Conference PaperPDF Available

Abstract and Figures

Timeseries line charts are a popular visualization technique but traditionally do not show many lines. We borrow concepts of tiny microtext and path dependent cartographic text to embed labels and additional text directly into lines on line charts, thereby making it easier to identify individual lines in a congested line chart, enabling more lines to be displayed and enabling additional data to be added to the lines as well.
Content may be subject to copyright.
Microtext Line Charts
Richard Brath
London South Bank University and Uncharted Software
Toronto, Canada
richard.brath@alumni.utoronto.ca
Ebad Banissi
London South Bank University
London, United Kingdom
banisse@lsbu.ac.uk
AbstractTimeseries line charts are a popular visualization
technique but traditionally do not show many lines. We
borrow concepts of tiny microtext and path dependent
cartographic text to embed labels and additional text directly
into lines on line charts, thereby making it easier to identify
individual lines in a congested line chart, enabling more lines
to be displayed and enabling additional data to be added to the
lines as well.
Keywords-visualization, line chart, typography
I. INTRODUCTION
Line charts are an extremely popular type of
visualization. In financial charting systems, line charts and
variations on line charts dominate over all other types of
visualizations of timeseries oriented data (based on
conversation at [1]).
A simple line chart with one line doesn’t require labeling:
the title can unambiguously indicate the line. But with
multiple lines, some way to distinguish between lines is
required, such as a use of colors which can be decoded by
cross-referencing a legend. There is cognitive effort required
to refer back and forth between a line in a chart and a
descriptive legend [2] and furthermore, the simple use of
color and/or dash patterns is difficult to scale beyond 10 or
so lines (e.g. see [3]).
Our unique contribution is 1) to move labels directly into
the plot area of timeseries charts to allow for many more
lines in a chart, and 2) to extend this approach to utilize
various typographic attributes to further disambiguate
between lines.
II. BACKGROUND
Type has been used extensively in information
visualization for labels. For example, many knowledge
visualizations on scimaps.org are label-dense: however the
labels is simply used to identify the underlying graphical
object. For example the Usenet Treemap [4] or Skupin’s
cartographic SOM [5] use labels on top of an underlying
visualization.
In addition to labels being plain, labels in information
visualizations tend towards largish sizes (8-12 point) as
opposed to fonts used in print. Historically, information
visualization has been associated with interactive video
displays, which until the late-2000’s were limited to 72-96
pixels per inch (PPI). New high resolution displays are
becoming much more common and these displays enable the
depiction of very small yet still legible fonts. For example,
some mobile phones now exceed 400 pixels per inch and
their applications may use fonts as small as 3 physical points.
Note, measurements in points herein use the traditional
typographic definition of points, i.e. one point = 1/72 of an
inch, or 0.35mm. In some modern programming contexts, a
point is defined in relation to the pixels, e.g. in CSS points
are relative to pixels, where 1 pixel = .75 points, and when a
font is defined in point size it will render at different physical
sizes based on the target device’s pixel density. For example,
Apple iOS guidelines recommend a minimum point size of
13 points, which in turn will render physically at 3.1 points
on a 400 PPI iPhone 6plus.
A. Small font sizes
In paper-based cartography, small fonts have been used
for centuries. Minimum font sizes in cartography are defined
as 3 point [6] or 4 point [7], with guidelines recommending 5
point [6] or 6 point [7] as a minimums. Similarly,
information graphics in print had very small minimum point
sizes, e.g. [8] recommended a minimum size of 4 point,
which was the smallest size supported by Monotype
machines at the time.
Figure 1. Sample of Bell Centennial font, specifically designed for use in
print at small sizes.
Text in print environments can go even smaller:
microtext is extremely small sized text and originated as an
anti-counterfeiting device for bank-notes and official
government documents, as well as archival texts. The quality
and legibility of text becomes more difficult to produce in
print as text becomes smaller. For example, Matthew Carter,
famously designed the font Bell Centennial to be legible
when printed in phone books - i.e. printed at small sizes on
high speed printers on poor quality paper - and had special
features such as notches in branches of letters to act as ink
traps to prevent the spread of ink to form blotches in crotches
and holes in letters (fig. 1).
B. Type as Texture
Microtext can be clear at extremely small sizes, although
the ability to decipher the text depends on the eyesight of the
viewer. Microtext can work as a texture due to a unique
feature of typography referred to as colour by typographers.
A well-designed font has an even distribution of text ink
across a sequence of letters regardless if the letters are sparse
(e.g. i,v) or dense (e.g. m, e): a blurred paragraph from such
a font will appear as an even distribution of intensity. In the
example in figure 2, the apparent light brown shading behind
the large word CANADA in the center of the banknote, is
tiny (one point) microtext when inspected closely.
Figure 2. Example banknote. Central stripe behind large text CANADA
appears to be smooth shading, but can be read as text
(BANKOFCANADA2) when inspected closely.
This feature of text as texture can be leveraged in
visualizations. As a texture, type has been used in some
novel visualizations such as typographic maps (e.g. [9],
which are based on human-designed typographic maps
created by Axis Maps, e.g. store.axismaps.com). These maps
uses text to completely fill areas on maps with labels,
removing need for edges and fills (fig. 3 top). The approach
is similar to the micro-text in fig. 2 above - i.e. labels are
repeated until the area is filled.
Figure 3. Top: Typographic map (axismaps.com). Bottom: FatFonts..
Fatfonts [10] are used to create a weighted texture: this is
a specially designed font where numerical glyphs are
weighted such that large numbers have proportionally
heavier weights (fig. 3 bottom).
C. Text aligned to paths
Some guidelines recommend against text at angles such
as axis labels (e.g. [11]). However, there are many
cartographic examples with text labels at angles or along
paths, for example, aligned to rivers, at angles to coast, or to
align with geographic areas, such as the example from
Stieler’s Atlas [12] in fig. 4 left (courtesy davidrumsey.com)
or Google maps (fig. 4 right).
Figure 4. Angled text and text along paths on maps.
Furthermore, there are many historic precedents of longer
phrases or sentences along paths such as early word balloons
for example from a late medieval book of hours [13] or an
early renaissance woodcut [14] as shown in figure 5.
Figure 5. Medieval and renaissance illustrations use word balloons with
lines of text flowing along the paths of curving scrolls.
More recent examples exist as well. While moveable type
set in rectilinear arrays may not have facilitated free form
text layouts, some authors experimented with text freed from
straight left-to-right lines such as poets. Apollinarie’s
concrete poems from the early 1900’s playfully adjust type
layout based on subject matter as shown in the poem Visée
[15] (Figure 6 left). Modern graphic design tools, such as
Photoshop and Illustrator provide easy-to-use tools to set text
along paths. The programmable graphics format SVG
natively provides functionality to place text along paths.
Figure 6. Organic layout of lines of text.
Within the field of information visualization Brad Paley’s
Map of Science [16] has long flowing labels, in this case the
nodes represent scientific topics and labels are a long list of
topic words: to avoid overlapping labels, the paths for labels
bend to avoid collision or cross each other at right angles to
better maintain readability (Figure 6 center). Ben Fry’s
Tendril [17] renders the text of webpages as twisting 3D
cylinders with hyperlinks spawning new orthogonal
cylinders of text (Fig. 6 right).
III. TIMESERIES ANALYSIS
In financial services, timeseries charts have been used for
more than 200 years, going back to William Playfair’s charts
and Japanese candlestick charts. By the early 1900’s,
organizations maintained and updated these physical paper
charts, potentially having many charts forming chart rooms
and chart libraries as seen in fig. 7 [18].
Figure 7. Pivoting boards for organizng many timeseries charts.
A. Non-interactive Charts
Although personal computers in the 1980's made it
possible to interactively plot financial timeseries charts,
variations of non-interactive charts have persisted through to
today. The current notion of a financial chart book or chart
library is a collection of time series charts which exist in a
non-interactive format. Chart libraries may consist of
hundreds of these charts. These charts may be consumed in
many different ways, such as paper output (e.g. a "chart
book"), a large PDF file, or used as content embedded into
commentary, such as financial research reports. In spite of
highly interactive computer systems, these non-interactive
charts continue to persist.
The author has had occasion to view, analyze and discuss
these chart libraries with users and experts in the capital
markets, including a large mutual fund company, a large
bank, a market data provider and so forth. Users and experts
claim benefits for these non-interactive charts including:
Ability to quickly flip through many charts: Flipping
through paper or pages in a PDF file is considered
faster than stepping through a software system that
dynamically retrieves data and updates charts.
Familiarity: The users have seen these particular
charts many times through their careers (e.g. weekly
or monthly) and may also return to them for
reference as needed. Off-the-shelf financial charting
software systems may not store all their
preferences, may not maintain consistency over
time (e.g. scales change), may not retain user
created notes and trend lines, and so forth are some
reasons provided for favoring these charts.
Interactivity is not required: The users want to be
able to see all the series and the full scope of data:
common computer-based visualization interactions
such as zoom, filtering, changing scales, and so
forth are expressly not desired. For example, they
have become familiar with the sizes of items and
may even refer to some aspects of such a chart with
physical dimensions, e.g. “a quarter inch movement
in a line”.
Annotation: Pen, pencil and ruler can be used to
achieve annotations more flexibly than some
interactive annotations.
Higher Resolution: Some practitioners will print
these charts out as a physical chart book. The print
version can be higher resolution than the screen
version. High quality 1200 DPI printers, in theory,
provide 15 times more resolution (120m dots) than
current state of the art 4k screens (8 megapixels).
TABLE I. NUMBER OF TIMESERIES PER CHART IN A FINANCIAL
CHART LIBRARY
NUMBER
OF SERIES
NUMBER
OF CHARTS
1-5
148
6-10
83
11-15
21
16-20
9
21-25
1
While these charts are not used interactively, they are for
the most part created computationally. They may be created
with heavily customized off the shelf software (e.g. Excel) or
other custom software. These chart libraries are considered a
proprietary confidential asset to a financial firm, therefore it
is typically difficult to get extended access to many of the
charts. The author was able to review all the timeseries
charts in one organization's chart library. This library
consisted of 262 timeseries charts. Most of the charts (57%)
displayed 5 or fewer timeseries. However, 31 of the charts
(12%) displayed 11 or more timeseries, with the current
maximum a chart with 23 timeseries, summarized in Table 1.
Note that this firm also had other timeseries charts, for
example one with 83 series, however these charts with a
higher number of series were consumed only in an
interactive format, typically toggling on only a few series at
a time for comparison; whereas in the chart library all the
charts were always plotted with all the lines.
There are various challenges with these charts when the
number of series is more than ten:
Differentiation: Charts typically differentiate
between series using hue, however, it can be
difficult to scale beyond ten unique hues [War13].
Strategies used included a combination of hues and
brightness (e.g. bright red, dark red, bright green,
dark green, etc); or hue and dash (e.g. red
continuous line, red solid line, green continuous
line, green dash line).
Labeling: Labels associated with each series is
typically indicated in a separate legend. With many
different series, it can become difficult to cross-
reference between the line in the chart and the
legend entry, for example the hue differentiation
may be subtle. Larkin and Simon [LS87], for
example, noted one benefit of diagrammatic
reasoning is reduced cross-referencing, however
this benefit is lost with a separate legend.
Other Layouts: Small multiples, horizon charts and
so forth were not acceptable alternatives: lines
needed to be superimposed on a common scale for
detailed visual comparison.
B. Line Labelling Alternatives
There are some known alternatives to legends for these
many line time series charts.
Labels at the line start or end. Rather than a legend,
labels can be depicted at the start or end of the line. This can
be challenging if there are many lines starting or ending at
similar values.
Labels in the plot area. Rather than rely on the legend,
some users add textual annotations in the plot area of the
chart, placing a straight line of text near the target line,
potentially with a leader line to visually connect the label
with the line. This has the benefit of reducing the cross-
reference to the legend (e.g. Figure 8 left [8]). However, it
does clutter the plot area and some users are adamant against
this. This is consistent with the arguments that labels can
make it more difficult to see patterns formed by data, e.g.
[19,20].
Labels aligned to paths. In historic hand-drawn examples
of timeseries charts - such as Playfair, or pre-computer
generated charts - the authors’ hand-letter charts and can thus
easily align text to shape of the line (Figure 8 right). This has
the benefit of reducing clutter and more directly associating
the label with a line as opposed to the straight line of text.
Labels aligned to graphical shapes are common in on-line
maps (e.g. Google street map), however, they are uncommon
in current time series visualizations. Directly labeling lines
has challenges with occlusion (labels obscuring each other or
the lines).
Figure 8. Labels directly in the plot area of line charts.
IV. MICROTEXT FINANCIAL TIMESERIES CHARTS
Microtext can be used directly label lines. Using local
labels means that cross-referencing a legend or visually
following a line back to an end label is not required - the
label is local to the area of inspection. There are alternatives
as to how the labels can be utilized:
Labels aligned to paths. Similar to river labels on maps,
labels can be aligned to lines. Within a congested line chart,
some care must be taken to reduce overlap. In Figure 9, a
line chart showing unemployment rate for 37 countries is
shown, with labels both at the end of the series and a few
labels along the length of the times series (in 7 point font). A
simple collision detection algorithm has been used to push
the labels apart to minimize overlapping labels.
Figure 9. Congested timeseries chart with labels along lines. Overview
top, closeup below..
Text as lines. With high resolution displays, line can
instead be drawn directly as continuous micro-text, similar to
the continuous text in Automatic Typographic Maps. In
Figure 10, lines are replaced with five point microtext and
appear perceptually similar to a dashed line.
Figure 10. Congested timeseries chart with labels fro lines.
Closeups in Figure 11.
Since lines cross and overlap, some consideration must
be given to text legibility at intersections.
Halo: In geographic maps, labels may be placed over top
various other graphics. A halo may be added around the text
to make the text more legible (e.g. TileMill text-halo-fill). In
a congested line chart, however, there may be significant
areas of overlap between lines. Adding a halo is feasible, but
many other microtext lines may be partially occluded
reducing line visibility and reducing legibility of partially
obscured text. See fig. 11 top.
Figure 11. Top: text with halos to improve legibility. Bottom: overplotted
text..
No halo: In the lower image of fig. 11, no halo is
provided around text. At holes in letters and gaps between
letters, the color of text beneath appears through the higher
text, and provides a better indication of line density than
halos. Assuming different colors of text, one can visually
trace other microtext lines and maintain some degree
readability through crossings, although not through areas of
high congestion. For example in fig. 11 compare Israel (red,
near bottom) in both the top image (with halo) and lower
image (without halo). The path of line is more easily
discerned through the many crossings in the lower, non-halo
representation. Or, Finland, in the middle image (brown) is
complete obscured, while in the lower image, although often
mixed with other text, can be somewhat discerned with some
effort.
To further enhance differentiation between the microtext
lines, each line is set with different font attributes: each line
varies in font weight, italics, case and font family, in addition
to colour.
Both microtext along a path or microtext as lines can be
combined with other labeling techniques, such as labels at
the ends of lines, or a legend. The viewer can choose to
ignore the microtext and use these other aids to decode the
lines.
Another benefit is that the labels can vary: rather than
repeat the same text, the microtext can be multilingual,
extending the usefulness of the chart across nationalities - as
seen in fig. 12.
Figure 12. Multilingual microtext lines.
V. TASKS, OBSERVATIONS AND DISCUSSION
The microtext timeseries were evaluated by six different
experts in capital markets who use non-interactive chart
books and chart libraries. All the subjects have at least 10
years expertise in capital markets specifically with financial
charts and visualization; and have affiliations or
certifications in professional financial analysis. All use
financial charts in their daily work. All work in a capacity
where they may be sharing their charts and observations with
other people, such as clients.
The youngest participant was 35, the oldest 65, with an
average 50. All were Caucasian men - as were most of their
clients. All had limited time availability: questions and tasks
were kept simple. No formal testing apparatus was set up:
these were meetings at the subject’s office.
A. Tasks and Questions
1. Initial Question: How many lines do you have at a
maximum in your charts? Why?
2. Then each of three Unemployment Rate as Percent of
Civilian Labour Force charts were presented in
sequence: first standard line chart with only labels
at the ends of lines; followed by one of the two
micro-text versions the followed with the other (i.e.
fig. 9, 10). For each variant the following two
questions were asked:
a. Which country was top (or bottom) in year x?
b. How did country x fare relative to its peers
through the 2008 recession?
3. Follow-up Question: Are these techniques relevant to
the kinds of analysis that your team does? Can you
identify use cases where you would consider using
this kind of chart?
B. Observations
With respect to the first question, the number of lines on
a chart, only one subject out of six claimed that he did not
need to plot more than 15-20 lines. Another claimed that
while most charts used a low number of lines, there were
cases where they could go much higher - beyond 50. The
most insightful answer came from an analyst: the community
was constrained by their tools and it was very difficult to
make an effective chart with more than 10 lines so people
have been conditioned to keep their charts simple - however
there are actually many cases where one would want to view
more lines if feasible, assuming the tools and representation
were appropriate. This user pointed out that many peers and
subordinates used Excel, which was poorly suited for
working with many simultaneous lines.
With regards to the tasks and responses to the three chart
variations: When presented with the first task (highest/lowest
country in year x with just a line chart), the response was
either a dismissive comment (e.g. that’s not going to work)
occurring for 4/6 subjects or taken as a challenge (i.e.
attempting to trace the line to find the right answer, requiring
more than ten seconds to do so). If the initial answer was
dismissive, the second task country x through recession) was
not asked.
For the second chart, the subjects responded almost
instantaneously with a visceral response, such as “Wow, this
is exciting”, or “I really love this.” One interesting aspect
was 4/6 subjects physically traced their response using their
finger, e.g. for the top in year x question, sliding across the
x-axis to the year, moving up to the target line, moving
across to the first local label to the answer (requiring no
more than 4 seconds).
The second task (how did country x fare through the
recession) was done typically by tracing vertically near the
beginning of the recession until the target was identified,
tracing across through the recession to a matching label on
the opposite side to confirm they were on the same line, and
then an observation relative to the rest of the peers, e.g. “I’d
say that Switzerland did quite well” or “I don’t think things
were good for Portugal.”
The third chart progressed much like the second, often
eliciting a second visceral response equal to the first. Two
subjects did not complete the task for the third variant. For
the remaining 4/6 there was not a significant difference in
time.
Most subjects felt compelled to compare and contrast the
two techniques. Opinion was divided:
The font sizes were slightly smaller for the text as
lines variant (5 point) compared to the labels on
path (7 point). For the oldest participant, the slightly
smaller size was borderline legible and he had some
difficulty reading the labels whereas he could read
the slightly larger path labels. He preferred the
larger labels but speculated that he might prefer the
smaller labels if size varied depending on the space
available.
One participant found the microtext as line variant
extremely compelling and referred to it as a very
clean layout. He specifically noticed and called out
the font variation (i.e. weight, case, typeface) as an
effective means for creating differentiation.
One subject had a strong preference to the path
labels. The approach represented the best
readability for the entire chart as the labels tend to
push out to the less dense regions on the chart while
the most congested areas of the chart retained high
legibility by having only thin, accurate lines
through these areas. He wanted labels at the end
removed to make the chart even cleaner. The
microtext as line approach was not preferred as the
labels were less legible when overlapping and
multi-lingual labels were a distraction to the core
content.
One participant noticed the multilingual labels and
hypothesized that the approach could make the
charts more broadly accessible to his global
audience.
One participant hypothesized that the labels on a
path may be appealing because of its familiarity
from cartography.
The microtext as line was identified by one viewer
as more stimulating than the other two: any local
anomaly or pattern could draw you into the chart by
having a meaningful label immediately visible
thereby engaging the viewer to spend more time
with the chart.
Overall, 3/6 preferred the microtext only approach,
2/6 preferred river labels and 1/6 did not express a
preference.
The final question provided responses that ranged across
user types, user locations, and data types. Users identified
various timeseries datasets that could benefit from using
these techniques such as index analysis, peer analysis,
economic analysis and state analysis. There was some
discussion as to the nature of the data - the example dataset
had low volatility (i.e. the lines did not zig zag up and down)
- and how would the approach fare under different datasets.
Various enhancements were suggested, such as:
Labels as lines could indicate other data, such as
values at high points and low points, rate of change,
or political leader during time periods.
Bold could be used to indicate additional
information part way along a path.
Interactivity could combine both the benefits of the
textual static view with selection for ability to easily
focus on any subset.
In addition, one insightful comment was that the
application area was much broader than capital markets -
understanding trends across peer groups are applicable in
many policy areas. In most of these cases, the reports will be
published and distributed and not interactive. Any decision
maker would benefit from these techniques.
C. Discussion
The application area of timeseries analysis with many
lines seems like a rich area for further investigation,
particularly given a much broader application area outside of
financial markets. However, a broader evaluation should be
done for this broader audience - unlike financial market chart
users, this broader community may have less familiarity with
charts and visualization literacy.
Minimal font size is also an area for additional
investigation. While one older subject found 5 point text
borderline legible another subject had undergone cataract
surgery one year earlier and had no difficultly reading the
small 5 point text. Lighting, glare, paper quality and other
factors can also affect legibility and this was not controlled.
Cartographic guidelines imply smaller sizes could be used
and thus data densities increased.
Figure 13. Number of retweets over time for popular tweets about Donald
Trump in mid 2015.
Experts suggested that lines could contain more
information than simple labels. Figure 13 is one possibility
of a line chart with more data. This figure plots the number
of retweets over time for some popular Donald Trump tweets
originating in late August 2015. Lines show the actual
content of the tweets, including the author of the tweet in
bold, the tweet content in plain text, and the lines
differentiated by color.
VI. CONCLUSION
Microtext is a promising extension to information
visualization. Microtext goes beyond previous type-based
approaches by:
1) Using text smaller than most visualization labels, but
staying large enough to remain readable without
interaction;
2) Uses lines to embed additional information beyond
simply labeling the area or line, e.g. items counted,
multilingual labels, or text content of the item
measured.
3) The approaches can be combined with other
visualization approaches (e.g. color)
4) The approach can be extended to use font attributes
to differentiate between categories (e.g. bold).
Furthermore, the approach is highly applicable to the
analysis of many timeseries, particularly in environments
with no or low interaction, which may be very broad.
REFERENCES
[1] R. Brath, L. Comrie, D. Keller, E. Knuth, and E. Sorenson.
"Challenges in Financial Visualization Panel." at VisWeek 2014.
Paris. http://bit.ly/2lFIm4H
[2] J. H. Larkin and H. A. Simon, “Why a diagram is (sometimes) worth
ten thousand words,” in Cognitive Science 11, no. 1 (Wiley: 1987):
65100.
[3] C. Ware. Information Visualization: Perception for Design. Elsevier.
2013.
[4] M.A. Smith and D. Fisher. 2004. Treemap View of 2004 Usenet
Returnees. http://www.scimaps.org/detailMap/index/treemap_
view_of_2004_57. Accessed 12//18/ 2015.
[5] A. Skupin, “The World of Geography: Visualizing a Knowledge
Domain with Cartographic Means.” PNAS 101 (Suppl. 1): 5274-
5278. 2004.
[6] A. Robinson, J. Morrison, P. Muehrcke, A. Kimerling, S. Guptil:
Elements of Cartography. Wiley, New York, NY, 1995.
[7] E.R. Hodges: The Guild Handbook of Scientific Illustration. John
Wiley, 2003.
[8] W.C. Brinton, Graphic Presentation. Brinton Associates. New York.
1939.
[9] S. Afzal, R. Maciejewski, Y. Jang, N. Elmqvist, & D. Ebert. “Spatial
text visualization using automatic typographic maps.” IEEE Trans. on
Visualization & Computer Graphics, (12). 2012.
[10] M. Nacenta, U. Hinrichs & S. Carpendale. “FatFonts: combining the
symbolic and visual aspects of numbers.” In Proceedings of the
International Working Conference on Advanced Visual Interfaces.
ACM. 2012.
[11] A. Wallgren, B. Wallgren, R. Persson, U. Jorner, J. Haaland.
Graphing Statistics & Data: Creating Better Charts. Sage
Publications. 1996.
[12] A. Stieler. Stielers Hand Atlas of Modern Geography. Justus Perthes,
Gotha. 1881. www.davidrumsey.com/luna/servlet/s/425okp
[13] Author Unknown. Book of Hours (Sarum Use). Early 1300s. East
Anglia. http://cudl.lib.cam.ac.uk/ view/MS-DD-00004-00017/16
Accessed Aug 23, 2016.
[14] Bruno Carthusianus. Brunonis Carthusianorum Patriarche sanctissimi,
theologi Parisiensis Scholae doctissimi: & Remensis ecclesiae
canonici moratissimi: Opera & vita post indicem serie literaria
indicanda. Venundatur Jodoco Badio Ascensio, Paris, 1524
https://archive.org/details/bub_gb_3FEGaBmWN0UC (page 999).
Accessed Aug 24, 2016.
[15] G. Apollinaire, « Visée », in Calligrammes. Poèmes de la paix et de la
guerre (1913-1916), Paris, Mercure de France, 1918.
https://archive.org/details/calligrammespo00apol (pg. 88).
[16] B. Paley. “Relationships Among Scientific Paradigms”. SEED
Magazine. 2007.
[17] B. Fry. Tendril. 2000. benfry.com/tendril/index.html
[18] W.C. Brinton, Graphic Methods for Presenting Facts. The
Engineering Magazine Company. New York. 1914.
[19] W. Cleveland, The Elements of Graphing Data. Hobart Press, Summit
NJ, 1994.
[20] S. Few. Beautiful Evidence: A Journey through the Mind of Edward
Tufte. 2006. www.perceptualedge.com/articles/b-
eye/beautiful_evidence.pdf.
Article
Instantaneously understanding the gestalt of thousands of words is achieved through the programmatic placement of the words and control of their presentation characteristics, such as size, repetition, and font. As early as the fourteenth century, words were used as building blocks for images. Hundreds of years later, this typographic experiment continues with the addition of raw computational power. The ability to place thousands of words in interesting forms gives rise to a quantitatively different form of expression. The resulting procedures are expressive enough to represent shapes, textures, and shading automatically. Though based on approaches for addressing the classic problem of algorithmic two-dimensional bin-packing, aesthetically pleasing results are achieved through the incorporation of a small set of rules to guide the layout.
Chapter
Microtext is an important aspect of security printing. At times, the embedding of copy protection in documents that are printed by a conventional office printer is required. In this chapter, we present a method for on-the-fly microtext generation. In general, the approach is applicable to any language and a wide set of typefaces. The technique consists of rendering a symbol to an intermediate image and creating a skeleton by iterative thinning and downsizing of the skeleton to target a micro-symbol bitmap. Optionally, the thickness of lines can be increased. Possible targets in the printing pipeline for implementation of the microtext generator are discussed and the advantages of the method presented over conventional text printing with an extremely small font size are demonstrated.
Article
Full-text available
Typographic variables were only briefly discussed by Bertin with no example applications. We review historic examples to extend Bertin’s framework with literal encoding, 10 typographic attributes, variations on scope including characters, words, phrases and paragraphs, and more layout types. We then apply the framework to Bertin’s population dataset to create 11 new typographic visualizations. The approach raises questions for new research such as readability, semantic association, interaction, and comprehension. © 2018
Article
Full-text available
We present a method for automatically building typographic maps that merge text and spatial data into a visual representation where text alone forms the graphical features. We further show how to use this approach to visualize spatial data such as traffic density, crime rate, or demographic data. The technique accepts a vector representation of a geographic map and spatializes the textual labels in the space onto polylines and polygons based on user-defined visual attributes and constraints. Our sample implementation runs as a Web service, spatializing shape files from the OpenStreetMap project into typographic maps for any region.
Conference Paper
Full-text available
In this paper we explore numeric typeface design for visualization purposes. We introduce FatFonts, a technique for visualizing quantitative data that bridges the gap between numeric and visual representations. FatFonts are based on Arabic numerals but, unlike regular numeric typefaces, the amount of ink (dark pixels) used for each digit is proportional to its quantitative value. This enables accurate reading of the numerical data while preserving an overall visual context. We discuss the challenges of this approach that we identified through our design process and propose a set of design goals that include legibility, familiarity, readability, spatial precision, dynamic range, and resolution. We contribute four FatFont typefaces that are derived from our exploration of the design space that these goals introduce. Finally, we discuss three example scenarios that show how FatFonts can be used for visualization purposes as valuable representation alternatives.
Chapter
Full-text available
Most designers know that yellow text presented against a blue background reads clearly and easily, but how many can explain why, and what really are the best ways to help others and ourselves clearly see key patterns in a bunch of data? This book explores the art and science of why we see objects the way we do. Based on the science of perception and vision, the author presents the key principles at work for a wide range of applications--resulting in visualization of improved clarity, utility, and persuasiveness. The book offers practical guidelines that can be applied by anyone: interaction designers, graphic designers of all kinds (including web designers), data miners, and financial analysts.
Article
We distinguish diagrammatic from sentential paper-and-pencil representations of information by developing alternative models of information-processing systems that are informationally equivalent and that can be characterized as sentential or diagrammatic. Sentential representations are sequential, like the propositions in a text. Diagrammatic representations are indexed by location in a plane. Diagrammatic representations also typically display information that is only implicit in sentential representations and that therefore has to be computed, sometimes at great cost, to make it explicit for use. We then contrast the computational efficiency of these representations for solving several illustrative problems in mathematics and physics. When two representations are informationally equivalent, their computational efficiency depends on the information-processing operators that act on them. Two sets of operators may differ in their capabilities for recognizing patterns, in the inferences they can carry out directly, and in their control strategies (in particular, the control of search). Diagrammatic and sentential representations support operators that differ in all of these respects. Operators working on one representation may recognize features readily or make inferences directly that are difficult to realize in the other representation. Most important, however, are differences in the efficiency of search for information and in the explicitness of information. In the representations we call diagrammatic, information is organized by location, and often much of the information needed to make an inference is present and explicit at a single location. In addition, cues to the next logical step in the problem may be present at an adjacent location. Therefore problem solving can proceed through a smooth traversal of the diagram, and may require very little search or computation of elements that had been implicit.
Article
Iniciales Xil. Port. a dos tintas y grab. xil. arquitectónico Texto a dos columnas Lugar de impresión tomado de colofón Grab. xil. de la vida de San Bruno en la última secuencia tipográfica Sign.: †8, a-z8, A-H8, Aa-Zz8, Aaa-Hhh8, AA-CC8
Article
Translated from the Swedish: Statistikens Bilder: Att Skapa Diagram
Article
From an informed critique of existing methods to the development of original tools, cartographic engagement can provide a unique perspective on knowledge domain visualization. Along with a discussion of some principles underlying a cartographically informed visualization methodology, results of experiments involving several thousand conference abstracts will be sketched and their plausibility reflected on.