Visualization: "I Like A Good Story Well Told. That Is The Reason I Am Sometimes Forced To Tell Them Myself." - Mark Twain

Download as rtf, pdf, or txt
Download as rtf, pdf, or txt
You are on page 1of 13

Visualization

What springs to mind when you hear the word


storytelling? For most of us, it conjures up images
of children gathered in front of a rocking chair, rapt
with attention as an elder narrates a fairy tale to them.
Unencumbered by the inhibitions of older years, they
are not afraid to interrupt and ask for details to satisfy
their curiosity, or clamor for more when the story
ends. ow can we, as !isuali"ation researchers and
practitioners, elicit this same engagement and wonder
in our !iewers? ow can we aid #r. $wain in his
plight, and ensure that he is not the e%clusi!e pur!eyor
of good stories well told?
&ll stories are se'uences of causally(related
e!ents. owe!er, the good ones tend to share se!eral
important features. First, they take time to unfold, and
their pacing matches the audience)s ability to follow
them. *econd, they hold the audience)s attention by
ha!ing interesting settings, plots, and characters.
Finally, they lea!e a lasting impression, either by
pi'uing the audience)s curiosity and making them
want to learn more, or by con!eying a deeper meaning
+wan(,iu #a, Uni!ersity of -alifornia, .a!is, -&, U*&
/saac ,iao, Uni!ersity of -alifornia, .a!is, -&, U*&
0ennifer Fra"ier, $he 1%ploratorium, *an Francisco, -&, U*&
elwig auser, Uni!ersity of 2ergen, 3orway
elen(3icole +ostis, 3&*& 4oddard *pace Flight -enter, 4reenbelt, #., U*&
"I like a good story well told. That is the reason I am sometimes forced to tell
them myself."
Mark Twain
than your normal e!eryday run(of(the(mill se'uence
of causally(related e!ents.
When using !isuali"ation to tell a story, what
does it mean for a !isuali"ation to ha!e good
pacing? & story that is well paced e%hibits deliberate
control o!er the rate at which plot points occur.
owe!er, any gi!en pace may feel too fast or too slow
to di5erent audiences, depending on their attention
spans and personal preferences. *imilarly, in
designing !isuali"ations, it is crucial to gauge the
intended !iewers) familiarity with both the subject
matter and with !isuali"ation con!entions. For a gi!en
dataset, distributing data across multiple line charts
might pro!e to be the most suitable approach for a
general audience, but domain e%perts might prefer to
combine data into a single parallel coordinates
diagram to facilitate comparison.
What are the settings, characters, and plots of
*e!eral interesting points emerged o!er the
course of the day. $here was a general consensus that
framing data as a narrati!e makes it more interesting
and memorable. Why might this be? -ogniti!e
science postulates the e%istence of separate types of
memory for storing di5erent types of information6
semantic memory, for remembering disconnected
facts, and episodic memory, for remembering
se'uences of e!ents. 2y presenting themsel!es as
narrati!es, !isuali"ations can tap into episodic
memory and establish themsel!es as cohesi!e entities.
/n addition, the issue of interacti!ity in !isuali"ations
came up repeatedly. $he style of storytelling present
in static !isuali"ations, such as info(graphics, is
fundamentally di5erent from that of interacti!e
!isuali"ations, in which users are allowed to na!igate
and modify !iews of data. #aking a !isuali"ation
more interacti!e gi!es users more freedom to e%plore,
but lessens the amount of control !isuali"ation
designers ha!e o!er how the story is told. /n the end,
we concluded that the interacti!ity of a !isuali"ation
should be carefully balanced against the need to guide
the !iewer through the data. & useful compromise
might be to start the !isuali"ation in a non(interacti!e
mode, ensuring that the most salient features of the
dataset are presented, and then allow users to e%plore
the rest of the dataset afterwards.
$he !isuali"ations presented and discussed in
this workshop fell s'uarely in the domain of
information visualization, which tends to use more
abstract representations and are usually targeted
towards more general audiences. 2y contrast, what
challenges does scientifc visualization face in
storytelling?
Scientifc Storytelling
7isuali"ation has become an important tool for
scientists in their daily work. *cientists create
!isuali"ations for !arious purposes6 to !alidate
e%periments, to e%plore datasets, or to communicate
8ndings to others. /f appropriately presented, such
!isuali"ations can be highly e5ecti!e in con!eying
narrati!es. $hus, using the abo!e criteria, let us
e%plore the possibility of telling stories using
scienti8c !isuali"ations.
$he narrati!e impact in information
!isuali"ation stems from !isual comparisons using
simple, abstract representations of data6 bar charts
show di5erences in length, scatterplots show
di5erences in position, treemaps and pie charts show
di5erences in area, and heatmaps show di5erences in
color and intensity. &s such, information !isuali"ation
stories are about comparison or change6 ,ook at how
much bigger & is than 2, or look at how - has
grown o!er time. /n contrast, much of the narrati!e
impact in scienti8c !isuali"ation comes from being
able to see real data that is normally in!isible. &t its
best, scienti8c !isuali"ation acts as an e%tension of
our senses, allowing us to percei!e and manipulate
data at otherwise impossible scales and perspecti!es,
such as !ector 8elds in weather systems, isosurfaces
in superno!a simulations, and layers of human
anatomy rendered semitransparently. Whereas
information !isuali"ations are allegories 9 abstractions
and summaries of raw data 9 scienti8c !isuali"ations
are more literal: they stri!e for realism and spatial
accuracy, sacri8cing details only to facilitate
understanding.
/n some ways, scienti8c !isuali"ation has it
easy. Usually, the intended !iewers are the scientists
who generated the data, and others in the same 8eld.
!isuali"ations? First, the setting of a !isuali"ation is
all of the background information a !iewer needs to
know in order to conte%tuali"e and comprehend the
!isuali"ation. /n theatrical productions, the stage is
generally set before the curtains rise: similarly,
!iewers should be introduced to the subject matter
before seeing a !isuali"ation of it. /n addition, !isual
elements representing data points are the characters
and centers of attention in !isuali"ations 9 they)re the
stars of the show. Finally, the plot, and the dramatic
tension, of a !isuali"ation arises from the
ju%taposition of its !isual elements, how they interact
and compare with one other, and how they e!ol!e
o!er time.
&rmed with these notions, let us discuss how
!isuali"ation can be used to tell a good story, and tell
it well. /n particular, we emphasi"e scientifc
storytelling 9 that is, telling stories using scienti8c
data 9 which is a topic that the !isuali"ation research
community has paid little attention to so far. /n
contrast, the subject of storytelling in information
visualization has been the topic of se!eral recent
workshops and panels, and pro!ides a starting point
for the discussion of scienti8c storytelling.
Storytelling in Information Visualization
&t 7isWeek ;<=< in *alt ,ake -ity, there was a daylong
workshop called $elling *tories with .ata6
Using !isuali"ation to create narrati!es and engage
audiences. osted by #att #c+eon, 0oan .i#icco,
and +arrie +arahalios, this workshop featured a
di!erse range of speakers, including journalists,
bloggers, literary analysts, and de!elopers of
information !isuali"ation software. $hroughout the
day, we saw numerous e%amples of how stories are
told with data6 a casualty map >ome and &way?
from -33.com: a political blogger >#atthias *hapiro,
aka =<<<<@ennies? using pennies to e%plain that a
budget cut of A=<< million, while sounding
impressi!e, is actually a tiny fraction of the de8cit: a
collaborati!e !isuali"ation website called
-omment*pace that allows users to create, share, and
comment on !iews of datasets. 7isuali"ation creators
shared their goals and design decisions, and breakout
sessions allowed for discussion in smaller groups.
$wo e%amples of scienti8c storytelling. Top:
*cience on a *phere, a presentation to Bueen
1li"abeth // and @rince @hilip at 3&*& 4oddard
*pace Flight -enter. Bottom: 7isitors at the
1%ploratorium, a science museum in *an
Francisco, e%amining oceanographic data
projected onto a topographical relief map.
$hus, they need !ery little introduction 9 in terms of
our storytelling metaphor, they)re already familiar
with the setting, and all that is left is to identify the
characters >for instance, what glyphs represent, and
how color is used?. /n fact, when we design scienti8c
!isuali"ations, the scientists are usually the ones
setting the stage for usC &dditionally, the fact that the
data are already highly rele!ant to them increases the
likelihood that !isuali"ations will lea!e a lasting
impression in their minds.
owe!er, diDculties arise when scienti8c
!isuali"ations are introduced to broader audiences.
1!en the best !isuali"ations are incomprehensible if
their concepts are alien, and scienti8c !isuali"ations
are often designed assuming !iewer familiarity with
the subject matter. #oreo!er, time constraints and
limited attention spans often preclude the possibility
of full e%planations. ow can we address these issues?
/n ;<=<, a one(day workshop on scienti8c
storytelling was held at the Uni!ersity of -alifornia,
.a!is. @articipants included !isuali"ation researchers
and practitioners as well as e%perts in animation,
scienti8c journalism, and science museum e%hibition.
$he rest of this article presents highlights and 8ndings
from this workshop.
Production Visualization at a Scientifc
Research Center
$he use of !isuali"ations to tell scienti8c stories is a
routine practice at the U.*. 3ational &eronautics and
*pace &dministration >3&*&?. Ebser!ational data 9
that is, data that can be recorded by instruments and
sensors 9 are continuously collected, archi!ed, and
processed from 3&*& airborne missions and
e%periments. &s of ;<==, there are FG airborne
missions operating within the *cience #ission
.irectorate of 3&*& >1arth6 =H, eliophysics6 =F,
&strophysics6 =I, and @lanetary6 =G? J=K. 1ach of these
missions is usually e'uipped with multiple sensors
and instruments, whose purpose is to ac'uire and
transmit data sets daily, hourly, or e!en e!ery few
minutes. .ata ac'uisition is an ongoing process, and
lasts for the duration of the mission. $he majority of
airborne missions are operational for more than a year,
and some can be operational for more than a decade
>e.g., ,and*at satellites?.
3&*& scientists, who are sometimes the
principal in!estigators of missions, need to process
and !isuali"e data ac'uired from airborne science
missions to ad!ance their research and to support
outbound communication and scholarly work, such as
by publishing in scienti8c journals. .ata !isuali"ation
is also needed for education and public outreach
acti!ities, to engage and educate the public about
3&*&)s research and science e5orts. *cientists and
mission teams ha!e their own tools to process and
analy"e data, but cannot easily de!elop and produce
high('uality !isuali"ations for the following reasons6
=? comple%ity and !olume of data: ;? comple%ity of
the tools and technology re'uired to perform high'uality
!isuali"ation production: L? lack of e%pertise
in the 8elds of art, !isuali"ation, and storytelling
production.
$he *cienti8c 7isuali"ation *tudio >*7*? J;K
at 3&*& 4oddard *pace Flight -enter >4*F-?
facilitates scienti8c in'uiry and outreach within
3&*& programs through !isuali"ation. $he *7*
works closely with scientists in the creation of
!isuali"ation products, systems, and processes in order
to promote a greater understanding of 1arth and *pace
*cience research acti!ities at 4*F- and within the
3&*& research community. $he *tudio also pro!ides
e%pertise in data !isuali"ation and science storytelling,
and it is part of the bigger 1arth *cience *torytelling
team, which comprises three entities6 $he *7*, the
-onceptual /mage ,aboratory >concept animators:
non(data(dri!en products?, and the 4oddard $7
#ultimedia >a team of producers, science writers,
editors, and web and social media e%perts?.
.ata !isuali"ations produced and de!eloped at
the *7* are cinematic('uality computer graphics
short 8lms, similar to productions by ollywood
computer animation studios. $he main characteristics
of *7* !isuali"ations are6 scienti8c integrity,
preser!ation of data, seamless blending of multiresolution
data from di5erent sources, aesthetics, and
a solid story that engages the public. $he successful
production of such !isuali"ations depends on freeform
collaboration among members from all three
teams and re'uires the following ingredients6
=? communication between all parties in!ol!ed,
including scientists:
;? data a!ailability and transparency regarding
limitations or problems in the datasets:
L? a conte%t that makes the science story rele!ant and
interesting to the public:
G? a!ailability of resources for the production of a
!isuali"ation story:
I? ability of participants to shift roles, wear multiple
hats, and work collaborati!ely.
*torytelling is a key component in the production of
e!ery !isuali"ation at the *7*. &lthough storytelling
manifests itself di5erently in !arious art forms,
whether they be literary, performance(based, aural,
!isual, or interacti!e, a storyteller should know the
story)s audience and take ownership of the story. /n
general, all forms of stories are made up of the
following ingredients6 perspecti!e, characters,
imagery, and language. &ll of these ingredients are
combined in a structure that de8nes the story from
beginning to end. 7isual storytelling, and speci8cally,
storytelling for animation, borrows from the
con!entions of photography, cinema, episodic comics,
and performance arts. $he structure in storytelling for
animation is established by camera work >!isual
perspecti!e, time and space of framing, composition,
point of !iew, lighting, color, form, and style?, audio
work >with or without, tone?, and the !isual, aural, and
editorial rhythm of the animation.
7isuali"ation(dri!en end products are
archi!ed in the *7* repository J;K, which is a free and
From a storyboard to the story of 3&*&)s ,unar
Meconnaissance Erbiter >,ME? #ission, told as a
se'uence of images. Left6 *toryboard. Right:
*elected images from top to bottom6 ,ME
launched from -ape -ana!eral F,, U*&: ,ME
approaching the #oon: ,ME orbit trail shown with
the sun and the dark side of the #oon: ,ME
mo!ing into orbit around the #oon. JLK
publicly accessible database with more than L,<<<
entries >as of 0une ;<==?. $he products span many
!isuali"ation forms including ;., L.(*tereoscopic,
*cience(En(&(*phere, yper(wall, .ome *how and
e!en touch(display. 1ach production includes !arious
formats including frame sets, still images, mo!ies and,
when appropriate, data in a wide gamut of resolutions.
Upon release the products may take on li!es of their
own, since the public can use them freely.
&lthough there is a streamlined process in
place for the production of !isuali"ations, there are
always challenges that might compromise the 'uality,
structure, and story of the end product. $hese
challenges are often rooted in data issues 9 for
e%ample, gaps in the data, insuDcient amounts of
data, low resolutions, or e!en data that do not show
the e%pected phenomena. Ether times, new
!isuali"ation techni'ues are re'uired in order to
highlight important, necessary information. $he need
for new techni'ues can occur either within the
technical infrastructure of the !isuali"ation production
pipeline >e.g., modi8ed shaders, transitions between
di5erent coordinate systems, or the de!elopment of a
new pipeline? or within the design domain >e.g.,
8nding the best ways to map comple% data to !isual
models?. /n short, the amount of resources and e5ort
re'uired to produce high('uality !isuali"ations for
scienti8c storytelling can o!erwhelm any indi!idual
scientist.
$he *7* is one e%ample of a successful
scienti8c storytelling and !isuali"ation studio. $o
create !isuali"ations suitable for consumption by the
general public, uni'ue challenges must be o!ercome,
and a dedicated team comprised of !ersatile, talented
indi!iduals is needed. /n short, scienti8c storytelling is
not a tri!ial endea!or, and the creation of successful
!isuali"ations re'uires the collecti!e e5ort of many
specialists working together.
Production Visualization at a Science
useum
*cience museums are places where people can
e%perience science in ways they can)t at school or at
home. #useum !isitors can swing on a giant
pendulum, stand under a life(si"e T. Rex fossil
skeleton, or watch the birth of a gala%y in L(..
#useums tell the stories of science, and 9 perhaps
more importantly 9 they also pro!ide a uni'ue !enue
for people across generations to play together, interact
with scientists, and use scienti8c tools.
#useums ha!e long used !isuali"ations to
show the public things they can)t normally see, such
as e!olutionary relationships or the structure of .3&.
2ut !isuali"ations are an increasingly critical medium
for science museums. &s the !olume of data collected
by scientists e%pands e%ponentially, !isuali"ation is
the tool that allows them to make obser!ations or
detect patterns. Whether comparing genomes,
mapping the structure of a !irus, or de!eloping new
models of 1arth)s climate, most scientists now do
some 9 if not all 9 of their work using !isuali"ed data.
$o tell the stories of modern science, museums must
use !isuali"ations.
$he growing importance of !isuali"ations in
science presents an e%citing opportunity for museums.
*cienti8c !isuali"ations can pro!ide stunning images,
engaging the public with phenomena theyN!e ne!er
seen before. 7isuali"ations can be displayed on large,
dynamic interfaces, pro!iding new ways for the public
to participate in interacti!e, social learning.
7isuali"ations can also be used to create authentic
tools for the public to make their own disco!eries,
analogous to microscopes or telescopes. 2ut for
scienti8c !isuali"ations to ha!e any signi8cant
meaning for the public, they must be carefully
interpreted or designed.
/nterpretation, or e%planation, is essential
when using scienti8c !isuali"ations in museums.
7isuali"ations often show comple% and abstract
phenomena at e%treme si"e scales using colors that
ha!e no inherent meaning. For instance, a research
study conducted at the 1%ploratorium, a science
museum in *an Francisco, showed that many !isitors
grossly misinterpreted the scale and use of color in a
nanoscale image JGK. *imilar studies ha!e
documented the diDculty of learners in interpreting
!isuali"ations from 8elds as disparate as genetics and
astrophysics. &s science museums increase their use
of scienti8c !isuali"ations, they are pro!iding more
interpretation through labels, !ideos, and li!e
e%planations. & complementary, though less
common, strategy is to redesign scienti8c
!isuali"ations with the public in mind. @hipps and
Mowe JIK conducted a study in which students were
better able to interpret !isuali"ations of
oceanographic data that had been redesigned with
more intuiti!e color schemes and recogni"able
>though unscienti8c? landmarks.
$he most signi8cant challenge for museums
is 8nding ways to transcend the use of !isuali"ations
as e%planatory animations or pretty pictures. /n many
museums, !isitors can watch stunning simulations of
1arthNs climate or the collapse of a star, but they
$he ay Model at the 1%ploratorium in *an
Francisco allows !isitors to interact with a
scienti8cally accurate !irtual model of how tides,
currents, and ri!ers combine to create the
comple% water Oows of the *an Francisco 2ay
estuary. Using a touchscreen, !isitors place
!irtual Ooats into a !ideo image projected onto a
three(dimensional geographic model of the 2ay
&rea. &fter a Ooat is launched, !isitors watch how
currents mo!e the Ooat to di5erent locations
according to predicted tide and ri!er Oow cycles.
-olor coding highlights !aried water conditions
during tidal phases. JFK
cannot control or e%plore these !isuali"ations. *uch
direct interaction would allow !isitors to control their
e%perience and to make disco!eries with data the way
that scientists do. $he 1%ploratorium is addressing
this challenge by creating !isuali"ation tools for the
public, where they can ask and answer their own
'uestions with real scienti8c data. /n a pilot project
funded by the 3ational *cience Foundation, the
1%ploratorium is collaborating with !isuali"ation
researchers. 2y tailoring their de!elopment process to
di5erent end users >the public? and iterating through
intensi!e prototype testing with !isitors, the
1%ploratorium hopes to create one of the 8rst
e%amples of a !isuali"ation tool that allows the public
to e%plore real scienti8c data without mediation.
/n summary, the increasing role of
!isuali"ation in scienti8c disco!ery presents a
tremendous opportunity for science museums to
engage the public with stunning images, no!el
interfaces, and authentic tools. owe!er, it is essential
to reali"e that transforming the rapidly growing
number of scienti8c !isuali"ations into meaningful
e%periences for the public re'uires thoughtful
interpretation, design, and collaboration.
Storytelling using Interacti!e
Visualization
Following the publication of the 3*F Meport on
!isualization in "cientifc #om$uting in =HPQ, the
early de!elopment of the 8eld of scienti8c
!isuali"ation was largely dri!en by the need to gain
insight into large, comple% datasets arising in science
and medicine. $his led to the in!ention of many new
!isual abstractions, rendering methods, and interaction
techni'ues. 7isuali"ations are used in scientifc
storytelling% but are generally created after the fact,
separate and independent from the data e%ploration
step. $here is no storytelling model built into the
!isuali"ation process: that is, there is no direct support
for creating a story based on the !isuali"ations made
in the process of data e%ploration and knowledge
disco!ery.
$he concept of incrementally creating a story
by depicting the forward progress of the !isuali"ation
process is intuiti!e and powerful. $he scientist is
immersed in the data domain and assembles pieces of
the story together as she learns more and more about
the data. &ni7i" JQK reali"es this concept by allowing
the user of a !isuali"ation system to do e%actly that,
and to present the story as an animation. &s the user
interacti!ely e%plores the data, he or she is able to
locate intrinsic !iews of the data, specify some of
these !iews as keyframes of an animation, re!iew the
animation constructed so far, add annotations and
!oiceo!ers, and edit the keyframes and transitions as
needed until the e%ploration is complete and the
resulting animation is satisfactory.
&ni7i" can e5ecti!ely support storytelling
within a !isuali"ation system based on the well(known
keyframe approach from the 8eld of computer
animation. /t is possible to think of this keyframe
approach as a story model if it is more intuiti!e to the
user. /n earlier work JFK, just such a story model is
described, which is composed of6
>a.? "tory nodes >major steps, or milestones, in a
story, where we can imagine the story coming
to a brief halt, perhaps for the purpose of
interacti!e e%ploration by the story consumer,
before resuming?, and
>b.? "tory transitions >which smoothly connect
story nodes, leading from one node to the
ne%t?.
/t makes sense to consider story transitions as
being composed of se'uential or parallel actions.
2ased on this model, it is possible to create
!isuali"ation stories that conform to the !isual
information(seeking mantra of 2en *hneiderman. &
!isuali"ation story could begin with an o!er!iew of
the data, followed by a focusing transition, leading the
user to a more detailed !isuali"ation of some
particular aspect, and conclude with a guided
se'uence of images that substantiate the message to be
communicated. Ether possible !isuali"ation stories
could be constructed around the aim of comparati!e
!isuali"ation >building up, for e%ample, a side(by(side
comparison during the story? or iterati!e !isuali"ation
>such as the se'uential !isuali"ation of all rele!ant
features in a selected region, following a repetiti!e
pattern like "oom onto a particular feature, rotate
around it, show conte%t, then continue?.
While storytelling, by nature, is not
completely interacti!e, we ponder how interacti!e
storytelling can be facilitated. ow can we stimulate
the participation of story consumers? -an we let them
inOuence not only how the story is told, but the
outcome of the story itself? For e%ample, ad!enture
games allow users to interact with and a5ect a
premade game story. &lso, in science museums, we
8nd many hands(on acti!ities, which may be
considered a form of interacti!e storytelling.
owe!er, once spectators become spect(actors >the
terminology of &ugusto 2oal, in $heater of the
Eppressed JPK?, a conOict of control emerges6 the
spect(actor di!erts the course of the story from the
original plan. $his is also known as the narrati!e
parado%, and di5erent suggestions ha!e been
presented regarding how to address it >e.g., by the use
of emergent narrati!es, as described by &ylett JHK?.
& ta%onomy of four di5erent modes for
splitting the control between the story author and the
story consumer to !arying degrees has been proposed
J=<K. First, the traditional $assive story telling mode
prohibits any interaction on the part of the story
consumer: the story author has full control o!er all
domains. *econd, in story telling with interactive
a$$roval, passi!e storytelling pauses at certain points
and allows spect(actors to take temporary control.
Users can change the !iew, representation, and e!en
content of the !isuali"ation. Ence users are satis8ed
with this interacti!e e%ploration, storytelling continues
as originally intended. $hird, in semi&interactive story
telling, the story consumer is allowed to take control
not just for an interim e%cursion, but for an entire
section of the story. Finally, in total se$aration from
the story, story consumers may completely detach
from the story and engage in interacti!e !isuali"ation
with total freedom.
/n terms of storytelling, interacti!e
!isuali"ation has the potential to help with three
issues that are important in the conte%t of
communication6 comprehensibility, credibility, and
in!ol!ement. First, by incrementally building up a
story, enhancing it with labels and annotations, and
allowing the user to interrupt the story and take
control of it, the risk of presenting an o!erloaded
!isuali"ation and ha!ing it poorly understood is
reduced, thus impro!ing comprehensibility.
&dditionally, the credibility of a !isuali"ation can be
impro!ed if obser!ers are allowed to interact with it
and !erify that the !isuali"ation actually shows what it
claims. Finally, allowing !iewers to interact with
!isuali"ations breaks the fourth wall, transforming
them from passi!e obser!ers to acti!e participants. 2y
acti!ely participating in the process of !isuali"ation,
!iewers will feel a greater sense of engagement with
the data being presented.
Conclusion
-learly, there is a need to consider how storytelling
and !isuali"ation can make scienti8c 8ndings more
comprehensible and accessible to the general public.
*cienti8c !isuali"ation has much to learn from
information !isuali"ation in this regard. -onsider that
information !isuali"ations are aimed at the general
public, and that they draw attention to di5erences and
changes in !isual elements. @erhaps scienti8c
!isuali"ations can take a similar approach in order to
reach broader audiences. /f we focus on important
features by emphasi"ing how they change across time
or e%perimental conditions, we may be able to tell a
compelling story without ha!ing to e%plain e%traneous
details.
/n addition, thinking about !isuali"ations in a
narrati!e conte%t can help make them more
comprehensible, memorable, and credible to the
general public. Whether we use !isuali"ations to tell a
story, or use a story model to make !isuali"ations
more compelling, we cannot neglect the fundamentals
of how to tell a good story. First, know your audience
9 assess their le!el of domain knowledge and
familiarity with !isuali"ation con!entions. 3e%t, set
the stage 9 make sure they ha!e enough background
knowledge about the speci8c dataset being !isuali"ed
to make sense of your !isuali"ation. /ntroduce the
characters 9 show them the !isual elements and what
they represent. .e!elop the plot 9 arrange your !isual
elements in a way that tells an interesting and
compelling story. Finally, lea!e the audience with a
lasting impression by showing them how the story is
rele!ant to them, and its greater implications.
*cienti8c storytelling using !isuali"ation is
not easy, and the successful e%amples highlighted in
this article are the e%ception rather than the rule.
#uch work remains to be done in establishing
guidelines and principles for successful storytelling.
&s !isuali"ation designers, we must ask oursel!es
how we can better support the e5orts of the scienti8c
community in reaching out to the general public.
*cientists ha!e ama"ing stories to tell, and we can
help ensure that they are not 9 to paraphrase #ark
$wain 9 forced to tell them themsel!es.
References
=. http6RRscience.nasa.go!Rmissions
;. http6RRs!s.gsfc.nasa.go!
L. http6RRs!s.gsfc.nasa.go!Rgoto?LF<L
G. 0oyce #a. !isitors' Inter$retations of Images of
the (anoscale. 3/*1 3et @roject Meport, *an
Francisco6 1%ploratorium. ;<<P(<Q(L<.
I. #olly @hipps and *hawn Mowe. "eeing satellite
data. @ublic Understanding of *cience% 7olume
=H, 3umber L, #ay ;<=<, pp. L==(L;=.
F. http6RRwww.e%ploratorium.eduRoutdoorRSRe%hibitRb
ay(modelR
Q. iroshi &kiba, -haoli Wang, and +wan(,iu #a.
)ni!iz* ) Tem$late&ased )nimation Tool for
!olume !isualization. /111 -omputer 4raphics
T &pplications, L<>I?, *eptemberREctober ;<=<,
pp. F=(Q=.
P. &ugusto 2oal. Theatre of the +$$ressed. 3ew
Uork6 $heatre -ommunications 4roup, =HQH.
H. *andy ,ouchart and Muth &ylett. (arrative theory
and emergent interactive narrative. /nternational
0ournal of -ontinuing 1ngineering 1ducation and
,ifelong ,earning, =G>F?, ;<<G, pp. I<F(I=P.
=<. #ichael Wohlfart and elwig auser. "tory
Telling for ,resentation in !olume !isualization.
/n 1uro7is ;<<Q @roceedings, pp. H=(HP.

You might also like