Sunteți pe pagina 1din 10

TRAN2A computer graphics program

to make sculpture
by ROBERT MALLARY
The University of Massachusetts
Amherst, Massachusetts

viewer. On the other hand, he has apparently made


no provision for constructing an actual sculpture from
this authentically three-dimensional information.4
Others have used 2-D plotter drawings to "suggest"
sculpture, or* have used computer graphic material
to embellish the surfaces of a sculpture, or have used
computer graphic output to determine the flat shapes
to be incorporated into a sculpture. But in none of
these latter programs has the computer generated,
processed or delivered up the real thing in the way of
three-dimensional form information.

INTRODUCTION
Historically the techniques of sculpture have tended to reflect the technological character and level
of the society in which the sculpture was made.
If primitive man carved bone and the Greeks cast
in bronze it should not be surprising that sculptors
today are using plastics, lasers, strobes, electronic
circuitry and transducers to link art with contemporary
technology. Even so, this still does not account for
why some of us have been using the computer, assigning it a status above other "art-and-technology"
possibilities and viewing it as nothing less than portentous in its implications for art. *
The computer is special among the technical resources previously available to the artist because for
the first time he has a tool, not only for executing a
work of art, but for conceiving one as well. Once a
computer has been programmed to generate a first
rate work of three-dimensional art with no direct assistance from a sculptor it will be legitimate to speak
of cybernetic sculpture in the fullest sense of the word.
Until then computer sculpture will qualify as cybernetic only in the sense that the design process is substantially facilitated by "intelligence amplification"
which is to say, by the use of advanced computer
graphic interactive systems.2
The core problem in computer sculpture is to program the machine to take in, manipulate and give
back three-dimensional information which can be used
to make sculpture. For example, a Massachusetts
sculptor, Alfred Duca, with the help of IBM programmers, has used a computer tape and an N / C
machine tool to carve out a large and intricate spherical
sculpture in metal. 3 Michael Noll of the Bell Telephone
Laboratories has programmed linear stereo-drawings
which appear three-dimensional when seen in a stereo-

TRAN2
In beginning about three years ago to work on
TRAN2 our intention was not so much to anticipate
the computer sculpture of the future, with its awesome
kinetic and form transformational capabilities, as to
take a small but real step ahead within our immediate
resources. However, TRAN2 does qualify as a basic,
or prototype, computer sculpture program in the sense
that it provides for a full description of volumetric
objects within the machine, processes this information
in a meaningful way, and generates a usable output.
Moreover, the program is workable in that it has actually been used to make a series of sculptures (see
Figure 1 and Figure 2). Now written in Fortran IV
for the IBM 1130 computer and plotter, when it has
been rewritten for the display it will be upgraded as
a more fully interactive program allowing for the almost
instant manipulation and transmutation of forms.
Form description
Crucial to the processing of three-dimensional form
informationbe it architecture, sculpture or indus451

452

Fall Joint Computer Conference, 1970

form and space description are more suitable than


others only in respect to the purpose of the program.
For example, MAGI uses a system called "combinatorial geometry" 7 to assemble forms within the computer, but it is difficult, using this system, to revise
and further develop the forms as freely as a sculptor
would like. Other methods, which refer to surfaces
rather than volumes, would seem to be more promising
for shaping and processing three-dimensional material
assuming, that is, that provision is made at some point
for fully enclosing and defining these surfaces as bounded and complete sculptural volumes. 8 But whatever
the system used, if the computer is to be involved with
sculpture in an authentic way it must be given either
a comprehensive numerical description of the material
it is to work with or the means to generate this material for itself.
In this respect TRAN2 does both, using contour
sectioning, or "slicing," as the basic method of form

Figure 1"QUAD III" TRAN2 computer sculpture in laminated


veneer. 60" high1968. The sculpture was made from plotter
templates provided by the Amherst College IBM 1130 computer

trial objectsis the 3-space depiction of the object


within the machine.5'6 Perhaps there is no "best" way
of doing this, keeping in mind that some methods of

Figure 2"QUAD IV" A TRAN2 computer sculpture in


laminated marble. 11" high1969

TRAN2

453

description and form generation. In effect, the form


is slicedmuch as an apple or a chunk or baloney
might be sliced, into a series of thin parallel cross sections of equal thickness. These two-dimensional slices
comprise a vertical set, or array, of modular form information units which are then graphed, digitized
and encoded onto computer punch cards. Each of the
slices has an axis point and a reference ("tick") mark
to position it on the vertical axis relative to all the other
slices comprising the set. It is by means of this "stacking" of two-dimensional data that the program converts
standard computer graphic capabilities to the requirements of three-dimensional form description.

Two modes for form description input


TRAN2 provides for two modes of form description
input, though others might be devised. The first, called
INITL, requires an antecedent hand-made "prototype" form which must be traced three-dimensionally
using a special contour grapher designed for this purpose (see Figure 4). The grapher has a swinging probe
which, when held gently against the slowly revolving
form, traces off the contour levels one-by-one and
transfers them to graph paper. These graphs are then
digitized with X / Y coordinates, using the vertical
axis as the Z coordinate, and transferred to punch
cards. Between 48 and 100 contours are needed, which
is insufficient to guarantee a smooth, continuous definition of the form (i. e., without a visible demarcation,
or "step," between each contourand the next), but is a
practical minimum considering the relatively small
capacity of the computer which has been available

INPUT

TRANSFORMATIONS

OUTPUT

TRAN2
SUBROUTINE SEQUENCES
AND OPTIONS

Figure 3Block diagram showing the basic program structure


of TRAN2

Figure 4Contour sections are being traced from a styrofoam


prototype form mounted in the contour grapher. The contour
"slices" are then graphed and punched onto cards to provide
the form description input for the INITL input mode

to us and the amount of hand work required to translate the computer output into an actual sculpture inthe-round.
The second input mode, called PROFIL, dispenses
with the prototype form, but in its place the computer
must be given coded profile drawings (see Figure 5).
Eight profiles are stored at a time, though the computer
needs only one to generate a form with radial symmetry. Two profiles generate a form having two planes
of symmetry and three generate a form with bilateral
symmetry. A set of four different profiles generates an
asymmetrical form, this being the number which is
normally specified unless one of the symmetrical
schemes is used.
By following the typed instructions given to it at
the console the computer "fills in" the form between
the profiles (see Figure 6). In calling up subroutines

454

Fall Joint Computer Conference, 1970

Figure 5Profile drawings for use with the PROSA input


mode. Needed are two facing profiles on the Y-Z plane, making
four in all. The profiles are generally designed as matching sets,
though eight are filed in the computer at a time and are interchangeable

such as ELIPS, OVAL1, OVAL2, and SUPER the


sculptor can generate forms for which all the cross
sections at all levels along the vertical axis are either
perfect circles, ellipses, super-ellipses, or are one of
two kinds of ovals. In other words, the coordinate values
taken from the profile drawings are used by the computer to shape these geometrical cross sections (see
Figure 7).
While this ability to generate an asymmetrical sculpture characterized by a uniform, and perhaps symmetrical, geometry throughout all the cross sections
is no guarantee of "beauty," it at least unlocks some
interesting possibilities for sculpture which deserve
further exploration. For that matter, the extensive
use (or should we say over-use?) of symmetry in so
much current computer art has yet to be applied to
three-dimensional structures and configurations, and
even in our use of TRAN2 we have hardly begun to

Figure 6Diagram showing how the computer uses the PROSA


input mode to combine 4 profiles and from them derive a contour
slice ADBC. 48 of these sections are stacked on the vertical axis
to define a complete form. It can be seen that if the same profile
were to be used in all positions ( + X and X, and + F and Y)
the form would have radial symmetry and all the sections would
be circles. Available as options are yet other systems of symmetry

TRAN2

scratch the surface in this area. Evidence of our continuing commitment to asymmetry, as against symmetry, is the fact that QUAD, which shapes an asymmetric contour section based on quadrants taken from
four different ellipses, is still the most used subroutine
within the PROFIL group of input subroutines.
The transformation subroutines
The computer, once it has either been given or has
generated for itself the form description data it re-

Figure 7This diagram shows how the computer, using OVAL


2, moves the minor chord AB to the center of the major cord CD,
making a cross section which is a perfect oval (AB and CD are
opposite points on two adjacent profile drawings). OVAL I
moves the major chord to the center of the minor chord, E L I P S
moves each chord to the center of the other to form a perfect
ellipse, and QUAD shapes a cross section comprised of quadrants
from 4 different ellipses

455

quires, in effect is converted into a sculptural modeling


and shaping tool. This is accomplished by calling up
one or more of the transformation subroutines such as
EXPND, EXPNT, ROTES, or MOVES. These generate permutations on the form description data by means
of various mathematical functions which stretch and
compress the form in a variety of ways. 9 The computer
requests instructions regarding the kind of transformation which is wanted and the specific values involved. For example, E X P N D is used to stretch or
compress a contour section on the X or Y coordinate
or on both. If 1.0 is typed for X and the same for Y
there is no change in contour section or in the over-all
form. But if 0.5 is typed for X and 2.0 for Y the form
is halved on X and doubled on Y. Because the computer calculates these values for only one contour at a
time, and in sequence, it is possible to specify incremental transformations, either of a positive or negative
kind. In this way the sculptor can induce a more drastic
transformation at the top of the form than at the bottom (or vice versa).
In understanding the action of these transformation "templates" it is important to bear in mind their
dependence on the underlying structure of the program,
based on the stacked layers of contour sections, which
limits them to the X / Y horizontal plane. As yet no
provision has been made in TRAN2 to introduce transformations on the Z, or vertical, coordinate. EXPND,
which is TRAN2's most basic "modeling tool," evenly
stretches or compresses the contour section over the
X / Y plane of transformation. In other words, the form,
in respect to its length and breadth, is uniformly
stretched or compressedwhich means that the kind
of transformation is constant even when the values
are changed.
E X P N T (for exponential) surmounts this invariance
in offering the sculptor a wider range as regards the
kind of transformation permutations available to him.
Crucial to this much enlarged transformational capability are various logarithmic and exponential functions and values which can be typed in.10
For example, using E X P N T the form can be compressed on one side and expanded on the other, or the
degree of expansion or compression can be made to
vary continuously along the X / Y plane of transformation. Once TRAN2 has been given a graphic console
and provided with proper instrumentation it should
be possible for the sculptor, using the typewriter and
function keys, to specify the class of transformation
he is seeking while he spins a knob in order to continuously vary the values. This will enable the sculptor
to scrutinize his creation, which is slowly swelling and
contracting on the display, while he waits to seize the
moment it "gells"either as the original image in

456

Fall Joint Computer Conference, 1970

Figure 8TRAN2 perspective plotter drawing of a sculptural


form, which can be thought of as a preliminary sketch or "study."
The sculptor decides on the basis of these drawings whether to
complete the sculpture in some durable material

his mind's eye or as an unexpected discovery. The


sculptor should also have the option of subjecting his
form to a series of transformation sequences which
are more or less automatic in their operation. As
programmed transformation "scenarios" they will be

Figure 10A contour section has been projected and traced


onto a %" thick piece of luaun veneer and is being cut out with
a bandsaw. Marked on the veneer are the center hole and a
registry mark used to properly orient the slices in the stack as
the sculpture is being assembled and laminated

Figure 9Photo of a complete set of plotter drawn contour


sections ready for projection onto the material to be used in
making the TRAN2 sculpture

only partially under the control of the sculptor in the


sense of his fully anticipating what happens next. They
will be useful, not only as an expanded computer-aided
design resource, but for their cinematic possibilities.
As a third option the sculptor should also be able to
"harden," or "fix," any of the transformation permutations at any given strategic point in the procedure,
using this new data set to replace the original form
description material as the basis for a further round
of transformation sequences.
Using ROTES incremental transformations can
also be used in order to rotate, or twist, the contour
sections sequentiallythe final result being to twist
the form as a whole. Using this subroutine forms have
been twisted 360, and even 720, about the axis (with
rather bizarre results, it might be added). In general
a subtle, less drastic, rotation is preferablefor ex-

TRAN2

457

ample, the 60 rotation used in designing QUAD III


(see Figure 1).
It should be emphasized that the TRAN2 transformations are cumulativei.e., in being added together in sequence they are in effect combined. For
example, the initial input form can first be expanded
using EXPND, then have its center point shifted on
the X / Y plan using MOVES, then be twisted using
ROTES. Sorely needed is a subroutine on the order
of MOVES, but more versatile and drastic in its transformations. This subroutine would completely reorient the sculpture relative to its axis, thereby assigning it a new top and bottom and a new set of contour sections. By enabling the transformations to work
on the form from any direction, and along any designated plane of transformation, the number of transformation possibilities would be multiplied many times
over.

Figure 12The cut out luaun sections are stacked on a steel rod
preparatory to gluing and laminating

The output subroutines

Figure 11The center hole is drilled

The group of output subroutines determines the kind


of drawing the computer is to make. PERSP specifies
a perspective drawing (see Figure 8), the sculptor
typing in the angle of vision he wantsat, above or
below eye level. He also specifies the view or views he
wishes, which is apt to be a complete set taken at
regular intervals around the form. By instructing the
plotter to make a series of drawingssay, at 15 or
30 increments, he in effect revolves the form before
his eyes and achieves a rough idea of what it would
look like if he were actually to construct it. It is in
this sense that TRAN2 is an example of how computer
graphic techniques can be exploited by the sculptor
to extend and enhance the usefulness of drawing as a
way of sorting and clarifying visual ideas preparatory
to executing them in a three-dimensional medium.
CONTR calls up a plot of the entire set of contour
sections as orthographic projections (see Figure 9).

458

Fall Joint Computer Conference, 1970

Capabilities of an improved TRAN2 program


TRAN2 is slow in its operation when measured
against the potential of a large c.p.u. and a true interactive program. It is also limited in that it can handle
solid, volumetric forms oriented around a central axis;
concavities are possible using the I N I T L input mode,
but undercuts are ruled out. Nor is it possible, using
PROFIL and its maximum of four profiles, to generate
a concavitythough the addition of, say, 12 profiles
might improve the situation in this regard (see Figure
14). It will be necessary either to enormously expand
the resources of TRAN2, or to develop a library of
specialized computer sculpture programs, if planar,
linear and open-form sculptures are to be made using
the computer.
An ideally interactive program along the lines of
TRAN2 would allow the sculptor to draw his profiles
directly on the display using a light pen11 or sketch
pad arrangement.12 He should then be able to evaluate
his work by referring to a graphic display which shows
the form slowly revolving about its axis. Next he should
be able to set in motion a series of transformation
scenarios of the type already described, then switch
back to the sketch modeperhaps to refine the details
of the form with the light pen or stylus. If he should

Figure 13The stacked sections have been laminated together


and the irregularities ground down and smoothed

Each of the contours includes both a center point and


a reference mark to orient the contours in the proper
position one to the other. The entire set of contours
is photographed as an 8 X10 inch positive transparency,
inserted into an overhead projector, projected onto
some appropriate material such as wood or plastic,
and traced. The set of traced contours is then cut out
(see Figure 10), the center holes are drilled (see Figure
11), and the contours are stacked over a metal rod
(see Figure 12). Finally the contours are glued, laminated together under pressure, ground down to remove
the "steps" and irregularities (see Figure 13), and
smoothed and polished. These, of course, are manual
operations at the handicraft level, but in principle,
inasmuch as the computer has generated all the essential three-dimensional information, the sculpture
could also be made using an N / C milling machine.

Figure 14A diagram showing how 12 profiles, instead of the


4 now used in PROSA, will permit the sculptor to introduce
concavities into his form and generate more complicated and
interesting surfaces. Needed is an elegant algorithm to generate
the smooth continuous curve which links the 12 points

TRAN2

decide he has taken a wrong turn at some point he


should be able to call upon the computer's memory
and return to an earlier stage in order to try something else. In other words, the program should have
what amounts to a quick rewind and playback capability based on a complete and permanent log of the
entire design process. Apart from its value to the
sculptor, a record of this kind might also be valuable
as a way of investigating systematically the dynamics
of the creative process and determining how one sculptor works as against another in creative problem solving.
It might also be helpful for the computer sculptor,
as for the architect as well, if he could rely on multiple
displays, each unit providing information regarding
a different aspect of the on-going problem. For example,
he might study several views of the same sculpture
at once, compare two or more current versions, or
refer back to an earlier version for comparison. Eventually a practical stereo display, and possibly a holographic display,14 will optimize the efficiency of computer graphic systems for communicating three-dimensional form information with maximum visual
clarity and precision. In fact, we might define the
eventual goal of computer graphic systems as providing the sculptor, architect or designer with a virtual
real-life experience of the form-in-progress so he may
design it better, more rapidly and with more assurance
that it will conform to his expectations once it has
actually been made.
A look over the horizon
The computer sculptor will make better use of
the computer to evaluate his work-in-progress when he
no longer has to rely on the rather crude wire cage
drawings to which graphic displays are now generally
restricted. A minimal step in the right direction would
be a drawing consisting of black lines against a white
field to replace the reverse image of white on black.
More remote, though beginnings are being made,13-14
would be a simulated light and shadow version of the
form on the display giving the sculptor the option of
simulating a procedure he is apt to follow in his own
studionamely, to adjust and vary the lighting on the form for the clearest effect. Sequencing
the shifting patterns of mobile lighting configurations,
either with real lighting equipment or as simulated
images on displays, is an obvious computer potential.

beyond computer-aided intelligence amplification (computer-aided design, in other words) into the more
creative aspects of the design process. The computer
will be more than a "slave"; it will be more like a collaborator or a virtual surrogate for the sculptor himself. According to his inclinations the sculptor will
vary the manner and degree of his involvement with
the computer. He will use and interact with it, monitor
it or leave it, as it were, to its own resources.

CONCLUSION
The sculpture now made with the help of TRAN2
does not forecast the look of the computer sculpture
of the future, which will be mainly kinetic, have multimedia features and will probably be based on a cinema
type projection system linking the computer with
holographic techniques. The relevance of TRAN2
in this connection is thatapart from the claims to
be made for it as one of the very first pioneering efforts
in the fieldit does forecast, in its use of mathematical
methods, an approach to form description, as well
as form manipulation and metamorphosis, which will
be crucial to these kinetic media of the future.
In conclusion I feel it an obligation to remind those
who know more about computers than they do about
art that I am at outs with some of my more conservative artist colleagues who deny that the computer
can make any constructive contribution to art at all.
But what is more nettling is that I am also at outs with
some of my more avant-garde colleagues who will
acknowledge (or even insist) that the computer can
play a role in art but that TRAN2, which according
to them is involved in an anachronistic kind of "object" art, is not a valid way to go about it. I differ
with these latter critics in holding that "object" sculpture (which is the kind most people think of, whether
it be realistic or abstract, when they think of sculpture
at all and which is the only kind we can as yet make
with TRAN2) still offers unexplored potentialities for
the computer to help uncover.
In any event, a beginning must be made at some
point, and for the present it may be a sufficient achievement to have demonstrated that the computer can
play a role in the making of sculptureall apart from
the question of how well it has done so.

REFERENCES
Cybernetic sculpture
Ultimately interactive programs will become truly
cybernetic in the sense that the computer will move

459

1 J REICHARDT
Cybernetic serendipity, the computer and the arts
Studio International London and New York 1968

460

Fall Joint Computer Conference, 1970

2 R MALLARY
Computer sculpture: six levels of cybernetics
Artforum Vol 7 No 9 pp 29-35 May 1969
3 R CHANDLER
Design for numerical control machining
Machine Design pp 4-24 February 15 1968
4 A M NOLL
The digital computer as a creative medium
I E E E Spectrum Vol 4 pp 87-95 October 1967
5 W FETTER
Computer graphics
Annual meeting of the American Society for Engineering
Educators 1966
6 C M THEISS
Computer graphic displays of simulated automobile dynamics
A F I P S Conference Proceedings Spring Joint Computer
Conference Vol 34 p 289 1969
7 MAGI
Description of the MAGI technique for accurate modeling and
graphic display of three-dimensional objects
Mathematical Applications Group Inc White Plains
New York 1967
8 T M P LEE
A class of surfaces for computer display
AFIPS Conference Proceedings Spring Joint Computer
Conference Vol 34 p 309 1969

9 C CSURI J S H A F F E R
Art, computers and mathematics
AFIPS Conference Proceedings Fall Joint Computer
Conference Part 2 Vol 33 p 1293 1968
10 C CSURI
Leonardo: circle to square transformation
The Magazine of the Institute of Contemporary Art
Number 5 London p p 27-28 August 1968
H I E SUTHERLAND
Sketchpad: a man-machine graphical communication system
M I T Lincoln Laboratory Technical Report N o 296 January
1963
12 M R DAVIS T O ELLIS
The rand tablet: a man-machine communication device
AFIPS Conference Proceedings Fall Joint Computer
Conference P a r t 1 Vol 26 p 325 1964
13 C W Y L I E G R O M N E Y D C EVANS*
A ERDAHL
Half-tone perspective drawings by computer
Technical Report 4-2 Computer Science University of
Utah-Salt Lake City Utah February 12 1968
14 H WILHELMSSON
Holography: a new scientific technique of possible use to the
arts
Leonardo Pergamon Press Oxford England Vol 1 Number 2
pp 161-169 April 1968

S-ar putea să vă placă și