Why Visualize with Fonts Now?

Traditionally, fonts have been treated rather simply in visualizations: simple labels with no differentiation. So, why should we even consider font attributes (such as bold and italic) in visualization. And why now?

1. Higher resolution: the resolutionary revolution

For 20 years screen displays were approximately 72-96 pixels per inch (PPI). However, mobile computing, started moving to higher resolutions, in part because the devices were physically closer to the viewer than desktop devices. Steve Jobs at Apple brought these higher resolutions various devices including phones, laptops, desktops and tablets under marketing terms such as retina displays and resolutionary. Previously at 72 PPI desktops, font capabilities were rather limited as shown in the figure below.

Font details

Enlarged nine point Futura and Baskerville fonts on a 86 PPI screen vs the same on an iPad3 with 264 PPI. With more than 9 times the resolution, the text is more legibile and fine details are clearly visible.

Previously, it would be difficult to effectively utilize font attributes on computers. Now, computers have the capability to display fonts in greater detail and at much lower resolutions.

In addition to higher resolution, two other advances further improve the visual crispness and differentiation among screen fonts. Specialized text rendering technology, namely Microsoft ClearType and MacIntosh Quartz, provide higher quality rendering of text than simple anti-aliasing. Also, a wider range of typefaces designed for the screen are now available and are more easily accessible (e.g. fonts.google.com).

User interface guidelines have evolved with the technology. In the 1990’s, it was pointed out (P. Kahn and K. Lenk. Principles of typography for user interface design)  that screen displays were ill suited to italics (due to low PPI), weight levels were constrained to two or three levels (again, due to low PPI), spacing was difficult to achieve with most font software, small caps were often unavailable and underline had a special meaning for web browsers. Heim (S. Heim. The Resonant Interface: HCI Foundations for Interaction Design) indicates that some fonts require more pixels for legibility and suffer at lower pixel densities, such as serif and cursive fonts.

Many of these constraints are less of an issue now. Current web and user interface guidelines may recommend a variety of type specific attributes including mixed font families, weight, italic/obliques, caps/small caps (but still recommend against underline – for example, see J. Teague. Fluid Web Typography).

With the proliferation of much higher pixel densities in the newest devices, better text rendering and a wide range of fonts, visualizations can now use a wide variety of typefaces (e.g. Futura, Baskerville, Didot, Helvetica, Glypha) and associated weights and styles (e.g. bold, italics, small caps, etc) to make more effective visualizations and potentially encode additional information via typeface attributes (e.g. font weight, font slope, caps/small caps/no caps, underline, superscript/subscript, spacing/tracking/leading, etc).

2. Color can reduce legibility

Color, specifically hue, is a popular visual attribute for encoding data. In the many visualizations on placesandspaces.org, if typography is modified, hue and size are used more often than any other visual attribute.

However, it has long been known that text legibility depends on the contrast between the text characters and their background (e.g. see Colin Ware’s Information Visualization: Perception for Design). When only a few different levels of categorization are required, then a few different hues (e.g. blue, dark red, dark green) can be used while still maintaining a high contrast. However, in visualization, more than a few categories may be required. There are many examples, where graphic designers or visualization researchers attempt to use more hues, potentially resulting in lower text legibility, such as yellow text on a cyan background or black text on a red background as shown in the figure below (the last figure being a published image by me – it’s so easy to use color in visualization).

HueContrast1HueContrast2HueContrast3 HueContrast4

If using many hues reduces legibility, other techniques should be considered. Looking at the history of typography, one can find a wide variety of different font faces, weights, italics and other properties that are used to differentiate text while maintaining high legibility – meaning that visualization should start to consider how some of these attributes could be used rather than reducing legibility with color.

3. There are 100,000+ of typefaces. Why?

Turning the exploration around in the opposite direction, from the perspective of typography, one might wonder why there are thousands of different fonts and so many techniques to create emphasis and differentiation with type. In fact, on fonts.com there are more than 150,000 fonts (!). And, while these are discussed elsewhere in more length, there are many different font attributes that can be relatively easily manipulated, including:

  • Font weight. Beyond bold. Some fonts support up to 9 weights.
  • Italic / Oblique. i.e. sloped font. Italic and oblique are different.
  • CAPS/Small caps/no caps.
  • Condensed / expanded.
  • S p a c i n g. i.e. tracking and leading.
  • Underlining.
  • Superscript/subscript
  • Individual glyphs. i.e. different letters, numbers and other characters #!?%
  • Paired delimiters. (e.g. [brackets],”quotes”,-dashes-).

One may wonder how and why all these evolved. There is a rich history and large design space associated with typography: at OCAD university there are a more than six undergrate courses on typography: starting with 1) letterform structure and relation to words; 2) typographic normative and conceptual principles; 3) typographic history; 4) organizational typographic structures; 5) advanced typographic syntactics, semantics and pragmatics; 6) experimental typography and 7) kinetic typography. Information visualization is yet another application for typography, with some new unique requirements.

One hint is that there are many different tasks and uses of typography ranging from quick, clear reading of highway signs; legible fonts on computer screens; packaging and advertising; specialized uses such as mathematics; varying requirements across languages; creating large blocks of text for reading; and on and on.

4. The even color of well design type

Using type in visualizations utilizes the notion that a well-designed font has an even distribution of text ink across a sequence of letters regardless if the specific letters are sparse (e.g. i, v) or dense (e.g. m, e). In type design this even distribution of ink is referred to as color (not to be confused with hue). For example, a blurred paragraph from a scan of a typeset book shows the relatively even distribution of intensity.


Blurred text. Note the relatively even distribution of ink: dense letters such as m or e do not stand out.

The font designer adjusts for different letterform shapes by varying the font shape (e.g. the stroke of a c may be thicker than the corresponding stroke of an e) and by adjusting spacing between letters and specific letter pairs (e.g. kerning, ligatures).


The letter C, compared to letter O (outline), has less ink and is narrower than O. G in comparison to C (outline) is slightly wider and has significant tapering and thinner forms at the spur on the lower right. From Lettering & Type by Willen and Strals.


As a result, given an expected even weight across a field of type using the same font, the viewer does not perceive any particular letters standing out. Conversely, concrete poetry adjusts the layout text to embed additional semantics, such as Calligrammes by poet Guillaume Apollinaire or the Mouse’s Tale by Lewis Carroll.
This evenness of color of text is a quality of text that can be manipulated to make some items standout or otherwise differentiated from other text to reveal, for example, distributions or proportional encoding (KMIR paper, fig 5.).

5. Type Attributes work together and are separable

One problem with visual attributes is that they are confounding. When multiple visual attributes are combined together, the full range implied by the two attributes does not work. For example, hue (e.g. red, yellow, blue) when combined with brightness can result in colors that do not fit into a user’s mental model of color space. While red and light red (pink) may be understood, yellow and dark yellow (i.e. brown) may not be understood to be related by brightness. Or, shape (squares, circles and diamonds) and size when combined results in indecipherable shapes when small. e.g. can you determine whether the periods at the end of sentence here is a small diamond or a circle?

Type attributes, however, have been designed to work together and be clearly separable and decipherable. In this sentence it is easy to distinguish between words in bold, italic, bold+italic, CAPS, CAPS+ITALIC, italic+underline,  {delimited},  or →DELIMITED+CAPS+BOLD+ITALIC+UNDERLINE←, and so on. One immediate guess, is that such clear separability could be useful in encoding data to indicate membership among many different sets.

6. Unique nameable glyphs

One challenge with the design of markers whether arbitrary shapes (e.g. circle, square, star); or glyphs (e.g. bird, cat, plane) is creation of a large number of uniquely identifiable glyphs when many (e.g. 20-50) unique glyphs are required. Most pre-packaged software provides only a limited set of different glyphs: for example Excel and D3.js both provide under 10 unique shapes as glyphs. Having a large number of unique glyphs creates the potential for higher data densities than plots using just a few unique shapes.

When the visualization designer attempts to create many unique glyphs, it becomes tricky to create unique shapes. This was one of the challenges in my earlier shape research. Using a morphological approach to generate shapes, it was possible to create interesting, unique shapes; and variants could be created by rotation or mirroring. However, it became very difficult to refer to these shapes in text or in discussion, because they were difficult to name and/or had visual similarities. For example, one shape could be a diamond ◊ and then rotated 45 degrees it is a square. However, another shape could be, say a lemon but rotated 45 degrees it is still a lemon.

Alphanumeric glyphs, are uniquely identifiable and uniquely nameable. There is almost no confusion between a, b, c, and so on assuming the characters are not rotated (e.g. an n rotated 180 degrees becomes a u). This provides up to 26 unique glyphs for lowercase, 26 unique glyphs for upper case and 10 unique numeric glyphs, although some care should be exercised, for example, potential confusion be O, o and 0 or between I and l, potentially reducing the candidate set 59 uniquely different and uniquely nameable characters.

7. Orderable glyphs

A unique property of alphanumeric glyphs compared to general glyphs is that alphanumerics are orderable – although this ordering is not pre-attentive. A comes before B and so on.

Imagine a visualization with 50 unique icons (e.g. square, circle, star, wheel chair, swishy marks, etc): unless all the icons are immediately intuitive, the viewer may need a legend to decode them. Since icons have no ordering, a viewer would need to linearly search through all the icons in the legend.

Alternatively, if the alphabetic characters a-z, A-Z are used instead, then a legend can be visually organized and searched in sort order.

8. Letters are mnemonics

Letters can be used to mnemonically encode data, as opposed to arbitrary sequences. For example, perusing “cattle brand charts” in Google image search will return a wide variety of cattle brand logos. A significant number of these brands (1/3-1/2? perhaps a survey would be useful) are letter combinations or stylized letters that represent the initials of the ranch owner or ranch name. Similarly, 2-letter and 3-letter country codes (ISO-3166-1) are mnemonics for the country names. These mnemonic encodings can aid the viewer in decoding the text. For example, scanning a list of websites for recipes one could easily understand that the recipes will be very different on websites ending in .fr vs websites ending in .jp.

9. Glyph sequences as words

Sequences of alphabetic glyphs can be combined to create words
which may be perceived quickly although not pre-attentitvely (e.g. see

Words may be less ambiguous than icons. Consider this chart of body weight vs. brain mass from Edward Tufte (and ongoing discussion about the history and variations of this chart).


In this chart, images of particular species represent their position on the chart. Which is the rat and which is the mouse? Similar species can be difficult to differentiate, which is why bird watchers have rather large field guides to help them identify different bird species with entire books on subspecies, e.g. hawks or warblers.

Alphanumeric glyphs may also include glyphs not-native to the
viewer’s language, but encode additional information that can be decoded
by the viewer and are orderable. Similarly, symbols (and punctation)
can also encode additional information, but these are not ordered.

There is a lot more that could be added here. I’ve mostly avoided any historic examples, from artists, advertising, charts and maps.

About richardbrath

Richard is a long time visualization designer and researcher. Professionally, I am one of the partners of Uncharted Software Inc. I have recently completed a PhD in data visualization at LSBU. The opinions on this blog are related to my personal interests in data visualization, particularly around research interests related to my PhD work- this blog is about exploratory aspects of data visualization not proven principles.
This entry was posted in Font Visualization, Text Visualization, Uncategorized, Visual Attributes. Bookmark the permalink.

1 Response to Why Visualize with Fonts Now?

  1. Pingback: Font Weight: The use of bold in visualization | richardbrath

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s