An Introductory Guide To Scientific Visualization

An

Introductory

Guide

to

Scientific

Visualization

R.

A.

Earnshaw

N.Wiseman

An

Introductory Guide

to

Scientific Visualization

With

72 Figures

Springer-V erlag Berlin Heidelberg GmbH

De.

Rae

A.

Eamshaw

Head

of

Computer

Graphics

University

of

Leeds

I.S2

9JT,

U. K .

Nonnan

Wiseman

NERC

Computer

Services

Kingsley Dunham

Cent~

Keyworth

Notts

NG12 5GG, U. K.

Front

covrr

plate

$hoW$

an image

of

Saa

Frandsco

with

sc:vc:ntl

multi-$pcctntl

analy5c:S

bc:1ng

c:mied

out.

Counesy

of

Silicon Graphlcslnc.,

AVIRlS

data, and

the

Jet Propulsion lJIbonnory.

ISBN

978-~2-63470-3

libnuy

of

Congress CatalogiJlg·in-Publication Data

Earnshaw,RaeA.

An

introductO!:")'

gulde

to

sclcntlficvlsualization/RA.Eamshaw,

N. Wlseman. p .

cm.

Includa

bibliographical refcrences

and

Index.

ISBN 978-3-642-63470-3 ISBN 978-3-642-58101-4 (eBook)

DOI 10.1007/978-3-642-58101-4

1. Science-Methodology. 2. YisuaUzation-Data proces!lng.

1.

Wlsc:maa, N.

(No

rman)

II

. Title. Q175.

E2H

1992

502.8-dc20

92-10987 CIP

This

work

is

subject

10

copyright

.

AlI

rights are reserved,

whether

the

whole

OI

part

of

the

material

is

concemed,

spcclfically

the

rtghts

of

translatlon,

reprinting, reusc:

of

ilIustrations, recitation, broadcasting,

rcproductlon

on

microfilms

or

in

aay

other

way,

and

storage

in

data banks. Duplication

of

this

publication

or

pans

thereof

ls

pcrmitted

only

under

the

provlsiofl5

of

the

~rman

Copyright

Law

of

Septembc:r

9,

1965,

In

II.!;

currcnt

version

,

and

pcrmission for

usc:

must

always

bc:

obtained

(rom

Springer-Yerlag. Yiolations

are llable for prose<:ution

under

the

~rman

Copyright

Law.

CI

Springer·Yerlag Berlin Heideibc:rg 1992

Originally

pub

li

.~hed

by Spri nger-

Yerlag

Berlin Heidclberg

New

York

in

1992

Softcovcr

repri

nI

of

lhe

hardcovt.'T

I

st

edilion

199

2

The

usc:

of

registercd names. trademarks,

etc.

In thls publlcation

does

nOI

imply,even

in

the

absence

of

a specific

statement

,

that

such

names

are

c:xempt

(rom

the

relevant

protectiv

e

bws

and

regulations

and

there!ore

free for general

"~

.

Cover

Design: H. l.oplca, Iiv

eshe

im

Typnctti.og:

K + Y

~'otosaIZ

GmbH,

Be

erfc1dcn

3313140-5432 -

Printed

on

acid-frec

paper

Foreword

Visualization has been the cornerstone of scientific progress

throughout history. Much of modern physics

is

the result of

the superior abstract visualization abilities of a

few

brilliant

men. Newton visualized the effect of gravitationa.l force

fields in three dimensional space acting

on

the center of

mass.

And Einstein visualized the geometric effects of ob-

jects in relative uniform and accelerated motion, with the

speed of light a constant, time part of space, and accelera-

tion indistinguishable from gravity. Virtually all com-

prehension in science, technology and even art calls on

our

ability to visualize. In fact, the ability to visualize

is

almost

synonymous with understanding.

We

have all used the

ex-

pression "I

see"

to mean "I understand".

Modern science part departs from the closed theories of

the last century and demands computer simulations

to

understand real world situations. Scientific Visualization

is

the

eyes

through which these simulations

are

viewed, from

electrochemical bonds to simulated interstellar jets associat-

ed with black holes.

Scientific Visualization

is

of value beyond strictly scien-

tific applications, however. The same technology

is

now

used in such diverse applications

as

clothing design, in-

dustrial design, automobile and airplane design, genetic

engineering, chemical and drug design, oil and mineral

ex-

ploration, chemical and nuclear power plant design, and

motion picture special effects and animation. It

is

rapidly

becoming a requirement for virtually all disciplines that

deal with geometric things.

What

is

Scientific Visualization?

It

is

a set of software

tools coupled with a powerful 3D graphical computing envi-

VI

Foreword

ronment that allows any geometric object

or

concept to

be

visualized by anyone. The software provides an easy to use

interface for the user. The hardware must

be

able to manip-

ulate complex, geometrically described, 3D environments

in motion, color and with any

level

of "realism" called for

to better communicate the essence of the computation.

Scientific Visualization

is

in its infancy, but the technol-

ogy

is

sure to revolutionize scientific education. I believe

that the requisite 3D graphical processing capability will

be

built into all personal computes within the next

five

years.

And by the year 2000, I am confident that even the home

digital television will combine such 3D graphical processing

capability with digital video and audio. Then, even complex

scientific textbooks will

be

viewed interactively

on

the

home screen, with video clips depicting a lecturer, mathe-

matical experiments run in and visualized on the "TV" and

the student able to guide the learning process. But until

then, such books

as

this will guide the

way.

May

1992

James

H.

Clark

Chairman, Silicon Graphics Inc

Mountain View

California, USA

Preface

Scientific visualization

is

concerned with exploring data

and information in such a way

as

to gain understanding and

insight into the data. This

is

a fundamental objective

of

much scientific investigation.

To

achieve this goal, scientific

visualization involves aspects in the areas

of

computer

graphics, user-interface methodology, image processing,

sys-

tem design, and signal processing.

This guide

is

intended for readers new

to

the field

who

require a quick and easy-to-read summary

of

what scientific

visualization

is

and what it can do. Written in a popular and

journalistic style

with

many illustrations, it will enable read-

ers to appreciate the benefits

of

scientific visualization and

how current tools can be exploited in many application ar-

eas.

It

will be invaluable for scientists and research workers

who

have never used computer graphics

or

other visual

tools before, and

who

wish to find out the benefits and

ad-

vantages

of

the new approaches.

This guide

is

concerned to answer the questions which

the newcomer to visualization may wish to ask concerning

what

it

is,

what it can do, what facilities are available, and

how much it costs. Points

of

contact for further informa-

tion

are also provided.

VIII

Dr.

R.

A.

Earnshaw

University

of

Leeds

UK

Mr.

Norman

Wiseman

Natural Environ-

ment Research

Council, UK

About the Authors

About

the

Authors

Rae Earnshaw

is

Head of Computer Graphics at the Uni-

versity of

Leeds,

with interests in graphics algorithms, scien-

tific visualization, display technology, CAD/CAM, and hu-

man-computer interface

issues.

He

has been a Visiting Pro-

fessor at Illinois Institute of Technology, Chicago, USA,

Northwestern Polytechnical University, China, and George

Washington University, Washington DC, USA.

He

was

a

Director

of

the NATO Advanced Study Institute on "Fun-

damental Algorithms for Computer Graphics" held in Italy,

England, in

1985,

a Co-Chair of the BCS/ACM Internation-

al

Summer Institute on "State of the

Art

in Computer

Graphics" held in Scotland in

1986,

and a Director of

of

the

NATO Advanced Study Institute on "Theoretical Founda-

tions of Computer Graphics and CAD" held in Italy in

1987.

He

is

a member

of

ACM, IEEE, CGS, EG, and a

Fel-

low

of

the British Computer Society.

Norman Wiseman

is

Northern Area Computer Services

Manager for the Natural Environment Research Council.

His special interests

are

in the application

of

visualization

in physical and biological sciences of the environment; edu-

cation and training

of

scientists in the

use

of graphical tech-

niques; and raster print technology.

He

has

been a systems

consultant for a number of

years

and

has

been involved in

several graphics hardware, software and training initiatives

in the

UK

Academic and Research Council communities.

Prior to this he has worked

on

a number of software pro-

jects involving the acquisition, storage, display and analysis

of seismic and borehole log data, primarily for

use

by scien-

tists in the British Geological

Survey.

He

is

a member of

Eurographics.

Contents

Acknowledgements

.................

XIII

Disclaimer . . . . . . . . . . . . . . . . . . . . . . .

..

XIV

Copyright Material

.................

XIV

Trademarks

........................

XV

Part I

Basics

of Scientific Visualization

Chapter 1 Introduction and Background

1.1

Introduction.

. . . . . . . . . . . . . . . . . 3

1.2

Background

..................

3

Chapter 2 What Scientific Visualization

can

do!

2.1

What

is

Scientific Visualization? 5

2.2

How

to do Scientific Visualization 7

2.3

Some Examples of Scientific

Visualization

.................

8

2.3.1

The March of Napoleon's Army 8

2.3.2 Cholera Outbreak

.............

9

2.3.3 Weather Maps from Meteorology

10

2.3.4 Molecular Modeling

...........

12

2.3.5 Pelvic Reconstruction

..........

12

2.3.6 Oil Exploration

...............

14

2.3.7 Designing Ship Propellors

......

16

2.3.8 Visualization of Forest Growth

18

Chapter 3 Explanation of Scientific

Visualization Terminology

3.1

Techniques...................

20

3.2

Volume Visualization

..........

25

3.3

Data Types

...................

27

3.3.1

Overview of Facilities. . . . . . . . . .

27

3.3.2

HDF

........................

29

3.3.3

NetCDF

.....................

30

3.3.4 Databases

....................

31

3.4 Current Application Areas

.....

31

x

Contents

3.4.1 Cartography

..................

31

3.4.2 Statistics

.....................

32

3.4.3 Remote Sensing

...............

32

3.4.4 Archeological Reconstruction

...

32

3.4.5 Molecular Modeling

...........

33

3.4.6 Medical Science

...............

33

3.4.7 Oceanography

................

34

3.4.8 Computational Fluid

Dynamics.

34

Chapter 4 Facilities for Scientific Visualization

4.1

Visualization Software Categories

35

4.1.1

Graphics Libraries and

Presentation Packages

..........

35

4.1.2 Turnkey Visualization

Applications

..................

36

4.1.3 Application Builders

...........

37

4.1.4 Choosing a Package

...........

37

4.2 Software Costs

................

38

4.2.1

Subroutine Libraries and

Presentation Packages

..........

38

4.2.2 Turnkey Visualization Systems

..

39

4.2.3 Application Builders

...........

39

4.3

Hardware Considerations

(including Hardcopy)

..........

39

4.4 Vendor Systems Versus Public

Domain Systems

..............

40

4.5 Summary

....................

4~

Chapter 5 Outputting Results

5.1

Hardcopy.

. . . . . . . . . . . . . . . . . . .

44

5.2

Video.

. . . . . . . . . . . . . . . . . . . . . . .

45

5.3

Other

Media

.................

46

Chapter 6 Current Developments

and Activities

6.1

USA.........................

47

6.2

UK..........................

49

6.3

Europe.

. . . . . . . . . . . . . . . . . . . . . .

50

Contents

XI

Part II Overview

of

Current

Systems

and Developments

Chapter 7 Current Vendor Systems in Use

7.1

Wavefront Technologies, Inc.

...

.

53

7.2 UNIRAS

A.S.

................

58

7.3

Precision Visuals, Inc. . . . . . . . . . .

63

7.4 Stardent Computer, Inc.

...

. . . . .

67

7.5 Silicon Graphics, Inc.

..........

72

7.6 Sun Microsystems, Inc.

....

. . . . .

79

7.6.1

Sun Vision -Sun's Visualization

Software Package

..............

79

7.6.2 Sun Vision Programming

Interfaces

..

. . . . . . . . . . . . . . . . . . .

79

7.6.3 Sun Vision Window-based Tools

80

7.6.4 The VX and MVX -Sun's

Visualization Accelerators

......

82

7.7 Sterling Federal Systems, Inc.

...

88

7.7.1

FAST (Flow Analysis Software

Toolkit)

......................

88

7.8

Dynamic Graphics Ltd.

........

89

7.9

Spyglass,

Inc.

.................

96

7.9.1

Spyglass

Transform

............

96

7.9.2

Spyglass

Dicer

................

97

7.10

LightWork Design Ltd. . . . . . . . . .

98

7.11

Ricoh Company Ltd.

..........

100

7.12

Vital Images, Inc.

.............

105

Chapter 8 Current Public Domain Systems

in Use

8.1

Khoros

......................

106

8.1.1

Overview

....................

106

8.1.2 Subsystem Component

Descriptions

..................

107

8.1.3

Current Status of Khoros

......

112

8.2

apE: A Dataflow Toolkit

for Scientific Visualization . . . .

..

117

XII

Contents

8.3

National Center for Super-

computing Applications (NCSA)

127

8.4

GPLOT, DRAWCGM, P3D

(Pittsburgh Supercomputer

Center)

......................

128

8.4.1

The GPLOT CGM Interpreter

..

129

8.4.2 The DrawCGM Graphics

Subroutine Library

............

131

8.4.3

The P3D Three-Dimensional

Metafile Project

...............

133

8.4.4 Software Availability

...........

135

8.5

RAYSHADE.

. . . . . . . . . . . . . .

..

135

8.6

NASA Ames Software

.........

137

8.6.1

PLOT3D

.....................

137

8.6.2 SURF

.......................

137

8.6.3

Graphics Animation

System

(GAS)

.......................

138

8.6.4 Applications in Computational

Fluid Dynamics (CFD)

........

138

8.7

Irisplot

......................

140

8.8

ISVAS.

. . . . . . . . . . . .

..

. . . . . .

..

140

Chapter

9

Other

Uses

of

Visualization Tools

9.1

Art and Design

...............

142

9.2

The 5th Dimension Animation

System.......................

142

9.3

Multimedia Environments

146

Chapter

10

Conclusions

10.1

Strategic Importance of Scientific

Visualization

.................

149

10.2

Current Developments

.........

150

10.3

More User-Friendly Facilities

...

150

10.4

Further Information

...........

151

10.5

What to do

next?

.............

151

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

..

152

Sources

of

Figures

............................

155

Acknowledgements

Phil Andrews (Pittsburgh Supercomputer Center), Mike

Bundred (UNlRAS Ltd.), Tat-Seng Chua (National Univer-

sity of Singapore), D. Scott Dyer (Ohio Supercomputer

Center), Basem EI-Haddadeh (University of Leeds), Todd

Elvins (San Diego Supercomputer Center), Mark Goossens

(Silicon Graphics Ltd.), Chris Green (British Geological

Survey), Simon Hansford (Precision Visuals), Dee Holmes

(Stardent Computer Ltd.), Peter Irwin (Dynamic Graphics

Ltd.), Teruaki Ito (Ricoh Company Ltd.), Tosiyasu L. Kunii

(University of Tokyo), Hideko

S.

Kunii (Ricoh Company

Ltd.), Chris Little (Meteorological Office),

Donna

McMil-

lan (Sun Microsystems Inc.), Nadia Magnenat-Thalmann

(University of Geneva), Eihachiro Nakamae (Hiroshima

University), Gordon Oliver (LightWork Design Ltd.),

Ai-

dan

O'Neill

(Ricoh Company Ltd.), Peter Quarendon (IBM

UK

Scientific Centre),

John

Rasure (University of New

Mexico), David

F.

Rogers (US Naval Academy), Peter Stot-

hart (Wavefront Technologies Ltd.), Yasuhito Suenaga

(NTT

Human

Interface Laboratories), Daniel Thalmann

(Swiss

Federal Institute of Technology), Hiroshi Toriya (Ri-

coh Company Ltd.), Craig Upson (Silicon Graphics Inc.),

Joel Welling (Pittsburgh Supercomputer Center), Jane

Wheelwright (Dynamic Graphics Ltd.), Michael Wood

(University of Aberdeen), Brian Wyvill (University of Cal-

gary), Geoff Wyvill (University of Otago).

The contributions of members of the the

AGOCG

Workshop

on

Scientific Visualization held in the UK,

22-25

February 1991, are gratefully acknowledged.

We

par-

ticularly appreciated

the

comments of the following

on

a

first draft of this guide: Ken Brodlie, Lesley Carpenter, Kate

Crennell, Todd Elvins, Hilary Hearnshaw, Roger Hubbold,

Chris Little, Anne Mumford, Howard Watkins, and Mike

Wood. However, responsibility for the final text remains

with the authors.

XIII

Many people have

supplied informa-

tion on their uses

and

applications

of

visualization

systems. Many de-

signers

and

imple-

mentors have sup-

plied details

of

their systems

and

also illustrations.

Others have sup-

plied details

of

as-

pects

of

visualiza-

tion, as well as sli-

des.

We

express

our thanks

and

appreCiation to:

XIV

We

are indebted

to

the following for

the use

of

copy-

right material and

illustrations:

Disclaimer I

Copyright

Material

Some companies were unable to supply information

or

illustrations of their products, despite being invited to do

so.

They

have

therefore been omitted from the information

on

current vendor systems. The list

of

vendor systems

is

not

therefore claimed to cover

all

the systems in the market

place

at

the time of writing. Those that

are

covered are the

ones where information

was

obtainable.

Disclaimer

The

views

expressed by the contributors of information on

products

is

believed to

be

accurate and given in good faith.

However, authors and publisher do not hold themselves

re-

sponsible for the views expressed in this volume in connec-

tion with vendor products

or

public domain products. In

addition, the authors and publisher do not hold themselves

responsible for the accuracy

or

otherwise of data extracted

from vendor specifications.

Copyright

Material

Chris Little, UK Meterological Office (UK Government),

Peter Quarendon,

IBM

UK Scientific Centre, David

F.

Rog-

ers,

US

Naval Academy, Regional Geophysics Research

Group, British Geological

Survey,

Peter Stothart, Wavefront

Technologies Ltd., Mike Bundred, UNIRAS Ltd., Precision

Visuals Ltd., Stardent Computer Ltd., Silicon Graphics Inc.,

Spyglass

Inc., Ricoh Company Ltd.,

San

Diego Supercom-

puter Center, Donna McMillan, Sun Microsystems Inc.,

Gordon Oliver, LightWork Design Ltd., Nadia Magnenat-

Thalmann, University

of

Geneva, Daniel Thalmann,

Swiss

Federal Institute

of

Technology.

Trademarks

UNIX

is

a trademark of AT & T Inc.,

OPEN

LOOK

is

a

trademark of AT & T Inc., X Window

is

a trademark

of

Massachusetts Institute

of

Technology,

Xll

is

a trademark

of

Massachusetts Institute

of

Technology,

Motif

is

a trade-

mark

of

the

Open

Software Foundation Inc., PostScript

is

a registered trademark

of

Adobe Systems Inc., Ethernet

is

a

trademark

of

Xerox Corporation, MS-DOS

is

a trademark

of

Microsoft Corporation, Stardent

is

a trademark

of

Star-

dent

Computer

Inc.,

AVS

is

a trademark

of

Stardent Com-

puter Inc.,

DORE

is

a trademark

of

Stardent

Computer

Inc., Silicon Graphics

is

registered trademark

of

Silicon

Graphics Inc., IRIS

is

a registered trademark

of

Silicon

Graphics Inc.,

POWER

series

is

a trademark of Silicon

Graphics Inc., Graphics Library

is

a trademark

of

Silicon

Graphics Inc., Image Vision Library

is

a trademark

of

Sili-

con

Graphics Inc., IL

is

a trademark

of

Silicon Graphics

Inc.,

GL

is

a trademark of Silicon Graphics Inc., Live Video

Digitizer

is

a trademark of Silicon Graphics Inc., Stereo View

is

a trademark

of

Silicon Graphics Inc., Explorer

is

a trade-

mark

of Silicon Graphics Inc

..

, Personal Visualizer

is

a

trademark

of

Wavefront Technologies Inc., Data Visualizer

is

a trademark

of

Wavefront Technologies Inc., Advanced

Visualizer

is

a trademark of Wavefront Technologies Inc.,

Sun Vision

is

a trademark of Sun Microsystems Inc., Sun-

View

is

a registered trademark

of

Sun Microsystems, Open-

Windows

is

a trademark

of

Sun Microsystems Inc.,

XDR

is

a trademark

of

Sun Microsystems Inc., XGL

is

a trademark

of

Sun Microsystems Inc., SPARC

is

a registered trademark

of SPARC International Inc., SPARCstation

is

a trademark

of SPARC International Inc., IBM,

PC,

PS/2 are trademarks

of

IBM Corporation,

PV-WAVE

is

a trademark

of

Precision

Visuals Inc.,

NAG

is

a registered trademark

of

Numerical

Algorithms Group Ltd. and, Numerical Algorithms Group,

Inc., RenderMan

is

a registered trademark

of

PIXAR, RIB

is

a trademark

of

Pixar, Spyglass

is

a trademark

of

Spyglass

Inc., MacIntosh

is

a trademark of Apple

Computer

Inc.,

xv

This

is an aggre-

gated list

of

regis-

tered trademarks

and trademarks

used

in

the

vol-

ume.

In order

to

identify products

unambiguously

it

is necessary

to

use these terms.

The

following are

known

to

be trade-

marks

or

regis-

tered trademarks

of

the companies

concerned.

We

trust that others

that may not be

noted are known

to

readers and are

referenced

in

a

manner accept-

able

to

the com-

panies concerned.

XVI

Trademarks

Laserwriter

is

a trademark of Apple Computer Inc.,

LightWorks

is

a trademark of LightWork Design Ltd.,

is

a trademark of

is

a trademark of Spatial Systems Inc., DataGlove

is

a trade-

mark of VPL, 3D Polhemus Digitizer

is

a trademark of

Polhemus, EyePhone

is

a trademark of VPL Research Inc.,

VoxelView

is

a trademark of Vital Images Inc., VoxelLab

is

a trademark of Vital Images Inc.

Part I

Basics of

Scientific

Visualization

Chapter

1

Introduction

and

Background

1.1

Introduction

This guide seeks to answer the following questions:

• What

is

scientific visualization?

• What can it

do?

• What do the technical terms and the jargon really mean?

• What products are currently available?

• What kind

of

hardware do I need?

• What are the costs?

• What do I

get?

• Where do I

go

next to find out more,

or

to explore cur-

rent possibilities?

• What are the prospects for the future?

The first part of this volume

is

concerned with introducing

the topic, definitions, terminology, techniques, methodolo-

gy,

and equipment. The second part contains an overview

of current systems and developments.

1.2

Background

The area encompassed by scientific visualization

is

defined

Information

in this guide, with the range of possible applications, and

the potential for the future. Considerable advances have

been made in the USA by dissemination of information and

by coordinated initiatives from industry and professional

organisations such

as

the Association for Computing

Ma-

chinery (ACM).

4 Introduction and Background

Objectives

The

objective

of

this guide

is

to inform the general reader

about scientific visualization, what it offers, and what it can

do.

It

should be useful to scientists and engineers

who

are

not

specialists in computing matters, but nonetheless wish

to use effective computer-based tools to further research ob-

jectives.

Developments and initiatives in the USA

are

summa-

rized. These demonstrate the relevance and importance

of

scientific visualization.

Facilities Software products are outlined and summarised -for

purposes

of

general information. These are indicative

of

the

kind

of

products available in the market place, and that are

supported

on

a variety of platforms. However, this

is

not

in-

tended to

be

exhaustive, and in certain application areas a

wide variety of software has been developed.

Current developments in animation are summarized be-

cause it

is

likely to become increasingly important for scien-

tific visualization.

Summary This guide provides an overall summary

of

the benefits

that accrue from scientific visualization and the methods,

tools and strategies that comprise its domain.

Chapter

2

What

Scientific

Visualization

Can

Do!

2.1

What

is

Scientific

Visualization?

"The purpose

of

computing

is

insight,

not

numbers" wrote

the much-cited Richard Hamming in Numerical

Methods

for

Scientists and

Engineers

(McGraw-Hill, 1962). Scientific

vi-

sualization

is

an amalgam

of

tools and techniques that seeks

to promote new dimensions

of

insight into problem-solving

using current technology.

Scientific visualization

is

concerned with exploring data

and information graphically -

as

a means

of

gaining under-

standing and insight into the data. Scientific visualization

is

a graphical process analogous to numerical analysis, and

is

often referred

to

as

visual data analysis. Scientific visualiza-

tion

systems are combinations

of

hardware and software

sys-

tems and techniques.

By displaying multi-dimensional data in an easily under-

standable form

on

a 2D screen, it enables insights into 3D

and higher-dimensional data and data sets that were

not

for-

merly possible.

Often data sets are very large, and this

gives

rise to prob-

lems

of

scale and

of

finding correlations and relationships

between different parts

of

the data.

Visualization

is

also a means

of

gaining a quick under-

standing

of

processes. This could

be

done in more classical

ways, but might take much longer.

The difference between scientific visualization and pre-

sentation graphics

is

that the latter

is

primarily concerned

with

the communication

of

information and results that

are

already understood. In scientific visualization

we

are seek-

ing to understand the data.

Insight not

numbers

What is it?

Multidimensional

Large volumes

Speed

Not presentation

graphics

6 What Scientific Visualization

Can

Do!

Lots

of

tools Visualization involves aspects in the

areas

of computer

graphics, user interface, cognitive science, image processing,

design, and signal processing. Formerly these were indepen-

dent fields, but convergence

is

being brought about by the

use of analogous techniques in the different

areas.

Visualiza-

tion

is

thus an additional tool for scientific research and in-

vestigation.

Help for Visualization highlights applications and application ar-

applications

eas

because it

is

concerned with providing the means for a

user to achieve greater exploitation of computing tools now

available. In a number of instances visualization has been

used to analyze and display large volumes of multi-dimen-

sional data in such a way

as

to allow the user to extract

sig-

nificant features and results quickly and

easily.

Tools and

techniques in this area are therefore concerned with data

analysis and data display, perhaps with provision for the

dis-

play of data changes with respect

to

time.

Examples Non-destructive and non-invasive examination of the in-

ternal structures of living organisms (e.g., reconstructions

from brain scan data), turbulence effects in fluid

flow,

and

genetic engineering are

all

examples that have caught the

public attention, and where scientific visualization has

brought substantial benefits. However, this

is

but one aspect

of the whole field,

as

indicated above.

Simulating nature Visualization fits into the overall process of numerical

simulation

as

indicated in Figure

2.1

below.

In the computational sciences the main goal

is

to under-

stand the workings of nature.

In

order to accomplish this,

the scientist proceeds through a number of steps from ob-

serving a natural event

or

phenomena to analyzing the

re-

sults of a simulation of the phenomena. Visual representa-

tion of this data

is

often indispensible in gaining an under-

standing of the processes involved.

Interactive steering Visualization systems can be used for the interactive

steering of computations. The user observes the progress of

the computation visually and alters parameter values accord-

ingly. These in

turn

determine the future computation.

How

to

do Scientific Visualization

~I--_r-l-

-,

,

---

Observations

~_~---

Physical

Laws

l""".=~==""

'"

Mathematica l

Model

Approx. to

Mathematical

Simulation

Specification

Simulation

SOlution

Images

Mathematical

Formulation

of

Laws

2.2

How

to

do

Scientific

Visualization

7

Fig,

2,1

Simulating Nature,

Simulating natural

phenomena: the

boxes represent

processes, the

circle's size

indicates the

relative volume

of

information

passing between

each

pair

of

processes

Visualization tools benefit from the availability of modern Computing power

workstations with good performance, large amounts of

memory and disk, and with powerful graphics facilities -

in terms of range of colors available, resolution, and speed

of

display by the workstation. This close coupling of graph-

ics

and raw computation power

is

a powerful combination

for those

areas

where visual insight

is

an

important part of

the problem-solving capability.

Such workstations now offer substantial computation 3 dimensions

power coupled with high-speed 3D graphics. These facilities

can

be

exploited to significant advantage in application

areas

such

as

modeling, simulation, and animation. Real-time dy-

namical simulation can involve the processing and display

of large amounts of data, and often the only effective analy-

sis

of the performance

or

validity of the model

is

through

visual observation.

Leading-edge applications will tend to require the most

powerful systems available. The vendors listed later in this

8 What Scientific Visualization Can

Do!

volume provide a range

of

systems to match a wide variety

of

applications, and are continuously improving computa-

tion power and graphics capabilities.

Supercomputers Such workstations provide the computation power to

process the data, and the high-speed graphics pipeline can

transform this into graphical images, often in real time. In

those

cases

where additional computational resources are

re-

quired, the calculation can be off-loaded onto a supercom-

puter,

or

other advanced workstations

with

spare capacity,

and the resulting image down-loaded for viewing (and per-

haps even interaction)

when

it

is

ready.

Output

is

often most useful

as

information

on

the work-

station screen, especially when the process

is

interactive.

More permanent copies

of

screen images can be printed in

full color

on

A4

to

AO

plotters and printers and

on

video

or

slides. However, this requires additional hardware and

software. Generally, the higher the cost the greater the vari-

ety

of

colors and quality

of

the final images.

2.3

Some Examples of Scientific

Visualization

2.3.1

The March of Napoleon's Army

Fig. 2.2 March

of

Napo/eon's Army

----_.'.-"

, -

,--

,

Some Examples

of

Scientific Visualization

The

classic

map/chart

of

Napoleon's march in Russia, and

the

retreat

of

1812, drawn

by

Charles Joseph Minard.

This

is

a good example

of

visualization

which

clearly

pre-dates scientific visualization!

2.3.2 Cholera Outbreak

The

work

of

Dr.

John

Snow (Gilbert 1958) provides an ear-

ly example

of

cartographic visualization

in

problem analy-

sis. While investigating

the

1853-54

cholera outbreak

in

London

he identified

what

he called a "cholera-field" in

the

Soho area.

He

had plotted

the

homes

of

the

500 victims

who

had died in

the

first

10

days

of

September 1854 and this

simple visualization (of quite a large and complex data set)

drew his attention to

the

previously unsuspected

link

be-

tween water supply and

the

disease. All victims had

drunk

from the Broad Street pump, in the middle

of

the

"field",

Fig. 2.3

Dr.

John Snow's

map (1855)

of

deaths from

cholera in the

Broad Street area

of

London in

September 1854

9

10

What Scientific Visualization

Can

Do!

which, it

was

later established, was being polluted by a leak-

ing cesspool. The "link"

was

confirmed by noting that a vir-

tually disease-free area

(a

large workhouse) within this zone

had its own clean water supply.

2.3.3

Weather

Maps

from

Meteorology

The

following three maps illustrate various aspects

of

weath-

er patterns.

This first map (Figure 2.4) shows the raw numerical fore-

cast data with contour lines showing the pressure. This

is

a

distillation

of

the information contained in thousands

of

numbers!

The

second map (Figure

2.5)

shows where the

fronts have been positioned. This

is

an interpretation

of

the

above map, and presents the data in a form that people can

understand more easily. The third map (Figure 2.6) shows

Fig. 2.4 a tailored short-hand form

of

the second map, and

is

the

Weather

Map

1 kind used by aircraft pilots.

12

Fig.

2.7

Molecular Model

What Scientific Visualization

Can

Do!

2.3.4

Molecular Modeling

This example shows a molecular model

of

liver alcohol

de-

hydrogenase -calculated by computer and

then

displayed.

The

mauve molecule fitting into the enzyme shows the

structure of the underlying molecule.

2.3.5

Pelvic Reconstruction

Karen Frankel A young man in his late twenties suffered a crushed pelvis

(1989) reported the in an auto accident. His orthopedists said that the fracture

following case

was

too complicated to operate

on

and elected to treat him

conservatively; he would be in traction for a

few

months.

The

doctors were certain that the young man would

be

per-

manently crippled.

Luckily the man's father, also a physician, knew

of

re-

search in 3D rendering of computed tomography (CT) scan

data.

He

sent his son's

CT

scan studies to the researchers,

a radiologist, and orthopaedic surgeon, and a computer

graphics expert,

who

studied the volumetric rendering

of

the pelvis that

was

created with specially designed hardware

Some Examples

of

Scientific Visualization

and software. Able

to

see

it from all angles, they determined

the extent

of

the fracture and locations of several key

frag-

ments.

The

pelvis was operable and the next day the sur-

geons set the fragments. Three

months

later the patient

re-

turned

for a check up and demonstrated full-range hip mo-

non.

13

Fig. 2.8

Fractured Pelvis.

Volume

renderings

of

the broken

pelvis using

CT

scan data by

Professor Elliot

K.

Fishman

of

Johns Hopkins

University Hospi-

tal, Dept

of

Radiology.

The

extent

of

the frac-

ture and location

of

the fragments

are clearly visible.

Although

radiologists have

been using

CT

data for almost

20 years, volume

renderings

of

CT

offer a new way

of

interpreting such

data

14 What Scientific Visualization Can

Do!

This

case

coupled great medicine and great computer

sci-

ence. The technique of volume rendering changed the

course of treatment by providing the physicians with more

data. This data ultimately

gave

them the confidence to oper-

ate

and thereby improve the patient's quality of

life.

While

volume rendering helped manage the medical complexities,

this

case

also represents departures from tradition for both

disciplines.

Text:

courtesy

of

ACM

2.3.6

Oil

Exploration

This example shows the

use

of

visualization in oil explora-

tion. The volumetric data

was

produced

as

part of a simula-

tion of a method for recovering oil from the tar sands of

northern Alberta, Canada.

This process

was

simulated by the Alberta Oil Sands

Technology Research Authority and the visualization

was

computed by Geoff Wyvill and Brian Wyvill.

Courtesy

of

Geoff Wyvill and Brian Wyvill.

Some Examples of Scientific Visualization

15

Fig.

2.9

Oil Explora-

tion

1.

Iso-temperature

contour surfaces

from Volumetric

Data produced by

simulation.

The

surfaces are tiled

using the "Soft

Object" algorithm.

The

red pipe

represents the

"injection well"

which pumps

superheated

steam into the

rocks

Fig.

2.10

Oil Explora-

tion

2.

As the simulation

progresses the

surface changes

shape as the

rocks are heated.

After a period

of

time,

oil is

precipitated into

the production

pipe

16

What Scientific Visualization

Can

Do!

2.3.7 Designing Ship Propellors

Sculptured Surface Fitting and Fairing

Shape design

The

fairness of a sculptured surface

is

important for design

applications

as

diverse

as

modern artistic sculptures and air-

craft

or

automobiles.

The

requirement for fairness can be

based

on

either aesthetic

or

technical considerations. Cur-

rently there

is

no universally accepted mathematical defini-

tion of fairness.

One

technique that aids in evaluating sur-

face

fairness

is

to look at the Gaussian curvature

of

the sur-

Visualizing the

face.

The Gaussian curvature

of

a surface

is

visualized by

us-

surface ing the values

of

Gaussian curvature to color encode the sur-

face.

If

the surface

is

fair, then the color hue smoothly tran-

sitions across the surface. U nfairnesses show up

as

splotches

or

lines

of

color within the surface.

The

three accompanying images show one side

of

the

surface and the fillet for a single blade of a ship propeller.

The

data supplied by the designer

is

shown in Fig. 2.11 visu-

alized with a dynamic three-dimensional rational B-spline

surface design program called Rbssd developed by Professor

David

F.

Rogers.

The data set

is

a combination of two independent data

sets comprising the surface and the fillet. Visualized in this

way it

is

obvious that the lines

of

data for the fillet and the

surface

are

not

aligned. This has implications

when

a ratio-

nal B-spline surface

is

fit to the data.

Figure 2.12 shows the rational B-spline surface generated

by the defining polygon net. The surface appears reasonably

smooth.

Figure 2.13 illustrates the color coded Gaussian curvature

surface. This shows that the surface

is

unfair. Here the green

indicates zero Gaussian curvature and the yellow positive

Gaussian curvature.

The

areas

of

yellow indicate ridges

or

bumps

or

hollows in the otherwise developable surface. No-

tice that many

of

these

are

concentrated in the area where

the fillet and the main surface data were joined. Visualiza-

tion

makes it obvious.

Contributed

by

Professor

David E

Rogers,

US. Naval Academy.

Some Examples of Scientific Visualization 17

Fig. 2.

11

Designer's Data

Visualized wi

th

Rbssd

Fig. 2.

12

Rational a-spline

Surface

Fig. 2.13

Color Coded

Gaussian

Curvature Surface

18

Growth

of

forest

Choice

of

forest

to

model

Physical properties

What Scientific Visualization

Can

Do!

Supporting

References

Rogers,

D.

F.

and Adlum,

L.:

Dynamic Rational B-spline

Surfaces. Computer Aided Design Journal, invited paper

in the commemorative

issue

honoring Pierre Bezier on

his 80th birthday, Computer Aided Design Journal,

Vo1.22,

pp.609-616,

1990.

Dill, ].c., and Rogers,

D.F.:

Color Graphics and Ship Hull

Surface Curvature. Proceedings of International Confer-

ence on Computer Application in the Automation of

Shipyard Operation and Ship Design

IV

(ICCAS '82),

7-10

June

1982,

Annapolis, Maryland, pp.197-205,

North-Holland.

2.3.8

Visualization

of

Forest

Growth

An

interactive tree model, FOREST,

has

been developed

at

the University of Tokyo to enable the processes of forest for-

mation to

be

visualized. The model includes a parallel

algo-

rithm of individual tree growth which considers both the

differences between the species and also the time-dependent

interactions among the trees through mutual shading.

A tropical rain forest in the equatorial zone

is

chosen

as

a typical

case

because other types of forest can

be

derived

from it by imposing a set of constraints, such

as

diminished

rain

fall

and lower tempratures, which slow down the speed

of growth processes. The results of the visualization of

algo-

rithmically animating a

few

hundred years

of

forest growth

processes using this model

have

been validated against the

data obtained in experimental observations in Pasoh on the

Malaysian peninsula.

The model of individual trees considers the internal

properties of trees such

as

the rate of the light/photosynthe-

sis

relationship, the death rate of the branches, and the pro-

portion of the foliage active in photosynthesis.

Figures

2.14

and

2.15

show two frames from a sequence

which show the entire

life

history of

the

forest. The first

shows the initial growth period

(1-60

years)

and the second

shows the higher layer formation processes starting

at

age

60.

Understanding a typical rain forest and its ecosystem

is

thus expected to lead to understanding

of

other types

of

for-

ests and their ecosystems.

Fig. 2.

14

Forest Visualiza-

tion.

Age 60 years

Fig. 2.15

Forest Visualiza-

tion.

Age 250 years

Such models and their visualization can be used to in- Growth

crease

our

understanding of the growth processes in nature processes

and the way these processes can

be

affected by apparently in nature

minor

disturbances in the environment. For example, fores-

tation affects the percentage

of

carbon dioxide in the atmo-

sphere, which in

turn

affects global warming, which in

turn

affects the total area

of

deserts. Understanding these effects

and the relationship between the variables

is

the key to un- Global modeling

derstanding how to influence the future of the planet. Visu-

alization can

playa

significant role in furthering this under-

standing.

In/ormation

supplied

by

Professor

T.

L.

Kunii,

University

0/

Tokyo.

Fig.

3.1

The

visualization

Mapping Space:

the mapping from

the computational

domain into the

visualization

domain

Chapter

3

Explanation

of

Scientific

Visualization

Terminology

3.1

Techniques

The schema in Fig.

3.1

outlines some of the current repre-

sentation techniques.

3D

<:

•

S!

11

<:

..

~

2D

....

~

..

<5

"5

.~

1D

OJ

<:

.S!

;g

..

. 5

~

OD

. , '

,

.......

~

....................................................

!

.............

"

......................

VolulII~

ReDderiDi

: : Solids Modellog

: :

C~11es

.....

1.

.................

HoI

....

ioldo

......

Tlied

Surr

...

.

: Pseudo Color Stacked Textures

: Contour Maps Rlbboas

~

Images

.•

Llnel

CUr:

flS

1D

........................

ContourM..,.

Field

Vectors

Space

C,unes

............................ Scatter Plots

Partldes

DotSurfKeI

2D

3DVKtor

Nets

..

.. .. .. .. .. .. .. .. ..

"Field Vectors

HedgeHop

Rlbbo

...

Scatter Plots

Particle Tracers

Dot

Sur:faces

3D

Dimensionality

of

the Computational Domain

lcoa,

Attribute

Maprog

................ AUribute

M.,pplog

....................

~

..............

.

nD

The

ongm

of several visualization techniques can be

traced back to line-based two-dimensional contour maps.

These extensions result in higher-dimensional

or

more con-

tinuous representations

as

shown in Fig. 3.2.

This chapter outlines techniques that can be used by a

scientist confronted with a

vast

amount of data generated by

computer models, remote sensing devices, and automated

re-

cording equipment.

'"

=

0

=

I:

--g

0

U

.!l

~

3D Cell

Rendering

3DVoxel )

Rendering

\

~

2.50

Continuous Tone

/'

Contour Maps

2D Continuous Tone

7"''''''''

2D Contour Maps

3D Polygonal

Surface Tilings

/

3D

Vector Nets

~

Q~------r------------------------r-------------

2D 3D

Dimensionality of the Domain

On

the following pages are some examples

of

the meth-

ods that are chosen to represent quantities in the visualiza-

tion domain.

Visual

Picture

This picture shows a drug molecule reacting

with

a large en-

zyme and illustrates the underlying protein structure.

Fig. 3.2

Three-dimensional

extensions

of

contouring

Representation

Method

Fig 3.3

Dot Surface.

Quick method for

small objects.

Time-consuming

and

counter-

productive for

larger objects

22

3D wire frame

representing the

shape

of

a

surface.

Fig.

3.4

Vector Net

Fig. 3.5

Polygonal Surface

Surface

represented

by

polygons

Explanation of Scientific Visualization Terminology

Perspective view

of

a calculated gravity field represented

as

a 3D smooth shaded polygonal surface, with lighting and

shading effects generated from an implementation of the

PHIGS PLUS model.

Techniques

This picture shows part

of

the

North

Atlantic where the

gulf stream

is

flowing.

The

sections show temperature at

three different depths

with

plumes

of

water demonstrating

mixing taking place in the vertical plane.

This shows airflow round a wing.

Color

shows one pa-

rameter; x, y shows direction; and twisting shows vorticity.

Fig.

3.6

Stacked

Contour

Map

.

Overlaying

of

20

cross·sections

to

represent

3D

volumes

Fig. 3.7

Ribbons

and

Streamers

23

24

Fig. 3.8

Hedgehogs

Fig.

3.9

Shaded

Contours

Explanation of Scientific Visualization Terminology

A method for showing a direction relative to the surface

(hence the term hedgehog spine!).

We

can of course show a

third variable (by the color

of

the vector).

This picture shows the

use

of

color to identify

areas

in

the plot between upper and lower threshold values specified

by the boundaries of the

areas.

Volume Visualization

Overlaying an additional data set

on

an existing 3D one.

Height represents axial velocity; shade represents radial

ve-

locity. As a turbine pushes fluid through an opening, scien-

tists can observe the density of a particular slice -shown

as

shading, with red being the most dense.

3.2 Volume Visualization

Volume rendering

is

used to view 3D data without the usual

intermediate step of deriving a geometric representation

which

is

then rendered. The volume representation

uses

vo-

xels (volume elements) to determine visual properties, such

as

opacity, color, and shading at each point in the computa-

tional domain. Several images are created by slicing the vol-

ume perpendicular to the viewing axis at a regular interval

and compositing together the contributing images from

back to front, thus summing voxel opacities and colors at

each pixel.

By

rapidly changing the color and opacity trans-

fer functions, various structures are interactively revealed in

the spatial domain.

25

Fig.

3.10

Attribute Mapping

Representation

Method

Voxe/s

not

geometry

26

Explanation of Scientific Visualization Terminology

Applications A

number

of

projects

in

the

USA have demonstrated

the

Fig. 3.

11

Rendered

Isosurfaces from

Slice Contours a

"nerve cell"

Fig. 3.

12

Cell Rendered

Volumetric Image

benefits to medical and surgical planning from these new

techniques.

Further

information may be found in Frenkel

(1989), Kaufman (1990), and

Upson

(1991).

Data Types

Figure 3.11 shows paired helical filaments (orange)

PHF

cracking a cell nucleus (blue). This

is

used in the study of

Altzheimer's

disease.

The digitized

slices

are

hand con-

toured; MOVIE· BYU mosaic connects the contours. The

"Marching Cubes" algorithm

is

used for PHF.

27

Fig. 3.13

Volume

Rendering

of

CT

Data

"Dolphin Head"

91

slices

3.3

Data

Types

3.3.1

Overview of Facilities

As

scientific visualization

is

often concerned with large Handling data

amounts of data, it

is

inevitable that methods for organizing

it, transferring it, manipulating it, and storing it

are

of great

importance. There

are

also

a wide variety of data formats

and utilities for translating between them.

Application data

is

concerned with information at the Application

application

level.

Data formats

are

often developed in

associ-

aspects

ation with particular application

areas.

Examples of such

formats

are

Hierarchical Data Format (HDF) developed by

the National Center for Supercomputing Applications

at

the University of Illinois, and Network Common Data

Form (netCDF) developed

at

NSSDC and NASA. Further

28

Explanation of Scientific Visualization Terminology

information

on

these formats

is

contained in Sections 3.3.2

and 3.3.3.

Graphics formats Graphics data

is

comprised of information output by the

graphics system (e.g., vectors, polygons) which

is

then con-

verted into appropriate image information. Examples of

graphics formats are Computer Graphics Metafile (CGM)

and Postscript.

Image data Image data

is

the information corresponding to the im-

age

on

the graphics display screen. For a display with

1000

by

1000

points

on

the screen

we

would need 1 million bits

to store the information, just for a simple black/white

dis-

play. With a wide range of colors this amount of informa-

tion increases, since

we

need to store a value representing the

color for each pixel. There

are

a number of formats for rep-

resenting image data, including GPF (Graphical Pixmap

Format), TIFF (Tagged Image File Format), Group 3 and

Group 4 Fax, VIFF (Visualization Image File Format for

the Khoros visualization software),

PICT

(MacIntosh for-

mat), and

PCX

(IBM format).

Networking In order

to

be able to transfer image data effectively (par-

implications ticularly over networks) it

is

important

to

reduce the size

of the image data set

to

manageable proportions without

losing essential information. The Joint Photographic Ex-

perts Group GPEG) and the Moving Picture Experts Group

(MPEG)

are

formulating proposals for standards for single

images and multiple frames respectively.

JPEG

and

MPEG

are proposals for standards in this area. In addition, fractal

compression techniques are being used with considerable

success.

Remote sensing Remote sensed image data contains real information

which may be extracted by image processing techniques.

This

is

a well-establised field.

Multi-media Multi-media systems combine software with facilities for

sound, images, graphics, video, and animation to create pow-

erful communication tools. Interest in the area

is

due to the

ability to incorporate data from many sources, and the bene-

fits arising from this. Multi-media products are widely

avail-

Data

Types

29

able

on

personal computers and are moving into the

UNIX

workstation environment.

In view of the wide variety of data formats currently in Choice

of

system

use, potential purchasers of scientific visualization systems

are advised to check that a given system will handle the data

formats required, and also have sufficient capability to han-

dle the volumes of data required.

3.3.2 HDF

The Hierarchical Data Format (HDF)

was

developed by the

National Centre for Supercomputing Application (NCSA)

and

is

available via anonymous ftp.

Hierarchical Data Format (HDF)

is

a multi-object

file

File format

format for the transfer of graphical and floating-point data

between different hardware platforms.

FORTRAN

and C

calling interfaces for storing and retrieving 8-bit and 24-bit

raster images, palettes (color tables), scientific data and

ac-

companying annotations have been developed.

HDF

allows

for the self-definition of data content and aims

to

be

extensi- Extensible

ble, thereby allowing for the inclusion of future enhance-

ments

or

compatibility with other standard formats.

HDF

provides a general purpose

file

structure that en- Facilities

compasses the following:

-makes it possible for the programs

to

obtain information

about the data directly from the

file,

rather than from

an-

other source

(e.g.

look-up table),

-enables the storage of arbitrary mixtures of data and relat-

ed information in different

files,

even when the

files

are

processed by the same application program,

standardizes the formats and descriptions of many types

of commonly used datasets, such

as

raster images and

sci-

entific data,

encourages the use of a common data format by all

ma-

chines and programs that produce

files

containing a spe-

cific dataset,

can be adapted to accommodate virtually any kind of

da-

ta

by defining new tags

or

a new combination of tags.

30

Explanation of Scientific Visualization Terminology

HDF

currently supports sharing data across machines and

systems such

as

CRAY (UNICOS), Silicon Graphics

(UNIX), Alliant (CONCENTRIX), Sun (UNIX),

VAX

(UNIX), Macintosh (MacOS), and IBM

PC

(MS-DOS).

3.3.3

NetCDF

National The Network

Common

Data Form (netCDF)

was

devel-

cooperation oped

as

part of U nidata - a U.

S.

national effort sponsered

by the Division of Atmospheric Sciences of

NSF.

The initia-

tive

is

managed by the University Corporation for Atmo-

spheric Research. The software

is

available via anonymous

ftp.

Storage

and

NetCDF

is

a data abstraction for the storing and retrieval

retrieval

of

data of scientific data, in particular multi-dimensional data.

NetCDF

is

a distributed, machine-independant software

li-

brary based upon this data abstraction which allows the

cre-

ation,

access

and sharing of data in a form that

is

self-de-

scribing and network-transparent. Both C and

FORTRAN

interfaces

are

supported.

Architecture

NetCDF

software utilizes the concept of an abstract data

type, which means that all operations to

access

and manipu-

late data in a netCDF

file

must

be

via a defined set of func-

tions provided by the C library interface.

As

the actual rep-

resentation of the data

is

hidden from the application, inter-

nal data representations can

be

changed without affecting

the program.

Network

To

achieve network transparency, netCDF

is

implement-

transparency ed

on

top of a layer of software for external data representa-

tion known

as

XDR.

XDR

is

a nonproprietary standard for

describing and encoding data developed by Sun Micro-

systems, Inc.

Availability The netCDF software provides common C and FOR-

TRAN

interfaces for applications and data. The C interface

library

is

available for many common computing platforms,

including UNIX,

VMS,

MSDOS, and MacOS environ-

ments. The

FORTRAN

interface

is

available

on

a smaller

set of environments (due to the lack of a standard for calling

C from FORTRAN).

Current Application Areas

XDR

has been implemented

on

a variety of platforms,

including SUNs,

VAXs,

Apple Macintoshes, IBM-PCs, IBM

mainframes, and

CRAYs.

3.3.4 Databases

The currently accepted storage method for most scientific Storage

of

data

is

the Relational Database Management System. Many

commercial examples are available (e.g., Oracle, Ingres). Da-

ta can be extracted using Standard

Query

Language (SQL)

based commands. Some scientific visualization systems have

these command interfaces built in (e.g., UNIRAS).

3.4

Current

Application

Areas

This section provides an overview of application areas

where visualization techniques

are

being used

on

input data Input data

from the real world, processed data, and computer-generated

data.

3.4.1

Cartography

is

rapidly moving from a discipline concerned

with the presentation of data (the map) to one concerned

with the storage and analysis of spatial data via Geographic

Information Systems (GIS). These systems may be used to

to store large amounts of information in a database and

al-

low retrieval and display based

on

user-specified criteria.

The visualization

is

used to select spatial features based

on

their attributes,

or

to observe topological relationships with

other features. For example, the user may wish to make

re-

quests such

as

the the following:

• Show me all the regions where forests are adjacent to

lakes and which have

access

by road.

• Show me all the principal roads which have houses with-

in

50

meters.

Analysis

of

spatial data

Interrogating

the data

31

32

Explanation of Scientific Visualization Terminology

• Display all the houses which have

not

had their

gas

ser-

vice supplies and electricity supplies renewed in the last

30 years.

Data relationships Obviously, there

is

more information stored in the database

than just the terrain. In particular, it illustrates how useful

it can be to have other information to do with the same ter-

ritory available so that it can be interrogated and overlaid

on

the terrain map.

Project planning The visualization can then be used

to

plan for work to

be done (e.g., by the service industries) in such a way

as

to

minimize costs. Equipment can be moved

to

an appropriate

point in the area and used to supply all the requirements for

the work to

be

done.

3.4.2

Statistics

Visual representation of statistical data

is

very useful for

providing insight and understanding into the data.

3.4.3

Remote Sensing

Satellite and other imaging

devices

are

producing

large

amounts of data. Many two-dimensional processing methods

exist for analyzing this data. New methods

are

being

devel-

Increasing oped to allow for increasing the dimensionality of the data

dimensionality

as

the number of frequency bands increases. Visualization

methods allow horizontal (two dimensions of

space

at

a

giv-

en frequency value) and vertical (one dimension of

space

and

one of frequency) sections through the data.

An

example of

the kind of picture produced

is

shown in Fig. 7.7.

3.4.4

Archeological Reconstruction

Rebuilding history Data from archeological excavations has been entered into

visualization systems to enable partial

or

major reconstruc-

tion to be done, and the resulting constructs to be viewed

interactively

on

a computer display screen. This enables the

archeologist to build up a picture of the original buildings,

objects, and their relationships.

Current Application Areas

3.4.5 Molecular Modeling

Chemists and biologists have been using physical models

of

Physical models

molecules for many years to enable the relationships

be-

tween the various components to be understood. This

is

33

now more easily and and effectively done by using a com- Computer models

puter-based model and interacting

with

it

on

a graphics dis-

play screen.

An

example

of

such a molecule

is

shown in

Fig. 2.7. Such molecules can be rotated and viewed from var- Studying the

ious angles,

as

well

as

providing relevant quantitative infor- molecule

mation (e.g., potential energies,

bond

distances, etc.). Analy-

zing X-ray diffraction maps

is

greatly facilitated by visualiza- Refining the model

tion

methods.

The

atomic positions in protein structures

can be adjusted by interaction until they best fit a given elec-

tron

density map.

In

the design

of

new drugs, existing mole- Design

of

cules can be modified by introducing new molecules into new drugs

the overall structure. Such systems often provide a stereo

view capability, where by means

of

special glasses the user

is

able to view the molecule

on

the screen in full 3D. This

can provide further understanding and insight into the over-

all structure.

3.4.6 Medical Science

Historically, radiologists have looked at a series

of

two-di-

mensional cross-sections and built up mental pictures

of

three dimensional structures. However, these are subjective

and can vary from one radiologist to another.

In

many

cases

more detailed and accurate three-dimensional information

can be very useful

when

planning surgical procedures for

complex

and/or

intricate structures,

or

in radiation treat-

ment

planning. Volume visualization techniques are increas-

ingly being used to provide three-dimensional information

from a series

of

two-dimensional slices. A

case

of

pelvic

re-

construction

is

outlined in Section 2.3.5, and volume meth-

ods are shown in Fig. 3.11-3.13.

From 2D to 3D

Accuracy in

planning

procedures

Visualizing

volumes

34

Explanation of Scientific Visualization Terminology

3.4.7

Oceanography

Large and Modeling the behavior of oceans

is

increasingly being done

complex natural using visualization techniques. Often there

are

a large num-

systems ber of variables involved,

such.

as

temperature, salinity,

depth, vorticity,

etc.

Representations are chosen to enable

Simulating this multi-dimensional data to

be

viewed on a display

ocean behavior screen. Internal structures can

be

shown and simulations

performed. This leads to an increase in overall understand-

ing of ocean behavior. Areas such

as

these are becoming in-

creasingly important

as

attention shifts from interplanetary

Protection

of

the investigations to earth-environment matters such

as

global

earth's resources warming and the ozone layer. Examples in the area of ocean-

ography

are

shown in Figs. 3.6 and 7.21.

3.4.8

Computational Fluid Dynamics (CFD)

Air flow over Visualization

is

being used

to

analyse complex flow systems.

aircraft wings Numerical simulations produce values of velocity, tempera-

ture, pressure, vorticity, and even tensor fields. A variety

of

techniques exist for displaying this data. Particle tracers

are

Fluid flow used in real-time to show aspects

of

the

flow.

Interactive

CFD

is

important for tracking and steering solutions.

Fig-

ures 3.7 and 7.21 show examples.

For further information on application

areas,

readers

are

referred to the more detailed reference work Scientific

Visu-

alization -

Techniques

and

Applications edited by

K.

W

Brodlie, L.A. Carpenter, R.A. Earnshaw, J.R. Gallop, R.J.

Hubbold, A.

M.

Mumford,

C.

D.

Osland,

P.

Quarendon,

also published by Springer-Verlag,

1992.

This volume also

has

more detailed reference information

on

data formats.

Chapter

4

Facilities

for

Scientific

Visualization

4.1

Visualization

Software

Categories

Visualization software has evolved over a period of time and Evolution

of

three distinct categories can

be

identified which appeared in facilities

succession. In general the older the category the

less

the

power, memory and storage required to

run

them. This

makes the software in the first category suitable for use in

PC

or

terminal-mainframe environments, and the most

re-

cent developments only suitable for the most modern

supercomputers

or

supercomputer workstations. At the

same time, because the first and second categories have been

around longer, more applications have been developed using

them and many products in the market

fall

into these

class-

es.

It

is

likely in the future that tools using the most modern

techniques will appear, but at the present time these tech-

niques

are

very much in the experimental and developmen-

tal stage and will need some time to mature.

4.1.1

Graphics

Libraries

and

Presentation

Packages

This

is

the traditional method for creating

ways

to view and

analyse data.

The libraries interface directly to graphics hardware

or

Libraries

of

provide graphics functionality in software. The user has to routines

supply nearly all the pieces of the application: the main pro-

gram, the user interface, data handling and geometry map-

ping. The most basic libraries only supply an interface to

the graphics devices (e.g., PLOT10 for Tektronix terminals,

HCBS for Calcomp plotters) and some higher-level libraries

36

Facilities for Scientific Visualization

handle more sophisticated graphic entities such

as

axes,

curve drawing, and

so

on. Typical examples in common use

are

the UNIRAS subroutine libraries, DISSPLA, GL (from

. Silicon Graphics), GKS(2D), PHIGS,

DORE

(Stardent),

and

NAG

Graphical Supplement.

Pros and cons The advantage of this type of software

is

its flexibility

and direct control but it suffers from the disadvantage of the

large amount of time necessary to write and support code.

PC

Graphics Many PC-based packages such

as

Harvard Graphics,

Slidewrite and CricketGraph

have

taken

on

board user-in-

terface functions to provide friendlier software, but still

re-

quire a great deal of user effort to achieve good results.

4.1.2

Turnkey

Visualization

Applications

These offer a fixed functionality to solve a limited range of

specific problems. The user supplies the data and the com-

putational instructions to the main program and possibly

some geometric mapping. The application supplies the

main program and rendering and usually

has

an attractive

user interface.

Dedicated

to

Many products in this category

are

extremely applica-

applications tion-specific and examples in oil exploration, molecular

modeling, and architectural modeling

are

common but of

limited

use

in other fields. They

are

also

very often only

available on a very

few

hardware platforms in common

use

in these industries -for commercial reasons. More general

examples

are

the UNIRAS interactives,

PV-Wave

(PVI), Data

Visualiser (Wavefront), Sun Vision (SUN) and VoxelView

(Vital Images).

No

programming The user does not

have

to program these packages and

needed! can obtain results very quickly. Their disadvantage

is

that

they

have

limited extensibility and therefore may often only

provide a part of the solution a user requires. They

have

all

reached a high

level

of maturity and many users applying

vi-

sualization to their work will probably

be

using one of the-

se

packages.

Visualization Software Categories

4.1.3 Application Builders

These offer a series

of

modules linked by interfaces which

are connected interactively at runtime.

It

combines features

from

both

of

the

other

two categories by providing turnkey

solutions for individual parts

of

the program and the flexibi-

lity to customise the final solution adopted.

The

supplied

modules can be replaced by user-written modules

as

re-

quired, providing they conform to the data

input/output

in-

terface requirements, therefore giving greater extensibility.

37

In these systems virtually everything the user needs

is

Select the options

provided by the program. The user has only to direct the

ex-

you need

ecution path

of

the program, provide the data, and optional-

ly,

their own computational modules if required.

Applications are constructed by a mouse-driven inter- Constructing

face,

manipulating icons

on

screens and linking them with applications

data paths. Once the required modules have been connected

and built the program can be executed.

New

applications

can be prototyped very quickly by connecting modules in

different ways but the user needs to know how to manage

the flow

of

data through the network, and how to extend

the module set.

Examples are

AVS3

(Stardent), Explorer (Silicon Graph- Some examples

ics

Inc.), apE (Ohio Supercomputer Centre) and Khoros

(University

of

New

Mexico). More advanced application

builders are currently under development and new advances

in visualization techniques and software will extend and im-

prove the application builder functionality. At the present

time these products are

not

mature enough to have had sub-

stantial numbers

of

packages built around them, but these

products should be appearing in the future.

4.1.4 Choosing a Package

Current work in scientific visualization tends to be done What should I

with

turnkey application tools -because

of

their function- use?

ality and their ability to process large data sets. However,

this does

not

mean that good visualization work cannot be

38

Facilities for Scientific Visualization

done with software libraries and

PC

packages, but merely

reflects the good understanding of these systems that exists

in the graphics community. Application builders

are

still

primarily used

as

research tools, but it

is

anticipated that

more applications will utilize them.

PC graphics PCs

are

more widely available in the academic/Research

survey Laboratory sphere than any other type of hardware and a

thorough review

of

currently available graphics software

has

recently been completed by the

UK

Inter-University Soft-

ware Committee Working Party and published

as

a Report.

Data transfer

It

is

probable that restrictions

on

internal data transfer

limitations bandwidth and graphics display facilities would limit the

usefulness of the current

PC

products

as

hardware platforms

for the majority

of

applications.

Access via X For those without workstations, PCs can

be

utilized

as

X-servers for such software running on superworkstations,

supercomputers,

or

mainframes, and connected via ethernet

or

X25

OANET). Users of PCs therefore can

have

access

to

visualization systems remotely and view the results at their

desks, albeit

at

lower graphics resolution and with a time

penalty introduced by current network performance.

4.2 Software Costs

4.2.1

Subroutine Libraries and Presentation

Packages

You

get what Costs of subroutine libaries and packages

are

their

you pay for functionality and degree of sophistication. Those which on-

ly provide interfaces to graphics

devices

(HCBS, PLOT10,

HPGL,

Xll)

are

bundled with the hardware. Packages with

a wide range of higher-level functionality, such

as

the UNI-

RAS subroutine libraries, can be fairly expensive. Packages

for PC-based operations reflect the prices that

PC

software

can command, and

are

typically in the range 100-500

pounds sterling. Much of this software

is

covered by educa-

tional

deals

and

is

available to the

UK

community and

Re-

Hardware Considerations (including Hardcopy)

search Councils under bulk discount arrangements (e.g., via

the

UK

Combined Higher Educational Software Team -

CHEST).

4.2.2

Turnkey

Visualization

Systems

The cost of these systems reflects the complexity of the

sys-

High-cost areas

terns and the comparative affluence of the targeted applica-

tion

areas!

For example, volume rendering products for the

petroleum industry are very expensive, but the users can

usually afford to

pay.

At the other extreme, products bun-

dled with,

or

designed for, particular hardware platforms

are

very reasonable. Educational deals exist for some of these

products.

4.2.3

Application

Builders

At present these systems are either available free (or almost Public domain,

free) in the public domain (e.g., Khoros, apE)

or

tend to be

or

proprietary

expensive if they are mainly machine-dependent and bun-

dled in with their hardware platforms (e.g.,

AVS).

In the

fu-

ture, more sophisticated systems may follow this scenario

or

may become more expensive, if unbundling takes place.

4.3

Hardware

Considerations

(including

Hardcopy)

Useful visualization work has been performed

on

very inex- Functionality

pensive equipment, but there

is

a growing requirement to needed

perform more complex analysis of increasingly large vol-

umes of data, using more and more sophisticated graphic

display techniques. When this

is

coupled with a demand for

much faster response it

is

obvious that hardware costs for

'ideal' systems could escalate. Although much valuable

work

has been done

on

superworkstations attached to a

su-

percomputer (e.g., CRAY) -this

is

a solution which

is

not

available to many people.

39

40

Facilities for Scientific Visualization

Costs

of

The amount of money in a hardware budget will deter-

components mine what kind of work can be performed,

or

the speed of

the computation. However,

we

can identify components

which are important. Systems

to

perform visualization with

subroutine libraries and packages are fairly modest and ter-

minal connection to a mainframe

or

a PC-based system cost-

ing

less

than 2 K pounds sterling

is

likely to be adequate.

Turnkey visualization systems usually require a

UNIX

workstation, with color screen and hard disk, which will

cost in

excess

of 6 K pounds. Higher-powered processors,

better quality, faster displays, and substantial hard disk stor-

age

will improve throughput but can increase the cost

be-

yond

20

K pounds. Application-builder software will only

work

properly

on

high-performance workstations. The

ab-

solute minimum configuration will cost around

15K-20K

pounds, but usable systems are likely to exceed

30

K.

High-

powered systems for complex analysis

are

likely to cost in

excess

of

70

K pounds.

Add-on extras Hidden costs should not be overlooked! These could in-

clude items such

as

high-speed networking

access,

archive/

backup facilities, sophisticated

I/O

devices, and -more im-

portantly -hardcopy.

A3

color Postscript devices suitable

for use with most workstations, PC's,

or

on

a network, are

available for

10-15K

pounds through educational deals.

If

it

is

desired to store graphic images in raster form, then large

amounts of disk space

are

required.

Costs for training and provision of expert advice and

support also need to be included.

4.4

Vendor

Systems

Versus

Public

Domain

Systems

Vendor systems Users are advised to note that vendor systems usually have

software support (e.g., for sorting out software faults and

other difficulties). As part of the licence agreement the user

will usually pay an annual software maintenance

fee

(typi-

cally

5-10%

of purchase price). Users also need to enquire

Vendor Systems Versus Public Domain Systems

whether such licence agreements entitle the user to receive

upgrades and new versions of the software

as

they become

available, if it

is

desired to continue with the same system

for a number

of

years. Users should note that usually up-

grades are included in the original licence agreement, but

that major developments

or

major new versions will often

be the subject

of

separate new licence agreements, for which

users have to

pay.

This

is

because the vendor has decided it

is

necessary to recoup research and development costs, and

therefore the new version has essentially been made into a

new item

of

software.

If

the user requires the additional new

functions in this software, then he

or

she

is

faced with a fur-

ther

licence charge.

In this area,

as

in others, users get what they pay for.

A variety

of

current vendor systems are detailed in

Chapter

7.

Public domain systems often have very limited support,

or

even none at all. Occasionally the authors and origina-

tors indicate they will receive reports

of

bugs

or

difficulties,

but

they cannot necessarily guarantee to provide remedies.

Also, future updates and revisions

of

the system may

be

mo-

re

uncertain. However, informal help

is

sometime available

via email discussion lists, where users report their problems

and

other

users

with

the same software can provide help and

assistance. Khoros has such an email discussion list. This list

also has the advantage that the Khoros Group at the U niver-

sity

of

New

Mexico are also

on

the list and so can provide

expert technical advice and help

as

appropriate. However,

such help

is

done

on

a voluntary basis; there

is

no contractu-

al

commitment.

Public domain systems often offer state

of

the art facili-

ties, and can be very useful for research and development

purposes. They are increasingly subjected to procedures

such

as

those applied to commercial vendor systems, for

ex-

ample, rigorous testing (e.g., Beta testing) before general

re-

lease to the community, and detailed on-line and hardcopy

documentation. They are also essentially free, though occa-

sionally a small distribution

fee

is

charged. Software can

41

New versions may

mean further costs

Public domain

systems

Addressing

problems

Quality is

improving

42 Facilities for Scientific Visualization

usually be obtained directly by anonymous

FTP

across the

international communications network. For users

with

no

funds, such software can be very useful indeed.

However, users and potential users should be aware

of

the following points before committing themselves to using

public domain software, especially for . large, on-going pro-

jects:

Check-list before • Software support

is

usually fairly limited.

deciding

to

use • Software support

is

usually in the hands

of

a small num-

the software ber

of

experts (often the designers and originators

of

the

project).

• Future developments (if any) are in the hands

of

the cur-

rent developers and the resources available to them.

•

The

current developers may decide to abandon the cur-

rent software and work

on

something completely differ-

ent,

or

their managers may move

them

to work

on

dif-

ferent projects.

•

The

software may be sold by the originators

or

their site

to a commercial vendor,

who

will then only release ver-

sions

of

it

as

for normal vendor software. The original

designers may no longer be involved, and the future

di-

rections the product will take become uncertain.

An

ex-

ample

of

this

is

scenario

is

apE, which began

as

a public

system and has been recently acquired by a corporation.

Further

information

on

this point

is

given in Chapter

8.

4.5

Summary

Comparing In addition to the obvious functional requirements that

us-

software ers need in the software to meet their specific application,

capabilities users should also bear in mind the following points

when

considering how the system

is

to be used:

• Software support

• Availability

of

source code

• Range

of

hardware platforms

on

which the software

is

available

Summary

•

Is

a library

of

graphical subroutines required?

•

Can

the software be distributed across a number

of

dif-

ferent platform types (e.g.; computational server and

workstation)?

•

Any

graphics interfaces that may be required by higher

level software

• X-support

• Data formats supported

•

Import

formats supported (for reading in information

from other sources)

• Export formats supported (for outputting information

to other systems)

Thus

users

of

scientific visualization systems need to consid- List al/ your

er very carefully

not

just the functions

of

the system, but requirements

also the environment in which

is

to be run, and the general

requirements associated with input

of

data and export

of

re-

sults to other systems.

43

Chapter

5

Outputting

Results

5.1

Hardcopy

Neglected topic This topic

is

included because it

is

often neglected in the

evaluation

of

scientific visualization systems and because

re-

quirements for it

only

surface

when

the user has started to

use the system and applied it to his

or

her problem.

Results Users should

think

carefully about the ultimate destina-

tion

of

their results. Often this includes written papers and

reports, and also presentations to funding bodies and con-

ferences.

Hardcopy Intermediate hardcopy

is

often required for the produc-

tion

of

draft reports for circulation within research groups

or

departments. Thus, although slides and video

are

usually

the preferred medium for the submission for publication, it

is

often very useful to be able to produce color paper hard-

copy for draft purposes.

Color Postscript

Color

Postscript printers and plotters are now widely

available in a variety

of

paper sizes from

A4

to

AO.

Most

vi-

sualization software contains drivers for producing color

Postscript, since it

is

a well accepted industry standard.

Stand-alone For a small research group, a dedicated printer connected

or

networked? to the workstation

is

a feasible option. For a larger laborato-

ry,

or

large groups

of

workstations and users, it

is

often

more economical to consider a larger printer connected via

the network. Users then have the capability

of

generating

output

of

larger size and often greater range

of

colors. How-

ever, it should be noted that large plotters are often expen-

sive to maintain and

run

(a

typical annual maintenance con-

tract can

be

7-10%

of

purchase price).

Areas such

as

geophysics and seismic applications appli- Specialized

cations often use

AO-size

maps

as

standard in the industry. applications

If

these are required

then

a top

of

the range color electrostat-

ic plotter will be required.

5.2 Video

Video

is

becoming an increasingly important medium for Color and

storage and display

of

real-time simulations for publication real-time images

of

research results and presentations at seminars and confer-

ences.

It

is

the

only

cost-effective medium for the publica-

tion

of

large amounts

of

color information

(1

hour

= 100,000 frames), and which

is

cheap and easy to copy.

It

has a natural interface to the

TV

technology domain and

provides a portable and easy-to-use medium.

The

use of video to demonstrate time-varying and dy-

namic processes

is

becoming increasingly important. A

se-

quence

of

still frames (e.g.,

on

slides) may

not

convey the

full information relating to

the

process involved, in a way

that

a sequence

of

moving images can.

The

display

of

real-time simulations offers

the

user a dy-

namic analysis tool to supplement

other

visualization meth-

ods.

Video technology

is

coming

within

the range

of

the Costs of video

workstation user by

the

increasing availability

of

low-cost

interface boards and also animation software.

pes

and

workstations can now be interfaced to a video recorder by

means

of

a video board, an animation controller, a PAL en-

coder, and a sync generator for around

7K

pounds sterling.

An

editing U matic

VTR

would cost a further 7 K pounds.

The

quality

of

the

finished product

is

proportional to Quality

of

the

the cost

of

the system.

If

broadcast quality

is

required,

then

result

the equipment required

is

currently expensive. However,

much more inexpensive systems

(as

the above) can produce

reasonable output.

Effective presentation

of

information via

the

video medi-

Viewers

can be

urn

is

a non-trivial task.

In

particular, almost all viewers

of

stern critics

46

Outputting Results

such information have become accustomed (unconsciously)

to a high level

of

presentation quality through the programs

presented

on

national and commercial television channels.

Professional Those new to video need to take advice from those with

advice substantial experience in this area, e.g., graphic artists,

TV

producers, video editors, etc. A close association with those

with experience in this area

is

likely to produce substantial

benefits in the quality of the work produced.

5.3

Other Media

CD

ROM

and

Other

media include

CD

ROM

and laserdisk. The typical

laserdisk cost of a laserdisk

is

700

pounds. This

is

capable of storing

large amounts

of

picture information, but

is

expensive un-

less

many copies are required.

Chapter

6

Current

Developments

and

Activities

6.1

USA

Initial impetus for scientific visualization

was

provided in NSF Report

1987 by a National Science Foundation (NSF) Panel Report

of a Workshop

on

"'Visualization in Scientific Computing»

(McCormick et

al.

1987).

The principal recommendations of the McCormick

Re-

Recommend-

port

were that national funding should

be

provided for ations

short and long term provision of tools and environments to

support scientific visualization, and to make these available

to the scientific and engineering community

at

large.

Such

provision

was

considered to

be

essential if the enabling tools

were to

be

effectively harnessed by current and future scien-

tists and engineers.

Such tools often require

access

to significant computa- Supercomputer

tion resources. A natural

focal

point for these developments facilities

has

been the funding of Supercomputer Centers -to pro-

vide both the facilities and

access

to them

by

the commu-

nity.

An

example of this

at

the

San

Diego Supercomputer

vi-

Network-based

sualization Center

is

the development of network-based visualization tools

general tools purpose visualization tools. These

are

accessed

by 2800 users with 350 different applications. Such users

ac-

cess

the facility by a variety

of

different routes including

di-

al-in lines, national networks, and dedicated high-speed

links.

In

addition to this broad range of provision there

are

Specialized

also more specialized tools for high-end applications (e.g., applications

molecular modeling, computational fluid dynamics).

Similar provision

has

also been made

at

other Supercom- Other centers

puter Centers

at

Cornell, Pittsburgh, and the University of

Illinois at Urbana-Champaign.

48 Current Developments and Activities

Workshops Workshops

on

scientific visualization

have

been estab-

lished by ACM SIGGRAPH and IEEE to address specific

as-

pects such

as

data facilities (to facilitate

ease

of

use

and trans-

fer

of information), and volume visualization (to enable rep-

resentation of

real

3 D information and to

give

inside views).

Representatives from the Department of Defense and the De-

partment of Energy

have

initiated a Working Group to

de-

fine a Visualization Reference Model. A conference of

CG

International

was

held at MIT, Media Laboratory, in June

1991

with the theme

"Scientific

Visualization

of

Physical

Phe-

nomena

~

The proceedings have been published

as

a book by

Springer-Verlag

(see

Chapter

11

-References).

Visualization A large number of major universities are establishing

vi-

laboratories sualization laboratories, and often such installations receive

supplementary funding for further proposals in specific

ap-

plication

areas.

Funding

is

provided by such bodies

as

NSF,

DARPA, and NASA. State supercomputers and associated

visualization facilities exist in Ohio,

North

Carolina, Min-

nesota, Utah, Alaska, and Florida.

Visualization

To

provide a forum for the presentation and discussion

conference of the latest advances in scientific visualization, the IEEE

Technical Committee

on

Computer Graphics has estab-

lished an international visualization conference, which

is

held

on

an annual basis.

NSF support In addition, the National Science Foundation

is

provid-

ing funds for the support and promotion of educational ini-

tiatives in scientific visualization by means of institutes,

workshops, and summer schools.

Network support Fast networks are required for distributed and remote

vi-

sualization. Developments in networking infrastructure are

planned to provide faster communication, interconnection,

and the ability to aggregate computing resources at different

locations

on

to one particular problem. For example, the

CASA test bed project

is

funded by the NSF to develop a

1 Gbit/sec network link between

Los

Alamos National Lab-

oratory, the California Institute of Technology, and

San

Die-

go

Supercomputer Center, to enable all three resources to be

concentrated

on

one application simultaneously.

UK

A multi-million-dollar grant

has

recently been awarded

by NSF to California Institute of Technology, Brown U ni-

versity, University of Utah, Cornell University, and the

University of

North

Carolina

at

Chapel Hill, to explore the

foundations of computer graphics and visualization.

6.2 UK

Foundation

aspects

49

A number of centers in UK academic institutions

are

con- Application areas

cerned with application

areas

such

as

molecular modeling

and computational fluid dynamics (CFD). There

are

a num-

ber of collaborative projects between academia and industry

in the

areas

of parallel processing and scientific visualiza-

tion.

One

example, GRASPARC, a Graphical Environment

for Supporting Parallel Computing,

is

a joint project

be-

tween

NAG

Ltd., the University of

Leeds

(School of Com-

puter Studies), and Quintek Ltd. The major objective

of

the

work

is

to improve the interaction between the computa-

tional scientist and the parallel computer through the

devel-

opment of interactive visualization software.

Vis

Lab at Sheffield University

is

engaged in

five

projects: VisLab

extending surface reconstruction to irregularly sampled

fields; rendering vector and tensor fields; building radiother-

apy planning tools; reconstructing cerebral blood

vessels

from a pair of x-ray projections; and

issues

surrounding per-

ception.

The

IBM

UK Scientific Centre in Winchester

is

primari- Industry and

ly concerned with scientific visualization and has a Visual- academia

ization Group, a European Visualization Group, a Medical

Imaging Group, and a Parallel Programming and Visualiza-

tion Group. There

are

a number of collaborative projects

with academia and industry in the

areas

of parallel process-

ing, user-interface aspects, and medical informatics.

Natural Environment Research Council (NERC)

has

a Research councils

Visualization Advisory Group concerned with evaluating

products for the

areas

of geological surveys and oceanogra-

phy. Science and Engineering Research Council Engineer-

50

Current Developments and Activities

ing Board has evaluated superworkstations in the

areas

of

hardware and software. The present

AGOCG

Scientific

Vi-

sualization Workshop which initiated this guide and the

Status Report arose out of proposals by the

UK

Universities

funding body for computing (the Computer Board) and the

Advisory Group on Computer Graphics (AGOCG).

Video facility The Rutherford Appleton Laboratory of the SERC,

Central Computing Division, has developed a video facility

for use by the academic and research community in the UK,

and

is

involved in projects in the

areas

of oceanography,

at-

mospheric physics, laser design, mechanical engineering,

ecological simulation, and CFD.

University

of

Leeds The University of

Leeds

has an interdisciplinary Scienti-

fic

Visualisation Group and promotes a wide range of soft-

ware on state of the art hardware platforms to support a

va-

riety of applications. (Reference: ACM SIGGRAPH Com-

puter Graphics, June

1992)

Other projects There

are

numerous other projects underway in this

field -the above

is

only an indication of the range of work

being done.

6.3

Europe

European centers IBM

has

a number of European centers actively involved in

projects involving Scientific Visualization. These include

the European Petroleum Applications Centre (EPAC) in

Stavanger, the Paris Scientific Centre which

is

involved in

visualization in the medical area, and the European Scientif-

ic

Centre in Rome which

is

involved in engineering and

modelling turbulent

flow.

IBM

also has a joint project with

the Centre of Competence in Visualization at the Universi-

ty

of Aix-Marseilles.

Volume FhG-AGD in Darmstadt

is

working

on

a number of

ar-

visualization

eas,

including tools for volume visualization

on

a variety of

platforms, and handling different kinds of data sets.

Workshops Eurographics arranged a Workshop on Scientific Visual-

ization in April

1990.

The proceedings will

be

available

from Springer-Verlag. Further workshops

are

planned.

Part

II

Overview of

Current Systems

and Developments

Chapter

7

Current

Vendor

Systems

in

Use

Readers

are

recommended to read Chapter 4 for a classifica-

tion and categorisation of scientific visualization systems

before reading this chapter. Chapter 4 sets out the overall

framework into which the products outlined in Chapters 7

and 8 fit.

Contact addresses

are

provided for each product at the

end of each section.

7.1

Wavefront

Technologies,

Inc.

Wavefront

was

founded in 1984 in California and provides

graphics products for

use

on a wide range of

UNIX

work-

stations including Silicon Graphics, IBM, Hewlett Packard,

DEC

and SUN. The company has a well established world-

wide

sales

and support network with its own offices in all

the key European countries, including the UK.

Wavefront's Visualizer software

is

designed to help engi-

neers, scientists, designers and graphic artists use the power

of today's 3D graphics workstations. There

are

three prod-

ucts providing professional visual communication to aid

rapid understanding.

-The Data Visualizer

is

designed to speed up the analysis

of

large volumes

of

3D data on any type of grid.

It

dis-

plays many variables at once and

uses

color and dynam-

ics

to create easily understandable images

or

animation

sequences of of complex processes.

-The Personal Visualizer

is

an

easy-

to-use image renderer

for

CAD

geometry.

It

provides photo-realistic images for

product designers, engineers and marketing groups.

Founded

in 1984

Visual

communication

Summary

of

Visualization

Products

54

Current Vendor Systems

in

Use

The

Advanced Visualizer includes interactive surface

modeling, animation and dynamics, and the ability to

render motion sequences and record them

on

tape

or

film for presentation.

It

has interfaces to a wide range of

CAD/CAE

and dynamics packages and

is

fully compati-

ble

with

the

other

visualizer products.

The Data Visualizer

Data

The

Data Visualizer

is

a graphic analysis toolkit for 3 D

sca-

Visualizer lar and vector data

on

any type

of

grid -including regular,

irregular, and unstructured grids.

Graphical interface

It

has a mouse-driven point-and-click interface that

al-

lows a wide variety

of

graphic tools including cutting pla-

nes, iso surfaces and iso volumes, particle systems, ribbons,

and sheets to

be

positioned and turned

ON

and

OFF

at the

click

of

a button. There

is

no limit to the number

of

tools

that can be created, grouped, and rendered simultaneously.

Interactive analysis

The

Data Visualizer

is

designed for the interactive analy-

sis

of

large 3 D volumes where many tools are required for

simultaneous analysis, and their rapid combination

is

a key

productivity factor. The user interface therefore provides

re-

fined management

of

tools and screen layout and lends itself

well to an environment in which many users require vol-

ume throughput.

Animation The data and all the graphic tools can be animated over

time, and there are also common image file formats such

as

color Postscript for use with

DTP

printers.

Analysis

of

fluids

The

ability to handle unstructured grids gives the Data

and

structures Visualizer a strong capability in the next major area

of

ad-

vancement in data visualization. Adaptive unstructured

grids allow complex shapes to be defined by the user

with

relatively simple and intuitive tools. This technology will

make numerical analysis

of

fluids and structures more

acces-

sible to the non-expert engineer

or

designer.

Objects and flows

One

of

the unique features

of

the Data Visualizer

is

its

round them ability to use the same analysis tools in the unstructured en-

vironment and for multi-block data where individual grids

Wavefront Technologies, Inc.

have been created around different components in a flow

field. This makes it possible to

see

a complex assembly such

as

an entire aircraft, together

with

the flow conditions

around it and its various parts.

Finite Element Analysis applications

run

exclusively

on

unstructured grids, and the Data Visualizer provides an

as-

sortment of tools for viewing data

on

the surface of these

grids. These include clipping planes and boxes that cut away

portions

of

data volumes and iso-surfaces to view interior

detail, while maintaining the exterior

view.

There

is

also a command language for users

who

wish to

build combinations

of

tools and results externally -for

ex-

ample from within their solver -and data reader source

co-

de

is

provided to allow on-line network transfer

of

results

from a host directly to the screen

of

the graphics worksta-

tion.

Figure

7.1

shows a transparent isosurface of rainwater

density in a cloud together with wind velocity flow ribbons

that are themselves mapped with a scalar density value. The

user interface includes interactive color map editing.

55

Viewing data on

unstructured grids

Command

language for tool

building

Fig.

7.1

Transparent

Isosurface

56

Fig.

Z2

Fighter

Fuselage

Current Vendor Systems

in

Use

Figure 7.2 shows

computational

fluid dynamics analysis

of

a fighter fuselage.

The

grid

is

irregular

and

contains mul-

tiple data blocks

and

approximately 250000 nodes.

The

pic-

ture

shows air pressure

on

the

fuselage

with

particle traces

mapped

with

mach

number,

spiralling

around

a vortex.

The Personal Visualizer

Personal

Visualizer

This

product

was developed for

the

casual user

who

needs

to

postprocess computer-aided design models.

It

interfaces

directly

with

many

leading

CAD

packages

and

allows inter-

active

control

of

lights, cameras,

and

surface materials

to

cre-

ate highly realistic images.

The

Personal Visualizer provides

texture mapping,

bump

mapping, transparency, refraction

and

ray tracing,

and

has a library

of

over 700 prepared mate-

rials

as

well

as

its

own

surface material

editor

with

which

new

surface textures can be created.

Wavefront Technologies, Inc.

The Advanced Visualizer

This product

is

for those

who

wish to create realistic motion Advanced

sequences, for example an engine assembly in motion,

or

a Visualizer

spacecraft docking sequence. It provides

all

the modeling

tools needed to create the geometry internally and can also

read geometric and stress data from external systems.

57

The Advanced Visualizer has sophisticated animation Sophisticated

tools and on-line motion channels that can be driven by animation facilities

ASCII data in real time. This latter feature makes it a useful

tool in displaying the results of computed dynamic analysis,

such

as

in vehicle crash performance testing. The Advanced

Visualizer includes all of Wavefront's state-of-the-art image

rendering technology, allowing the user to create virtually

any effect that may be required.

Figure

7.3

is

taken from an animated engine sequence

showing all the major parts in synchronised motion seen

through the transparent engine block.

Contributed

by

Peter Stothart,

UK

Managing Director,

Wavefront

Technology

Ltd.

Fig. Z3

Engine

58

High quality color

graphics

Raster based

Pioneering

developments

Real world

application

of

visualization tools

Current Vendor Systems

in

Use

For further information contact:

US.A.

Wavefront Technologies, Inc.

530

East Montecito Street

Santa Barbara

CA

93103

US.A.

Tel:

805-962-8117

Fax:

805-963-0410

United Kingdom

Europe

Wavefront Technologies

Guldenspoorstraat

21-

23

B-9000

Gent Belgium

Tel:

32-91-254555

Fax:

32-91-234456

Wavefront Technologies Ltd.

Oakridge House; Wellington Road

High Wycombe

Bucks HP12 3PR

UK.

Tel:

0494-441273

Fax: 0494-464904

7.2 UNIRAS

A.

S.

UNIRAS

was

established in

1980

with the objective of satis-

fying the growing need among computer users for a range

of high-quality color graphics software. This need arose

partly from current requirements to analyse ever increasing

volumes of data and partly to utilize fully the high perfor-

mance graphical output devices coming

on

to the market.

UNIRAS software

uses

raster techniques to deliver a

broader spectrum of colors, improved resolution, and great-

er throughput. UNIRAS therefore makes the most of new

hardware technology, while still effectively supporting the

traditional vector output devices of an earlier generation.

Key management and development people at UNIRAS

have worked with color raster graphics since the early

1970s

and were closely involved in the design of software for the

first inkjet plotters. The current UNIRAS product range

has evolved from this pioneering work.

Reflecting the rapid growth of the offshore oil industry

at that time, UNIRAS' first commercially available product

was

a software package to aid exploration companies in their

UNIRAS A.S.

search for oil and

gas.

This

was

an early example of the

use

of

scientific visualization tools in real-world applications

with considerable strategic benefits. Since then UNIRAS

has taken its technology into a number of other application

areas, including automotive and aerospace manufacturing,

pharmaceutical industry, communications, defense, energy

generation and distribution, and environmental manage-

ment.

UNIRAS software technology comes in two forms. In-

teractive, user-friendly packages permit non-specialist com-

puter users -scientists, engineers, and managers -to learn

to use the extensive facilities quickly and easily; while the

range of subroutines provides a choice of powerful tools to

help the application programmer integrate high quality

graphics with new and existing applications. All UNIRAS

products are computer and device independent and comply

with accepted international standards.

Today UNIRAS

is

a truly international organization and

can list many famous companies, research institutions, and

universities among its hundreds of customers. UNIRAS has

its headquarters in Denmark, with wholly owned subsidiar-

ies

in the USA, UK, France, Germany and Italy, a

sales

of-

fice

in Tokyo, and representatives in other parts of the

world. Research and development takes place in Denmark

and the USA. The major shareholders in UNIRAS

are

the

Danish financial services group Hafnia, the Dutch invest-

ment company Halder Holdings, and UNIRAS' own man-

agement.

UNIGRAPH

+ 2000

is

a powerful, fully interactive data

visualization system which enables users

to:

• retrieve their technical and scientific data from a

file

or

database,

• edit and operate

on

it in a variety of ways,

• analyse and visualize it quickly and in many forms, from

very simple charts to advanced multidimensional sur-

faces,

•

give

plots an extra touch of professional presentation

quality,

• present the information graphically

as

hardcopies of the

highest quality.

59

Interactive

modules

or

library

routines

International

vendor and clients

UNIGRAPH

facilities

60

Visualizing

datasets

Wide

variety

of

output

devices

Integrating

visualization

and

presentation

Windowing

environments

Networked

facilities

Platforms

Benefits

and

advantages

Current Vendor Systems

in

Use

Datasets can

be

accessed, edited, and analyzed using mathe-

matical, logical,

or

statistical operators. A comprehensive set

of interpolation techniques correctly handles such complex-

ities

as

regions and barriers, allowing datasets to

be

visual-

ized

as

2

D,

3

D,

or

4 D surfaces in color

or

monochrome.

The

UNIGRAPH

+

2000

hardcopy system produces

high-resolution hardcopies on a wide variety of output

de-

vices

including raster and vector

devices

as

well

as

the new

generation of Postscript printers. Pictures can

also

be

saved

for later

use

or

exported to other systems by the creation of

ISO Standard Computer Graphics Metafiles (CGM)

or

en-

capsulated Postscript

files.

agX/TOOLMASTER

is

a suite of high-level graphics

tools which allows the software developer to easily integrate

visualization and presentation techniques into application

programs.

agX/TOOLMASTER programming tools

are

callable

from C and

have

been developed and optimized for

use

in

an X Window environment. The open systems architecture

of agX/TOOLMASTER allows it to

be

fully integrated

with the X Window, OSF/MOTIF and

OPEN

LOOK win-

dowing environments.

With agX/TOOLMASTER the application program-

mer can combine high-level graphics functions with the best

features of X such

as

multiple windows and event handling

for interactivity, pixmap generation for animation, and

cli-

ent/server techniques for network computing.

agX/TOOLMASTER runs on

all

major

UNIX

worksta-

tions and supercomputers

as

well

as

the VAX/VMS environ-

ment, making applications portable across the network.

The virtual color system in agX/TOOLMASTER provides

X-server independence with its support of monochrome

and color displays and various bitplane depths.

The benefits of building an application with agX/

TOOLMASTER

are:

• The amount of time and code required for building

ap-

plications in the X environment

is

greatly reduced,

UNIRAS A.S.

• Code maintenance and support over the lifetime

of

an

application

is

also reduced,

• Productivity

of

programmers

is

increased by

access

to

high-level single function calls for visualization

of

nu-

merical data,

• Hardware investments are protected.

61

Fig. Z4

agXJCONTOURS

example

Fig. Z5

agXJVOLUMES

example

62

Current Vendor Systems

in

Use

Commitment

to

UNIRAS technology

is

based on standards, and in the

standards future UNIRAS will continue to develop products that

of-

fer

flexibility together with computer and device indepen-

dence. The further development and enhancement of both

the UNlRAS interactive packages and subroutine libraries

will continue to provide high-quality graphics solutions to

both

end users and application programmers.

Ongoing Research and development now takes place in both Eu-

developments rope and the USA in order to reinforce the global scope of

the company's products. Future developments will also take

place with the cooperation of both UNIRAS users and the

computer hardware vendors. The UNIRAS network of lo-

cal

subsidiaries in Europe, USA, and Japan will enable the

company to continuously strengthen local

sales

and support

activities.

Contributed

by

Mike Bundred,

General Manager UNlRAS Ltd.

For further information contact:

Denmark

UNIRAS A.S.

376 Gladsaxevej

DK-2860 Soborg

Denmark

Tel:

45-31-672288

Fax:

45-31-676045

United Kingdom

UNlRAS Ltd.

Ambassador House

181

Farnham Road

Slough SLl 4XP

u.K.

Tel:

0753-579293

Fax:

0753-821231

Germany

UNIRAS

GmbH

Niederkasseler Lohweg 8

W-4000

Dusseldorf

11

Federal Republic of Germany

Tel:

0211-5961017

Fax:

0211-5961019

Precision Visuals,

Inc.

7.3

Precision

Visuals,

Inc.

Engineers, scientists and researchers have a common need Visual data

for visual data analysis (VDA) in order to understand and analysis

use their data. A common requirement

is

for large datasets

and fast graphics

with

analytical capabilities such

as

mathe-

matics, statistics, signal processing, and image processing.

Precision Visuals Workstation Analysis and Visualiza- Visualization

tion Environment

(PV-WAVE)

is

a powerful software system environment

which lets users display, reduce, analyse and re-display large

multi-dimensional data sets.

As an example, Figure 7.6 exhibits temperature, carbon

monoxide, and sulfur dioxide contents in a city's air for one

year.

The

3 D surface allows observation of how all the pa-

rameters interact, and integrates the three variables in one

plot. Thus relationships between parameters can be visual-

ized. Menus also allow highlighting and magnification of a

specific dataset.

Visual Data Analysis and

PV-WAVE

improves

upon

tra- User interaction

ditional data analysis by allowing the user

to

control data

Fig. Z6

Test

Engineering

63

-

Air

Quality Data

64 Current Vendor Systems

in

Use

analysis by user interaction with visual representations. The

features of

PV-WAVE

are:

• Reads and defines large, multi-dimensienal datasets

• Tools for fast manipulation and subsetting

• Quick graphical displays of immediate results

• Immediate user interaction

• Advanced graphic tools for animating and displaying

multidimensional data

Increase

in

VDA techniques

are

invaluable for scientific discovery and

productivity engineering analysis. They offer impressive advantages, in-

cluding increases in productivity and a means for visual

communication with colleagues.

PV-WAVE

is

interactive software for visualizing and ana-

lyzing technical data.

It

consists of a set of high-level, inter-

pretive commands and procedures that provide:

Features • Data

access,

reduction and analysis

• 2 D and 3 D graphics

• Dynamic graphics

• Image processing and manipulation

• Application development

Application areas

PV-WAVE

is

being successfully applied in the following

ap-

plication

areas:

Laboratory Science

to visualize and analyze data from analytical instruments

to develop instrument automation systems

to create custom laboratory information systems

Test

Engineering

to visualize vibration, heat transfer, and emissions test

data

to compare prototype performance with theoretical

ex-

pectations

-to implement quality control systems in manufacturing

environments

Precision Visuals,

Inc.

Real-time Data Acquisition and Control

-

to

build automated remote sensing systems

to

manage water quality and

sewage

systems

to

monitor atmospheric conditions

to create simulations based

on

real data

Space Exploration and Astrophysics

-to reconstruct planetary and stellar environments

-to study geodynamics in planets and satellites

-to study seismology

to simulate astronomical events and objects

Computational Fluid Dynamics

to identify flow patterns such

as

shock

waves,

vortices,

and shear layers

to

apply

CFD

research to aeronautics, automotive

de-

sign, weather forecasting, and oceanography

-to analyze data from thermal dynamics, fluid dynamics

and nuclear reactions

Finite element modeling and analysis

-to assure quality and reliability in computations involv-

ing field equations

to apply finite element pre- and postprocessing methods

to such areas

as

aircraft design and stress analysis in build-

ing components

Imaging -medical and remote sensing

-to postprocess remote sensing data

-to display and analyse bioscience imagery, including

NMR/MRI,

X-Ray, CAT, and electron microscopy

Earth Resources

to interpret seismic data

to analyze well logs for locating mineral deposits

-to make meteorological predictions

-to compile and combine raw data for mapping

65

66 Current Vendor Systems

in

Use

. ..

[-

Fig. 7.7

Remote Sensing

-Landsat Image

In

Figure 7.7.

PV-WAVE

Point and Click subsets this satel-

lite image in multiple windows using the toolbox. The pic-

ture shows the Boulder

Valley

east of the Rocky Mountains.

The Boulder Reservoir

is

highlighted in the smaller win-

dows.

Other products

Other

members

of

the

PV-WAVE

family of visual data anal-

ysis software products include:

Numerical analysis

PV-WAVE:

NAG

-features the powerful numerical

analysis capabilities of the

NAG

Workstation Library with

the sophisticated visualization and data analysis functions

of

PV-WAVE

to create a single tightly integrated system,

al-

lowing

access

to

172

subroutines and functions from the

NAG

library through a seamless link.

Programming

PV-WAVE

Point and Click -combines the power and

not needed functionality of

PV-WAVE

with an easy to use Point and

Click mouse-driven interface that allows technical profes-

sionals to

access,

analyse, and visualize their data, without

the need to program.

Stardent Computer, Inc.

For

further information contact:

US.A

Precision Visuals, Inc.

Lookout Road

Boulder

CO

80301

US.A.

Tel: 303-530-9000

Fax: 303-530-9329

United

Kingdom

Germany

Precision Visuals

International

GmbH

Lyoner Stern

Hahnstrasse

70

W 6000 F

rankfurt/Main

71

Federal Republic

of

Germany

Tel:

49-69-6690150

Fax: 49-69-6666738

Precision Visuals International, Inc.

Royal House

1-4

Vine Street

Uxbridge

Middlesex UB8 1XF

UK.

Tel: 0895-235131

Fax: 0895-272299

7.4

Stardent

Computer,

Inc.

AVS

is

an advanced interactive visualization environment

for scientists, engineers, and technical professionals.

AVS

supports geometric, image and volume datasets -

the

user

can interactively select the appropriate menu option.

No

programming

is

required.

For

the

more sophisticated user, the

AVS

Network

Edi-

tor

can be used to build processing networks into which us-

er-developed modules and computational programs can be

integrated easily.

Modules can by dynamically added, connected, and de-

leted. Modules are

only

re-executed

when

new data

is

re-

quired

or

an

input

parameter

is

changed. Modules have con-

trol panels for interactive control

of

input

parameters by on-

Interactive

visualization

Visual Network

Editor

67

Modular approach

68

Current Vendor Systems

in

Use

screen sliders,

file

browsers, dials and buttons. The control

panel

is

automatically generated when a module

is

connect-

ed into the network.

Building The complete network can be

saved

with all the user

de-

applications fined interactive controls and layout specifications

as

a com-

plete application. This can then

be

invoked directly, bypass-

ing the standard

AVS

menus and the Network Editor.

Integration

of

User programs can

be

coupled into the network to allow

user programs real-time visualization of dynamic simulations. This allows

the user to transform existing batch programs into interac-

tive visual applications.

Further modules

AVS

has a wide range of data input, filter, mapper and

renderer modules. User-written programs

or

subroutines in

FORTRAN

or

C can

be

easily converted into

AVS

modules.

Filters Filters transform data into data

or

geometry into geome-

try. Some filters convert the output data of widely-used

ap-

plications into displayable form.

AVS

includes filters for

ap-

plications in engineering analysis, computational fluid dy-

namics (CFD), chemistry and other fields.

Other

filters pro-

cess

commonly used data formats such

as

the Brookhaven

Protein Databank (PDB) molecular structure format,

or

graphics formats such

as

MOVIE.BYU and Wavefront

Tech-

nologies.

New filters New filters can be developed by means of templates and

geometric conversion utilities.

Mapping Mappers transform data into geometry. Multiple visual-

ization techniques can

be

selected to suit the problem being

studied. Examples of mappers include: isosurfaces of a 3 D

field;

2D

slices

of a

3D

data volume;

3D

meshes from

2D

elevation datasets.

Renderers Renderers display geometry, images, and volumes on

screen.

AVS

networks can incorporate multiple rendering

modules, including a fully-featured 3 D geometry renderer,

an image display renderer, and a range of volume renderers.

Graphic images may also

be

output to hardcopy devices

or

video tape.

Image processing

AVS

provides a complete image display capability, in-

cluding real-time pan and zoom, rotation and transforma-

Stardent Computer,

Inc.

69

tion,

flip

book

animation, and

support

for 8-bit, 24-bit, and

floating

point

images. Imaging filters include look-up table

operations such

as

contrast stretching, pseudo-coloring, and

histogram balancing,

as

well

as

data resizing operations such

as

interpolation.

AVS

takes image processing a step further

by

generalizing Volume imaging

these modules for 3 D volume imaging.

AVS

provides a vari-

ety

of

tools for rendering volume data; a real-time isosurface

generator; a unique transparent volume renderer

which

cre-

ates real-time, semi-transparent images

with

full rotational

and lighting control; generation

of

geometric objects such

as

arbitrary slicing surfaces,

dot

surfaces and vector nets;

and VBUFFER, a unique, high-quality volume renderer.

Fig. Z8

Stardent

AVS

70

Current Vendor Systems

in

Use

Geometry The

AVS

Geometry Viewer

gives

full control with sim-

viewing

pIe

menu-driven parameter selections.

It

offers wireframe,

Gouraud,

or

Phong shading;

16

individually controlled

col-

ored light sources, selectable

as

point, directional

or

spot

lights; surface properties such

as

specularity and transparen-

cy;

real-time texture mapping and anti-aliasing.

Different views

AVS

allows creation of multiple windows with different

views of the same geometric object

or

simultaneous display

of

multiple objects.

Hierarchies

Scenes

with hierarchies of objects can

be

created and

ma-

nipulated individually

or

as

one

or

more groups.

Scenes

can

be

saved,

with

all

viewing selections preserved for later

re-

display. Sequences of images can

be

created and

saved,

and

Animated views the sequence cycled through to provide animated views of

dynamic behavior in real time.

Platforms

AVS

is

designed for portability and multi-platform sup-

port, from desktop systems to supercomputers. Written in

C,

AVS

runs in a

UNIX

X-Window environment. The

ge-

ometry renderer

is

designed to support a variety of graphics

display subsystems including the standards PHIGS and

PHIGS+, Stardent's advanced rendering and display envi-

ronment,

DORE

(Dynamic Object Rendering Environ-

ment), and other display list

or

immediate mode graphics

interfaces.

AVS

modules

are

a convenient means of exchanging new

computational and visualization software.

AVS

is

supplied

free

with every Stardent visualization

system. Licensing of the software

is

available for other plat-

forms.

Figure

7.8

shows

3D

terrain elevation mapping using

Stardent's

AVS

package. The area shown

is

Orange County,

southern California.

Figure

7.9

shows the visualization of electron orbital

within a hydrogen atom using Stardent's

AVS

package.

From information supplied by Stardent Computer Ltd.

Stardent Computer, Inc.

For further information contact:

USA

Stardent Computer, Inc.

6

New

England

Tech Center

521

Virginia Road

Concord

MA

01742

U.S.A.

Tel:

508-287-0100

Fax: 508-371-7414

United Kingdom

Stardent Computer Ltd.

7 Huxley Road

The

Surrey Research Park

International Headquarters

Stardent Computer

Hagenauer Strasse

42

W-6200

Wiesbaden 1

Federal Republic

of

Germany

Tel:

49-611-22037

Fax: 49-611-260181

Guildford Surrey

GU2

5RE

Tel:

0483-505388

U.

K.

Fax: 0438-505352

Fig.

7.9

Visualization

of

Electron Orbit

71

72

Visual processing

Four levels

of

products

Upper

and

lower levels

Graphics

Library GL

Current Vendor Systems

in

Use

7.5

Silicon

Graphics,

Inc.

Silicon Graphics

is

the world's leading manufacturer of

vi-

sual processing systems. Visual processing allows staff and

researchers to work with data in a more natural, intuitive

way -graphically. In addition, Silicon Graphics designs,

manufactures and markets computational systems used by

engineers, scientists and animation professionals for design

and analysis of 3D objects and for general purpose technical

computing. By employing

RIse

technology and propri-

etary VLSI components to provide both high-performance

computing and high-performance graphics, Silicon Graph-

ics

continues to deliver amongst the most powerful systems

available for engineering and scientific applications.

As

part of its visual processing software, Silicon Graphics

is

offering four modular software products which fit into

different

levels

between application and source code. Figure

7.10 shows how these products relate to each other. Moving

up the chart shown in Fig. 7.10 away from the source code

and toward the application makes it easier for the user to

use

a product without specialist knowledge, but this

loses

some

of

the flexibility at the source level.

Often this difference between the upper and lower

levels

is

partly

or

wholly obscured by the proliferation of tools

and facilities currently available in the market place.

The products depicted in Fig. 7.10

are

as

follows: GL

is

the Graphics Library that has been available from Silicon

Graphics for a number of years, and

is

now gaining accep-

tance with other hardware suppliers. This provides the basic

system calls required to address the 3D graphics hardware in-

corporated into every Silicon Graphics system.

It

is

a set of

calls that can

be

included into source code, which may

be

written in a programming language. Above this

level

is

a

graphics toolkit which provides a library of standalone GL

calls that have been developed to address a wide range of

graphics problems and which

is

available to be incorporated

into software developers' products.

Silicon Graphics,

Inc.

Application at User

Level

Explorer )

Flexibility Image Vision Library Ease of Use

GL Toolkit

GL (Graphics Library)

Source Code at Programmer

Level

The more significant products

are

at the highest

level.

Image Vision Library

(IL)

is

an object-oriented, extensible

toolkit for creating, processing and displaying images on all

Silicon Image Graphics workstations. The

IL

toolkit pro-

vides image-processing application developers with a robust

framework for managing and manipulating images. The

toolkit

is

specifically designed to provide a constant soft-

ware interface to hardware that may change underneath,

thus ensuring that applications can continue to

run

un-

changed in the future.

The Image Vision Library consists of a shared library

de-

veloped in

C++,

with interfaces for C and FORTRAN.

It

has a core set of more than

70

image processing operators,

and

is

user-extensible for specific needs. Silicon Graphics

provides a set of data abstractions and

access

functions to

make it easy to augment the IL toolkit's image operators

and design new ones.

Image data sets have a wide variety of formats. The IL

toolkit allows new

file

formats to

be

integrated into the

li-

brary. The toolkit currently supports three standard

for-

mats: SGI, and extended version of TIFF, and a simple tiled

format called FIT.

73

Fig.

Z10

Levels

of

products

Higher level

products

Image Vision

Library

Image processing

operators

Image data set

formats

74

Image

manipulation

Data processed

on demand

Visualization

facilities

Customised

application

building

Modular approach

Current Vendor Systems

in

Use

The

IL toolkit provides an efficient model for the manip-

ulation

of

image data and image attributes. The toolkit's im-

age

model includes a configurable cache

to

allow

access

to,

and processing of, the very large images

common

in many

disciplines. IL provides a

common

interface for image ma-

nipulations, while requiring little

or

no programmer knowl-

edge

of

the image's internal structure

of

format. IL imple-

ments a demand driven execution model, such that data

is

processed only

on

demand. This model

is

based

on

the same

cached-image model

as

file images. This technique enables

an application to process just the area

of

interest, providing

significant benefits in terms

of

reduced

I/O

and improved

system performance.

The

fourth product

is

called Explorer. This

is

a true

ap-

plications developer package.

It

provides visualization and

analysis functionality for users whose needs are

not

met by

commercial software packages,

or

who

want to extend exist-

ing systems with their own algorithms and techniques.

The

software environment falls into the category

of

Ap-

plication Builders -environments that consist

of

function-

al

program pieces called modules which

are

visually con-

nected together through a point-and-click user interface into

a data flow style network. This flexible and interactive envi-

ronment

of

building application programs by choosing

from a suite

of

functional modules

is

the true power

of

the

system.

Modules

are

the building blocks

of

the visualization soft-

ware and cover a wide range

of

functionality. Because the

software environment encompasses a distributed execution

model, modules may execute

on

the local workstation

or

on

other

platforms

on

the network. Users can easily integrate

their existing algorithms into the system in the form

of

new

modules. Through point-and-click selections, a model-

building facility

is

provided which generates the code need-

ed

to

make the user's algorithm into a visualization software

module. Modules generally fall into the following catego-

nes:

Silicon Graphics,

Inc.

• Input

Modules that read data files

• Feature Extraction and Analysis

Modules that produce data from data (e.g., extract a pla-

nar slice from a volumetric dataset)

• Geometric Representation

Modules that produce geometry-based display lists from

data

• Renderers

Modules that produce images from geometry, volumes

or

Images

•

Output

Modules that write to disk

System Data Types are the data formats for passing data Data types

through the system.

The

system data types are powerful and

abstracted, and each can represent an entire

class

of

data

as

well

as

a very specific instantiation

of

the data. Date types

are

as

follows:

• Parameter

This data type conveys widget interaction to the module.

Parameters are scalar quantities including long integer,

double precision, floating point, and character string.

• Lattice

This

is

the most widely used data type.

It

is

essentially

a multi-dimensional array

with

two major components:

data stored at nodes

of

the lattice, and a coordinate map-

pmg.

• Pyramid

This data type combines lattices with connectivity in a

hierarchical structure. The depth

of

this structure

is

arbi-

trary.

• Geometry

This data type contains a hierarchical geometrical scene

description. The geometric description contains all in-

formation concerning geometric objects and their attri-

butes, cameras, lights, etc.

75

76 Current Vendor Systems

in

Use

•

Unknown

The unknown type

is

an uninterpreted array of bytes.

The organisation and interpretation

is

left to the pro-

grammer.

Widening the use IRIS Explorer

is

a key part of Silicon Graphics' new techni-

of

visualization

cal

computation environment, aimed at making the compa-

ny's industry-leading visualization technology more

accessi-

ble to the broad range of workstation users. With IRIS Ex-

plorer, users view data and create applications by visually

connecting software modules into flow chart configurations

called module maps. Modules, the building blocks of IRIS

Explorer, perform specific program functions such

as

data

reading, data analysis, image processing, geometric and vol-

ume rendering, and many other tasks. Modules can be

exe-

cuted across heterogeneous platforms, delivering powerful,

resource-efficient distributed computing capabilities to

ap-

plication users and developers.

The following pictures were created by Silicon Graphics

using

AVIRIS

data, courtesy of the Jet Propulsion Laborato-

ry.

AVIRIS

data

is

made up of 224 spectral bands. The imag-

es

were created by traversing through the data in the spectral

dimension, and show spectral responses and histograms of

the data.

Figure 7.11

is

a general view of the user-interface show-

ing an image of San Francisco with several multi-spectral

an-

alyses being carried out. (Created by Silicon Graphics using

AVIRIS

data, courtesy of the Jet Propulsion Laboratory.)

Silicon Graphics,

Inc.

Figure

7.12

contains an isometric view showing spectral

signatures for a number of pixels along a specific transect of

the original image. Vertical scale

is

the radiance, horizontal

scales

are position and wavelength.

Fig.

Z11

Multi-spectral

Analyses

Fig. Z12

Spectral

Signatures

77

78

Fig. Z13

Chair Diagram

Current Vendor Systems

in

Use

Figure

7.13

is

a chair diagram

of

multi-spectral image

da-

ta.

The

horizontal plane shows an image at a particular

wavelength. The vertical planes are slices at a given X,Y loca-

tion

showing radiance at the spectral wavelengths which

have been recorded.

From information supplied by

Mark

Goossens, Education

and Research, Silicon Graphics Ltd.

For further information contact:

USA

Silicon Graphics, Inc.

2011

N.

Shoreline Boulevard

P.O. Box

7311

Mountain View

CA

94039-7311

U.S.A.

Tel:

415-960-1980

Fax: 415-961-0595

Europe

Silicon Graphics

International

18

Avenue Louis Casai

CH-1209 Geneva

Switzerland

Tel:

41-22-7987525

Fax: 41-22-7988230

United Kingdom

Silicon Graphics Ltd.

Forum 1 Station Road

Theale Reading

Berks RG7 4RA

U.K.

Tel:

0734-306222

Fax: 0734-302550

Sun Microsystems, Inc.

7.6

Sun

Microsystems,

Inc.

7.6.1

SunVision

-

Sun's

Visualization

Software

Package

Sun Vision

is

a software platform with which application

de-

velopers and sophisticated end users can develop visualiza-

tion applications.

It

includes two programming interfaces

(for image processing and high-quality rendering), and

Open

Windows-based tools for image processing, volume

rendering, manipulation of 3 D geometric data, high-quality

rendering and movie loop display. The tools and libraries

are highly integrated so that data and images can

be

shared

among them.

SunVision

1.1

runs

on

any 8-bit color SPARCsta-

tion/Sun-4 Workstation and

on

Sun workstations equipped

with the true color

VX/MVX

visualization accelerators.

No

special purpose hardware

is

required.

If

a VX

or

VX + MVX visualization accelerator

is

present in the

sys-

tem, the Sun Vision applications

are

transparently acceler-

ated.

7.6.2

SunVision

Programming

Interfaces

Visualization

applications

79

No

special

purpose hardware

add-ons required

Sun Vision provides two libraries for visualization tasks (im- Libraries

age

processing and high-quality rendering), and one utility

library. Future releases will include an additional library in-

terface for volume rendering. In addition, Sun Vision

is

de-

signed to work in an integrated fashion with XGL, Sun's

3 D interactive graphics library.

SunIPLib

is

Sun's image processing library, providing

ex-

Image processing

tensive image processing functionality. It consists of C-call-

able functions for:

• arithmetic and logical operations

• spatial filtering

• Fourier domain processing

• image analysis

• geometric operations

80

Current Vendor Systems

in

Use

In addition, there

are

library functions to create and manip-

ulate subimages and regions of interest.

Images can have multiple bands, and can

have

unsigned

byte, signed 16-bit short,

or

32-bit floating point data types.

Additional imaging functions can

be

added to the library.

High-quality For high-quality rendering, Sun Vision provides a Ren-

pictures derMan function library and a

RIB

(Renderman Interface

Bytestream) protocol interpreter. The RenderMan interface

provides a way to describe geometry, scenes, the camera, and

lights so that computer images can

be

generated from this

information.

RenderMan Users of the RenderMan interface specify a set of proce-

dures that describe a scene. Object color and location, light-

ing, and viewer perspective can all

be

specified. The Render-

Man shading language provides a way to create shaders

spe-

cific

to

a given scene.

The SunVision RenderMan function library

is

compliant

with version

3.1

of the RenderMan Interface Specification.

It

also supports the following optional features:

• solid modeling

• programmable shading

• displacements

• texture mapping

• environmental mapping (partial implementation)

• bump mapping

• volume shading (partial implementation)

The following surface shader options

are

also provided:

• general (the default)

• the 6 standard RenderMan surface shaders (plastic, paint-

ed plastic, metal, shiny metal, matte, and constant)

• the 2 standard RenderMan atmosphere shaders (depth

cue and

fog)

7.6.3

SunVision

Window-based

Tools

Visualization tools Sun Vision also provides Open Windows-based visualization

tools for image processing (SunIP), volume rendering (Sun-

Voxel)

, 3 D graphics manipulation (SunGV), high quality

Sun Microsystems,

Inc.

81

rendering (SunART), and movie loop display (SunMovie).

Additionally, there

is

an interactive colormap editor that

is

accessible by each tool. These tools can

be

used

"as

is" by

sophisticated end users,

or

as

"application prototypes"

which developers can tailor to a specific application.

Each window-based tool

is

an independent program that Shared information

communicates with a shared parameter database program

(PMGR), which, in turn, communicates with a user inter-

face

management program (SunVIF). The user interface for

each tool can

be

changed

at

run

time, with no program-

ming. Additional programs can

be

easily added to the Sun-

Vision user interface.

SunIP and SunART

are

tools that implement the image

processing and RenderMan functions described above. The

source code for SunIP and SunART

is

provided in the form

of examples for using these libraries, along with SunVIF and

PMGR.

SunVoxel

is

an interactive tool for the generation of imag- Images from

es

from volume data.

It

consists of window-based rendering volume data

and analysis functions. With

SunVoxel,

volume data can

be

manipulated and viewed in two modes:

(1)

manipulate the entire volume, using orthogonal

or

obli-

que slicing planes where needed, and view internal

structures using ray-casting, and

(2)

extract and view

2D

slices

of

the volume data in a "light

box" mode. There

is

also

a "cloud" mode for displaying

data stored in point cloud format.

SunVoxel

supports unsigned byte data on uniform rect-

angular grids. Data filters

are

provided to convert

TAAC-1

volume data to the Sun Vision data format.

SunGV

is

used to interactively view 3D geometric data. Viewing

3D

It

can also

be

used to edit scenes that can be input to

Su-

geometric data

nART for final rendering. SunGV provides wire frame and interactively

Gouraud shaded display of polygon and patch data types.

Scenes can

be

composed of a series of objects which

are

or-

ganized hierarchically in a tree structure. Editing functions

allow the user to select,

copy,

paste, cut, and delete objects

82 Current Vendor Systems

in

Use

in the hierarchy. Objects can be transformed and assigned

attributes, such

as

color, opacity, specularity, texture,

etc.

Functions

are

also

provided for changing the viewing and

projection parameters, and for defining the lighting model,

which supports up to

32

light sources. The source code for

SunGV

is

provided

as

an example for using the XGL graph-

ics

library along with SunVIF and PMGR.

Movies

SunMovie

is

a tool for the display of image and movie

loop data. Images and sequences of images generated by oth-

er components of Sun Vision can be viewed using this tool.

7.6.4

The VX and MVX -

Sun's Visualization Accelerators

Accelerators Sun's new visualization workstations

are

powered by the

new VX and MVX accelerator boards. The VX accelerator

is

a successor to the TAAC-l, incorporating added features,

including twice the memory, double the performance, mul-

tiple windows, and a lower price.

The MVX

is

a multi-processor accelerator that can

be

added to a VX system, providing performance of more than

4-6

times the TAAC-1. Both accelerate SunVision's visual-

ization tools and libraries, and XGL, Sun's graphics library.

Applicatioh

areas

VX and MVX systems

are

for developers who require a

combination of imaging and graphics to develop

visualiza~

tion software. Target markets include medical imaging,

re-

mote sensing, earth resources, scientific visualization, and

AEC.

7.6.4.1

Features and Benefits

VX

Characteristics • Accelerates SunVision and XGL.

• High-performance Intel

i860

processor

(40

MIPS,

80

MFLOPS) for high-speed integer and floating point com-

putation required by visualization applications.

• Dual frame buffer architecture with 32-bit VX and 8-bit

GX

accelerated frame buffers

on

one board; a digital

key-

ing technique

is

used to cleanly integrate the VX win-

dows into the system display.

Sun Microsystems, Inc.

• Transparent integration of multiple VX windows into

the

GX

Open

Windows environment; four independent

colormaps

are

available for use by the multiple VX win-

dows, with a fifth colormap allocated for the GX.

• Reconfigurable VX frame buffer can

be

used to display

24-bit true color plus 8-bit alpha, or four independent

8-bit channels.

• Supports Sun's new

1280x1024

@

67Hz

format

as

well

as

the existing 1152x900 @

66

Hz

format.

• Single 9U VME board.

MVX

• Provides additional acceleration of SunVision and XGL.

• Four Intel i860 processors, offering a total of

160

MIPs

and

320

peak single-precision MFLOPS

(240

peak dou-

ble-precision MFLOPS).

• Four Mbytes of memory per processor for

fast

data and

Image

access.

• High-speed data bus for fast, smooth data and image

transfer between the MVX and VX.

• High-speed control bus for transfer of commands

be-

tween the MVX and VX, eliminating VME overhead.

Software

Included

• SunVision and XGL for the most integrated, easy-to-use

visualization environment, with the widest range of

graphics and imaging functionality, in the industry.

• Complete set of C-development tools, including compil-

er and debuggers, for application developers.

For further information, contact:

Doug Schiff

Sun Vision Product Manager

p.

O.

Box

13447

Research Triangle Park

NC

27709

U.S.A.

Tel:

919-469-8300

Email: doug.schiff@East.Sun.COM

83

84

Fig. Z14

VX

and

MVX

Overview

Current Vendor Systems in Use

Sun's new VX visualization accelerator delivers high-perfor-

mance across the full range of visualization techniques, in-

cluding image processing, volume rendering, 3D graphics,

and high-quality rendering. A multiprocessor MVX board

can be added to the VX model to boost performance to

160

MIPS and

320

peak single precision MFLOPS. The VX and

MVX accelerate Sun's XGL and Sun Vision software

li-

braries. Shown here are a Gouraud shaded teapot, a volume-

rendered air duct, and a 2D Landsat image, running in Sun's

Open

Windows environment. (Figure

7.14)

Sun Vision

is

a software toolkit for integrated, desktop vi-

sualization

on

any SPARC-based Sun Workstation with a

GX

or

VX accelerated color frame buffer. It delivers image

processing, volume rendering, interactive 3 D graphics view-

ing, and high-quality rendering in Sun's

Open

Windows en-

vironment. Figure

7.15

shows a wireframe model of a tea

set, a volume rendered

CT

head, Sun Vision's colormap edit-

ing tool, filtered images, and a photorealistic rendering of

geometric objects.

Sun Microsystems, Inc.

......

-.

-----

A new volume rendering technique called "splat",

devel-

oped by

Lee

Westover while at the University of

North

Car-

olina,

was

used to render different views of a 256x256x96

CT

scan of a human head (Figure 7.16). The four images

de-

pict the skin and bone surfaces. The transparency of the

skin

is

changed in each picture. The upper left view shows

an intermediate step

as

the data

is

being rendered from back

to front. The algorithm

is

easy to parallelize; these images

were generated

on

a Sun VX + MVX visualization accelera-

tor, with four processors doing the data shading and one

processor compositing the samples into the final image.

Sun Vision's volume rendering tool,

SunVoxel,

generates

images from 3D volumetric data. The data can be manipu-

lated and viewed in one of several modes, including slice

mode, ray-casting, surface editing mode, point cloud mode,

and light table mode. In Figure 7.17, a 256x256x96

CT

scan

of a head

is

rendered in ray-casting mode with semi-trans-

parent substances. The user can easily assign color and opac-

ity properties to ranges of data values that represent

areas

of

interest.

Fig.

7.15

SunVision

Overview

85

86

Fig.

7.16

Four Views

of

CT

Head

Fig. 7.17

Semi-transparent

CT

Head

Fig.

7.18

2 D slice from

CT

Head

Fig.

7.19

Filtered 2 D slice

from

CT

Head

Current Vendor Systems

in

Use

Sun Microsystems, Inc.

SunVision's volume rendering tool,

SunVoxel,

lets the

us-

er extract

2D

slices from the

3D

volume. In Figure

7.18

a

single slice from the 256x256x96

CT

scan of a head (shown

in Fig.7.17)

is

selected and displayed. The image can be

scaled to any size, and regions can be marked and deleted.

The image can be saved to a file for future

use.

Sun's image processing library, SunIPLib, provides soft-

ware for analysis and manipulation of images. SunIPLib

provides arithmetic operations, logical operations, spatial

filtering, Fourier domain processing, morphological opera-

tions, geometric operations, statistics, and more. The image

saved in Fig.7.18 has been processed with morphological

functions to

give

Fig.7.19.

All Sun pictures are reproduced by permission of Sun

Microsystems, Inc.

From information supplied by

Donna

McMillan, Sun Mic-

rosystems, Inc.

For further information contact:

West

US.A.

Sun Microsystems, Inc.

2550 Garcia Avenue

Mountain View

CA

94043

US.A.

Tel:

415-960-1300

Fax:

415-969-9131

United Kingdom

Sun Microsystems Europe

Bagshot Manor

Green Lane

Bagshot

Surrey GU19 5NL

UK.

East

US.A.

Sun Microsystems, Inc.

P.

O. Box 13447

Research Triangle Park

NC

27709

US.A.

Tel:

919-469-8300

Email: donna.mcmillan@

east.sun.com

87

88 Current Vendor Systems

in

Use

7.7

Sterling

Federal

Systems,

Inc.

7.7.1

FAST

(Flow

Analysis

Software

Toolkit)

Aerodynamics Visualization of computational aerodynamics requires

flexi-

applications ble, extensible, and adaptable software tools for performing

analysis tasks. Full

scale,

3

D,

unsteady, multi-zoned fluid

dynamics simulations

are

common features of typical prob-

lems

at

NASA Ames' Numerical Aerodynamic Simulation

(NAS). NAS scientists perform calculations on CRAY 2

and

CRAY-YMP

supercomputers and then graphically visu-

alize the results

on

IRIS workstations. New developments

in the scientific computing environment warrant a new

ap-

proach to the design and implementation of analysis tools

with multiple processor workstations available in the

2-8

Mflop

range.

FAST

is

a software environment for analysing

such computational fluid dynamics (CFD) data.

Modular approach The FAST environment consists of a collection of

sepa-

rate programs (modules) that run simultaneously. Using the-

se

modules, the NAS

CFD

scientist can efficiently examine

the results of numerical simulations.

FAST

functions FAST provides functions which include:

• Loading data

files,

• Performing calculations on the data,

• Constructing scenes of 3 D graphical objects that may be

animated and recorded.

Superior to While these capabilities existed to some extent, they were

earlier approaches spread across many specialized programs such

as

RIP,

SURF,

GAS and PLOT3D with overlapping functionality. These

specialized programs were problematic because the data

was

only partially compatible, and user interfaces varied widely.

The approach used in FAST solves these problems. FAST

creates an environment of compatible modules, each with

its own purpose and functionality. In addition, each module

has a consistent, easy-to-use, highly interactive user interface

(using the Panel Library developed by David Tristram, NA-

SA,

Code RNR). A programmer can add a new module to

Dynamic Graphics

Ltd.

89

FAST by making use of the data in shared memory, the

PANEL LIBRARY interface, and the NAS (input/output)

module. With these features, the FAST team has worked to

make the environment

as

extensible

as

possible.

Complex fluid dynamic simulation problems created a Enhanced

need for new visualization techniques not possible with the techniques

existing software programs. These techniques will change

as

the supercomputing environment (and hence the scientific

methods employed) evolve even further. Flexibility means Flexibility

the ability to handle a diverse range of problems. Extensi- Extensibility

bility means the ability to interact at

all

levels

of the soft-

ware hierarchy, either through existing built-in functional-

ity

or

through the implementation of custom 'plug-in'

modules. Adaptability means the ability to adapt to new Adaptability

software and hardware configurations through the use of

modular structured programming methods, a graphics

li-

brary standard, and common network communication pro-

tocols (like

UNIX

sockets) for distributed processing.

For further information contact:

Sterling Federal Systems, Inc.

1121

San

Antonio Avenue

Palo Alto,

CA

94303, U.S.A.

Tel:

415-964-9900

7.8

Dynamic Graphics

Ltd.

Interactive Volume Modeling (IVM)

is

a product from Dy-

namic Graphics, Inc., which models, displays, and interac-

tively manipulates measured property values P in three

di-

mensions located by X,

Y,

and Z.

The modeling procedure in IVM takes scattered data

points of a physical property value (e.g., porosity, tempera-

ture, salinity, chemical concentration) and calculates a three-

dimensional grid. This grid represents the modeled distribu-

tion of the property in three-dimensional space.

Multi-dimensional

modeling

Data pOints

90

Calculation

methods

3D

gridding

technique

New methods

Displaying the

information

Display

requirements

Current Vendor Systems

in

Use

The property model can be calculated by one of three

methods. The first method calculates the grid throughout

the volume defined by the input data distribution,

or

by the

user. The second method restricts the calculation laterally to

the area enclosed within a predefined polygon,

i.e.

limits in

X and

Y,

but not in Z. The third method allows the user

to specify previously calculated faulted

or

unfaulted two-di-

mensional structural surfaces

as

hard boundaries which lim-

it the modeling process in X,

Y,

and Z. This third method

enables the user to calculate, for example, a model of porosi-

ty

or

permeability within a zone while not allowing the

model to

be

influenced by measured values in underlying

or

overlying

layers.

This procedure provides a much more

real-

istic model of property variation within a zone.

The procedures for calculating property models are

based

on

a three-dimensional extension of the robust two-di-

mensional gridding techniques used by Dynamic Graphics'

Interactive Surface Modeling

(ISM)

program. These routines

utilise a variation

of

the minimum tensions surface

algo-

rithm.

A newly developed strategy

is

currently being tested.

This procedure

is

designed to enhance lateral continuity

within layers. The user

is

given control over the degree of

horizontal continuity that the gridder tried to establish.

The calculated property model can be made to conform to

the shape of either an underlying

or

overlying structural

surface. The results

so

far

have

been promising, especially

with thin laterally continuous

or

discontinuous beds.

A three-dimensional grid

is

of limited value without

dis-

play techniques that allow the user to rotate, slice, peel, and

otherwise manipulate the model in real time. This rapid in-

teractive ability

is

vitally important because no single view

can adequately reveal the complex geometric relationships

contained within any model. IVM provides these capabili-

ties.

Before any manipulation can occur the user must build

a display

file

from the three-dimensional grid. This display

file

is

in

essence

the three-dimensional equivalent of a two-

Dynamic Graphics

Ltd.

dimensional contour map. This

file

contains three-dimen-

sional isovalue surfaces drawn at selected intervals through-

out the modeled volume. These surfaces

are

displayed

as

smooth color-filled Gouraud-shaded bodies.

Once built, the user can manipulate the display

file

in

numerous

ways

to better understand the internal relation-

ships contained within the model. The model can

be

sliced

along the X, Y

or

2

axes

at

specified intervals. The model

can also

be

sliced first along

Y,

then

X,

then

2,

or

any com-

bination thereof to produce color-filled sections along

all

three

sides.

The model may

be

rotated to any combination

of user specified azimuth and inclination, or may

be

"grabb-

ed" and turned to any desired orientation.

While slicing and rotating, the user can select a particu-

lar range of isovalue surfaces to

be

displayed (or not

dis-

played). For example, the user may decide to display only

those porosity values between 6% and 9%,

or

to display

all

porosity values except those between 6% and 9%.

Alternatively the user can

use

the "chair mode" mode

display which cuts out only a piece of the model parallel to

the X,

Y,

and 2

axes.

The chair void

is

bounded by vertical

walls (X and Y

axes)

and a horizontal floor

(2

axis)

on

which

are

displayed color-filled sections of the property

dis-

tribution. The user can interactively adjust the width,

depth, and height of the void along any of the

axes.

Also,

the user can elect to display, within the chair void, a range

of isovalue surfaces. These surfaces

give

the appearance of

being extruded into the missing volume.

The user can also rapidly flicker between two different

property models which

gives

the effect of superimposing

the two models. The user loads one model and selects the

proper orientation that best displays the property distribu-

tion. Then the other model

is

loaded with the same display

parameters. With the touch of a single

key,

the user flicks

between each display

as

rapidly

or

slowly

as

desired. This

is

very useful for comparing distributions of such important

properties

as

porosity and permeability.

Understanding

aspects

of

the model

Slicing and

rotating

Chair mode

Superimposing

models

91

92 Current Vendor Systems

in

Use

Choice

of

colors

The

Color

Table

Editor

allows any

combination

of

co-

lors

to

be selected.

Analysis

IVM

contains a full

complement

of

the

extensive analy-

sis capabilities

found

within

Dynamic

Graphics' two-di-

mensional

mapping

package, ISM. .

Calculating

The

user can calculate volumes

in

a variety

of

ways. Vol-

volumes

ume

can be

determined

for

the

entire model, between isova-

lue surfaces,

within

a surface polygon, between two-dimen-

sional structural surfaces, above

and

below specific depths,

or

any

combination

of

these possibilities. A typical

problem

could

be: "Calculate

the

pore volume between 6%

and

9%

porosity

within

Zone

B inside

of

Lease J above

the

oil/water

contact

W".

Trend

grids Three-dimensional

trend

gridding

is

available for poly-

nomial

surfaces between 0

and

14th order. These

trend

grids

can be subtracted

from

property

grids

to

calculate three-di-

mensional residual surfaces. These residual grids can be used

for display

or

volumetric calculations, if desired.

Grid operations

IVM

contains three-dimensional extensions

of

nearly all

of

the

standard grid operations

found

in

two-dimensional

mapping

packages. A

property

grid can be modified

through

such mathematical operations

as

addition, subtrac-

tion,

multiplication,

or

division

by

a constant

or

another

property

grid.

Other

grid functions are also available.

Data extraction

Data

can be extracted

from

the

three-dimensional grid

in

the

form

of

individual data

points

or

as

two-dimensional

grids.

For

individual

points

the

user

must

supply

X, Y

and

Z coordinates for

the

locations at

which

values will be back

interpolated

from

the

property

model. Two-dimensional

grids can be extracted

either

along a slice

through

the

body,

or

at

node

locations defined

by

a previously calculated two-

dimensional grid.

The

extracted two-dimensional grid val-

ues can be

either

the

discrete back interpolated values

or

the

column

averages between

two

structural surfaces.

The

data

extracted from three-dimensional grids can be used for simu-

lation

purposes, additional analyses,

or

for standard two-di-

mensional

computer

mapping.

Dynamic Graphics

Ltd.

93

IVM has been successfully used to display, verify, and

ed-

Applications

it three-dimensional seismic velocity

files.

IVM

is

being

used with reservoir simulators both

as

a front-end processor

(build and verify geometric relationships) and

as

a back-end

processor (display and manipulate results from time steps).

IVM

is

being used to monitor and evaluate various thermal

and chemical enhanced oil recovery projects, and

is

being

used to better understand reservoir geometries and proper-

ties.

One

key area that

is

being studied

is

the distribution of Studies

in

permeability within a reservoir. Within certain reservoirs the environment

tilted permeability barriers have been well defined and stud-

ied. This knowledge can lead to changes in established drill-

ing programs, and hopefully, significantly increased oil pro-

duction. IVM

is

also used to monitor distribution and

movement of pollutants within aquifers, and to assist in

cle-

an-up efforts. Concentrations of ozone within the Earth's

atmosphere are being studied with IVM,

as

well

as

salinity

and temperature distributions within oceans.

Distribution

and

concentration

lev-

els

of

PCE organic

solvent contamina-

tion below a

WW11

airbase site.

Data from test

boreholes.

Fig. Z20

PCE Plume

94

IVM modeled and

displayed tempera-

ture data collected

for

11

years

to

re-

veal the higher

temperature water

from the Gulf

Stream does not

pass into the Arc-

tic Ocean.

Fig.

Z21

Fram Strait -

Connecting the

Atlantic and Arctic

Oceans

IVM was used

to

depict the simulat-

ed

concentration

and dispersion

of

pollutants in a

plume from a mu-

nicipal garbage in-

cinerator stack in

Minneapolis.

Fig.

Z22

Simulated Plume

from Smoke Stack

Current Vendor Systems

in

Use

Dynamic Graphics

Ltd.

Geoscientists can now study and analyse three-dimen-

sional geometric and property relationships in ways that

previously have been impossible with two-dimensional

mapping techniques. IVM's modeling, display, manipula-

tion, and analysis capabilities are indeed applicable to a large

number

of

geoscience problems.

95

Summary

Advantages

of

3D

IVM was used

to

model and display

porosity data from

borehole readings

within discrete

lithologic units

which were then

combined into a

single model.

This

provided a unique

visualization

of

po-

rosity distribution

and enabled im-

proved volumetric

calculations and

recovery tech-

niques

to

be em-

ployed.

Fig.

7.23

Porosity Modeling

for Oil Recovery

Porosity variations

in a particular res-

ervoir.

Isoporosity

shells are peeled

back

to

reveal

those areas with

porosity greater

then 6%.

Fig.

7.24

Porosity Variations

96 Current Vendor Systems

in

Use

From information supplied by Peter Irwin, Dynamic

Graphics Ltd.

For further information contact:

US.A.

Dynamic Graphics, Inc.

1015

Atlantic Avenue

Alameda,

CA

94501, US.A.

Tel:

415-522-0700

Fax: 415-522-5670

7.9

Spyglass,

Inc.

7.9.1

Spyglass

Transform

United Kingdom

Dynamic Graphics Ltd.

Addison

Wesley

Building

Finchampstead Road

Wokingham

Berks RG

11

2NZ,

UK.

Tel:

0734-774755

Fax: 0734-774721

Analysis and Transform

is

a comprehensive tool for analysis and visuali-

visualization zation of two-dimensional data

on

the Macintosh. It can

generate contour plots, surface plots, vector plots, line

graphs, polar images, animations, overlays, and raster images

(assigning colors to data values).

Data input The Import command reads and converts

2D

and

3D

HDF

datasets,

2D

and

3D

generic datasets (byte, integer,

long integer, float),

HDF

image

files,

PICT

files,

TIFF

files,

FITS

files,

2 D ASCII data, and

X-Y

column data not already

in array form. Transform reads 3 D data one slice at a time.

Output

of

results Every image, plot,

or

dataset can be printed to any color

or

black-and-white Postscript printer,

or

any Macintosh-

compatible color printer. Exporting graphics to other Mac-

intosh applications,

or

to produce

35

mm

slides

is

easy.

Data

and images created in Transform can be exported

as

PICT

or

HDF

files,

or

via the Clipboard.

Spyglass, Inc.

7.9.2 Spyglass Dicer

Dicer

is

a comprehensive tool for visualizing volumetric

da-

Visualizing

ta

on

the Macintosh.

It

can perform 3 D blocks, slices in volume data

three planes, 3 D cubes and cutouts, variable orientation,

da-

ta re-sampling, variable color maps, and animation

se-

quences.

Dicer reads 3 D

HDF

(float and byte), netCDF, and

ge-

Data input

neric (byte, short and long integer, single- and double-preci-

sion floating point) file formats directly. A utility can con-

vert and import folders of

2D

files

in ASCII byte, integer,

and floating-point formats, and construct 3 D datasets from

the

2D

files.

The utility also imports

3D

ASCII data

files,

as

well

as

PICT, TIFF, FITS, and

HDF

image

files.

Dicer offers a menu of over

20

color tables,

or

user tables Color options

can be imported. Interactive tools enable any color to be

made transparent

to

enable the user to 'see through' the vol-

ume it previously occupied,

or

to substitute colors in select-

ed regions.

Any configuration created

on

screen can be saved with Output

or

without corresponding data, and snapshot images can be

saved alone

as

PICT

or

HDF

image

files.

Dicer images can

be exported

as

PICT

files

using the Macintosh Clipboard,

and 2 D slices can be exported to Spyglass Transform for fur-

ther manipulation. Data sets can be re-sampled and saved

as

either

HDF

or

netCDF. Any image created

on

screen can

be printed to color (or black and white) Postscript printers,

or

to Macintosh-compatible color printers.

Dicer also offers two

ways

to create and

save

animations. Animations

After defining parameters, the user can generate and

save

se-

quences

of

3 D frames,

or

a sequence of 2 D slices from 3 D

frames created in Dicer. The sequenced images are saved in

folders and can be viewed

as

animations using Spyglass

View.

Information courtesy of Digital Studio.

97

98 Current Vendor Systems

in

Use

For further information contact:

US.A.

Spyglass,

Inc.

701

Devonshire Drive,

C-17

Champaign

IL 61820,

US.A.

Tel:

217-355-1665

Fax:

217-398-0413

United Kingdom

Spyglass,

clo

Digital Studio

Clifton

Mews

Saffron Walden

Essex

CB10

lEE,

UK.

Tel:

0799-513773

Fax:

0799-513454

7.10

LightWork

Design

Ltd.

High-quality LightWorks

is

a powerful new image generation system

images from LightWork Design Ltd. that enables users of modeling

software to create high-quality images

of

their models,

showing accurate surface finishes and lighting effects. It

is

optimised for interactive operation in computer-aided

de-

sign environments.

Natural Images of photographic quality are produced by simulat-

phenomena ing natural phenomena such

as

reflective surfaces, transpar-

ency and shadows from any number of light sources. Arbi-

trarily complex material characteristics can be defined to

create realistic finishes such

as

marble, wood, brick, chrome

and steel. A wide range of geometric modeling primitives

are supported

on

to which scanned images can

be

mapped.

Applications LightWorks can be used in many application

areas.

Ex-

amples

are

visualization for product designers; visualizing

simulation data for scientists and engineers; building

de-

signs for architects; accurate simulations of color and light-

ing schemes for interior designers; print and animation for

graphic designers.

User interface The system has an interactive mouse and window-based

scene editor module which presents an easy-to-use yet pow-

erful user interface with which the visual characteristics of

a model can be controlled. Incremental rendering tech-

niques enable many color and lighting combinations to

be

compared in a short time.

LightWork Design Ltd.

This engineering component shows different kinds of

metal surfaces -the machined chrome, the cast red, and the

threads. The "bumps"

on

the background and also

on

the

red surface are done by displacements. Lighting and shad-

ows are also present.

An

example of an architectural application. This shows

different finishes, soft shadows, and a perspective

view.

Fig. Z25

Engineering

Component

Fig. Z26

Architectural

Application

99

100

Representing

3D

shapes

3D

modeling

Current Vendor Systems

in

Use

Written in C, LightWorks has been designed to

be

easily

ported across a wide variety of computer platforms includ-

ing

UNIX

workstations, MS-DOS PCs, and cost-effective

parallel processing computers. LightWorks

is

supported on

Sun-4/SPARC, IBM RISC System/6000, Silicon Graphics

IRIS/4D, Hewlett Packard 9000 300/400/700/800, Sony

NEWS, and

80386

and

80486

IBM

PC

compatibles running

MS-DOS with Phar Lap's 386/Dos-Extender

or

Microsoft's

Windows 3.0.

Information supplied by

Dr

R. Gordon Oliver,

LightWork Design Ltd.

For further information contact:

Lightwork Design Ltd.

Sheffield Science Park

Arundel Street

Sheffield S 1 2NS,

U.

K.

Tel:

0742-724126

Fax: 0742-720379

7.11

Ricoh

Company

Ltd.

Use

of

Visualization

in

Modeling

and

CAD/CAM

Solid modelers, designed

to

represent 3 D shapes

as

solids,

have become essential tools in computer-aided design and

manufacturing (CAD/CAM) systems. The benefits of solid

modelers vary from designing to molding, and from struc-

tural analysis to robot simulations. Unlike surface modelers,

however, existing solid modelers

have

had limitations in the

representation of complex surfaces.

DESIGNBASE

is

a 3 D solid modeling system developed

by Ricoh Company Ltd. This

is

designed to represent com-

plex free-form surfaces in

UNIX

workstation environ-

ments. Using a surface representation method called the

"Gregory patch", DESIGNBASE enables free-form surfaces

to

be

smoothly connected and locally operated.

Ricoh Company

Ltd.

DESIGNBASE provides various surface representation

methods: the Gregory patch, the rational boundary Grego-

ry

patch, natural quadric surfaces (spheres, cylinders, cones),

and n-th degree rational Bezier patches. Featured with these

surface representation methods and translation libraries,

DESIGNBASE

is

capable of bi-directional data exchange

with other

CAD/CAM

systems. DESIGNBASE translates

its surface data to non-uniform rational B-splines (NURBS),

the

de

facto standard surface representation method, and

sends the data through Initial Graphic Exchange Specifica-

tion (IGES) to other

CAD/CAM

systems.

The Gregory patch

is

the more suitable surface represen-

tation method for the smooth connection of free-form sur-

faces

compared to the Bezier patch. A bicubic Gregory

patch, which

is

defined by

20

control points,

is

an extension

of the bicubic Bezier patch, which has

16

control points.

101

Fig. Z27

Scene produced

by

DESIGNBASE

Surface

representation

Connecting

surfaces smoothly

102

Fig.

Z28

Free-form Surface

represented

by

Gregory Patch

Fig.

Z29

Irregular Mesh

represented

by

Rational Boundary

Gregory

Patch-1

Fig.

Z30

Irregular Mesh

represented

by

Rational Boundary

Gregory

Patch-2

Current Vendor Systems

in

Use

The extra 4 internal control points of the Gregory patch are

used in the smooth connection of surfaces and enable the

interpolation of irregular meshes.

Figure 7.29 shows the control points of the rational

boundary Gregory patch. A six-sided irregular mesh

is

inter-

polated by 6 RBG patches.

In Figure 7.30 the contour lines show the smoothness of

this surface representation method.

Fillets Filleting

is

a crucial capability for

CAD

software for

de-

signing products such

as

automobile engines

or

electric

ap-

pliances.

Ricoh Company

Ltd.

103

Boolean operations (union, difference, and intersection) Boolean

are critical in the representation

of

complex shapes. Howev- operations

er, most solid modeling systems approximate the shape of

the free-form surfaces with facets when executing these op-

erations.

When natural quadric surfaces are intersecting,

DESIGNBASE uses a special library for high-speed surface-

to-surface intersection calculations.

This shows an example of a body generated by applying

Boolean operations. The geometric accuracy

of

the resultant

solid based

on

these operations

is

within the margin of

10-6

on

intersection vertices and 10-3

on

intersection

curves.

To

support the trial and error implicit in the design pro-

cess,

DESIGNBASE provides high-speed Undo, Redo and

Re-execution commands. Each command of DESIGNBASE

is

subdivided internally into primitive operations, and each

primitive operation

has

a reverse operation. By executing

the reverse operations, previous shapes can be generated by

tracing the tree-type history backwards (Undo) and forward

(Redo). In addition, shapes input previously can be modi-

fied easily by giving different parameter values (Re-execu-

tion); this facility

is

useful for designing analogous shapes.

Fig.

7.31

Engineering

Component

created by

applying Boolean

Operations

Trial

and

error

in the design

process

104

Current Vendor Systems

in

Use

Summary Computer-aided design (CAD) began with

2D

systems

which replaced drafting instruments. However, total

CAD/CAM

or

computer-aided engineering (CAE) systems

require the facility to process 3 D data, and it must be possi-

ble to freely exchange this data between systems.

Solid modeling systems

are

assuming a more significant

role

as

industry introduces

CAD/CAM/CAE

systems.

High powered workstations enable such systems to perform

efficiently and also cost-effectively.

Applications DESIGNBASE has been applied to the automobile and

electricity industries

as

the basis of

CAD/CAM/CAE

re-

quirements.

It

has also been used for rendering, stereo

li-

thography, and pre-processing for the production of finite

element meshes.

Reference

Chiyokura H.: Solid Modeling with DESIGNBASE: Theory

and Implementation. Addison

Wesley,

Reading, MA,

1988.

Information supplied by

T.

Ito, A. O'Neill, and

H.

Toriya,

Ricoh Company Ltd.

For further information contact:

Ricoh Company Ltd.

1-1-17

Koishikawa

Bunkyo-ku

Tokyo

112

Japan

Tel:

81-3-3815-7261

Fax:

81-3-3818-0348

Vital Images, Inc.

7.12

Vital Images,

Inc.

Vital Images began developing software

as

part of a research

VoxelView

project at Maharishi International University to visualize

la-

ser-scan confocal microscope data of living nerve cells. This

research

was

funded by Iowa Department of Economic De-

velopment and the National Science Foundation.

VoxelView

is

a general purpose, high-performance vol-

ume rendering package. Data values of voxels can be map-

ped to corresponding opacity values. This enables the user

to view faint

or

small details inside the volume. It

is

also

possible to do thresholding to remove voxels,

or

redistribute

voxel values from one range to another. Sequences

of

render-

ings can be stored and then played back in real time. The

following features

are

also included:

• graphical data

base

system,

• full surface shading with lighting,

• user control of animation parameters,

• gradient operations to extract and selectively display

nested surfaces within the volume,

• autoconfiguring for multiprocessor systems.

VoxelLab

is

an entry-level version of VoxelView and

is

avail-

VoxelLab

able

on

the Silicon Graphics workstation.

It

enables begin-

ning users to appreciate the power and potential of volume

rendering systems.

For further information contact:

Vital Images, Inc.

P.O. Box

551

Fairfield

10

52556

U.S.A.

Tel:

515-472-7726

Fax:

515-472-1661

105

Tool

for

research

and

development

Chapter

8

Current

Public

Domain

Systems

in

Use

Readers are recommended to read Chapter 4 for a classifica-

tion and categorisation of scientific visualization systems

before reading this present chapter. Chapter 4 sets out the

overall framework into which the products outlined in

Chapters 7 and 8 fit.

In

addition, readers should note the points

on

public do-

main systems that are set out in Chapter 4 (Section 4.4).

Contact addresses

are

provided for each product at the

end of each section.

8.1

Khoros

is

an open environment for data processing, visual-

ization, and software development. This summary describes

how the Khoros software system can be utilized

as

a founda-

tion

or

platform to improve productivity and promote soft-

ware reuse in data processing applications. First, a high-level

description of Khoros

is

given, then the current status of

Khoros

is

discussed.

8.1.1

Overview

The Khoros system integrates multiple user interface modes,

code generators, instructional aids, data visualization, and

information processing. The result

is

a comprehensive tool

for computational research and development. The Khoros

infrastructure consists of

five

major subsystems:

• a general visual language,

• a user-interface development system (UIDS),

Khoros

• an interoperable data exchange format

(vif£),

• application-specific data processing libraries,

• interactive data display/manipulation programs and a

vi-

sualization toolkit.

The software structure that embodies this system provides Portable

and

for extensibility and portability, and allows for easy tailor- extensible

ing to target specific application domains and processing

en-

vironments. Khoros

is

a successful example of how research

programming, end-user applications programming, infor-

mation processing, data visualization, instruction, docu-

mentation, and maintenance can

be

integrated to build a

state-of-the-art software environment.

8.1.2

Subsystem Component Descriptions

a)

X Windows Applications

The interactive graphical user interface programs

are

based

on

MIT

XllR4

and the Athena widgets. They

are

all

de-

signed to

have

a simple and consistent look and

feel.

Program Name

cantata

editimage

animate

xprism2 and xprism3

vlewlmage

warplmage

concert

Description

extensible visual programming

language

interactive image display and

manipulation

interactive image sequence

dis-

play

comprehensive

2D

and

3D

plotting packages

surface visualization (imagery

draped over elevation data)

image registration and warping

distributed user interface con-

troller

107

108 Current Public Domain Systems

in

Use

b)

Visual Language

Dataflow graphs The visual language of Khoros, cantata,

is

a graphically

ex-

pressed, dataflow-oriented language. The user builds a canta-

ta application program by connecting processing nodes

to

form a dataflow graph. Nodes are selected from an applica-

tion specific library of routines created using the Khoros

UIDS, and may have arbitrary granularity, from fine to large

grain. Control nodes and a parser extend the functionality

of the underlying data flow methodology. Visual proce-

dures, representing a hierarchy of subgraphs, add structure

to the visual language and help

to

manage the complexity

often associated with visual programming. A dynamic

exe-

cution scheduler allows the user to interactively execute the

entire flow graph across a heterogeneous computer network.

The execution can be set to either a demand-driven

or

data-

driven model depending on the application and desired level

of interactivity.

Many applications Cantata has been targeted at a variety of application do-

mains: a visual language interface has been completed for

the LINPACK/EISPACK libraries, an image processing

li-

brary, a digital signal processing library and a remote sens-

ing/geographical information system. The design of cantata

promotes code reuse and modular design of libraries.

c)

User Interface Development System

The Khoros system combines interactive graphical user in-

terface specification/editing and code generation to

give

the

user a programmer's assistant. This UIDS can be used to

cre-

ate general X Windows applications

or

to extend cantata.

Dialog The central component of the UIDS

is

a high-level user

interface specification that represents a formal description

of the dialog between the user and the application, indepen-

dent of the user interface mode. The specification

is

used

di-

rectly to generate the code for either a graphical

or

com-

mand-line user interface. When the user interface specifica-

tion

is

combined with a formal Khoros program specifica-

Khoros

tion, the entire application (documentation and code) can

be maintained via a set

of

automated source configuration

tools.

The software development tools that are provided allow an

end-user to act

as

a developer to extend the system.

Program Name

preVIew

composer

conductor

ghostwriter

kinstall

Description

graphical user interface display tool

interactive graphical user interface

editor

code generation tool for a graphical

user interface

code generation tool for a command

line user interface

source configuration and manage-

ment tool

The user interfaces created with the Khoros UIDS

all

utilize User interfaces

the same layered libraries. This provides features common to

all applications, such

as:

• journal recording and playback,

• distributed user interface,

• reconfigurability without recompiling,

• consistency

of

use.

Perhaps the most powerful and innovative item in the list

is

the distributed user interface

or

groupware capability. All

of

the graphical user interfaces created within the Khoros

sys-

tem allow for multiple user interfaces to be running

on

dif-

ferent machines. This allows a group

of

researchers (either

as

master and slaves

or

as

all

masters) to simultaneously in-

teract across a network using the same data and application

software. This groupware capability motivates users to share

resources.

109

110

Current Public Domain Systems

in

Use

The

UillS

also promotes consistent structured program-

ming methodology and styles

as

well

as

code reuse. The ho-

pe

is

that

as

Khoros

evolves,

many reusable libraries can

be

provided in various languages for various applications that

have

a consistent and powerful user interface.

d)

Interoperable Data Exchange

Data formats The Khoros data structure

or

visualization model supports

general geometric objects, multidimensional data, and a ro-

bust mapping scheme. Storage type conversion between

dif-

ferent architecture platforms

is

automatically performed by

the read/write utilities,

i.e.,

DEC

VAX

floating-point data

is

automatically converted to IEEE format if read

on

a

SUN

computer. The consistent

use

of

the Khoros data structure

promotes an algorithm library that can

be

used in many

dis-

ciplines and supports data sharing.

Standard formats

It

is

important to state that

as

Khoros expands, there will

be

a need to support a set of "standard"

file

formats. Cur-

rently, Khoros provides for data interchange with other

sys-

tems via

file

format converters. Khoros supports the follow-

ing

file

formats: TIFF, pbm, BIG, DEM, DLG,

ELAS,

FITS,

Matlab, sun raster, TGA, and xbm.

e)

Data Processing Libraries

Functions Khoros includes a library of programs that can operate on

point data, one-dimensional data, two-dimensional data and

multi band

or

vector data. These operators

are

designed to

be

polymorphic,

i.e.,

they function on bit, byte, short, integer,

float and complex data types. This also implies that the

functions will operate differently depending

on

the dimen-

sionality of the data.

Interface levels There

are

two interface

levels

defined in the library

levels

functions in Khoros; the program

or

process interface and

the function

call

or

procedure interface. The program inter-

face

is

determined completely by the high-level user inter-

face

specification described

above.

The procedure interface

is

currently not

as

well defined, but allows the procedures

Khoros

111

(functions) to

be

easily combined into a single program.

Vi-

sual programs

are

built by executing a set of programs using

the program interface.

The library contains over

260

programs, in the following Library programs

categories: arithmetic, classification, color conversion, data

conversion,

file

format conversion, feature extraction,

fre-

quency filtering, spatial filtering, morphology filtering,

geo-

metric manipulation, histogram manipulation, statistics,

signal generation, linear operations, segmentation, spectral

estimation, subregion, and transforms.

f)

Visualization

Toolkit

A visualization tool

is

of limited utility if it cannot

be

modi- Adapt

and

extend

fied to view and process data in a new user-specified

way.

This will only

be

possible if the user can modify and extend

the software system. A scientist should not

be

required to

modify a

large

C program to do this; a high-level language

or

specification should

be

provided

as

in the Khoros UIDS.

Khoros includes generic interactive X Windows applica- High-level tools

tions for image (2D data) visualization and three-dimen-

sional surface rendering. But more important

are

the high-

level

graphics and display libraries that

are

accessible from

the UIDS to build custom visualization programs. The

fol-

lowing libraries

act

as

a visualization toolkit that

is

layered

on top of Xlib, Xtk and the Athena Widgets.

Library Name

forms

utils

graphics

display

Description

hierarchical user interface based

on

forms and panes

browsers, error reporting, pop-up

lists, and help

2D

and

3D

drawing library; supports

X, Postscript,

HPGL

image display and editing; manages

color allocation and editing

112 Current Public Domain Systems

in

Use

8.1.3

Current Status

of

Khoros

Many users The Khoros user community

is

applying Khoros primarily

to image and signal processing research and development

projects. In addition, sites

are

retargeting Khoros to applica-

tion domains such

as,

three-dimensional volume rendering,

relational databases, and telecommunications. Khoros

is

be-

ing used

as

a teaching tool at several universities.

Help The documentation for the system

is

a combination of

on-line help and printed manuals. The manual comprises

2200

pages

in three separate volumes: User's Manual, Pro-

grammer's Manual, and Reference Guide. Also, journal

playback

files

are

provided to

give

the new user "live" dem-

onstrations of the various applications.

Platforms Khoros currently runs on SUN, DEC, APOLLO/HP,

SGI, IBM, NeXT, and CRAY platforms and there

are

port-

ing efforts for 386/486 and Apple platforms. Khoros

is

now

available via anonymous ftp at no charge,

or

a tape and

printed documentation can

be

ordered for $250.00. Hun-

dreds of Khoros users participate in a mail user's group,

email khoros-request@chama.eece.unm.edu for more infor-

mation. The software can

be

obtained on tape by mail

or

by ftp. Email khoros-request for an order form,

or

mail the

request to the address below. Orders can also

be

faxed to

505-277-1439.

The ftp address for the software IS pprg.eece.unm.edu

(129.24.24.10).

Login: anonymous (or ftp)

Password:

user~ame@machine

cd

/pub/khoros

For users in Germany, the ftp

is

ftp.uni-erlangen.de

Login: anonymous

Password:

user~ame@machine

cd /cyber/khoros

The documentation can also

be

printed by usmg the

"prnmanual" program.

Khoros

In Figure 8.1, the cantata visual language

is

being used

for two simple applications: blending two images and then

pseudo-coloring (top), enhancing a magnetic resonance im-

age

of a human spine (bottom).

In

Figure

8.2

the cantata visual language program

is

used

to synthesize, filter, and display a one-dimensional signal.

C

1\1\~""""'

___

"'_

__

jr"

...

-_-u.u---'111

...

" II

CO"

II

..

· DO

II

I

~~-

II

II II

..

'V"

II .. - I

Fig. 8.1

Cantata Visual

Language

113

Fig. 8.2

Filter

and

Display

114

Fig. 8.3

3D

Capabilities

Current Public Domain Systems

in

Use

The left xprism2 plots shows the signal before and after

fil-

tering and the right xprism2 plot shows the filter response.

The three-dimensional scientific data plotting package

xprism3 can be used to interactively visualize surfaces, con-

tours. and meshes (Fie:ure 8.3).

The Khoros system can

be

used to integrate satellite im-

agery with ground elevation and map data to produce a

three-dimensional scene of the earth's surface. The viewima-

ge

program

is

used to interactively change the perspective

view of the surface, and the animate application can

be

used

to create a 'fly-by' sequence (Figure 8.4).

Khoros

Warp image

is

an interactive application for registering

and then warping images to produce integrated data sets

(Figure

8.5).

Fig. 8.4

Integration

Methods

Fig. 8.5

Image Warping

115

116

Fig. 8.6

Noise

Removal

Current Public Domain Systems

in

Use

The cantata visual language

is

being used to remove shot

noise from an image of the moon. This

is

done by using a

count loop containing a median filter (Figure

8.6).

~

-...

(

'.14~"'_---"''''';:==---.

.-_----,

~

......

I I

~:".

'''I'

'"

,

..

I

0"

1'1. I . -

.,,.,,.

I·

.•

an I

Contributed

by

Dr.

John Rasure, University

of

New

Mexico.

For further information contact:

The Khoros Group

Department of Electrical and Computer Engineering

University of New Mexico

Albuquerque

NM

87131

U.S.A.

Fax:

505-277-1439.

Email queries: khoros-request@chama.eece.unm.edu

apE: A Dataflow Toolkit for Scientific Visualization

8.2

apE:

A

Dataflow

Toolkit

for

Scientific

Visualization

In

1984,

the Ohio State University competed with institu-

tions across the United States to host a National Supercom-

puter Center. While its proposal

was

highly ranked, Ohio

State lost its bid for National Science Foundation funding

to obtain a center. However, the highly motivated group of

computational chemists who spearheaded Ohio State's

chemistry efforts then received help from the state. In

1987,

the legislature appropriated funds for the Ohio Board of

Re-

gents' supercomputer initiative to create a center serving

ac-

ademic and industrial users in the state of Ohio. In June

1987,

a Cray

X-MP

was

installed

at

the Ohio Supercompu-

ter Center, followed by a Cray

Y-MP

in June

1989.

One

of the early supporters of the

Ohio

Supercomputer

Center

was

Professor Charles Csuri, a pioneer in the field

of

computer graphics and Director of the Advanced Com-

puting Center for the Arts and Design

at

Ohio

State.

He

foresaw the rise of scientific visualization in the early

1980s

and built a significant graphics research component into the

base

of the then-fledgling Ohio Supercomputer Center.

Thus in late

1987,

the newly formed Ohio Supercomputer

Graphics Project set out to design an effective software

sys-

tem for visualization, apE.

Rather than dictate to the scientific community a partic-

ular methodology, extensive time

was

spent with potential

users to understand the

real

needs

of

scientific research. The

apE group listened to users of

all

kinds of current graphics

software and hardware, discovered the realities of fixed bud-

gets that permit only modest hardware acquisition and the

effects of slow network connections

on

high speed comput-

ing. In short, they tried to

face

the

real

world, and to design

and build a product that would outlast current hardware

platforms while providing a high degree of flexibility to to-

day's users.

Background

Computational

chemistry

117

Developments

in

computer graphics

Tools

for

supercomputers

What do potential

users really need

from computers?

Hardware

independence

and

flexibility

118

Current Public Domain Systems

in

Use

Dataflow model Efforts in the mid-1980s at the Computer Graphics

Re-

search Group (now known

as

the Advanced Computing

Center for the Arts and Design) at the

Ohio

State Universi-

ty

led to selection of a dataflow model for the apE system.

Steps in A dataflow system maps very well to the general steps

fol-

understanding lowed in visualizing scientific data. Most researchers follow

scientific data a five-step process, beginning with a computational

or

ex-

perimental simulation, and concluding with interpretation.

Intermediate steps include preparation, mapping, and ren-

dering (the preparation stage

is

occasionally omitted

or

merged with the map stage). Ideally, the results of interpre-

Feedback tation can be

fed

back into the original experiment

or

simu-

lation. This kind of feedback

is

known

as

steering, and has

been used with great success in some limited applications.

New

software technology, beyond the reach of apE,

is

need-

ed

to

investigate the steering issue completely.

Utilizing The dataflow abstraction

is

ideal for remote execution

networked and parallel operation.

Network

computing environments

computers are commonplace, and distributed computation

is

a require-

ment for maximum resource utilization. Dataflow systems

can naturally distribute each execution element

on

a

sepa-

rate machine

or

processor. The apE dataflow

is

data-driven

and not demand-driven, which offers the benefit

of

parallel

execution for time-dependent

or

multi-frame data sets. Each

element operates not under the control of some central

au-

thority

but instead only

as

input data, frame boundaries,

Local

and

and other local conditions dictate. Successive elements in a

centralized visualization pipeline can

be

operating

on

separate groups of

computers data, all in concert, without any additional user interaction.

This notion of distributed computing maps well to the reali-

ties encountered among researchers,

as

they are often located

at distant sites, far from their supercomputing resource, but

may have some local computing power available.

Dataflow language Once the group

was

firmly committed to the data flow

concept, they examined the requirements for a data lan-

guage.

Incompatible binary formats are common in a heter-

ogeneous network environment. While transmission of data

as

text

files

would mitigate this problem, the operational

apE: A Dataflow Toolkit for Scientific Visualization

overhead for such transmission

was

out of the question in

an interactive system. Thus a dataflow language

was

born,

designed to represent not only common data elements from

the scientific domain (such

as

grids and variables) but

also

common graphical forms, such

as

objects, images, and

ge-

ometries.

There

is

a great difference between writing a small piece

of personal software and constructing a

large

software envi-

ronment. Additional complexity appears when consider-

ation

is

given to machine and device independence and por-

tability. Still more demands come from software which

is

to

be distributed not

as

a closed system but

as

source code, to

be

modified, improved, and extended

as

required. The

group tried to build

as

portable a graphics environment

as

is

possible using existing software and hardware technology.

Clearly, the analysis of this requirement results in a different

answer today than it did in

1987

when this project

was

be-

gun. However, many of the design decisions faced then

are

also faced today by large

scale

developers. These decisions

can

be

summarized

as

three primary turning points: the

se-

lection of an operating system, the selection of a graphics

li-

brary, and the selection of a user interface.

The apE system

is

built on the

UNIX

platform. The

mid-1980s

saw

an explosive growth in a new breed of per-

sonal computer known

as

the "workstation". Performance,

power, and software resources that were once only part of

large mainframe systems rapidly became available on the

desktop, and the

UNIX

operating system quickly became

the

de

facto standard operating system. Manufacturers that

did not respond to this trend

saw

their

sales

diminish.

The group chose not to embed any graphics library in

the

basis

of the system. They were criticized for not build-

ing their software upon a graphics software layer such

as

CORE,

GKS,

PHIGS, PHIGS+, PEX,

or

others. In late

1987,

when this decision

was

made, the number

of

compet-

ing standards

was

large, and no clear winner had emerged.

None of the standards available then were really sufficient

for scientific visualization. Constructing the software on

119

Representing data

elements

and

graphical forms

Software

engineering

principles

Extensibility

by users

Portability

Design decisions

UNIX platform

De facto standard

Independent

of

graphics libraries

Proliferating

standards

Visualization

requirements

120

Current Public Domain Systems

in

Use

such a platform would be a tacit endorsement of one of the-

Add-on costs for

se

standards and would require users to obtain the necessary

graphics libraries licenses to actually program within apE. Most workstation

vendors do not currently ship a PHIGS product, for

e~am

pIe,

as

no-cost, bundled software with their systems,

so

addi-

tional cost

is

incurred in purchasing, installing, and main-

taining a graphics library in addition to apE. All that

is

needed to run apE

is

apE.

Independent

of

apE incorporates a new interface layer on top of existing

window systems "standards". The group chose not to build upon one of the

existing window systems. Clearly today the only "standard"

window system

is

the X Window System. However, in

1987,

a number

of

alternative threatened to steal the glory from

X. Despite the claims, the intense battle between such com-

Motif and peting higher-level standards

as

Motif and Openlook will

Openlook continue this uncertainty.

Developing With source code control, interface, and (lack of) a

the software graphics library in hand, the group

was

ready to implement

the application software. This phase of the development

was

divided into three logical elements: the construction of the

libraries, the construction of the individual dataflow

ele-

User interface ments (or modules), and finally the construction of the

Look

and

feel tools and interface that would comprise the look and

feel

of

apE to the

average

user. The library implementation

was

done in phases, with the UNIX-level hiding functions com-

pleted first. The data language,

usel"_

interface library, and

graphics functions were done in preliminary test forms pri-

or

to full implementation. The resulting test software

was

released

(as

version

1.1)

and used to help motivate the full

implementation of apE version 2.0.

First version Version

1.1

of apE, released in early

1989,

had many of

the important aspects of the dataflow design, but lacked the

full implementation needed to make a truly useful software

package. Many changes occurred in the

18

months from the

public release of apE

1.1

to the first glimpses of apE 2.0.

Data language The data language "flow" that had been developed for

extensions apE

was

enhanced, extended, and rechristened flux. Specifi-

cally designed to deal with

large

amounts of data in user-de-

apE: A Dataflow Toolkit for Scientific Visualization

signed grouping, flux

is

a powerful information manage-

ment tool. All data entities, from images to variables to pipe-

line descriptions to icons, are represented in the flux.

The generic user interface, first presented in apE 1.1,

was

expanded and renamed

face.

The

face

libraries provide a

complete, window-system-independent interface for pro-

gram development.

Face

elements include most of the stan-

dard interface items, such

as

buttons, menus, sliders, scan-

ners, and text

entry

boxes.

On

top of this layer more com-

plex elements are provided

as

well, such

as

alerts, browsers

(for selecting a text element from a list), and collectors (for

selecting several text elements from a list). Face provides a

generic, application-based interface for interactive tool

de-

sign which allows a single application to execute under Sun-

View, X, and GL without significant source code changes.

The operational tools provided in the first release were

also significantly reworked. The pipeline construction tool

has been reworked to increase interactivity and to handle

different connection methods between the elements (apE

1.1

used

UNIX

pipes to connect the dataflow elements; apE

2.0 uses both

UNIX

pipes and sockets for connections). A

central console provides an outlet for error

messages

and

ac-

cess

to documentation.

An

interactive image viewer allows

manipulation of single and multiple images and real-time

"playback" of image sequences. A geometry viewer allows

interactive viewing of geometries.

While apE

1.1

was

limited to nearly-linear pipelines, apE

2.0

is

designed to allow complex pipeline configurations, in-

cluding multi-input and multi-output and cyclic graphics.

This cyclic capability provides apE 2.0 users with the ability

to investigate connections between graphics and supercom-

puter simulations, and to attempt to "steer" a simulation

through visual feedback. These additions are all natural

ex-

tensions of apE

1.1.

Finally, the filters/modules have been extended to in-

clude three-dimensional elements

as

well

as

the traditional

two-dimensional elements found in the first release of apE.

121

Flux

Advances

in

the

user interface

Face elements

Interworking with

SunView

X,

GL

Operational tools

Pipelines

Errors

and

help

Playback

of

sequences

Complex pipelines

Supercomputer

simulations

3 D element filters

122

Visualization

techniques

Volume methods

Extensions

Distribution policy

Criteria

for success

Release

of

code

Modifications

Changes

Understanding

Source code

availability

Current Public Domain Systems

in

Use

Visualization techniques include carpet and contour plots,

surface detection, terrain generation, and

all

forms of ren-

dering from scanline polygonal techniques to ray tracing. A

volumetric rendering system based on methods methods

de-

veloped by

Levoy

is

also included. Particle tracing,

advec-

tion, and surface feature detection (such

as

stream lines)

are

also included. In addition, full prototypes

are

provided to

al-

low extension of the system by the addition

of

new filters,

data types, tools, and interface elements.

One

of the real

keys

to the

success

of the apE software

effort

has

been the distribution policy. While a corporation

must

be

concerned about profits, competition, market anal-

ysis, and other factors, the group concentrated solely on

providing the best tools for the research community, know-

ing that

success

would

be

judged by user productivity, not

the corporate bottom line. The best and only result hoped

for

was

widespread

usage

and increased productivity among

Ohio's researchers.

The first version of apE

was

released in binary form on-

ly.

For many users this

was

insufficient, because it prevented

them from fully utilizing the software. First, many people

needed to modify the code to suit particular needs

or

de-

mands in a particular application

or

field of interest. Some

needed to make changes to suit local equipment

or

configu-

rations. Finally, for many, not being able to

see

the source

code caused a lack of confidence in the final results. Even

if the code

is

not modified, it

is

of great value to examine

sections to understand how a particular function

is

imple-

mented

or

why an unexpected result

is

seen. University

en-

vironments typically enjoy source code for most applica-

tions for precisely this reason.

The group can now distribute the second version of the

software in source code form. All of the apE system, includ-

ing window system layers, program development layers,

da-

ta format layers, and

all

existing filters and tools will

be

re-

leased in source code form with the software. Academic in-

stitutions can request this software (with manuals) at no

charge, although a license (prohibiting redistribution) must

be

signed.

apE: A Dataflow Toolkit for Scientific Visualization

For commercial and non-profit users, the university has

chosen to pursue commercialization of the software with

the TaraVisual Corporation in Columbus, Ohio. This com-

pany offers maintenance, installation, and consulting servic-

es

related to apE.

It

is

not

affiliated in any way with the

Ohio

State University

or

the developers of apE and thus the

version of apE offered through TaraVisual

is

expected to

di-

verge

from that offered by

Osu.

While this may seem

counter to some of the philosophies of apE, it

was

a decision

made by the university and

is

not representative of the

gen-

eral feelings of the developers.

The apE system does not represent a breakthrough in

computer graphics. Most of the technology that

has

been

harnessed to construct apE has been in existence for a num-

ber of years, and precious little of it could in any way

be

considered to

be

state of the art. However, the apE system

does represent a significant new step in placing sophisticated

tools into the hands of users. The project has helped to push

industry toward a greater realization of the nature of the

sci-

entific visualization problem. The potential of visual meth-

ods for data analysis

is

enormous, and

we

need to recognize

that the grand challenge that

faces

us

today

is

not in making

faster silicon but in finding new

ways

to improve the pro-

ductivity of

our

research community.

apE Runs

on

Convex C-l, C-2, Cray, SGI, SUN, HP,

NeXT, DEC, Stardent, and IBM (RS6000, AIX).

apE supports data manipulation, data mapping, image

rendering, and animation. Data manipulation includes a

da-

ta flow language (flux), creation and editing of polygonal

da-

ta, image format conversion, image processing, and image

la-

belling. Data mapping includes isosurface construction of

volumetric data, support for

RGB,

HSV,

HLS colorspace

mapping, color palette editing, data in uniform and non-

uniform grids, rectangular, polar, spherical and geocentric

coordinate systems in 2,

3,

and n dimensions.

The user interface supports Sun

View,

X Window System,

and Silicon Graphics GL.

It

has a visual language paradigm,

and fully indexed on-line user documentation.

It

also allows

distributed processing over a computer network.

123

Commercial users

Aggregation

of

existing tools

and

methodologies

Tools

for real users

Power

of

visual

methods

Platforms

Summary

of

facilities

User interfaces

124

Current Public Domain Systems

in

Use

Image rendering Image rendering includes

lD

frequency plots,

2D

con-

tour line images,

2D

continuous-color contour images,

2D

colour contours with gradient shading (bump mapping),

carpet plots, photorealistic rendering of polygonal data,

vol-

ume rendering, particle animation of vector fields, and

"glyph" rendering.

Reference

"A

Dataflow Toolkit for Visualization" by Scott Dyer in

information IEEE Computer Graphics and Applications,

Vol.

10,

No.4,

1990,

pp.60-69

,

gives

further information on apE .

Fig.B.7

Isosurface

Rendering

mCb'O' .

~

I •

Isosurface rendering using apE. A pipeline has been

cre-

ated to examine a 3 D volumetric dataset consisting of MRI

data from a human subject.

Two

isosurface values

have

been

selected -the outer one

has

been made transparent.

apE: A Dataflow Toolkit for Scientific Visualization

11111",,,

~

.1.,

~I'III

...

,

....

(_.-t

~.I.I'

"""-

...

_11-1

...

-'.-1

125

.111'."'"

I

r=====~:::::::::::=~

_,,_I

,.,.

..

"-"-I

Volumetric rendering using apE. A pipeline has been cre-

Fig.

B.B

ated to examine a 3 D dataset

of

temperature in the Atlantic

Volume

Rendering

Ocean and has been rendered using a ray-tracing technique.

126

Current Public Domain Systems

in

Use

Acknowl- This project has been possible only because of the dedi-

edgements cation and commitment of the members of the

Ohio

Super-

computer Graphics Project. The listed authors would like to

thank Barb Dean, Manager of the

Ohio

Visualization Labo-

ratory, and Michelle Messenger, our Project Coordinator. In

addition, the participation

of

the Advanced Computing

Center for the Arts and Design, in the form of graphics

spe-

cialists

Steve

Spencer and Jeff Light, has added features and

capabilities to apE that would have otherwise been absent.

The authors also thank Prof. Charles Csuri, who mar-

shalled the resources to bring this project into being and

supported it throughout its history, Dr. William McCurdy,

for providing the resources and being the catalyst for many

of

apE's scientific concepts, and Dr. Charlie Bender, Direc-

tor of the

Ohio

Supercomputer Center, for believing in the

project, and continuing to supporting it, during its most

crucial hours.

Support This work

was

supported by the

Ohio

Board of Regents,

through the

Ohio

Supercomputer Center, and by the

Ohio

State University. This work

was

supported in part by a grant

from Cray Research, Inc., by an equipment grant from Ap-

ple Computer, Inc., and by an equipment loan from Silicon

Graphics, Inc.

Contributed by the

Ohio

Supercomputer Graphics Project:

Scott Dyer, Project Leader

Steve

Anderson

John Berton

Pete Carswell

John Donkin

Jeff Faust

Jill Kempf

Robert Marshall

Since this information

was

submitted, apE has been taken

over by TaraVisual Corporation, who

are

now responsible

for its distribution and support. Although it

was

initially

made available

as

a public domain product from

Ohio

Su-

National Center for Supercomputing Applications (NCSA)

percomputer Center, this

is

no longer the

case

at the time

of writing. However, discounts on the software

are

available

for academic

use.

All enquiries on apE should now

be

di-

rected to the address below, and not to Ohio Supercomputer

Center.

TaraVisual Corporation

929

Harrison Avenue

Columbus

OH

43215

U.S.A.

Tel:

(800)

458-8731

Tel:

614-291-2912

Fax: 614-291-2867

8.3

National

Center

for

Supercomputing

Applications

(NCSA)

The NCSA

Tools

for

the Macintosh

The National Center for Supercomputing Applications Interactive data

(NCSA) offers a number of tools that

are

available in the analysis

public domain. NCSA Distributed DataScope

is

an interac-

tive data analysis tool that displays 32-bit scientific data

val-

ues

in spreadsheet form

or

as

simple scaled, interpolated,

or

polar color raster images. NCSA Image

is

a color imaging Color imaging

and analysis application that permits manipulation of two-

and

analysis

and three-dimensional image data sets. Distributed capabili-

ties across TCP/IP network connections allow image pro-

cessing on powerful computers such

as

a CRAY

1.

Specific

data manipulation features in NCSA Image include histo-

gram equalizations, contrast enhancements, and useful utili-

ty

functions. NCSA Layout

is

a presentation tool that

al-

Layout and

lows the and user to display and annotate two-dimensional annotation

data annotation images so that users can photograph their

Macintosh screen display with a

35

mm camera and produce

presentation-quality slides. NCSA Telnet provides a link

be-

127

128

Current Public Domain Systems

in

Use

tween the Macintosh and the TCP/IP networks. It includes

File transfer a standard

file

transfer server (FTP), which allows

file

shar-

ing with other machines.

All these tools

are

available

free

via the Internet. The

software and documentation are also available for purchase

through the NCSA Technical Resources Catalog.

Contact:

NCSA Documentation Orders

152

Computing Applications Building

605

East Springfield Avenue

Champaign, IL 61820, U.S.A.

Tel:

217-244-0072.

(This information

is

supplied courtesy of visualization

Technology: an Introduction", by Anne Kaplan-Neher, in

Syllabus, Summer

1991,

Number

17,

P.O.

Box

2716, Sunny-

vale,

CA

94087-

0716, AppleLink: SYLLABUS, Internet:

SYLLABUS@APPLELINK.APPLE.COM; Phone and

Fax:

408-773-0670.

It

was

initially obtained from the "Arti-

cles

database of CCNEWS, the Electronic Forum for Cam-

pus Computing Newsletter Editors, a BITNET-based ser-

vice of EDUCOM").

8.4

GPLOT,

DRAWCGM, P3D

(Pittsburgh

Supercomputer

Center)

GPLOT can interpret

CGM

metafiles and can

run

anima-

tion hardware. DRAWCGM produces rasters from

2D

ar-

rays of integers

or

reals. P3D creates and views 3 D models.

Such models can

be

viewed on SUNs and SGIs.

It

can build

isosurface models and molecular models, and create anima-

tions directly

on

video tape.

It

is

available via anonymous

FTP

from calpe.psc.edu. Further information

is

given

be-

low.

GPLOT, DRAWCGM, P3D

8.4.1

The GPLOT CGM Interpreter

The Pittsburgh Supercomputing Center began in 1986

as

Supercomputer

one

of

several sites charged by the U.

S.

National Science center

Foundation with providing supercomputer (and other)

ca-

pabilities to NSF researchers. Unlike other such sites, the

us-

er base (over 2000 researchers)

is

very diverse, both

geo-

graphically and technologically, with almost all users physi- Remote access

cally remote.

Providing a graphical capability to these users

was

prob- Graphics

lematic; but being a new center provided the option to

de-

standards

sign systems from scratch.

It

was

decided to standardise

completely

on

the Computer Graphics Metafile (CGM) for-

CGM

mat for two-dimensional image storage. Accordingly only

graphics packages that could produce

CGM

files

were pur-

chased. Each of these packages came with a

CGM

translator,

but it

was

found that these translators could only reliably

translate

CGM

files

from the corresponding graphics pack- Which

CGM?

age!

It

was

unacceptable to distinguish between types of

CGM

files

and it

was

decided to write a

CGM

translator:

GPLOT. In addition to homogenizing the

CGM

file popu-

GPLOT

lation there were several other advantages to using this soft-

ware. Firstly, GPLOT could be freely distributed to remote Distributed

sites and users were encouraged to produce their

CGM

files

software

at the PSC and ship them home for viewing with a local

co-

py

of

GPLOT. This improved response for the users dramat-

ically compared to viewing the

CGM

files

across their net- Reduced

work connection and also reduced the load

on

the network network load

connections. Secondly it allowed the rapid addition of new

output devices and facilities.

GPLOT

was

more successful than expected; there are 275 sites

now over

275

sites

on

the list of users, including most of the

major universities and research laboratories in the USA and

several sites in other countries. GPLOT now supports out-

put

to many different output devices with three different

us-

er interfaces.

129

130

Current Public Domain Systems

in

Use

Video GPLOT

was

also used to greatly facilitate the creation of

videos by remote users. Users

are

encouraged to create

CGM

Frame checking

files

with many (possibly very many) frames. They can

ex-

amine individual frames

at

their home site using their copy

of

GPLOT and when satisfied with their "look" submit

them for animation. A local copy

of

GPLOT at the PSC

is

Animation then used to create a full animation either on a Sony U matic

sequences recorder

or

on

a Sony laser disk recorder.

It

can then

be

dubbed to VHS tape and mailed to the user, with a turn-

around time of a

few

days.

In fact this

is

the only output

that PSC mails to users. In this manner

several

minutes of

animation can

be

produced, possibly spanning

several

indi-

vidual animations, a night. Presently there are three parallel

animation systems, two U matics and one laser disk

re-

corder.

Cost savings In addition

to

purely remote

use,

three of the heaviest

an-

imation users decided to purchase their own hardware, and

used GPLOT software

to

produce their own animation

sys-

tems. Since

all

necessary software

was

provided, the cost to

these users

was

much

less

than a commercial system.

Random access The size of the

CGM

files

required for animations

(fre-

quently close to a gigabyte) motivated extension

of

the

CGM

standard to include a random

access

capability. This

was

done in cooperation with the SLATEC supercompu-

ting community and it

has

worked very well to date.

The original GPLOT system

was

written entirely in the

C programming language, compilable under either the

UNIX

or

VMS

operating systems.

As

the move began to-

Graphical user wards sophisticated graphical user interfaces and users began

interface to request more capabilities for GPLOT (including on-

screen animations for workstations) it

was

decided to com-

Object oriented pletely rewrite GPLOT in an object-oriented fashion using

the

C+

+ language. This greatly simplified the addition of

some features, including on-screen animations using the

X-

Window system and a Motif user interface.

Port

to

This

C+

+ version also simplified the

port

to the Apple

Apple Macintosh Macintosh operating system. The great majority of the code

is

common with two Macintosh specific modules, one for

GPLOT, DRAWCGM, P3D

the user interface and one for the Quickdraw imaging

sys-

tem.

The object-oriented design

was

also intended to allow

the easy integration

of

all

of

GPLOT's capabilities into oth-

er packages. Work

is

underway to perform this in combina-

tion with a documentation system to allow true text-graph-

ics

integration in a single system.

8.4.2 The DrawCGM Graphics

Subroutine Library

Access

to

other packages

During the development of GPLOT, it became necessary to

Test

facilities

generate

CGM

files

for test purposes. A simple library called

CGMGen

was

written to do this. It had the ability to pro-

duce indexed

or

direct color

CGM

files,

the interface being

set up

so

that a single call generally produced a single

CGM

element.

At the same time it

was

becoming obvious that graphics Graphics from

packages available then, like Disspla, DI-3000, and the supercomputers

NCAR

Graphics Library, lacked features needed to do some

of the graphics required in a supercomputing environment.

In particular, it

is

very common for a supercomputer user

to wish to produce a color image from a two-dimensional

regular array

of

data. This

is

done simply by mapping the

data into a range of integer values, and drawing the image

using those values

as

indices into a color

map.

It

was

re-

quired to provide this functionality to the users in a simple

way.

This led to the development

of

the DrawCGM graphics Raster images

library, which

is

particularly well suited to the generation of

raster images. The package provides facilities for manipulat-

ing color maps, scaling and quantizing rasters of reals, and

drawing multiple images within a single

CGM

frame. The

ability to produce other

CGM

primitives

was

added, such

as

markers, lines, polygons, and text

as

well, because these

functions were readily available in CGMGen. Utilities to

easily draw color bars and labels were

also

included.

Because

the primary task of DrawCGM

was

to handle color mapped

131

132 Current Public Domain Systems

in

Use

images, only the indexed color facilities in CGMGen

were used. DrawCGM

is

written in FORTRAN, while

CGMGen

is

written in

C.

Eventually confidence in the interpretation of the

CGM

standard increased

so

that support for the independent

CGM

generator in CGMGen could

be

dropped. This

CGM

Interface

to

GPLOT generator

was

replaced with a direct interface to the GPLOT

device driver library, making it possible to

use

the

CGMGen interface to do graphics interactively

on

any

de-

vice supported by GPLOT. Since GPLOT

also

supports

de-

vice drivers which create binary

or

clear CGM, the ability

to produce metafiles from DrawCGM

was

not lost. Thus

DrawCGM became an interactive package, supporting a

wide range of devices.

DrawCGM and CGMGen

are

distributed with the

GPLOT library, and

are

now quite widely used. DrawCGM

Applications has been particularly successful for producing animations,

for example of hydrodynamic systems. Users will preview

animations either interactively

or

via a

CGM

metafile and

GPLOT, and will

pass

a

CGM

file

containing the entire ani-

mation for recording when their results

are

satisfactory.

DrawCGM provides the ability to

have

multiple images

with distinct color maps on screen simultaneously, which

can greatly improve the information content of this sort of

animation. The underlying CGMGen layer still supports

the ability to

use

direct color, so it

has

been used to interface

a number of 24-bit color applications to the GPLOT device

driver library.

One

feature which DrawCGM does not currently sup-

port

is

the ability to

have

multiple output devices open

si-

multaneously. It

is

hoped to correct this when CGMGen

is

recoded to interface to the new object-oriented version of

GPLOT.

GPLOT, DRAWCGM, P3D

8.4.3 The P3D Three-Dimensional

Metafile Project

Success with a metafile-based environment for

2D

graphics

2D

to

3D

led to the consideration of a similar system for three-dimen-

sional models. The goal would be

to

produce a format

which all programs generating 3 D models at the PSC would

produce, and which could be rendered in three dimensions

on

a wide variety

of

platforms. The existence

of

this format

would allow models to be transferred between the central

si-

te and user sites, and would simplify the software support

situation analogous to the simplification provided by CGM.

This

is

not

to belittle the importance of interactive 3 D

graphics; it

is

simply believed that a complete environment

requires both interactive and metafile forms and that it

is

ap-

propriate to investigate the metafile approach.

Unfortunately, unlike the

2D

case

in which the standard Format for 3D

CGM

format already existed, there

was

and

is

no standard

format for 3 D scientific graphical models. The available

non-standard formats were examined and none of them

were found very satisfactory for current needs. Therefore a

further format

was

developed, called P3D -the P denoting

'Programmable'.

P3D

is

based

on

a subset of the

Common

Lisp language,

3D

language

with a small number of extensions to describe geometry.

This makes P3D a complete programming language in itself,

allowing it the same flexibility which programmability pro-

vides to the Postscript page description language. Like Post-

script, it

is

never necessary for a user to actually write a pro-

gram in P3D. A model generating program (for example a

molecular modeler) produces a P3D model, and a P3D

viewer translates the model into images which the user can

view.

The P3D viewer includes a locally written Lisp inter-

preter; there

is

no

need for the site using P3D to license an

interpreter from a third party. Because the full 3 D structure

of

the model

is

stored in the P3D

file,

the model can be

viewed from any direction

or

incorporated into more com-

plex models.

133

134

Current Public Domain Systems

in

Use

Needs

of

scientific P3D

was

designed to support the needs of scientific visu-

visualization alization, rather than those of photorealism or, for example,

computer aided design. The current implementation sup-

ports ten geometrical primitives, fairly complete lighting

and camera information, arbitrary transformations, and a

very extensible attribute structure in a hierarchical model

environment.

Rendering options P3D models can

be

viewed

on

seven different renderers,

with more under development. These range from a simple

mouse-driven renderer for the X Window System environ-

ment to renderers for solid modeling workstations and a ray

tracer. A number of generators for P3D models now exist,

including translators for molecular dynamics output,

marching cubes algorithms, a general purpose subroutine

li-

brary, and a simple tool for generating fly-bys of P3D mod-

els.

The programmability of P3D makes it easy to modify

existing codes to produce models in P3D format, since the

metafile can essentially

be

tailored to the needs of the code

rather than the other way round.

Animation

As

with CGM, it

is

possible to produce animation from

a P3D model

file

in which a number of

views

are

specified.

This allows animations to

be

previewed on a user's worksta-

tion, and then nicely rendered (possibly ray-traced) and

re-

corded

as

high quality video animation. The animation

sys-

tem

uses

the same hardware and some of the same software

as

the CGM-based system.

Exchange

of

3D

It would

be

useful for the P3D format to become a com-

mode/ data mon medium

of

exchange for 3D models. There

are

current-

ly about

40

sites

on

the mailing list of those using

or

inter-

ested in P3D,

so

some progress

is

being made toward this

goal. Current work

is

on

designing interfaces which will

al-

low general interactive visualization packages like Stardent's

AVS

and the Ohio Supercomputer Center's apE to read and

write P3D models, and to incorporate additional renderer

interfaces such

as

Pixar's Renderman.

Other

development

projects include a translator to generate P3D from finite

ele-

ment models, and general modifications to improve the

functioning of the P3D renderers.

RAYSHADE

8.4.4

Software Availability

Software developed at the Pittsburgh Supercomputing Cen- Free

of

charge

ter, including GPLOT, DrawCGM, and the P3D software

suite,

is

available

free

of

charge by anonymous

FTP

from

the machine ftp.psc.edu. DrawCGM

is

included in the

GPLOT package; the P3D software

is

a separate distribution

but (depending

on

the configuration chosen) may require

the GPLOT software.

If

you take GPLOT, please send mail

to Anjana Kar (kar@psc.edu) to be added to the appropriate

mailing list. Advanced questions regarding GPLOT,

as

op- Information

posed to simply taking and installing the software, can

be

directed to Phil Andrews (andrews@psc.edu).

If

you take

P3D, please send mail to Joel Welling (welling@psc.edu)

to

be

added to that mailing list.

Contributed

by

Joel

Welling

and

Phil Andrews.

Further information from:

Dr. Joel Welling

Pittsburgh Supercomputer Center

4400

Fifth Avenue

Pittsburgh,

PA

15213,

U.

S.

A.

Tel:

412-268-6352

Email: welling@psc.edu

8.5

RAYS

HADE

135

This

is

an excellent ray-tracing program for scene rendering. Scene rendering

It handles many different kinds of object.

It

is

available from

the University of

Yale

and the University

of

Utah.

Rayshade reads a multi-line ASCII

file

describing a scene Format

to

be

rendered and produces a

Utah

Raster RLE format

file

of the raytraced image.

136

Facilities Features include:

• Primitives:

boxes

cones

cylinders

height fields

planes

polygons

spheres

Current Public Domain Systems

in

Use

triangles (flat-

or

Phong-shaded),

• Composite objects,

• Point, directional, and extended

(area)

light sources,

• Solid texturing and bump mapping of primitives, ob-

jects, and individual instances of objects,

• Antialiasing through adaptive supersampling

or

"jit-

tered" sampling,

• Arbitrary linear transformations of primitives, instances

of objects, and texture/bump maps,

• Use of uniform spatial subdivision and/or hierarchy of

bounding volumes to speed rendering,

• Options to facilitate rendering of stereo pairs,

• Support for the Linda parallel programming language.

An

awk script

is

provided

to

translate

NFF

format scripts

to rayshade format.

C language Rayshade

is

written in C with parsing support provided

through

lex

and

yacc.

The C,

lex

and

yacc

files

comprise

ap-

proximately

8000

lines of code.

Sites

without lex and

yacc

can make

use

of the C source

files

produced by

lex

and

yacc

which

are

included in this distribution.

Platforms Rayshade

has

been tested on a number of UNIX-based

machines, including

Vaxes,

Sun Workstations, Iris 4D Work-

stations, Encore Multimax,

AT&T

3B2/310, CRAY

XMP,

and

IBM

RTs.

In addition, support

is

provided for the Ami-

ga

using the Aztec C compiler.

Getting a copy Rayshade makes

use

of the Utah Raster toolkit, a pack-

age

consisting

of

a large number of useful image manipula-

tion programs, test images, and a library to read and write

NASA Ames Software

images written using the toolkit's RLE format. The toolkit

is

available via anonymous FTP from cs.utah.edu

or

from

weedeater.math.yale.edu.

Those sites that cannot

or

do not want to

use

the Utah

Raster toolkit can make

use

of a compile-time option to

produce images written using a generic

file

format identical

to that used in Mark

Van

de

Wettering's "MTV" raytracer.

Rayshade

is

copyrighted in a "Gnu-like" manner.

Rayshade

is

available via anonymous ftp from

weedea-

ter.math.yale.edu (192.26.88.42) in pub/Rayshade.2.21.tar.Z.

The Utah Raster toolkit

is

available in pub/UtahTool-

kit.tar.Z.

8.6

NASA

Ames

Software

8.6.1

PLOT3D

is

a computer graphics program designed to visual- Computational

ize the grids and solutions of computational fluid dynamics. fluid dynamics

Eighty-five functions

are

available, and versions

are

available

for many systems. PLOT3D can handle multiple grids with

many grid points, and can produce varieties of model ren-

derings, such

as

wireframe

or

flat shaded.

Output

from

PLOT3D can be used in animation programs.

PLOT3D User's Manual and PLOT3D software can

be

Availability

distributed

free

of charge and without copyright to any in-

stitution

or

business in the USA.

Contact:

Workstation Applications Office

NASA Ames Research Center,

MS258-2

Moffett Field,

CA

94035,

U.

S.

A.

8.6.2 SURF

137

A further program, SURF, allows the user to input

PLOT3D grid and solution

files

and interactively create wi-

reframe, shaded, and function mapped parts to

view,

and

Interactive viewing

of

PLOT 3D files

138

Current Public Domain Systems

in

Use

then output to ARCGraph

files

which can

be

animated

us-

ing the Graphics Animation System (GAS). Shaded parts

are

created based

on

user specified lightsources (at least

20),

viewpoint, and the ambient light level. The function map-

ped parts can have their color spectrum adjusted interactive-

ly.

Legends can be created to show the correlation

of

color

and normalised function values (i.e., pressure, density, tem-

perature, and mach number). Also, function-mapped parts

can

be

clipped so that they only show contours within a

specified range of function values (e.g., normalised pressure

between 1 and

2).

Further facilities

Other

features

of

SURF include the ability to work with

several grids and solutions, grid/solution deletion, support

of multi-grid

files,

input/output for colormaps, matrices,

light sources, function extrema, screen dump pixel in-

put/output, display of current grid and part attribute data,

and a

UNIX

shell

escape.

8.6.3

Graphics Animation System

(GAS)

Animation system GAS

is

a graphics animation software system that

is

menu-

driven and provides fast, simple viewing capability

as

well

as

more complex rendering and animation features.

It

is

used to display two- and three-dimensional objects along

with computed data, and also to record animation sequenc-

es

on

video digital disks, videotape, and

16

mm

film.

8.6.4

Applications in Computational

Fluid Dynamics (CFD)

Illustrations

of

use Some example applications

have

been the following: pres-

sure distribution inside the space shuttle main engine, vor-

tex

flows

over the wing/strake surface of

F-16

aircraft, pres-

sure distribution on an oscillating

F-5

wing, simulation of

turbine engine rotor-stator interaction, particle traces over

the space shuttle orbiter, and pressure distributions over the

high-speed National Aerospace Plane.

Getting a solution The

CFD

software analysis cycle begins with the design

of

a test geometry 'grid' (e.g., forward-swept wing model

NASA Ames Software

with surrounding airspace), specification of simulation con-

ditions (e.g., angle of attack, mach number, reynolds num-

ber) and coding and execution

on

supercomputers of 'flow

solver' programs that solve the higher-order mathematical

equations governing the flight characteristics. Then the nu-

merical solution data

is

collected and converted to graphics

images of fluid

flows,

pressure distributions, shock

waves,

and particle traces using workstations running PLOT3D

and other specialised graphics programs. Then SURF can

be

used to add shadinglcoloring enhancements to the

images.

Finally, animation sequences

are

generated and recorded

with

GAS.

The results produced

are

animated

16

mm films and vid-

eotapes showing solid, pressure mapped aircraft models

with wing/body vortices, particle traces, temperature distri-

butions and shock

waves.

The ability to display the physical

properties in aerodynamic flight

is

a tremendous aid in un-

derstanding and designing aircraft geometries for specific

flight characteristics.

By

using interactive computer graphics

in the aerodynamics study, the critical

areas

(e.g., high tur-

bulence, high temperature, reverse

flow)

immediately

be-

come obvious

so

they can

be

studied more closely using a

finer grid, more particle traces, and higher-resolution map-

ping of pressure

or

temperature contours. For example,

graphical studies revealed high turbulence and pressure in-

side the

Space

Shuttle Orbiter main engine hot

gas

mani-

fold, and further analysis led

to

a redesign

of

the engine

with reduced decisions internal pressure and turbulence.

(Courtesy of PLOT3D User's Manual by

P.

Walatka and

P.

G. Buning, SURF User's Guide by

T.

Plessel, and GAS

User's Manual by

T.

Plessel).

Contact:

NASA Ames Research Center

Mail Stop

258-2

Moffett Field,

CA

94035,

U.

S.

A.

Tel:

415-694-4052

139

Use

of

computer

graphics

Animation

Visualization aids

understanding

Studying physical

properties

Design decisions

140 Current Public Domain Systems in Use

8.7

IRISPLOT

Display

of

surfaces IRISPLOT

is

an extended version of

GNUpiot

for the

Sili-

con Graphics workstation that allows algebraic surfaces to

be displayed, shaded, etc.

IRISPLOT allows the user to define some of the graphi-

cal objects built from surfaces and curves, which in

turn

are

defined from mathematical functions, discrete maps, differ-

ential equations and data

files.

The

user has full control

of

Viewing the the graphical attributes, which includes viewing, orthogonal

surface

or

perspective projection, object transformation, object slic-

ing, 8 different light sources with different color and loca-

tion, and different material properties for each object in

plot.

It

also allows contouring

on

the surfaces.

Interactive

visualization

Finite element

data

For more information, contact system@math.arizo-

na.edu

8.8

ISVAS

FhG-AGD in Darmstadt provides a tool called

ISVAS,

an in-

teractive volume visualizer for

SUN

(Xll/0SF

Motif) and

SGI (GL/Motif).

ISVAS

stands for Interactive Software for

Visual AnalysiS of fracture mechanics. The system has been

developed for the visualization of the results of three-dimen-

sional finite element simulations in fracture mechanics and

other application areas.

The main aim in the development of

ISVAS

is

interact

i-

vity. Finite element analysis produces large amounts of data

and the graphical presentation

of

the data

is

a computer-in-

tensive process. Therefore there

is

a need for presenting data

at speed, but with low amounts of detail. The tools allow

a rough preview of the data and interactive specification of

parameters, such

as

viewpoint and cut planes. The user can

thus produce a quick picture

of

what the data looks like,

specify the parameters of the required image,

then

render it

for a higher-quality presentation.

ISVAS

141

The

Xll

version, which

is

running on SUN, DEC, and

other machines, has already been installed at the Technical

University of Munich, the University of Lisbon, and the

University of Mexico. Sun (SPARe) code

is

available, and

al-

so

source code

of

the data filters for adaption to the user's

own volume data. Data filters for

several

FE-data types

are

provided. Version

1.2

of

ISVAS

allows the user to visualize

scalar volume data. The next version will

also

allow vector

data to be displayed.

ISVAS

is

built upon X Windows and

OSF/Motif

toolkit Portability

to

be

portable to any

UNIX

colour workstation.

Further information:

Dr. Martin Goebel

Fraunhofer-Arbeitsgruppe

fur Graphische Datenverarbeitung

Wilhelminenstrasse 7

W-6100

Darmstadt

Federal Republic of Germany

Tel:

49-6151-155123

Fax:

49-6151-155199

Email: goebel@de.fhg.agd

Creative and

artistic

applications

Artists and

designers

Chapter

9

Other

Uses

of

Visualization

Tools

9.1

Art

and

Design

Tools such

as

those described in this guide can

be

also used

by those whose primary interest

is

not in the scientific con-

tent

of

the information presented, but rather the creative

or

aesthetic value.

Barlow et

al.

(1990) outlines how artists create effects and

explores

issues

at the interface between art and science.

Artists and sculptors

have

been using computer-assisted

tools for a number

of

years (Lansdown, 1989) and these

tools often promote new and unexpected

ways

of creating

and developing images and objects. Architects, designers and

engineers also use these methods. Thus visualization tools

are

not

confined to scientific visualization but can be used

in all

areas

where the user

is

seeking to create and manipu-

late information via visual means.

9.2

The

5th

Dimension

Animation

System

3D

animation and The 5th Dimension Project

is

a large research project in

visualization three-dimensional animation and visualization. The main

objective of the project

is

the animation of synthetic actors

in their environment, which involves a number of related

areas

of

computer animation and scientific visualization. In

Applications particular, the following applications

are

being developed:

• animation of articulated bodies based

on

mechanical

laws,

• vision-based behavioural animation,

The 5th Dimension Animation System 143

• hair rendering and animation,

• object grasping,

• facial animation,

• personification in walking models,

• synchronization in task-level animation,

• deformation of flexible and elastic objects,

• cloth animation with detection of collision.

To

coordinate efforts and allow good communication

be-

High-level toolkit

tween the various applications, a toolkit of high-level dy-

namic

classes,

both two- and three-dimensional,

has

been

constructed. This toolkit, called the 5th Dimension Toolkit,

uses

a uniformly object-oriented design for

all

its data struc- Object oriented

tures, resulting in a high degree

of

integration between vari-

ous applications.

The 5th Dimension animation system

is

intended to

of-

3D

interaction

fer to the animator a full 3 D interaction including the possi-

bility of entering into the virtual world and communicating Virtual worlds

with the synthetic actors. The hardware used consists of

21

Silicon Graphics IRIS Workstations including three Power-

vision (VGX) models. Most 5th Dimension applications

take advantage of visual 3 D interfaces using the various 3 D

devices

available in the laboratories: two datagloves, several

SpaceBalls, an EyePhone, a 3-D Polhemus digitizer, a

Live

Video Digitizer, a Stereo View station, and a synthesizer

key-

board controlled by a NeXT Cube workstation.

In the current version, six applications provide a user in-

3D

devices

terface based on 3 D

devices:

• the sculpting program SURFMAN,

• the Muscle and Expression editor in the SMILE Facial

Animation system,

• the cloth design software,

• the hand gesture recording system GESTURE

LAB,

• the program to create 3 D paths for cameras, objects and

light sources,

• a communication program animator-actor (in develop-

ment).

144

Other Uses of Visualization

Tools

3D

input The first three programs

are

mainly based

on

the ball and

mouse metaphor. SURFMAN may

also

take advantage of

Stereo View and the 3 D Polhemus digitizer. Hand gestures

are recorded using the DataGlove and 3 D paths

are

mainly

generated using the

SpaceBall.

We

are

developing a way of

creating camera paths based on the EyePhone. The commu-

nication program animator-actor

uses

the Living Video Dig-

itizer to capture the animator

face.

Fig.

9.1

Cloth Animation

Other

applications in the 5th Dimension system

are

on-

ly based

on

mouse interaction. They include:

• an interactive system to design individual walking,

• the BODY-MOVING human keyframe animation

sys-

tem,

• a hair modelling and rendering program.

Submitted

by

Nadia Magnenat Thalmann, University

of

Geneva, and Daniel Thalmann,

Swiss

Federal

Institute

of

Technology.

Figure

9.1

shows cloth animation from the film Flashback,

by

B.

Lafleur,

N.M.

and

D.

Thalmann, University of Gene-

va

and

Swiss

Federal Institute of Technology.

The 5th Dimension Animation System

Frame from the

film

Still

Walking

by

A.

Paouri,

R.

Bou-

lie,

N.

M.

and

D.

Thalmann, University of

Geneva

and

Swiss

Federal

Institute of

Technology.

Hair rendering

by

A.

leBlanc,

A.

Paouri, N.M.

and

D.

Thalmann, University of

Geneva

and

Swiss

Federal

Institute

of

Technology.

145

Fig. 9.2

Human Walking

Fig. 9.3

Hair Rendering

146 Other Uses of Visualization

Tools

9.3

Multimedia

Environments

Project

KICK

Multi-media KICK

is

an interactive multimedia environment designed to

support industrial training applications (Serra et

al.

1991).

The main component of KICK

is

an authoring environ-

ment for the designers to organize the contents and orches-

trate the presentation of multimedia information. The

me-

dium types supported include text, image, 3 D graphics, and

video.

Hierarchy The effective organization of multimedia information

for interactive

access

is

of major concern.

In

KICK, multi-

media information

is

organized primarily using the natural

hierarchy of the physical objects to be modeled. In order to

permit associative

access

to information

as

advocated in hy-

permedia, auxiliary

access

paths are provided by means of

other media, such

as

text, image, and video. The resulting

information structure thus facilitates both structured and

associative

access

to information.

In KICK, the same direct manipulation interface

is

used

to manipulate information of any medium type.

To

achieve

Video

and

image this for video and image media, techniques have been

devel-

oped to allow the synchronization of 3 D graphics models

and animation of video sequences and images. In addition,

techniques to model object motions, constraints and rela-

tionships have also been developed.

Training The system developed

is

intended for training applica-

tions, where users can learn about the structure and opera-

tions of a complex mechanism by interactively:

-accessing its component hierarchy,

experimenting with how various components interact

with each other in dynamic operating conditions,

-studying the effects of various externally applied forces

on

different parts of the mechanism.

Examples

To

demonstrate the usability of system KICK, two applica-

tions for industrial training have been developed.

One

is

ba-

Multimedia Environments

sed on an aircraft and the other

on

a

1/8

scale model car

with a

0.21

cc

engine. Figure 9.4 shows an engine compo-

nent of the aircraft application in four different media -3 D

graphics, text, video and image. Figure 9.5 shows the engine

component of the model car application. Through the

dis-

play, the user may interact with any medium type to

re-

trieve further information

or

to view the operations of the

engine in context.

KICK

is

developed

on

a Silicon 4D/210

GTX

worksta-

tion with a Live Video Digitizer. The video input

is

ob-

tained from a Sony LDP-1S00 laser disk player. KICK

is

im-

plemented using Starship, a frame language developed at

ISS

(Loo

1991).

The figure shows the concept of a turbine engine in four

dif-

ferent media - 3 D graphics (top right), text (top left), video

(bottom right) and image (bottom left). Users may interact

directly with any medium type to retrieve further informa-

tion about its subcomponents, view the video about the

assembly of the engine,

or

generate an animation of a 3 D

model.

Fig.

9.4

Turbine Engine

from Aircraft

Application

147

148

Fig. 9.5

Concept Engine

from Car

Application

Other Uses of Visualization

Tools

This figure shows the interface of KICK without the vid-

eo window. The users may again interact with any medium

type directly to obtain further information.

References

Loo J.p.L.

(1991)

The Starship Manual (Version 2.0),

ISS

In-

ternal Technical Report,

TR91-54-1

Serra, L., Chua,

T.

S.

and Teh,

W.

S.

(1991)

A Model for Inte-

grating Multimedia Information around 3 D Graphics

Hierarchies. The Visual Computer (in press)

Information supplied by Luis Serra and Wei-Shoong Teh,

Institute of Systems Science, National University of Singa-

pore, and Tat-Seng Chua, Department of Information

Sys-

tems and Computer Science, National University of Singa-

pore.

Chapter

10

Conclusions

10.1

Strategic

Importance

of

Scientific

Visualization

In view of the believed strategic importance of scientific

vi-

Tools

and

sualization it

is

timely to consider what tools and tech- techniques

niques should

be

provided in this

area

for the community, needed

and also what kind of initiatives and objectives should be

supported and promoted.

It

is

important that scientific visualization

be

developed Promotion

and promoted. Here

are

some of the strategic

issues

that

need addressing if scientific visualization tools

are

to

be

ef-

fectively utilised.

Forum for:

• Planning, discussion, and problem-solving,

• Coordinate developments that may

be

required in the

area,

• Exchange experiences in different application

areas,

• Share common software tools, where available,

• Disseminate information,

• Assist with teaching materials,

• Education and training

issues.

Plan for:

• Any new products that may

be

needed,

• Any general network developments that may

be

needed,

• Any general video facilities that may

be

needed,

• Providing visualization tools for the community,

• Supporting research into scientific visualization.

150 Conclusions

10.2

Current

Developments

More detailed These advances will allow mathematical models and simula-

modeling tions to become increasingly complex and detailed. This

re-

sults in a closer approximation

to

reality, thus enhancing the

possibility

of

acquiring new knowledge and understanding.

Scientific visualization

is

concerned with methods of under-

standing

large

collections of numerical

values

containing a

great deal

of

information. The scientist

has

to be able to

make effective

use

of this information for analytic purposes.

Interactive A further aspect

is

that increases in computer perfor-

3D

design mance allow

3D

problems in simulation and design to be

done interactively. In addition, processes that formerly

sepa-

rated out simulation and design can now bring them togeth-

er (e.g., in CAD,

or

in the design of new drugs). This in turn

moves the user into a new

era

of methods of design.

Handling mUlti- Control over fine simulations, interactivity, and comput-

dimensional data er performance mean that

vast

amounts of multidimension-

al

data can

be

generated. Superworkstations allow this data

to be displayed in optimum

ways.

These features and

capa-

bilities

are

driving the current

wave

of

interest in scientific

visualization.

New techniques Work

is

proceeding in evolving new techniques for data

for analysis display

as

the data

is

being analysed.

What you drive is

what you see

(WYDIWYS)

10.3

More

User-Friendly

Facilities

A further current trend

is

to make software tools for visual-

ization more user-friendly and accessible to a wide variety

of

application

areas,

thus increasing their potential and

us-

ability.

Improvements to the graphical user interface and the

way the user interacts with the model

are

likely to provide

more effective

ways

of communicating relationships and

other aspects of data to the user.

What to do next?

10.4

Further

Information

The book by Nielson et

aI.

(1990) contains a wide variety Applications

of current applications of scientific visualization and also an

excellent bibliography of scientific papers. A 2-hour video

tape

is

available with the book. This tape

gives

effective

demonstrations of the projects described in the book. The

format

is

NTSC but can

be

played

on

a dual-format player

(such

as

Panasonic

J35)

in the

UK

without any problem.

This player can take both PAL and NTSC formats.

Frenkel (1988) provides a general introduction to basic Introduction

visualization techniques.

Thalmann (1990) contains a number of papers in the

ar-

eas

of scientific visualization and graphical simulation.

For further details of any of the software mentioned

ear-

lier, please contact the local office of the vendor. In addi-

tion, Chapter

11

contains a list of references which may

be

of further interest.

There

are

numerous electronic mail subscription lists

and bulletin boards on topics in scientific visualization and

also on specific items of software.

10.5

What

to

do

next?

The majority of users

have

some experience of tools for pre-

sentation graphics. Some users are currently using visualiza-

tion tools of one kind

or

another, in that the

use

of such

tools facilitates the process of understanding more about the

data. However, very

few

users

have

access

to visualization

systems of the kind described in this guide.

It

is

expected

that

access

to such systems will become more common

as

costs decrease and experience in application

areas

increases.

The developments in the U.

S.

A.

outlined in Chapter 6

indicate the trends

as

of 1991, of course!

A more detailed reference volume

on

scientific visualiza-

tion

is

available entitled

Scientific

Visualization

-

Techniques

and

Applications

edited by Brodlie et

aI.,

Springer-Verlag,

1991.

Understanding

data

151

References

H. Barlow,

C.

Blakemore,

M.

Weston-Smith (eds.): Images

and Understanding: Thoughts about Images, Ideas and

Understanding. Cambridge University Press, 1990

Outlines how artists create effects and explores

issues

at the inter-

face

between art and science.

K.

W Brodlie, L. A. Carpenter,

R.

A. Earnshaw,

J.

R.

Gallop,

R.J. Hubbold, A.M. Mumford,

c.D.

Osland,

P.

Quaren-

don

(eds.):

Scientific Visualization -Techniques and Ap-

plications. Springer-Verlag,

1991

This volume represents a full consideration of the subject of scien-

tific visualization and

is

intended to be a reference guide for the

community on the technical aspects of the subject. The topics

covered include: the framework, visualization techniques, data

fa-

cilities, human computer interface, applications, products, a

glos-

sary of terms, enabling technologies, and an extensive bibliogra-

phy.

D.

Scott Dyer: A Dataflow Toolkit for Visualization. IEEE

Computer Graphics and Applications, 10:4,

60-69

Ouly

1990)

This article describes the design principles behind the apE visual-

ization software (apE stands for Animation Production Environ-

ment).

E.J. Farrell (ed.): Visual Interpretation of Complex Data.

IBM Journal

of

Research and Development, 35:1/2

Oan-

March 1991)

A special issue of the

IBM

Journal of Research and Development

on visualization. There

are

papers on visualization of volumetric

data, image display and interpretation, and animation for data in-

terpretation. There

is

also a companion video "Understanding

Complex Data with Computer Animation".

References

K.A. Frenkel: The Art and Science of Visualizing Data.

Communications of the ACM, 31:2, 110-121 (1988)

An

introductory paper which looks at a range of application

areas

and the

uses

of visualization tools and techniques. A wide range

of pictures illustrate some of the techniques currently being used.

K.A. Frenkel: Volume Rendering. Communications of the

ACM, 32:4,

426-435

(1989)

Outline of volume visualization, with application areas.

A.

Kaufman (ed.): Volume Visualisation. IEEE Press, 1990

A collection of most of the key papers in the area of volume visu-

alization.

R.J. Lansdown, R.A. Earnshaw

(eds.):

Computers in Art,

Design and Animation. Springer-Verlag, 1989

Collection of papers

on

the area of creative uses of computer

graphics and associated tools and techniques.

B.H. McCormick, T.A. DeFanti, M.D. Brown

(eds.):

Visu-

alization in Scientific Computing. ACM SIGGRAPH

Computer Graphics, 21:6 (November 1987)

The original Panel Report which outlines the political, economic,

educational, and technological aspects of scientific visualization

as

an emerging discipline.

B.H. McCormick, T.A. DeFanti, M.D. Brown

(eds.):

Visu-

alization in Scientific Computing. IEEE Computer, 23:8

(August 1989)

An

updated version of the original McCormick

(1987)

report,

outlining current progress and advances in scientific visualization.

G.M. Nielson,

B.

Shriver, L. Rosenblum

(eds.):

Visualiza-

tion in Scientific Computing. IEEE Press, 1990

A collection of papers in the

areas

of techniques and applications

of scientific visualization from a variety of academic, government,

and industrial organisations in the USA.

153

154 References

N.

M.

Patrikalakis (ed.): Scientific Visualization of Physical

Phenomena. Springer-Verlag,

1991

Proceedings of the 9th International Conference of the Computer

Graphics Society on the theme

of

scientific visualization. This

volume contains a number of key invited papers in the areas of

applications of scientific visualization, including engineering

de-

sign, spacecraft exploration of the solar system, and remote subsea

exploration.

D.

Thalmann (ed.): Scientific Visualisation and Graphics

Simulation. Wiley, 1990

Computational and graphical techniques that

are

necessary to

vi-

sualize scientific experiments

are

surveyed in this volume, with a

number of

case

studies in particular application

areas.

N.

Magnenat-Thalmann,

D.

Thalmann

(eds.):

New Trends

in Animation and Visualization.

Wiley,

1991

A collection of papers covering state-of-the-art topics in the areas

of scientific visualization, animation, graphical simulation, mod-

eling, hypermedia, facial animation, natural phenomena, human

modeling, and applications.

E.

R. Tufte: The Visual Display of Quantitative Informa-

tion. Graphics Press, USA, 1983

E.

R.

Tufte:

Envisioning Information. Graphics Press, USA,

1990

Two

introductory texts with guidelines

on

displaying information

effectively.

C.

Upson,

T.

Faulhaber,

D.

Kamins,

D.

Laidlaw,

D.

Schle-

gel,

J.

Vroom, R. Gurwitz,

A.

van Dam: The Application

Visualization System: A Computational Environment

for Scientific Visualization. IEEE Computer Graphics

and Applications, 9:4,

30-42

(1989)

This paper describes the design principles of

AVS.

C.

Upson: Volumetric Visualization Techniques. In:

D.E

Rogers, R.A. Earnshaw

(eds):

State of the

Art

in Com-

puter Graphics -Visualization and Modeling. Springer-

Verlag,

1991

Description of volumetric techniques from one of the designers

of

AVS

and the designer of Explorer.

Sources

of

Figures

Rendered by DESIGNBASE

Reproduced by permission

of

Dynamics Graphics Ltd.

Reproduced by permission

©

IBM

UK Scientific Centre

Reproduced by permission

British Geological Survey

Reproduced by permission

Ricoh Company Ltd.

Reproduced by permission

F.

Rogers,

1991

Reproduced by permission

of

Stardent Computer Ltd.

SUN

Microsystems, Inc.

Rendered

on

a

SUN

SPARCstation

470VX+MVX

©

1990

SUN

Microsystems, Inc.

Rendered

on

a

SUN

SPARCstation

using SunVision software

©SUN

Microsystems, Inc.

Rendered

on

a

SUN

SPARCstation

VX+MVX using Westover's splat software

Data courtesy

of

Frank Bryan

at

NCAR

Figure 7.27

Figure 7.20,

-7.24

Figure 2.7,

3.3

Figure 7.25, 7.26

Figure 3.10, 7.6,

7.7

Figure 3.5

Figure 7.27, 7.28,

7.29,

7.31

Figure 2.11-2.13

Figure

3.12

Figure

7.14

Figure 7.15,

7.17-7.19

Figure

7.16

Figure

8.8

156 Sources

of

Figures

Courtesy of Dr.

B.

EI-Haddadeh, Figure

3.9

University of

Leeds,

and UNIRAS software

Courtesy of Todd Elvins,

San

Diego Figure

3.11

Supercomputer Center (SDSC)j

Data: Mark Ellisman, University of

California,

San

Diegoj Visualization:

Dave Hessler, SDSCj Software: SYNU

from SDSC

Courtesy of Todd Elvins,

San

Diego

Supercomputer Center (SDSC)j

Data:

Ted

Cranford, University of Califor-

nia, Santa Cruzj Visualization: Todd

Elvins, Phil Mercurio, SDSC, Software:

SUN

Microsystems

Voxvu

Courtesy of Prof. Elliot

K.

Fishman MD

Courtesy of Prof.

T.

L. Kunii

Courtesy of the Lamont Doherty

Geologic Observatory

Courtesy of the Lawrence Livermore

National Laboratory

Courtesy of Silicon Graphics Ltd.

Courtesy of Simultec, Switzerland

Data courtesy of Dr. Michael Torello

Courtesy of Craig Upson

Reproduced by permission

M.

Thalman,

D.

Thalman,

1991

Reproduced by permission of

©

u.

K.

Meteorological Office

Reproduced by permission of

Courtesy of Geoff Wyvill &

Brian Wyvill

Figure

3.13

Figure

2.8

Figure 2.14,

2.15

Figure

7.21

Figure 7.20

Figure 7.11-7.13

Figure 7.22

Figure

8.7

Figure

2.1,

3.1,

3.2, 3.4

Figure 9.1-9.3

Figure 2.4-2.6

Figure 7.4,

7.5

Figure 3.6-3.8,

7.1-7.3

Figure 2.9,

2.10

An Introductory Guide To Scientific Visualization

Navigation menu

Versions of this User Manual:

Views

Navigation