m13 IF5170 IsuVisualisasiData 150414

IF5170 Visualisasi Data
Issues on Visualization:
Color, Large-Scale Data Visualization,
and Perceptual Issues
12/1/2014
FNA/IF5170/Issues on Visualization
Tujuan
Mahasiswa memahami berbagai isu yang
terkait dengan penggunaan warna (color)
dalam visualisasi data
terkait dengan visualisasi data dalam jumlah
besar
terkait dengan persepsi dalam visualisasi data
12/1/2014
COLOR
12/1/2014
Sources
C. Ware: Information Visualization:
Perception for Design, Chapter 4: Color,
Morgan Kauffman, 2004
Some pictures are from:
J. Estelle, N. Illinsky: Beautiful Visualization:
Looking at Data Through the Eyes of Experts,
Chapter 4: Color: The Cinderella of Data
Visualization by Michael Driscoll
12/1/2014
12/1/2014
The use of color in visualization

It is useful to think of color as an attribute of
an object rather than as its primary
characteristic.
It is excellent for labeling and categorization,
but poor for displaying shape, detail, or space.
12/1/2014
Color Measurement
We can match any color with a mixture of no more
than three primary lights the basis of colorimetry
12/1/2014
12/1/2014
CIE System of Color Standards
12/1/2014
Opponent Process Theory
12/1/2014
10
Application of Color in Visualization
Color specification interfaces and color spaces

Color for labeling
Color sequences for data maps
Color reproduction
Color for exploring multidimensional discrete
data
12/1/2014
11
Color Specification Interfaces and

Color Spaces (1)
Types of control that can be given to the users
to choose their own colors:
a point in a three-dimensional color space
a set of color names to choose from
a palette of predefined color samples
12/1/2014
12

Color Spaces (2)
HSV color space (Smith, 1978) hue, saturation,
and value (HSV) coordinates to RGB monitor
coordinate
12/1/2014
13
12/1/2014
14

Color Spaces (3)
Color naming systems, e.g.:
Natural Color System (NCS)
based on Herings opponent color theory (1920)
widely used in England and other European countries
The Pantone system

used in the printing industry
The Munsell system

an important reference for surface colors
12/1/2014
15

Color Spaces (4)
NCS
12/1/2014
16
Color for Labeling (1)

Nominal information coding: technical name
for labeling an object
Color is extremely effective as a nominal code
Perceptual factors to choose a set of color

labels:
1. Distinctness: the degree of perceived difference
between two colors that are placed close
together
(kemampuan untuk menentukan perbedaan)
12/1/2014
17
12/1/2014
18

Perceptual factors to choose a set of color labels
(cont.):
2. Unique hues: red, green, yellow, blue, black, white
Natural choices when a small set of colors is needed

No two colors should be chosen from the same category
3. Constrast with background

Color-coded objects can be expected to appear on a
variety of backgrounds
A method for reducing contrast effects place a thin
white or black border around the color-coded object
12/1/2014
19

Perceptual factors to choose a set of color
labels (cont.):
4. Color blindness: most color-blind population
cannot differ colors in a red-green direction
Almost everyone can distinguish color in yellow-blue
direction
(tergantung user, buta
Reduces the design choices warna atau tidak)
5. Number: Only a small number of color can be

rapidly perceived five to ten codes (Healey,
1996)
12/1/2014
20

Perceptual factors to choose a set of color labels
(cont.):
6. Field size: the larger the area that is color-coded,
the more easily colors can be distinguished
Small objects highly saturated colors for maximum

discrimination
Large areas low saturation, differ only slightly from one
another
7. Conventions:
12/1/2014
Common conventions, e.g. red = hot, red = danger, blue =

cold, green = life, green = go
21

12 colors recommended for use in coding:
12/1/2014
22
12/1/2014
23
Color Sequences for Data Maps (1)

Pseudocoloring: the technique of
representing continuously varying map values
using a sequence of color
Used in: astronomical radiation charts,
medical imaging, geography, etc.
12/1/2014
24
12/1/2014
25

Nominal pseudocolor sequence:
pseudocolor sequence designed to enable rapid
visual classification of regions where the values
within the regions have no particular order (i.e.,
no greater than relationship holds for the
values)
12/1/2014
26
12/1/2014
27

Ordinal pseudocolor sequence:
pseudocolor sequence in which the monotonic
ordering of data values in different parts of the
display can be perceived
Can be achieved by using:

a black-white, red-green, or yellow-blue sequence
a saturation sequence or with any relatively
straight line through opponent color space
a spectrum approximation
12/1/2014
28
A map of ozone concentrations in the

atmosphere is shown:
(a) As a blackwhite sequence.
(b) As a saturation sequence.
(c) As a spectrum-approximation sequence.
Images courtesy of Penny Rheingans (Rheingans,
1999).
12/1/2014
29
12/1/2014
30

Interval pseudocolor sequences:
pseudocolor sequences in which each unit step of the
sequence represents an equal change in magnitude
of the characteristic being displayed across the whole
range of the sequence.
Can be achieved by:

Using a uniform color space in which equal perceptual
steps correspond to equal metric steps (Robertson
and OCallaghan, 1988)
Introducing steps deliberately in the color sequence (a
banded color sequence)
Using isovalue contours
12/1/2014
31
12/1/2014
32
12/1/2014
33

Ratio pseudocolor sequence:
An interval sequence that has a true zero and all that
this implies: the sign of a value is significant; one value
can be twice as large as another
No known visualization technique is capable of

accurately conveying ratios with any precision
A sequence can be designed that effectively
expresses a zero point and numbers above and
below zero
12/1/2014
34
12/1/2014
35

Sequences for the color blind:
Some color sequences are not perceived well by
people with color blindness red to green
Sequences on black-to-white or yellow-to-blue
dimension will still be clear for color-blind people
12/1/2014
36
12/1/2014
37

Bivariate color sequences:
it is possible to display two or even three
dimensions using pseudocoloring
e.g. yellow-blue variation combined with lightdark variation and saturation
12/1/2014
38
12/1/2014
39
12/1/2014
40
12/1/2014
41
Color Reproduction (1)

The problem transferring color
appearances from one display device to
another
E.g.: from a computer monitor to a sheet of paper
The gamut of a device: the set of all colors

that can be produced by a device
12/1/2014
42
12/1/2014
43

Heuristic principles to create good mapping from
one device to another (Stone et al.,1988):
The gray axis of the image should be preserved
Maximum luminance contrast (black to white) is
desirable
Few colors should lie outside the destination gamut
Hue and saturation shifts should be minimized
An overall increase of color saturation is preferable to
a decrease
12/1/2014
44
12/1/2014
45

Heuristic principles to create good mapping from one
device to another (Stone et al.,1988) (cont.), steps:
Calibration: to calibrate the monitor and the printing
device in a common reference system
Range scaling: to equate the luminance range of the
source and destination images
Rotation: to equate the monitor white with the paper
white, the monitor gamut is rotated so as to make the
white axes collinear
Saturation scaling: the monitor gamut is scaled radially
with respect to the blackwhite axis to bring the monitor
gamut within the range of the printing gamut
12/1/2014
46
Color for Exploring Multidimensional

Discrete Data (1)
One of the most interesting but difficult
challenges for data visualization is to support
exploratory data analysis
Problems can arise in exploring data when
more than two dimensions of data are to be
displayed
12/1/2014
47
12/1/2014
48
Conclusions (1)
To show detail in a visualization, always have
considerable luminance contrast between
foreground and background information
Use only a few colors if they are distinct codes
Black or white borders around colored symbols
can help make them distinct by ensuring a
luminance contrast break with surrounding colors
Red, green, yellow, and blue are hard-wired into
the brain as primaries
12/1/2014
49
Conclusions (2)
When color-coding large areas, use muted colors,
especially if colored symbols are to be superimposed
Small color-coded objects should be given highsaturation colors
When a perceptually meaningful ordering is needed,
use a sequence that varies monotonically on at least
one of the opponent color channels
If it is important to show variations above and below
zero, use a neutral value to represent zero and use
increases in saturation toward opposite colors to show
positive and negative values
12/1/2014
50
Conclusions (3)
Color contrast can cause large errors in the
representation of quantity
Beware of oversaturating colors, especially
when a printed image is to be the end product
12/1/2014
51
Tugas
Untuk dikumpulkan dan dibahas minggu depan
(Selasa, 22 April 2014)
Carilah 1 buah contoh visualisasi data (bebas, boleh
interaktif, boleh statik)
Jelaskan:
Data apa yang divisualisasikan dan bagaimana teknik
visualisasinya
Diskusikan berdasarkan aspek-aspek evaluasi (lihat catatan
di belakang ini)
Diskusikan penggunaan color dalam visualisasi tersebut
(kaitkan dengan saran-saran pada kuliah hari ini)
Dikerjakan berkelompok 2 orang, ditulis dalam

laporan hardcopy di kertas A4
12/1/2014
52
Evaluation
Things to watch out:
Data and statistical accuracy
Visualization accuracy
Functional accuracy
Visual inference
Formatting accuracy
Annotation accuracy
12/1/2014
FNA/IF5170/Interactive Visualization and

Virtual Reality
53
Post-Launch Evaluation
To seek to assess the visualization's
effectiveness and impact in a post-launch
setting position yourself as the user
12/1/2014
FNA/IF5170/Interactive Visualization and

Virtual Reality
54
LARGE-SCALE DATA VISUALIZATION
12/1/2014
55
Sources
C. Johnson, C. M. Hansen: The Visualization
Handbook, Elsevier, 2004:
Chapter 27: Large Scale Data Visualization and
Rendering: A Problem Driven Approach by P.
McCormick and J. Ahrens
Chapter 28: Issues and Architectures in LargeScale Data Visualization by C. Pavlakos, P. D.
Heermann
12/1/2014
56
Introduction
Large-scale data the datasets are larger
than can be processed by a single computer
4 fundamental techniques for processing the
visualization of large-scale data:
Data streaming
Task parallelism => data sama, proses beda
Pipeline parallelism
Data parallelism (penjumlahan matriks cocok untuk ini)
+ Hybrid systems
12/1/2014
=> proses sama
57
Data Streaming (1)

Data streaming: Most commonly used to process
independent subsets of a larger dataset, one subset at
a time
Often, the only feasible approach in situations where the
size of a dataset exceeds the capacity of the available
computing resources (memory and swap space)
The key advantage any size dataset can be

successfully processed
The drawback often requires a substantial amount
of execution time and does not allow for the
interactive exploration of the data
12/1/2014
58
Data Streaming (2)

In order to produce the correct solution, the
algorithms must be result invariant:
the results must be consistent regardless of the
number of subsets into which the data is split
the algorithms must be able to divide the original
dataset into piece
12/1/2014
59
Data Streaming (3)

Case study: Terascale Supernova Initiative (TSI)
project
12/1/2014
60
http://www.csm.ornl.gov/newsite/group_astro.html
12/1/2014
61
12/1/2014
62
12/1/2014
63
Data Streaming (4)

Conclusions of case study:
It takes longer to process a single piece of data than to
process multiple subsets of data
By combining the performance benefits of data
streaming and the reduction in overall memory usage,
it is possible to use a computer system with a much
lower price point than might initially be expected to
successfully process a large dataset
A straightforward approach to data streaming fails to
meet the performance goal of 5 to 30 frames/s
Data streaming has introduced the ability to correctly
process subsets of a large dataset
12/1/2014
64
Task Parallelism (1)

Task parallelism: independent modules in an application
execute in parallel
An algorithm is required to be broken up into independent
tasks and that multiple computing resources be available
The key advantage it enables multiple portions of a
visualization task to be executed in parallel
The main disadvantage
Difficult to calculate the number of independent tasks that can
be identified and the number of CPUs available, limits the
maximum amount of parallelism
Difficult to load-balance the tasks it can be very challenging
to take full advantage of the available resources
12/1/2014
65

Case Study: Datasets on earths oceans by
Parallel Ocean Program (POP) at Los Alamos
National Laboratory
12/1/2014
66
12/1/2014
67
12/1/2014
68
12/1/2014
69

The bottleneck in gathering the results on a single
node
Limited to 16 total isosurface tasks theoretical
limit of factor of 16 times faster than a single
processor implementation
Performance limitation as the result of the cost of
sending the data from each of the tasks to the
display process as well as the load imbalance
between the various tasks
12/1/2014
70
Pipeline Parallelism (1)

Pipeline parallelism: a number of modules in an application
execute in parallel, but on independent subsets of data
Advantages:
In situations where there are multiple, heterogeneous tasks.
It allows parallel use of the overall computing resources
Disadvantages:
It can make it difficult to balance the execution time required by
the individual stages; in an unbalanced pipeline, the slowest
stage directly impacts the overall performance.
The length of the pipeline directly limits the amount of
parallelism that can be achieved (i.e., you must have as many
processors as there are pipeline stages).
12/1/2014
71
Pipeline Parallelism (2)

Case study: A simple PC-based animation
application that reads the image files from
disk and displays them on a single monitor
Conclusion:
the two tasks require similar time for small images
but the read operation becomes more costly as
the image size increases
12/1/2014
72
12/1/2014
73
12/1/2014
74
Data Parallelism (1)

Data parallelism: the code within each module of
an application executes in parallel
requires that a dataset be subdivided and multiple
processes run the same algorithm on the resulting
pieces concurrently
can be implemented as an extension of the datadecomposition technique described in the streaming
section
Commonly referred to as a single program

multiple data (SPMD)
12/1/2014
75

The advantages:
It can achieve a high degree of parallelism;
solutions tend to scale well as the number of
processors increases.
When there is a large number of processors
available, this approach is often one of the best
ways to achieve increased performance.
Possible drawback:
Scalability can be limited by interprocess
communication costs
12/1/2014
76

Case study: Salinity of the Atlantic by creating and
viewing an isosurface of salinity colored by
temperature (by by the Parallel Ocean Program
(POP))
Conclusions:
Performance improves by a factor of two with each
doubling of the number of processors for all dataset
sizes.
Due to the memory requirements of the full dataset
and the resulting graphics primitives, visualization is
only possible when we use 16 or more processor
12/1/2014
77
12/1/2014
78
12/1/2014
79
Summary
12/1/2014
80
Hybrid Systems (1)

Although using multiple techniques can lead
to a more complex implementation, it is often
a very powerful way to fully utilize available
resources
12/1/2014
81
Hybrid Systems (2)

Case study: TRex is a hybrid system that uses
data streaming, pipelining, and data
parallelism
a large volume of data is broken into individual
subsets in order to efficiently render the data
data parallelism
overall series of parallel tasks: Read, Render,
Composite, Display parallel pipeline
streaming multiple time-steps of data through the
Read module
12/1/2014
82
12/1/2014
83
Hybrid Systems (3)

The specific details of a hardware architecture into
consideration and leveraging the advantages of
three of the techniques can be advantageous
The size of data that can be processed is limited
by the amount of texture memory
12/1/2014
84
The process of analyzing and

visualizing large-scale scientific data
12/1/2014
85
Some Problems (1)

Enabling Data Exploration and Discovery
The need for tools and environments that support
the efficient, effective exploration of data
A robust interactive environment is needed that
enables the scientist or analyst (the user) to
receive timely and useful feedback in the search
for answers.
This environment must be accessible from the
office, which is where day-to-day work is done
12/1/2014
86
Some Problems (2)

The Need for Scalable Solutions
The computing resources demanded by complex,
high-fidelity simulation applications inherently imply
the use of parallel computing
Common problems: the visualization of data
does not fit in the memory of a conventional office
visualization system
is large enough that conventional office graphics packages
are inadequate
may not fit on the local disk
is large enough that traditional high-performance graphics
systems lack the rendering performance
12/1/2014
87
An End-to-End Architecture for

Large Data Exploration
12/1/2014
88
12/1/2014
89
Conclusions (1)
Features of tools and environments that enable
effective and efficient data analysis and
visualization of large-scale data:
The ability to process large data without moving it all
around
The ability to manipulate data at an object level
rather than at a complete dataset level
A rich functional set of data and visualization services
that enable data exploration and discovery
12/1/2014
90
Conclusions (2)
visualization of large-scale data (cont.):
The ability to get the right data to the right place for
further processing when needed
Wide availability of cost-effective, scalable, highperformance infrastructures
Parallelism and the ability to support end-to-end
parallelism throughout high-performance parts of the
environment
12/1/2014
91
Conclusions (2)
visualization of large-scale data (cont.):
The ability to apply a broad set of diverse tools,

leveraging interoperability
The ability to leverage the interactivity and power of
increasing desktop computing and visualization
resources
The ability to drive the whole environment from the
computational-science laboratory of choice, namely
the office
12/1/2014
92
PERCEPTUAL ISSUES
12/1/2014
93
Source
C. Johnson, C. M. Hansen: The Visualization
Handbook, Elsevier, 2004:
Chapter 39: Extending Visualization to
Perceptualization: The Importance of Perception in
Effective Communication of Information by D. S.
Ebert
Chapter 40: Art and Science in Visualization by
Victoria Interrante
C. Ware: Information Visualization: Perception

for Design
Appendix C: The Perceptual Evaluation of
Visualization Techniques and Systems
12/1/2014
94
Introduction
Traditional visualization must evolve into
perceptualization of information
conveying information through multiple
perceptual channels and perceptually tuned
rendering techniques
The choice of visual rendering techniques

should be driven by characteristics of human
perception
perceptual channels are the communication
medium.
12/1/2014
95
Human Perception
Categories:
Preattentive processing: some of the information
is processed at a very low level in parallel without
conscious thought (persepsi alam bawah sadar)
Attentive processing: other information requires
attention, or conscious thought, to perceive the
information. (persepsi yang dilakukan dengan berpikir)
12/1/2014
96
Preattentive Visual Processing (1)

Preattentive Visual Processing
Visual processing that is apparently accomplished
automatically and simultaneously for the entire
visual field of view
An understanding of which visual features are
preattentive can be used to more effectively
design visualization systems
12/1/2014
97
Preattentive Visual Processing (2)

Classification of features:
12/1/2014
98
Types of Channels
Types of channels:
The visual perceptual channel the most widely
used communication channel
The auditory and haptic channels being
incorporated
to convey additional information
as redundant forms of communication to increase accuracy
or speed of communication
others: olfactory, vestibular, gustatory
Understanding the way humans perceive

information vital to the effective conveyance
of information through perceptualization.
12/1/2014
99
The Use of Art dan Science in

Visualization
There are significant potential benefits in:
Seeking inspiration from previous graphical work
in art, illustration, visual communication, and
design
Seeking insights from research in vision and visual
perception
12/1/2014
100
Using gaps to effectively portraying dense collections of overlapping lines
Since prehistoric times, artists have used gaps to indicate the passing of one surface behind
another, as shown in the image of a horse from Paleolithic cave paintings in Lascaux, France
12/1/2014
101
Using gaps to effectively portray dense collections of overlapping lines
12/1/2014
102
Using Feature Lines to Emphasize the Essential 3D Structure of a Form
12/1/2014
103
Using Texture on Surfaces to Clarify Shape
12/1/2014
104
Using Texture on Surfaces to Clarify Shape
12/1/2014
105
An illustration of the texture interference effects that arise when multiple

overlapping transparent surfaces are rendered with principal direction
texture strokes
(http://www-users.cs.umn.edu/~interran/tvcg/tvcg.html)
12/1/2014
106
Perceptual Evaluation Techniques

Empirical research methods can be applied to
evaluate perceptual aspects of visualization:
Psychophysics
Statistical Exploration
Detection methods
Method of adjustment
Cognitive Psychology
Structural Analysis
Testbench applications
Structured interviews
Rating scales
12/1/2014
Principal Component
Analysis
Multidimensional Scaling
Clustering
Multiple Regression
Cross-cultural Studies
Child Studies
107
Psychophysics (1)
Psychophysics:
A set of techniques based on applying the methods of
physics to measurements of human sensation
Extremely successful in defining the basic set of

limits of the visual system and discovering the
important sensory dimensions of color, texture,
sound, and so on, e.g.:
How rapidly must a light flicker before it is perceived
as steady?
What is the smallest brightness change that can be
detected?
12/1/2014
108
Psychophysics (2)
Problems with psychophysics:
Often carried out using only one or two observer
generalized to the entire human race
But, some experiments require hundreds of hours of
careful observation large subject population is out of
question
It is usually assumed (sometimes wrongly) that

instructional biases are not significant in the
experiment
12/1/2014
109
Psychophysics (3)
Information psychophysics: a new variant of
psychophysics
To apply classical psychophysics to common
information structures, e.g. elementary flow
pattern, surface shape, paths in graphs
12/1/2014
110
Psychophysics (4)
Common psychophysical methods (may also
be applied in information psychophysics):
Detection Method
The goal of the experiment: determining the error rate
how many errors people make when performing a
certain task
E.g.: aircraft inspection process expected error rate
of an inspector is critical
Error rate is commonly used to determine threshold
Staircase procedure
Signal detection theory
12/1/2014
111
12/1/2014
112
Psychophysics (5)
Common psychophysical methods (cont.):
Method of Adjustment
Give application domain experts control over some
variable and ask them to adjust it so that it is optimal in
some way for them
Can also be used to answer questions about perceptual
distortion
12/1/2014
113
Cognitive Psychology
Goal of experiments: testing a hypothesis
about a cognitive model
Methods:
Measuring reaction time or measuring error
E.g. Determining whether or not a particular object is in
a display
Measuring interference between visual patterns

Increase in the resulting errors is used as evidence that
different channels of information processing converge
at some point
12/1/2014
114
Structural Analysis (1)

In structural analysis, theories of cognitive processing
are constructed using direct observation as evidence
Studies are conducted via interviews rather than formal
experiments
Often especially appropriate for studies of computer
interfaces fast moving and can take variety of factors
into account
Research tools:
Testbench applications
Semistructured interviews
Rating scales
12/1/2014
115

Testbench Applications:
The testbench application: a flexible tool capable
of producing a range of visual mappings of the
data and a range of interaction possibilites
E.g. Problem: find the best way to represent the
shape of a surface, then the testbench app should
be able to:
Load different surface shapes, change lighting, change
surface texture properties, turn stereoscopic viewing
on/off, provide motion parallax cues, etc.
12/1/2014
116

Structured Interviews:
Construct an interview with a structured set of
questions of elicit information about specific task
requirements
In visualization:
To evaluate what aspects of visualization actually are
important for potential users
To evaluate a number of different solutions for strengths and
weaknesses
Advantage: to gain information about a wide range of

issues with relatively little effort
12/1/2014
117

Rating Scales:
A method for turning opinions into numbers, e.g.:
We have six visual representations of a flow pattern, we might ask

subjects to rate how well they are on a scale of 1 to 5
Subjects tend to bias the rating scale toward either the

lower or upper end
No absolute meaning should be given to rating scale data
Rating scale an excellent tool for measuring relative
preferences
12/1/2014
118
Statistical Exploration (1)

Statistical discovery techniques can be useful to
learn about some class of visualization methods
Using statistics to discover how many dimensions that
can be conveyed by a visualization
Major techniques:
12/1/2014
Principal Component Analysis

Multidimensional Scaling
Clustering
Multiple Regression
119

Principal Component Analysis:
The goal: to take a set of variables and find a new set
of variables (the principal components) that are
uncorrelated with each other might be used to
reduce a high-dimensional dataset to lower
dimensions
Multidimensional Scaling:
A method explicitly designed to reduce the
dimensionality of a set of data points to two or three,
so that these dimensions can be displayed visually
12/1/2014
120

Clustering:
A statistical technique designed to find clusters of
points in a data space of any dimensionality
Two basic kinds: hierarchical and k-means
Multiple Regression:
A statistical technique that can be used to
discover whether it is possible to predict some
response variable from display properties
12/1/2014
121
Cross-Cultural Studies
Cross-cultural studies can be used to test
whether sensory codes are interpreted easily
by all humans
E.g. Color naming are compared across more than
100 languages (Berlin & Kay, 1969)
It is becoming impossible with the globalization of
world culture
12/1/2014
122
Child Studies
Using behaviorism techniques:
To discover things about a childs sensory
processing even before the child is capable of
speech revealing basic processing mechanism
12/1/2014
123
Practical Problems in Conducting User

Studies (1)
Experimenter bias
There are many opportunities for experimenter
bias in both the gathering and the interpretation
of results
How many subjects are used?

Statistically, the number of subjects and the
number of observations required depend on the
variability of responses with a single subject and
the variability from one subject to another
12/1/2014
124

Studies (2)
Combinatorial Explosion:
In visualization design problems, there are often
many possible independent variables leads to
combinatorial explosion cannot be
experimented using brute force approach
Task Identification:
In order to provide a useful measure of
performance, it is also important that the task can
be set up to have a clear and simple user response
12/1/2014
125

Studies (3)
Controls:
A control is a condition that is used to provide
some basis for comparison
In evaluating a new visualization method, the
most reasonable control is the current best
practice display method
Getting help:
Studies in information visualization are
fundamentally multidisciplinary
12/1/2014
126

m13 IF5170 IsuVisualisasiData 150414

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

m13 IF5170 IsuVisualisasiData 150414

Hochgeladen von

Copyright:

Verfügbare Formate

IF5170 Visualisasi Data

The use of color in visualization

CIE System of Color Standards

Opponent Process Theory

Application of Color in Visualization

Color specification interfaces and color spaces

Color Specification Interfaces and

Color Specification Interfaces and

Color Specification Interfaces and

The Pantone system

The Munsell system

Color Specification Interfaces and

Color for Labeling (1)

Perceptual factors to choose a set of color

Color for Labeling (2)

Natural choices when a small set of colors is needed

3. Constrast with background

Color for Labeling (3)

5. Number: Only a small number of color can be

Color for Labeling (4)

Small objects highly saturated colors for maximum

Common conventions, e.g. red = hot, red = danger, blue =

Color for Labeling (4)

Color for Labeling (5)

Color Sequences for Data Maps (1)

Color Sequences for Data Maps (2)

Color Sequences for Data Maps (3)

Can be achieved by using:

A map of ozone concentrations in the

Color Sequences for Data Maps (4)

Can be achieved by:

Color Sequences for Data Maps (5)

No known visualization technique is capable of

Color Sequences for Data Maps (6)

Color Sequences for Data Maps (7)

Color Reproduction (1)

The gamut of a device: the set of all colors

Color Reproduction (2)

Color Reproduction (3)

Color for Exploring Multidimensional

Dikerjakan berkelompok 2 orang, ditulis dalam

FNA/IF5170/Interactive Visualization and

FNA/IF5170/Interactive Visualization and

LARGE-SCALE DATA VISUALIZATION

=> proses sama

Data Streaming (1)

The key advantage any size dataset can be

Data Streaming (2)

Data Streaming (3)

Data Streaming (4)

Task Parallelism (1)

Task Parallelism (2)

Task Parallelism (3)

Pipeline Parallelism (1)

Pipeline Parallelism (2)

Data Parallelism (1)

Commonly referred to as a single program

Data Parallelism (2)

Data Parallelism (3)

Hybrid Systems (1)

Hybrid Systems (2)

Hybrid Systems (3)

The process of analyzing and