need the notion of the centerline and tool in my mind when I draw, because an outline describes only a part of form.

Outline descriptions (c) and (d) are made with parametric curves, numerical constructs defined by related points in space. Instance (c), a trace of (a), is a strict digital interpretation. This is currently done in font editors by either importing the scanned image and auto-tracing (eg. FontLab Studio), importing and manually tracing (eg. Glyphs), or copy-pasting outline descriptions from tracing done in other applications like Adobe Illustrator or third-party modules (eg. Robofont). A Bézier curve, arguably the most widely used form of parametric curve, requires a start point, a finish point, and curvature-control points for each to create an outline segment. Outline descriptions are made up of many connected but unrelated outline segments. Instance (d) is a recreation of (a), made up of 17 outline segments. To make (d) without an image of (a) available, I will need to have a strong understanding of (a)’s form and edge because an outline segment describes only a part of an outline.

We see that as we progress from form through outline to outline description, we pass through two rounds of the form’s conceptual deconstruction, each round more distanced from form and requiring a different object and duration of mental focus. Creating the strokes in (a) is a form-wise process that takes four to five seconds. The outline in (b) requires about ten seconds spent on thinking and drawing line-wise; the intended form comes into being only about twenty seconds later, after the outline is filled. The outline description in (c) involves three to four minutes to scan (a) and import it into a font editor, a form-wise translation process. Creating (d), however – the method that many of my colleagues follow, and which is convenient for students and hobbyists – requires several minutes to focus on, think about, and create (not draw) the separate point-wise outline segments.

If we apply the time, focus, and effort in working with only outline segments within one letter, extend this to the amount of characters in a typeface, and add the kind of personal intensity this type of work attracts, it’s not hard to imagine a day-to-day conditioning that takes place from creating forms as bits of curves. Indeed such a conditioning relates to the original scope of the Bézier curve: a technical tool for engineering the shape of cars.[4]

In the 1960s, a decade after the introduction of the computer into manufacturing, Pierre Bézier, an engineer in Renault’s Design Office, introduced a system for defining curved surfaces into the company’s industrial process. The system, UNISURF, used parametric curves to describe the shape of a car’s exterior more precisely than previous techniques. This required defining blocks of form from the stylist’s small-scale model for creating and assembling full-size models.[5] Styling and design were separate offices, however, and the tool was conceived for resolving engineering problems. Though this system was later used by stylists as an aid for setting up preliminary models, it is doubtful that this tool for precise definition was used as a main sketching and conceptual tool.

These virtues of economy and precision were carried from computer-aided design through computer graphics into font editors, where we see our options for drawing form constrained to the simple geometry of the outline segment: the point, straight line, and curve. The important step of sketching and conceptual drawing is not addressed. Furthermore, font editors’ conversion of stroke-like input with the mouse or stylus remains basic and, ironically, imprecise. The perceived freedom of creating form through outline segments – being able to drawing anything with a mouse! – becomes a problem.

Figure 2. Degrees of freedom in handwriting
Figure 2. Degrees of freedom in handwriting

Figure 2[6] demonstrates in the writing process the concept of degrees of freedom, the independent ranges of motion that an object can experience in space. The writer here has three degrees of freedom: the shoulder, the elbow, and the fingers (grouped here for simplicity). To make smaller strokes, the writer reduces movement in the shoulder and elbow and increases movement in the fingers; the scale of fluidity is reduced for finer control. For greater scale of fluidity of sweeping strokes encompassing larger areas, this relationship is reversed. We see that the resulting stroke, and thus form, is a result of the movement through each degree of freedom in the chain. The fingers cannot make strokes with the pen in regions of space where the elbow will not allow them to travel, nor the elbow by way of the constrained shoulder. The writer’s situation produces a limited range of natural outcomes.

Figure 3. Degrees of freedom in an outline description of S
Figure 3. Degrees of freedom in an outline description of S

This is not the case for the in Figure 3, in which we see that each point is independent of one another, resulting in an outline description with 49 degrees of freedom. However, they remain conceptually related and constrained to the stroke, as we saw in Figure 1. Should one of these points not adhere to this conceptual constraint, the outline segment containing it will be affected, which in turn is likely to be perceived as an erratic outline description and a weak form. No wonder this manner of ‘drawing’ is difficult!

A related effect we see between movement and the outline segment is the absence of any spatial relationship between the movement of the hand and the defined outline description. Figure 4 shows likely movements of the hand in creating (a) strokes with a pen, (b) outline with a pen, and (c) Bézier curves with a mouse. The movements in (c) are based on the position of control points and the scale of movement in the mouse, bearing no relationship with the curves themselves. As a result, no real learning of form through hand/eye movements takes place.

Figure 4. Traces of the hand in forming S as (a) stroke, (b) outline, and (c) Bézier curves
Figure 4. Traces of the hand in forming S as (a) stroke, (b) outline, and (c) Bézier curves

From my experience, a combination of marking tool and hand movement creates a connection with form and outline that is often necessary to get the right degree of feeling of the stroke or sweep of the line. Manipulating only Bézier control points often obstructs this, as one sweeping line may consist of several outline segments editable only with many independent points.

We have taken a look at modern font editors as tools that emphasize precise definition, standardization, and output. But they cannot replace the volatile and important conceptual stages of letterforms. Unfinished ideas funneled through them early risk being tweaked using the more abstract and specific grammar of the outline segment. Furthermore the convenient options font editors provide, both micro- and macro- in scale (eg. alignments, curve standardization, alternate character forms) can cheat us from spending time – occasionally struggling – with our results, encouraging us to trade normalization of unique elements for quicker output. The designer is subtly primed to address design problems as development problems. What results is the font editor participating in design decisions. I’m sure many type design professionals are familiar with the malaise that accompanies the unsettled point-pushing and vague wobbliness of certain outlines through their development. This is the result of a fruitless search for deep form in shallow outline description. The very advantages font editors propose can constrain the development of a cohesive and effective collection of letterforms, particularly those based on the stroke.

I am aware that some designers feel more comfortable creating letterforms with font editors than with stroke-based tools. The path and choice of tools for the results we seek is a matter of personal preference. A solid conceptual phase to any typeface is important, and I have argued that the tools influence the results. Penned forms, however, aren’t the final goal in typography. At one extreme of their utility, they can precisely define the final letterforms until the moment of digitization. At the other extreme, they can serve as conceptual aids, much like the study sketches for tone and composition created by fine artists in preparing a final painting. However they are used, sketch- and form-based tools can provide additional opportunities to generate and refine ideas.


Now that we have identified the ways in which font editors and Bézier curves influence our thinking and execution, we can look at ways to minimize this influence in our designs.

One option is to restrict font editors to conversion and clean-up of our final drawn letterforms. If we seek an all-digital workflow in creating letterforms, a decision is necessary between conserving the notion of strokes in our outline descriptions or abandoning them for outline-based ideas. An in-between approach – vague notions of stroke-based forms represented with outline descriptions – produces weak, restless results.

If we choose to replace the stroke with outline segments as the smallest conceptual element, than in freeing ourselves from the constraints of the stroke in form, we are no longer bound to the experience and expectations of readers. Typefaces could become repertoires of self-contained systems or follow some other unifying criteria, like counters or mass.

However, if we choose to retain the stroke concept, we have two options. One is regular training with marks from traditional tools. This can keep our minds nimble and loaded with a rich repertoire of form that we can mentally translate into convincing outline descriptions. The other option relies on font editors offering real drawing possibilities beyond the rudimentary, inexact results they currently provide. They could integrate stylus orientation and pressure from the drawing tablet, for instance, and interpret stylus movements and virtual pen constructs more faithfully. An eraser tool could remove portions of outline descriptions (and not just outline segments, which they currently do), providing a welcome equivalent to the white paint and scalpel used by lettering artists and type designers in more traditional media. In other words, accurate parallels for adding and removing marks in the Bézier paradigm are necessary in our font editors if we are to truly draw with them. Adobe Illustrator, with its sophisticated options, is one example in the right direction.

Designers can retain greater control over their results when they aren’t needlessly limited to the specific paradigms presented by mainstream tools. This involves exposing users to a wide variety of paradigms and learning the advantages and constraints of each. In doing so, we can resist conforming our ideas to the nature of a narrow set of tools, and select tools that work with the nature of our ideas.


Special thanks to those who participated in the discussion at Face Forward, as well as John de Szendeffy, Gerry Leonidas, Christophe Rabut, and Dan Reynolds.



[1] ‘TypeTool is our basic font editor for Mac OS and for Windows. For students, hobby typographers and creative professionals who occasionally need to create or customize fonts.’
‘TypeTool,’ FontLab Ltd., accessed 16 February 2017, link.

[2] Rainer Erich Scheichelbauer and Georg Seifert, Glyphs 2.3 Handbook (July 2016), accessed 2 December 2016, link.

[3] Fred Smeijers, Counterpunch: making type in the sixteenth century, designing typefaces now, 2nd edition (London: Hyphen Press, 2011), 19.

[4] Christophe Rabut, ‘On Pierre Bézier’s life and motivations,’ Computer-Aided Design 34 no. 7 (2002): 493–510.

[5] Pierre E Bézier, ‘Example of an existing system in the motor industry: the Unisurf system,’ Proceedings of the Royal Society of London A 321 (1971): 207–18.

[6] Adapted from Parker Zaner Bloser, Lessons in Ornamental Penmanship (Columbus: The Zaner-Bloser Company, 1948), accessed 8 November 2015, link.