UNIVERSITY of CALIFORNIA
Santa Barbara
DEPARTMENT of GEOGRAPHY
MA Thesis
June 2004
Human psychological response to landscape visual
filtering in animation design
Marco Ruocco
Landscape is a valuable visual resource characterized by a
specific degree of visual accessibility, which affects the
exposure to the resource, human perception, and the outcomes of
exploration. The motion along a trajectory allows the viewer to
obtain a more specific view of the landscape, enhancing her
information-seeking activity through a negotiation with the
multiple visual aspects of the environment. As a conceptual tool,
we consider this form of mediated exposure to landscape as a
visual filter that selects only part of the landscape and
temporarily hides the rest. This experimental study aims at
verifying the hypothesis that the trajectory of the viewer on the
landscape is a determinant factor in what perceptions and
experiences are finally achieved. The objective is to show that
the visual accessibility properties of landscape are not
isotropic, but rather patterned in landscape- and
trajectory-specific ways. The experiment consisted of the
development of six computer graphics fly-by sequences of three
different landscapes (an agricultural plain, a narrow valley, and
a steep hill), chosen to represent different terrain types. Each
landscape was animated at low altitude in a terrain-following
mode, and at high altitude in uniform mode. In a between-subject
design, two groups of participants were asked to evaluate and
self-report their perceptions, aesthetical insights, spatial
knowledge and sense of place impressions on the three landscape in
the two altitude conditions. The results of the experiment suggest
that the visual landscape is patterned in terms of how
accessibility determines experience, since there are differences
in specifically predicted classes of responses. For example, the
effect of mountain sheltering is felt only at low altitude in a
sheltered terrain, and not in any other condition. The landscape
seems to offer a different "face" (in the many dimensions that
are considered) according to the trajectory of motion from which
it is seen.
Contents
1 Introduction
2 Literature Review
2.1 Landscape
2.1.1
History
2.1.2
Aesthetics of landscape
2.2 Psychological
approaches
2.2.1
Ecological psychology
2.2.2
Psychology of emotion
2.2.3
Psychology of motion
2.3 Spatial
knowledge
2.4 Visual
landscape resources and design
2.5 Cartographic
visualization
2.6 Sense of place
2.7 Approaches in
Cinema and Photography
2.8 Planetary
exploration
3 Conceptual Framework
3.1 Introduction
3.2 Theoretical
aspects of landscape
3.3 Human-landscape
psychological relationships
3.3.1
Focusing on aesthetics
3.3.2
Spatial knowledge
3.3.3
Sense of place
3.4 Research
strategy
3.4.1
Trajectories
3.4.2
Landscape categorization
4 Pilot Experiments
5 Methods
5.1 Design
5.2 Participants
5.3 Materials
5.3.1
Landscape stimuli
5.3.2
Instruments: questionnaire
5.4 Procedure
6 Results
7 Discussion
8 Conclusions
A 1 Appendix 1 : Questionnaire
Chapter 1
Introduction
The visual properties of landscape are a valuable resource that is
manifest, for example, in the aesthetical reaction that most of us
experience by viewing striking scenery. In fact, the current
interest in visual impact and scenic quality indicates the
increasing awareness of the community in landscape visual
preservation. Acquiring the ability to access the resource of
landscape is as important as the actual presence of the resource
itself. Accessibility can be seen as the main factor of an
epistemological quest that underpinned the evolution of
psychological human-landscape relationships.
Landscape has been considered as "a way of seeing"
[Lowenthal, 1961].
This intrinsic quality might indicate that it
is necessary to consider the process of observation as a
conceptual basis for landscape. This is especially true when
observation is considered as a process of selection of visual
elements from the totality of the visible environment.
The interest in dynamic visualization can be related to the recent
attempts to find novel and more effective methods of using
increasing amounts of geographic data. Scientific visualization
has brought forward the idea of "seeing the unseen", an
attempt to expose the hidden properties of data, and extracting
their information content through visual representation
[Buttenfield and
Mackaness, 1991]. Similarly, it is argued here that the
"seeing the unseen" in a landscape context consists in
increasing the degree of exposure of landscape to the viewer. The
process responsible for exposing landscape is strongly related to
the dynamics of observation: it involves the selection of hidden
or fragmented visual elements from a series of views and their
combination in integrated experiences.
It is suggested here that the trajectory of the observer on the
landscape is critical in determining the characteristics of the
experience of landscape. However, reasonably similar trajectories
could suggest to us how comparable sets of landscape elements give
rise to different experiences. Such difference will be patterned
enough to help us determine their specific influence on the
observer's perceptions. The interesting aspects of comparing the
trajectories considered in this study is that they vary as a
function of the form of the terrain. Such adaptivity is what
characterizes our trajectories as terrain-based trajectories,
instead of the other trajectories that are only an expression of
independent movement in space.
Since the interest is placed on trajectory and not on viewpoint,
the traditional approach of measuring preference using static
landscape photographs was enhanced by means of introducing dynamic
(although passive and pre-rendered) video animation. In fact, the
animated landscape-stimulus allows us to address the problem of
considering the experience of landscape with reasonably effective
instruments. The better experience-inducing instrument of
animation also allows us to expand the front of investigation,
which in previous studies was limited to preference and spatial
cognition.
Finding evidence of the influence of trajectory characteristics on
landscape experience would assign to trajectory the crucial role
of enabling the access to the visual resource of landscape
discussed above. By analyzing the reactions of participants to
animations it will be possible to know whether experience varies
together with the trajectory of approach to landscape, and whether
such relationship varies, in turn, with the type of landscape
being shown.
The contributing elements to landscape experience are constituted
by spatial knowledge, landscape aesthetics, the emotional impact
of environments, and the sense of place by which locations are
characterized. The experimental framework accounts for such wide
range of intellectual and emotional facets that characterize the
psychological relationship between landscape and human beings.
More specifically, in the abundance of visual data available from
the landscape, the information that makes us experience an
aesthetical reaction is a deeply intertwined combination of visual
factors. The cognitive framework adopted in this study relates the
aesthetical reaction to the spatial information about the
landscape available to the observer.
The creation of trajectories over and on a landscape can be
conceptualized as a filter that selects the possible ones from the
impossible ones (humans cannot fly without technological aids
such as airplanes), and second, those eliciting specific
perceptions over others. Filtering is thus considered to be a
first conceptual step in determining the abstract conceptual
structure that influences the epistemological access to landscape.
A filter is primarily a technological device. However, it should
not be seen as a constraint to the epistemological scope of this
thesis.
This Chapter will consider in sequence the history of the concept
of landscape and the theories considering landscape aesthetics
(Section 2.1); the psychological
approaches from
ecological conceptualizations to ideas about emotion and motion
(Section 2.2); behavioral geography
concepts
about spatial knowledge (Section 2.3); a
review of
visual design approaches (Section 2.4); the
cartographic and visualization perspective on this research
problem (Section 2.5); some
considerations
about sense of place (Section 2.6); and
finally some
material from potential fields of application of this study, and
specifically cinema and photography (Section 2.7)
and
planetary exploration (Section 2.8).
1 Landscape
Landscape has been considered an elusive concept. The landscape
concept overlaps in part with the concepts of region and scenery.
We can approach the concept of landscape from two main directions
of investigation: it can be seen as an object, that defines a
particular physical domain and a natural system supporting life;
but also subjectively as scenery to contemplate from a particular
viewpoint [Tuan, 1979].
Lowenthal [1961]
explicitly defines landscape as "a
way of seeing", which therefore has much to do with the viewer
as with the viewed, a mediation of the external world through
subjective human experience in a way that the concepts of region
or area do not readily suggest, also indicating an epistemological
perspective to the problem of definition, dependent on the
individual who is approaching landscape.
The subjective mediation of objective reality, carried out through
"people's eyes" [Lowenthal,
1966], means that the
combination of objective and subjective takes place in the mind,
or "in the mind's eye" [Tuan,
1979].
Granö [1929] provides an
interesting historical example of how
the subjective/objective differentiation of landscape can be
systematized. He formulated a new discipline called Pure
Geography, in which the region was adopted as the basis of
scientific investigation. In particular he suggested to define
regions in the environment on the basis of the subjective and
perceptual experience of the individual, thus proposing an
egocentric conceptualization of the environment. Visual, auditory
and olfactory regions were referred to a perceiving observer, thus
the subject's experience defined the object of study in the world.
Landscape, in Granö's view, was a region defined by a
threshold of egocentric distance from the observer, and extending
up to the horizon, making in practice the concept correspond to
the background component of terrain of a scene (this specific
definition is not supported in this thesis).
1.1 History
The history of the term "landscape" begins in the 16th century
when the Middle Dutch word "landskip", at that time
used to indicate the works of Flemish painters, was translated
into English [Lorzing, 2001].
The word "landskip" referred specifically to a painting
of a prospect involving elements like hills, woods, ruins,
valleys, and towns [Shepard, 1991].
16th century Flemish painters
did not promote the making of a faithful depiction of the
environment: rather, their paintings had an overwhelming spiritual
and allegorical component, although their symbolic content was
articulated in a landscape view. Only in the 17th century Dutch
painters presented a more "realistic" and "documentary"
attitude than their Flemish counterpart.
The Dutch "landskip" still resulted from a considerably
creative composition of directly experienced scenery with other
landscape elements (like buildings and churches). Those elements
were placed in semantically "strategic" positions according to
an underlying desire to express their interest for their land, and
to extend wishful control on the menacing flooding waters,
especially by depicting land which was not flooded [Adams, 1994].
The term "landscape", which derived linguistically from
"landskip", similarly refers to the definition of
"view or prospect originated from one point of view" as
given in the context of artistic painting later in the 18th
century [Cosgrove, 1984].
gure
Figure 1: William Turner. Fall of the Tees, Yorkshire.
c.1825-1826.
Watercolour on paper. Private collection, UK
In the context of 19th century English painting, William Turner
was making use of the variable of viewpoint location in order to
distinguish his works from those of other contemporaries, that
were making portraits of picturesque scenes adopting a point of
view located at ground level, which corresponded to the typical
(and loathed by some) tourist experience of the landscape. For
example, the painting "Fall of the Tees, Yorkshire" of
1827, by William Turner (see Figure 2.1), is
based on a choice of
viewpoint location that is placed in mid-air in front of the
waterfall, instead of adopting a standard and ground level view
[Helsinger, 1994].
The contemporary interpretation of the term "landscape" took
shape in the late 19th century as a portion of territory that is
comprised in one view, including its constituting objects in their
pictorial aspect [Harrison, 1994].
This transition signified the
transformation of the term from the status of identifier of a
representation to the specific content matter of such
representation. Later it became attached to the cultural elements
comprised in the view [Cosgrove,
1984].
According to Harrison [1994],
the effect of a painting is not
necessarily dependent upon the location of the viewpoint or the
latent landscape content that is disclosed when that viewpoint is
adopted. Rather it has to be found in a "coincidence
between thought and making" that is beyond viewing and latent
content. By "effect" it was meant the ability to convey a
naturalistic impression to the viewer.
1.2 Aesthetics of landscape
In philosophy, aesthetics are the study of the meaning and the
nature of art, but the term has a different meaning when applied
to the environment [,see]]berleant or to media
[,see]]zettl. The study of the aesthetics of the
environment finds a theoretical justification in the original
interest of aesthetics for the natural world, even if historically
only a few philosophers diverted their attention from art to
nature.
First, an informational interpretation of aesthetics is based on
seeing the environment in its collative properties, that is, the
properties that link structurally the elements of the environment
together. Aesthetic response is a function of those collative
properties (such as novelty and complexity) and of the explorative
behavior that they consequently elicit. In particular, aesthetics
interprets exploration in two ways: specific, such as a
search in the landscape for information (i.e., for a specific
view) to diminish uncertainty and satisfy arousal; or
diversive, that is, aimed at finding a stimulus
configuration that grants optimal levels of uncertainty and
arousal [Hartig and Evans, 1993].
Kaplan [1987] expanded the
informational theory of aesthetics
suggesting that our preference for landscape depends on the kind
of information processing that is necessary to perceive and relate
with the environment. According to this theory, we have evolved as
human beings by preferring those landscapes characterized by a
particular balance between order and uncertainty, that are in turn
organized in four separate factors: coherence,
complexity, legibility and mystery (see
Table 2.1). The element that helps the
understanding in the
immediate time frame is the coherence of the environment.
When the observer is instead involved with the environment, such
as during exploration, it is the complexity of the
environment that engages aesthetically the individual. In case the
observer is engaged in a prolonged interaction with the
environment, the understanding effort is coupled with aesthetical
satisfaction if the environment is legible, and, in case
of exploration, if it generates mystery [Kaplan, 1987].
|
Understanding |
Exploration |
| Immediate |
Coherence |
Complexity |
| Inferred |
Legibility |
Mystery |
Table 1: The information model of environmental preference,
from
Kaplan [1987]
The mystery factor is exemplified by a landscape picture
representing a road in the foreground that turns and disappears
from view, indicating that further environmental information is
available as soon as the observer changes her location and moves
towards the hidden part of the landscape where the road turns.
Another classical example of promised information is a view on a
landscape that is partially occluded by some foreground foliage,
suggesting that more satisfying environmental information is
available as soon as the intervening element is removed from view
by walking forward. Also, the mystery factor appears in the
literature as the factor explaining the greatest variance when
compared with the effect of the other three, thus indicating its
dominance in explaining landscape aesthetics.
Appleton [1996]
proposed the theory of
Prospect-Refuge, based on the non-experimental analysis
of landscape paintings, according to which one likes or dislikes a
scenery or landscape depending on biological considerations of
survival. Prospect-Refuge theory is more specific than
habitat theory and is dependent on the imagination and experience
of the observer, as well as on environmental conditions.
In particular the two situations of seeing and being seen are the
fundamental building blocks of an aesthetical consideration of
landscape. From an evolutionary point of view we have evolved by
preferring those areas that afford prospect on the prey and at the
same time refuge from a possible predator ("to see and
not to be seen"). Landscape preference as suggested by
Prospect-Refuge is not a conscious activity but in some
ways it underscores our sensations of pleasure in the landscape.
Mitchell [1994] argues that
Appleton's theory presupposes the
presence of a universal and "natural" observer of landscape,
leaving out other categories of observer, such as woman, gatherer,
tourist, etc. However the author also suggests how the predatory
and violence-based (hunting, war, surveillance) observation
underlies any category of observer. This is in line with a
theoretical linkage between Appleton's theory and evolutionary
psychology, that defines evolutionary biases as the basis for
landscape aesthetics.
In summary, the variable land surface of Prospect-Refuge,
with places to hide and prospects controlled by topography, finds
aesthetical justification by means of an evolutionary preference
for survival. Topographical features, when they control the
observer's viewpoint, afford specific aesthetical reactions.
Aesthetics is an important contribution to our experience of
landscape. It has also been suggested that landscape aesthetics is
not to be considered as an applied form of a more general
aesthetics (i.e. the one studied in Philosophy), but rather the
most fundamental aesthetics for human beings, mainly because
humans have learned to evaluate the external world (i.e. the
environment) before anything else.
2 Psychological approaches
2.1 Ecological psychology
Ecological perception
Gibson [1979] proposes the
concept of ecological perception,
that rejects the previous "classical" view adopted in Psychology
that considered the human being to be operating as a passive
receptor of strictly sensorial stimuli from the environment.
According to that view, those stimuli were processed and
recomposed internally to make sense of the external world. Instead
he suggests that the process of perception is in reality one of
dynamic interaction between the human being and the environment:
humans walk around the landscape, turn their heads in different
directions, in a word they attune to the environment in their
continuous search for information. They are active receptors, and
they can improve at that by means of experience.
At base level, perception is different from sensation. Although we
perceive invariants because they are correlated with sensorial
stimuli, we can have perceptions without sensations. In fact, an
edge of an object that disappears behind another one does not
cease to exist for our perceptual system: although we don't
receive direct sensorial stimuli from it anymore, we know that it
is simply hidden.
Perception is also considered by Gibson to be strictly related to
the natural motion of the observer in the environment, as in
reality there is no such a thing as a static observation, since,
even without moving our bodies, our eyes are always in continuous
motion [Gibson, 1979].
Affordances
Gibson's ecological approaches to the psychology of perception
also suggest a functional interpretation of the environment that
can help explain the processes underlying the human-landscape
relationships considered in this study.
The concept of affordance was proposed by Gibson to
explain the fact that humans can directly perceive the functional
properties of the elements of the environment without having to
completely reconstruct those from individual percepts. For
example, we don't need to reconstruct a complete mental model of a
chair from individual percepts in order to deduce that we can sit
down on it. According to the ecological framework of mutual
human-environment interaction, humans are able to function
according to the affordances they perceive while interacting with
the environment.
This functional interpretation offered by Gibson also indicates
how perception might underlie all human-landscape psychological
relationships, extending the concept of information pick-up beyond
the single percept to other more complex organizing principles
[Gibson, 1979].
Perspective and invariant structure
A useful distinction to make is between the perspective structure
and the invariant structure of the environment. The perspective
structure refers to the changing appearance of the environment as
an observer moves through it, and can be described in terms of
optic flow as indicated in the next Section. For example,
when we approach a chair, the actual shape of the legs of the
chair change and deforms according to the continuously changing
angle at which we are looking at it.
The invariant structure, on the other end, consists of the
information content of the environment, articulated in terms of
the so-called invariants. The invariant structure is perceivable
through the continuously changing perspective structure, but it is
not any particular instance of the arrested optic flow
that we can abstract from our experience of continuous motion in
the environment by freezing time to one instant
[Gibson, 1979]
God's view
Another concept that will be useful in this study is the so called
God's View of the environment. It consists in the
integration of all the possible perspective views available on a
given environment, and it is not simply the "sum" of those
views. For example, considering a landscape, the God's
view will consist in the integration of views available along all
possible trajectories (which is an infinite set) described in the
space above the terrain.
Isovist
A concept related to the visibility-based view of landscape is the
isovist [Benedikt and
Burnham, 1985]. The isovist is an
extension of the Gibsonian concept of optic array and it could
serve as a foundation of studies dealing with the perception of
space.
In practice, it consists in all of the information arriving to the
eye by direct light rays (that is, without considering
reflections), containing cues about the distance of objects and
the layout of the environment. Variations in the isovist
induce variations in the perception of the environment, in a
manner that can be controlled experimentally.
The geographical concept of viewshed, as explained in the
Section "Visual landscape resource and design" below,
is related to isovist, and indicates in particular the
area of terrain that is visible from one defined viewpoint.
Viewshed is perhaps more directly applicable to landscape
modeling approaches, but the isovist is more suitable to
investigate perceptual and aesthetic response.
2.2 Psychology of emotion
Environmental psychology is concerned with the impact of physical
stimuli on human emotions and on behavior. It is based on the
existence of a metric and taxonomy for the description of the
ordinary physical environment. This description is based on the
two factors of sense modality and information rate, which
are simplified by the mediating variables pleasure,
arousal and dominance, correlated empirically
with stimuli and behaviors [Mehrabian
and Russel, 1974].
Information rate represents a unified measure to
integrate the diverse concepts of complexity, randomness,
heterogeneity, novelty, etc. Those aspects relate to information
in the sense that they are based on the uncertainty that
characterizes the displays. In complex situations (including video
recordings) verbal measures can extract the information rate of
the environment [Mehrabian
and Russel, 1974].
The desire to explore a situation combines several aspects
including liking, desire to seek out and not avoid a situation
[Mehrabian and Russel, 1974].
This idea probably transfers to the desire to
explore a landscape, which will be highest when arousal is at
intermediate levels and when the landscape being explored is
particularly liked.
2.3 Psychology of motion
The investigation into the psychology of motion provides a
framework to relate the experience of landscape to its fundamental
psychological underpinnings. The objective is to find a
theoretical basis to analyze video animations according to visual
structure, as captured by human vision.
An important comparison to make is between static displays and
motion. The information carried by static displays is considerably
high, even without relying on motion. The study of distance
perception identified gradients of textures in the optic array as
determinant elements in perceiving exocentric and egocentric
distances. Texture can be stochastic or regular, and aligned
parallel or orthogonal to the picture plane, generating
respectively linear perspective and compression. The perception of
surfaces is based on the implicit structure of the optic array
that allows detecting geographical slant independently of optical
slant [Gillam, 1995].
Static depth cues include occlusion, relative size and density,
and aerial perspective, while motion parallax characterizes the
environment in motion [Cutting
and Vishton, 1995]. Motion has special
implications for the perception of three-dimensional structure
[Todd, 1995] while the framework
for self-motion identifies the
type and amount of motion information specified by the visual
field of a moving person [Warren,
1995].
In particular the optic flow consists in the change in structure
of the optic array due to the displacement of the point of
observation before it is sampled by the eye. The optic flow
contains information about the 3D layout of the environment and
about the process of self-motion. The optic flow specifies also
the effect of motion parallax, based on the different optical
velocities of objects located at different distances, thus
generating velocity gradients [Warren,
1995].
The invariant structure of the optic flow is important in the
formation of survey knowledge, since it specifies object-to-object
relations instead of the self-to-objects relations specified by
the perspective structure. This idea results useful in
establishing a link between visual information and spatial
knowledge [Sholl, 1996].
3 Spatial knowledge
One aspect of this study is to assess the spatial knowledge of the
environment that develops after exposure to animation. To achieve
this goal it is important to discuss the nature of the process by
which landscape information becomes spatial knowledge.
The spatial experience of a landscape through the medium of
animation might involve the development of a mental map or schema
that includes visual memories of the appearance of the surface,
the location of major natural landmarks and the shape of the
topography, as well as the spatial relationships between the
visual elements of landscape. Sholl
[1996] suggests that while
much animal navigation takes place without visual information by
the process of dead reckoning, instead in human beings vision is a
fundamental sense modality for spatial knowledge acquisition.
The second approach considers wayfinding to be a process deriving
from the perception of the environment, without requiring the
construction of cognitive maps. This ecological approach states
that, while moving, instead of perceiving static snapshots of the
environment that are later integrated into a cognitive map, we
rather perceive the continuous optic flow and the invariants of
the environment as they are picked-up over time. This allows us to
acquire a holistic, higher order perception that is not dependent
on the original viewpoints, and which also does not require the
existence of a cognitive internal representation [Heft, 1996].
However, the contribution of cognitive factors is acknowledged in
the sense that memory has a role in certain forms of spatial
behavior that perception alone cannot explain [Heft, 1983].
Each view and each path in the environment is unique, and we can
distinguish in the perspective flow an alternation of vistas (a
set of unhidden surfaces seen from a vantage point) and
transitions (portions of a route where an occluded view replaces
the current view), respectively characterized by a low magnitude
and a high magnitude of change. This defines the temporal
character of navigation on which perception is based
[Heft, 1996]. This is important
in landscape animation because
it allows us to formulate the idea that spatial knowledge is a
higher order perception of landscape when perceived within a
temporal structure.
The ancillary effects of cognitive processes on perception might
account for the influence of memory, experience and culture in
spatial knowledge acquisition, storage and utilization. Cognitive
maps might be considered as a representation of spatial knowledge
rather than as a fundamental structure. This solution is
interesting because it relies directly upon perception to explain
spatial knowledge, thus presenting, together with the previous
discussion on aesthetics, an occurrence of the fundamentally
perceptual nature of landscape experience.
The animation used in this study can be classified as an indirect
source of spatial knowledge, since it conveys spatial information
indirectly through a pictorial representation, and it is
contrasted with direct sources that involve apprehension of
spatial knowledge directly from the environment via sensorimotor
experience.
4 Visual landscape resources and
design
The idea of visual resource stems from a particular interpretation
of landscape that considers directly the outcomes (benefits, or
negative influences) made visually available to us by landscape.
These outcomes might depend on the characteristics of both the
physical elements observed and the cognitive processes within the
observer, but including also the particular dynamics of
observation.
It is considered difficult to identify the visual resource of
landscape. One direction of investigation is to shift the
attention from the quality of the landscape to the quality of the
person impacted by a scenic view. In turn, the properties of
viewers, from a visual resource management standpoint, are
considered products to be managed. The products include mood,
mental health, physical health, and learning [Bishop and Hull, 1991]. A
management solution is to develop a construct of visual quality
instead of relating each product to a manageable characteristic of
the environment, a procedure that is impossible at the actual
state of research [Bishop and Hull,
1991].
The possibility of learning about the landscape deserves special
attention as an interesting form of visual resource
[Bishop and Hull, 1991]. This
aspect can be related to the process of
learning about the physical elements of landscape, which, in the
context of this thesis, depends on the trajectory along which the
landscape is seen. Landscape as a resource can be based, in other
words, on the idea of knowing the landscape. The learning
component of resource suggested by the authors attaches an
economical value to the degree of epistemological access to
landscape that has been reached.
Another interesting aspect involving the visual properties of
landscape can be found in the area of landscape design. Here the
interest shifts toward the structure of landscape and the elements
that compose what we see in the landscape. Visual elements like
point, line, plane and volume have been applied to the visual
interpretation of landscape. The pattern of change defines several
variables of interest. Shape (or form when
considered in 3D) is concerned with the variation of lines and the
edges of planes and volumes, and describes the irregularity of
landforms. Time varies in terms of change in landscape
attributes, but it is also involved in motion and the position of
a moving observer: landscape is in fact often observed from a
moving position (such as a car or an airplane). Different speeds
of motion affect perception: with high speeds the eye takes only
general picture and focus on distant parts of the environment.
Another variable of interest is visual force, which
describes lines of visual forces in the landscape that suggest
particular observation patterns. Genius loci is a design
concept similar to sense of place by which landforms are key
defining element when the landscape is predominantly natural. The
issue of scale influences our feeling of enclosure in a
landscape that depends on the height of the enclosing element and
its distance from us. Perception of scale changes from
distant view to middle ground and foreground where texture is well
visible, and the height of observers affects the perception of
scale. In fact, down in the valley the landscape is
characterized by short distances, limited views and strong sense
of enclosure, while from a summit the valley is a part of
landscape at a wider scale [Bell, 1993].
Shape is of particular interest among the visual variable
for the way it defines landscape profile and landscape form. The
manipulation of height has important consequences on the feeling
of enclosure and the scale perception.
Visibility is a central concept in the visual studies of
landscape. The concept of viewshed can be related to the
analogous psychological concept of isovist. Despite the
fact the two were originated in different disciplines, they both
refer to an objectification of spatial inter-visibility,
respectively in a terrain context and in an architectural context.
It might be possible to link visibility to landscape aesthetics,
but the method will not be developed in this thesis. On this
front, the work of Llobera [2003]
leading to the concepts of
visual exposure and visualscapes allows us to structure
spatio-analytically the idea of perceptually-filtered visibility.
The idea hinges upon an observing entity that necessarily makes
explicit the spatial structure of landscape. Llobera [2003] also
offers technical solutions to represent and visualize those total
properties of landscape picked up by human perception and
previously left implicit in obsolete 2D GIS models.
5 Cartographic visualization
The display of three-dimensional cartographic objects, when the
variable is a single Z surface, is called surface exploration.
When immediate control on the visualization is provided, the
exploration becomes dynamic. While there has been considerable
interest in attempting to discover an optimal viewpoint over a
static terrain, a dynamic display offers instead a moving
viewpoint: the cartographer is not limited by a fixed view, but
she has an almost infinite range of options for representing a
terrain surface. The advantage is evident when the complexity of
the surface is so high that no single viewpoint can be sufficient
for understanding the surface. All the dynamic changes implicit to
this approach are to be considered part of cartographic design
[Moellering, 1980].
In the literature there are examples of the use of scale and
orientation/viewing parameters variables to study the effects of
spatial knowledge acquisition from maps [Taylor, 1984], especially
in their cognitive component [Eley, 1992].
They suggested that
these variables have a strong influence on the cognitive processes
of map reading.
investigates whether the three-dimensional map can be considered
within the same theoretical framework as the ordinary
bi-dimensional map. 3D maps avoid the problem of interpretation of
more abstract devices like contour lines and offer depth cues to
the observer to interpret correctly the image. A differentiation
is made between a map that can be understood at a glance without
demanding great cognitive resources, and a map that requires more
careful and non-instantaneous interpretation. In practical terms a
3D perspective map with realistic textures drawn on top might be
more cognitively demanding than a 3D perspective map with just the
fundamental depth cues. When considering the addition of the
fourth dimension of time, Kraak's framework would probably predict
a cognitive overload. Instead, real-time cartographic animations
can be "consumed" without particular overloads [
Landscape animation can be considered as a form of landscape
visualization in motion, and therefore as an extension of
three-dimensional cartographic representation. Kraak [1988]Kraak, 1990]
In fact, at this point it is worthwhile asking if those animated
visualizations need to be considered as maps requiring
interpretation or rather as the synthetic counterpart of real
world video images, in a cinematic framework. Supposedly this
depends on the kind of interpretation required of the participant.
Spatial knowledge acquisition requires a sophisticated visual
analysis from the participant (including distances, locations,
slope, etc.), while aesthetic evaluation probably requires a more
holistic emotional interpretation, which has little to do with the
interpretation of a map as it is normally conceived.
Geovisualization as an emerging discipline has the "time
invariant" objective to provide a framework to extend the scope
of research to other disciplines and to other non-traditional
collaborators including the entertainment industry [UCGIS, 2000].
In this light, the analogies between certain types of animation
and film can be seen as examples of situations in which
geovisualization might be related to other media other than the
cartographic-based representations.
Geovisualization literature uses the important concept of
exploration to represent the earliest phase of the process of
visualization, namely the one responsible for obtaining a sense of
the existing patterns in a dataset before proceeding to the later
stages of confirmation, synthesis and presentation
[MacEachren, 1992]. It
is a process of revealing the unknowns of a
dataset, and thus it can be considered part of the process of
information discovery. In this light the "seeing the
unseen" of visualization can be considered as a way to gain a
new perspective on data and developing new concepts based on the
discovery of new information.
In recent years animation has received increasing attention in
geovisualization. The three dynamic variables of animation
proposed by DiBiase et al.
[1992], namely duration,
rate of change and order, were devised mainly
for abstract representations but they can work with fly-by
animations such as those used in this study.
The use of animation for the exploration of three-dimensional
terrain surfaces, as exemplified by the Jet Propulsion Laboratory
(JPL) planetary fly-bys, has military and geological applications,
but also more human-related ones, such as the representation of
human movement and interaction. Campbell
and Egbert [1990] also reported
that in the design of terrain animations it is necessary to
control the effect of the overwhelming novelty of the medium. For
example the animation "L.A.: The Movie" by JPL, showing
a fly-by on the city of Los Angeles and on its most important
features and landmarks, arguably offers only a limited chance to
orient oneself in the represented environment due to the high
speed of motion and the rapid turns of the trajectory of the
camera.
Cartography does not consider landscape only as a terrain object.
Burt [1995] presents the issue of
considering maps as devices for
developing a sense of place (see also Section "Sense of
place"). The emotional involvement that we can experience with a
carefully designed map works as a stimulus to gain new knowledge.
When considering a work of art, we show empathy caused by feelings
related to what we find. In the same way a map, characterized by
clear presentation of information and a particular "mood" due to
tone and compositional arrangement, might provide a new experience
that links memories and concepts related to place. A combination
of map, graphic image and photograph in a multimedia type virtual
map may be used in enhancing the sense of place (see below).
There are several examples in the literature concerning 3D
visualizations of landscape, considering technical aspects of
actual methods of computer graphics [Graf
et al., 1994] and remote
sensing [Graf, 1995], and an
overview of techniques
[McLaren and Kennie, 1989]
and applications [Berry et al.,
1988]. Other specific
studies on environmental visualizations can be found in the
Landscape Planning literature.
The representation of terrain in three-dimensional cartographic
maps introduces the problem of finding a suitable multiplying
factor, called vertical exaggeration, for the z
value (elevation) of the terrain on the map. In fact it was
suggested that 3D maps look more realistic when the
vertical exaggeration is chosen according to a measure of relative
relief determined by the contour interval. This aspect was
investigated experimentally in Jenks
and Caspall [1967]. In one experiment,
pairs of 2D maps with the corresponding 3D topographic maps were
presented to the participants together with a scale from which the
degree of over- or under-exaggeration had to be evaluated.
Indicatively, the greater the relief, the greater the maps had to
be vertically exaggerated in order to look realistic.
The underlying assumption of the latter study is that cartographic
realism is not a property of the geometrical characteristics of
the map alone, but is a function of what the viewer expects to
find in the map itself. The psychological process used by the
viewer (or by the map maker) in evaluating the realism of a 3D map
is suggested to have an aesthetical component, even if that is not
investigated further.
However, it might be argued that it is not convincing to determine
whether a map is realistic or not by verifying if the condition of
"it pleases the eye" is satisfied. In fact, the
reliance on aesthetics to evaluate the soundness of the spatial
structure of a 3D map, even if compared to a 2D map, might be
related to the ideas on aesthetics previously illustrated. Human
aesthetics simply do not respond to the level of realism of a
spatial arrangement, but rather to those conditions that appear to
guarantee adequate chances of survival, albeit restructured to be
applied to a display that does not look like the hunter's
savannah. The differences in aesthetics between the non-
vertically-exaggerated map and the vertically-exaggerated map have
nothing to do with realism, because anything convenient to
survival would "please the eye".
In general we might suggest that aesthetics (i.e., strictly what
we like and what we dislike) constitutes a source of confounds as
a measure of realism. For example, vertical exaggeration has been
extensively used in Art to produce aesthetically pleasing
landscapes, dramatically showing greatly enhanced cliffs and
extreme formations in portraits of areas that in reality were more
scaled-down.
In the context of this study, the vertical exaggeration applied to
all landscapes used as materials in the experiment is always equal
to 1 (i.e., no vertical exaggeration). In fact, it is argued that
the realism of a first-person perspective view of a geometrically
realistic landscape visualization is analogous to its real
counterpart, that is, the three-dimensional projection on the
picture plane of the unmodified topographic dataset, where all the
geometric properties of the terrain are preserved.
It must be stressed that the only source of what here is called
realism in the geovisualizations used in this study is
geometrical, provided by the GIS software, maintaining
proportions, spatial structure, and soundness of methods of
projection in a coherent framework that covers all landscape
design instances used in the study. In turn, this relies on the
terrain dataset used, which was obtained from physical reality.
6 Sense of place
Sense of place is a vague umbrella concept incorporating many
different aspects related to place. It combines the ideas of
location, landscape and personal involvement in place, includes
concepts of identity and attachment to an area and, overall, it
offers a stronger unity than the region concept. Phenomenologists
do not define the term and leave the meaning to the user, while
operational definitions were tried for empirical studies
[Shamai, 1991].
Muir [1999] addresses the issue of
landscape and place in the
broader context of landscape studies. Sense of place derives from
two main factors: 1) the intrinsic personality of places which are
visually striking and produce powerful images and 2) the emotional
attachment to localities when considered as home settings.
Landscape makes a substantial contribution to the sense of a
place, and determines many qualities of it, including the
character of the scenery.
Tuan [1975] proposes a scale of
classification: at one end
there are places that are remote from sensory experience
considered as points in a spatial system; on the other end there
are places eliciting visceral feelings and rooted in a locality.
Sense of place is expressed at different scales, from home to
nation, constituting multiple centers of meaning. Sense of place
has also been reported to be possibly varying with differences in
age, upbringing, class and gender. The sense of place for an area
might vary with the view characteristics, such as the perspective
of a traveller from the top of a hill versus a farmer in the
valley below [Muir, 1999], although
there are few, if any,
experimental studies investigating this aspect of
topographically-dependent sense of place. However, the height of a
place as a factor in the perception of the environment was
considered from a cultural point of view: for example, the
experience of seeing a city from the top of an observation tower,
after being used to a ground perspective, is suggested to change
the relationship with the city itself. Other examples are the
artistic panoramas in painting and photography [Dubbini, 1994].
Tuan [1975] considers that, in
terms of experience and time,
sense of place is rarely acquired in passing, and in order to know
a place well a long residence and deep involvement is required.
Visual qualities, however, are appreciated in a short visit.
An empirical self-report measurement scale of sense of place was
developed by Shamai [1991],
and it is based on a scale from lower
(alienation, homelessness, not belonging) to higher (identity)
sense of place, subdivided in steps of knowledge, feeling of
belonging, attachment, identification, involvement and, finally,
sacrifice. Smith and Brown [1996]
elaborate a sense of place concept in the
context of education amongst schoolchildren, where place is listed
among the core elements of Geography. Their version of the concept
does not include an aesthetic component and it seems entirely
based on a notion of environmental awareness, a form of deeper
knowledge of the surroundings and the ecologic system behind our
daily lives. In both examples the role of knowledge as the base of
sense of place is stressed.
7 Approaches in Cinema and
Photography
Sitney [1993] reports several
examples of cinematic techniques.
Early in the history of cinema the panoramic sweep (or pan shot)
emerged to convey to the viewer the impression of a boundless
viewpoint. The long shot, hinging on the subjective view of deep
space, is distant in relation to the center of human activity and
has an establishing function, locating an individual in a wider
landscape, emphasizing human dominance and diminishing human
scale. The long shot has a cinematic meaning in the context of
other shots and alternate perspectives, whereby it serves as an
establishing shot for other subordinated ones. Zooming is another
element of Cinema and consists in a virtual movement of the camera
to traverse landscapes and indicate possible trajectories for
exploration.
What can be gained from considering Cinema in this study is not
only the use of the specific techniques, but a level of complexity
of dealing with moving elements for communicating content that is
unparalleled in the context of dynamic visualization. The concept
of spatial articulation [Johnson,
1974] is important in Cinema, and
it is based on the notion that spatial relations expressed on the
screen by a combination of camera movement and implied filmic
space are able to produce content and to elicit emotions. Spatial
articulation
comprises the main factors: the first is proximity of the camera to the
target,
generating a set of contrasting reactions in viewers such as
removal-involvement and conceal-reveal. The second factor is the angle
of
vision, with the subtle aspect of relative closure, creating patterns
able to
communicate different messages such as rationality or spontaneity
according to
their intrinsic movement. The two factors
are influenced by the timelines of motion that communicate
additional messages, especially using the device of montage, i.e.
composition and editing of scenes to confer acceleration and
eliciting particular experiences in the viewer.
A parallel can be made between landscape animation and these
general cinematic elements, because they are particularly
significant in suggesting a direct implementation in the context
of this study. For example, proximity generates involvement
similar to a close-up view of a landscape, which can then elicit
sense of place in the viewer (the opposite is also true, with
aerial views suggesting a sense abstraction from a place). At this
point it is useful to go beyond analogies and consider the way in
which Cinema combines all the different elements in communicating
a precise message. Since Lumiere the very location of the camera
or the lens used were the devices for communicating a message,
which became more complex as more elements were added to the
picture [Huss and Silverstein, 1968].
8 Planetary exploration
By considering research in planetary exploration we are able to
gain an insight into the concepts of presence, place and
exploration. McGreevy
[1993] reports that the first Lunar
Orbiter images of the Moon (years 1966-1968) offered the first
oblique perspective views of the Moon, making it seem
"more of a place". Perspective views add the component
of place to an otherwise impersonal notion of terrain, and
therefore a sense of presence is possible in such representations.
A similar sense of presence is produced by the lunar photo
panoramas, whereby mosaics of photographs were displayed on
spherical screen and gave an impression of presence to the
observer located in the center.
In the terrain exploration tradition, McGreevy [1993] adopts
the view according to which the environment must be experienced
concretely and directly through personal experience in order to
appreciate the affordances of an environment. Such personal
experience can be surrogated by virtual reality or by other forms
of representation in varying degrees of efficacy. By exploring a
digital representation of a planetary terrain it is possible to
gain a specific understanding of the place represented, enabling
an expansion of the capability of exploration of the scientist.
The main difference between orbital and surface views is that
satellite pictures are perceived as a 2D texture, not a habitat or
environment, while surface views are the kind of views that humans
have evolved to perceive. As far as motion is concerned, it might
be argued that without "moving around" there is a diminished
understanding of the spatial characteristics of the place
[McGreevy, 1993].
The JPL/DIAL (Jet Propulsion Laboratory/Digital Image Animation
Laboratory) in Pasadena, California, produced several fly-by
animations of the surface of several planets (specifically, Earth,
Venus, Mars and Miranda, a satellite of Uranus) [DiBiase et al.,
1992].
McGreevy [1993]
considers that all viewers reported a greater
visual and spatial appreciation of the planetary environments
after seeing the videos, and later stressed that the nature of
this appreciation is to be found in the concept of
presence (personal communication). In these respects,
McGreevy [1994] further
investigates what is the nature of the
understanding of geologic environments from the point of view of a
field geologist. The main characteristics of such field experience
are the continuity of presence, that is, the possibility of
traversing the field and observing an object without any
discontinuity in the personal action space. In other words,
presence is related to continuous natural locomotion and seamless
shift of attention from the environment to the individual sample
collected in the field.
Chapter 3
Conceptual Framework
1 Introduction
The main focus of this Section is to provide sound theoretical
arguments supporting the experimental design and in particular the
choice of independent variables. The fundamental factors
distinguishing the two experimental conditions from each other are
reviewed and evaluated on the basis of experimental evidence and
theoretical considerations. In fact, a differentiation between
camera trajectories is proposed based on the variable of camera
altitude, which in combination with camera elevation angle can be
used to generate two qualitatively different viewing conditions
(i.e. ground view and layout view). The objective of this study is
to evaluate whether the two viewing conditions elicit different
responses. A discussion will be articulated by comparing the
specific ability of the two trajectories to cause different types
of environmental information to be displayed.
2 Theoretical aspects of landscape
Historically, landscape has been seen as the result of an act of
seeing, and of framing the environment in a painting realized from
a specific point of view and characterized by a particular viewing
angle, making an area of land visually accessible to a viewer. In
the history of landscape painting this selection operated not only
on concrete landscape elements, but also on compositions of
materials taken from spiritual and conceptual spaces (see 16th
century Flemish painting in the Literature Review).
It is convenient to retain the definition of landscape originated
in artistic painting. Instead of focusing on the characteristics
of the constituent elements of landscape, the approach is instead
to consider the preliminary step prior to any human-landscape
relationship, that is, the idea of exposure to landscape.
Landscape exposure
Landscape exposure can be seen as the spatio-temporal description
of the information-seeking process illustrated by Gibson by means
of which humans learn about the environment by walking on the
ground, turning their heads towards interesting areas, and moving
their eyes around. The adaptive exploration of the environment is
suggested as similarly important as the intake of sensorial
correlates for information pick-up.
The adaptive perception of a human being is limited to the height
of the eyes when walking, to a turning angle of the head of
maximum half a circle, and to a viewing range allowed by the
resolution of the eyes. A pilot of a helicopter, instead, has a
different set of intrinsic limitations, such as the field of view
allowed to her by the cockpit, limiting in turn the potential
range of visibility initially offered by the eyes.
The strategy suggested in this study is to force the participant
into very specific conditions of limited landscape exposure, such as
a high altitude and uniform flight, reproduced in a fixed
resolution animation screen. Since the perceptual information
pick-up depends on those constraining aspects related to
adaptation, we conclude that particular constraints on movement
might generate different patterns of information pick-up.
For example, if we fix in the real world the orientation of the
head of a person in a central position using a special device, so
that she cannot turn it around freely, we might impair the
efficiency of information pick-up for that person in most
environments. In a sense, it is like ruling out all possible
choices of head orientations except for one. It might be argued,
on one hand, that any constraints on adaptive perception can
change how well information is picked-up. Similarly, on the other
hand, imposing changes on any variable on which adaptive
perception is articulated, affects the process of information
pick-up.
It would make sense, in other words, to investigate whether a high
altitude fly-by is perceived differently than a low altitude
fly-by, which is the main tenet of this study. In fact, it is not
only a matter of predicting how the visual properties of the
visible landscape, and in turn the participants' perceptions of
them, change between the two trajectories. The issue might be
related to the more fundamental fact that a restricted visual
access, imposed by defining an observer's trajectory, might be a
constraint on adaptive perception, or even something that extends
the normal possibilities of adaptive perception (i.e. flying above
the landscape allows the observer to gain a layout view which is
not limited in breadth as a ground level view).
The ecological and Gibsonian idea of considering of great
importance the role of adaptive behavior in perception, can be
seen as supporting the fundamental idea of this study, that is,
the dependency of the outcomes of the process of information
pick-up on the spatio-temporal dynamics of body, head and eye
movement, and thus on the characteristics of those dynamics when
constrained and "packaged" in defined observer's trajectories.
Filtering approach to landscape
The framework proposed in this thesis indicates that the
God's View of a given environment is the conceptual
starting point for any further analysis of the visual landscape.
The totality of views afforded by a given terrain, and articulated
in infinite viewpoint locations and viewing parameters, can be
seen as the complete representation of all the possible ways we
can look at a given terrain.
When we look at that terrain from a single viewpoint, defining it
from the infinite set of viewpoints available on or above that
terrain, the God's View, originally defining a viewpoint
class of entities, is reduced to a single
instance of viewpoint, from which a particular view of
the terrain can be obtained.
This process of defining a single viewpoint from the totality of
viewpoints of a God's View can be considered as a kind of
selection. The concept of filter, developed in engineering and
used also in other disciplines such as ecology and psychology,
might be useful in describing the mechanism that, in its more
abstract sense, maps the entire set of source viewpoints onto one
single final viewpoint.
The important idea is that the immediate result of the
God's View-to-single-viewpoint mapping process is the
fact that the terrain, from being implicit and object-like,
becomes viewable in a general sense. The terrain becomes
accessible through the first-person, egocentric framework of the
observer. At the same time, besides becoming "viewable", it also
instantiates a particular view, or optic array, characterized by
precise viewing frames. In other words, with the act of defining a
view, a specific perspective structure is also explicitly defined,
through which the invariant structure of terrain can be perceived.
From a philosophical point of view, a filtering process applied on
the God's View determines the perceptual accessibility of
landscape, which defines the first step to come in touch with the
visual resource. In general, in order to perceive landscape, we
need to be immersed in its perspective structure.
In this thesis it is proposed that the term "landscape"
signifies the instance of a view on the terrain generated by the
filtering process. This definition is in line with the historic
definition of landscape as "a view on the environment".
3 Human-landscape psychological
relationships
Within the landscape framework, the nature of the human-landscape
relationship is measured as the psychological response to the
particular conditions of viewing determined by a specific instance
of landscape after filtering.
Human psychological response is based on the ecological nature of
the human-landscape relationship. Humans are seen in interaction
with the landscape, developing relationships based on aesthetical
evaluation, spatial knowledge acquisition and development of sense
of place. The fundamental aspect is aesthetical evaluation, that
is a synthetic element, while spatial knowledge is integrated with
the other two by providing a knowledge base for evaluation and
feeling. Sense of place is considered here in its aesthetical and
evaluative component.
Those three elements are based on heterogeneous concepts, thus
their investigation in combination might present methodological
problems, such as considering affect and cognition at the same
time, that are difficult to tackle in a single study. However, the
approach used in this study makes the research problem
conceptually more focused and manageable, while preserving the
diversity of angles from which landscape is investigated.
This research approach emphasizes the perceptual and cognitive
dimensions of the human-landscape relationships. Aesthetic
preference is considered to be the result of unconscious cognitive
information processing (see the relationship of this idea with
landscape preference in Kaplan
[1987], despite the interesting
hypothesis that affect is independent of cognition
[Zajonc, 1980]). The two views
reported here agree on the fact
that aesthetics is probably not solely the result of conscious
information processing of environmental information. Spatial
knowledge consists of the result of encoding of environmental
information in knowledge structures, and it is considered a result
of cognitive processing, though the literature presents discordant
views (see the Literature Review and Heft [1996]). Sense of
place, as explained below, is considered here only in its
cognitive component of place identity and character, closely
related to aesthetics and to an evaluative relationship with
landscape. Non-cognitive interpretations of both aesthetics and
sense of place are not directly considered. The reason for
emphasizing cognition comes from the idea that a fundamental layer
of cognitive information processing might underlie spatial
knowledge, aesthetics and sense of place. Therefore it is
interesting to evaluate the proposed aspects of visualization
primarily considering this layer.
Besides the cognitive emphasis, this study is also centered on
aesthetics, which consists in the evaluative component of the
human-landscape relationship. Spatial knowledge is considered here
only as an auxiliary source of information about the
human-landscape psychological relationships, since the study is not
designed to specifically investigate the process of spatial
knowledge acquisition. It is nonetheless argued that having an
insight into people's knowledge of the spatial structure of
landscape might inform us on the extent and kind of information
base used for their aesthetical evaluation of landscape. Sense of
place is instead considered in its cognitive and evaluative
component, rather than in its specifically affect-based component
of place attachment. Central in this context is the idea of
landscape character (see below) and in particular the idea of
distinctiveness and uniqueness of the information base contained
in a landscape. This aspect of sense of place extends the scope of
the concept of aesthetics while introducing more holistic and
identity-based evaluations. In summary, emphasizing aesthetics in
this study is justified on the grounds of the exceptional
characteristics of the evaluative human-landscape relationship.
This relationship is based on the fact that aesthetics stems from
the cognitive processing of environmental information (thus
sharing a common root with spatial knowledge) and also on more
holistic judgments that tend to be captured more by sense of
place, although they are fundamentally aesthetical and evaluative
in nature. In other words, by considering aesthetics as a central
aspect of this thesis we can have also a convenient perspective on
the related cognitive dimensions of spatial knowledge and sense of
place.
3.1 Focusing on aesthetics
Aesthetical experiences are, in part, the outcome of an innate
human ability of relating with the environment, which depends on
evolutionary considerations of adaptation. Other contributing
factors are the human cognition of stimuli from the environment,
and culture.
In Gibsonian terms, the Prospect-Refuge theory is a
description of landscape in terms of prospect and refuge
affordances, that is, the functional values of landscape
[Hartig and Evans, 1993]. Gibson [1979] in fact
suggested that the
affordances of the environment are perceived by an observer by
means of a process of direct perception that even preceded the
process of classification. Although Gibson never specifically
referred to aesthetics, he hinted at the concept of higher
order invariants that arguably stem from the first-order
perception of the environment. This suggests how aesthetics could
in principle be conceptualized as a kind of higher order invariant
making use of environmental information such as the affordances of
the environment.
The idea of landscape aesthetics hinges upon an evaluative
relationship with landscape that refers to the fundamental and
very complex (although experientially simple) process of liking or
disliking a scene. However it also includes a range of
appreciative relations with the environment, such as for example
interest and curiosity stemming from the visual properties of
landscape. The informational interpretation of landscape
aesthetics proposed by Kaplan
[1987] and articulated in the
four informational factors of landscape preference lies at the
foundations of this research approach. According to this line of
thought, landscape aesthetics are a process of environmental
evaluation that is based on the unconscious cognitive judgment of
the information content of a landscape. Of particular interest to
this thesis is the informational mystery factor that
specifically refers to the amount of promised information
contained in a landscape view. In other words, promised
information in a landscape originates from actual environmental
information that indicates the availability of further information
after a slight change in the vantage point.
This research aims at extending the concept of mystery suggested
in the literature by considering the promised information effect
generated by the occlusion of topographic forms. For example, a
form of topographic mystery might be the effect of a foreground
hill that, by means of occluding the view on the mountain beyond,
actually generates an attractive view based on the promised
information implicitly made available about the mountain. This
form of topographic mystery is coherent with defining mystery as a
condition in which some environmental information is promised by
means of actual information cues in the landscape: for example, in
topographic terms, a cue might be the highest tip of a mountain
visible beyond the foreground landscape.
In summary the idea is to use not only preference as a measure of
aesthetics, but also other self-report variables that are built
around the information factors of landscape aesthetics and that
capture aesthetical perception from multiple perspectives (for
instance promised information as desire for further exposure to
landscape animation). As explained in Chapter 5,
the
experimental design consists of a measurement of landscape
preference (in line with the literature) and of lower-level
variables (as an extension of the research strategy) in particular
conditions of actual and promised information such as those
offered by specific landforms.
3.2 Spatial knowledge
The interest in spatial knowledge in this study is mainly related
to the need of establishing a relationship between landscape
information and aesthetical perception. In fact, investigating how
spatial knowledge is acquired after exposure to a landscape
animation is important to determine the extent of the information
base used by the viewer to carry out her aesthetical evaluations.
In other words, as the viewer perceives the environment she
develops an information base that is unconsciously processed
during aesthetical perception. A way to assess the extent of that
perceptual information base is to verify how it helped develop
spatial knowledge. From this point of view, investigating the
extent of spatial knowledge by means of analyzing its
externalizations is like giving a different look at the landscape
information that contributes to the viewer's cognitions and
feelings. For example, knowing the level of detail of the viewer's
memories of the spatial properties of landscape can be indicative
of the things she noticed and that might have affected her
aesthetical evaluations.
Another component of spatial knowledge is related to the spatial
awareness of the characteristics of the viewer's trajectory. Being
able to remember the type of trajectory in relation to the ground
is an ability that can be related to the preference for the mode
of landscape exploration used in an animation. In flight
simulation there is interest in modeling the way the pilot
perceives the motion of the plane [Rolfe
and Staples, 1986]. While emphasizing
there the importance of safer and more efficient flight, in our
case it is interesting to study the perception of the
characteristics of a trajectory as a factor that completes the
experience of landscape in more general terms of self-motion
awareness.
The methods used in this study to investigate spatial knowledge
are based on self-report sketch maps of the plan and profile view
of the landscape being viewed, and the plan and profile projection
of the trajectory along which the exploration occurs. The sketch
maps of the landscape contain the topographical structure of the
terrain, including major topographic landmarks and landforms (see
Chapter 5 below). While most spatial knowledge
sketch-maps used in the literature refer to the built environment,
drawing a sketch map of the natural environment is a less common
task, especially when referred to a landscape lacking traditional
navigation features, such as roads and nodes. However the pilot
study has shown that participants are able to encode a good amount
of information about the landscape in their maps.
3.3 Sense of place
Sense of place captures the feelings of belonging to landscape,
the cultural and emotive attachment to landscape and the special
emotional bonds that develop between observer and landscape due to
the influence of memories of past experiences. Ideally, sense of
place would extend the range of human-landscape relationships
since not only could landscape be memorized (spatial knowledge)
but also evaluated (aesthetics) and become the object of
attachment (sense of place).
The concept of sense of place has specific implications with
respect to time and memory. According to the cultural tradition in
Geography, sense of place is rarely acquired in passing and
requires a long residence and deep involvement. From a temporal
point of view it seems unlike the quick evaluative aesthetical
relationship with landscape, which instead operates in much
shorter (although not instantaneous) time frames.
However, it might be argued that while a cognitive map of the
environment starts to be built at a first exposure, in absolute
temporal terms it takes more time to develop memories (in
affective sense) and feelings of attachment to place. The current
experiment design, based on very short exposures to landscape
(mostly 30-second sequences), does not allow us to examine the
affective attachment dimension of sense of place, as the latter
would probably not develop in such a short period of time. In
other words, sense of place cannot be included in full in the same
experimental design together with aesthetics and spatial
knowledge.
The idea of landscape character is not only related to an
appreciative (aesthetical) relationship to landscape, but also to
an early form of attachment, better represented by sense of place.
The questionnaire developed for the experimental design focuses on
capturing the early forms of sense of place that do not require a
long exposure to landscape, but are rather attached to the idea of
character and uniqueness, which almost fall into the aesthetical
categories of human-landscape relationships.
4 Research strategy
The general strategy is to compare the psychological response of
participants to two instances of viewing trajectories on three
different landscapes, to see if the different filtering process of
the two results in different responses, and how those responses
are patterned. The specific strategy is to define qualitatively
different trajectories, with a specific framing and camera
attitude, but still in part related in terms of length on the
ground covered and velocity. In the Sections below the criteria of
trajectory and landscape design will be illustrated in light of
the research strategy.
4.1 Trajectories
The aesthetical mystery factor (see Literature Review) indicates
the importance of the visual structure of landscape, related in
particular to the characteristics of actual and promised
information. The visual structure depends in part on the
appearance and arrangement of features in the landscape; also, it
depends on the way the visual elements are presented in a view and
are available to the viewer's perception. Those two factors
consist respectively in absolute and subjective structure, which
are in reciprocal interaction when the landscape is explored. We
can say that, for example, the mystery effect of the mountain
partially hidden by the hill is not only a matter of absolute
hill-mountain configuration, but also a matter of observer's
location.
In the exploration of landscape there are certain trajectories
that have the potential to trigger particular responses due to
specific configurations of actual and promised information as
suggested for example by the mystery factor. The essence of this
experimental study is to compare very similar but still
qualitatively different trajectories. In general the research
question is to verify whether a slight variation in the viewing
dynamics is able to influence significantly the perception of
landscape.
It is not trivial to choose the proper level of difference between
trajectories in order for them to still be comparable (i.e. able
to show reasonably overlapping landscapes) but not so similar as
to provoke inadequate perceptual differences. Also, designing
qualitatively different trajectories means that we have to select
examples of trajectories that have a particular meaning in the
exploration of landscape: that is, we need to discriminate
particular aspects of the landscape space that instill quality in
otherwise infinite and randomly generated trajectories. Quality,
in this study, comes from the characteristics of the terrain, and,
in particular, from topographic variation, but in principle can be
originated from any existing landscape element (i.e., vegetation
distribution). However, topography is here the main aspect to be
considered.
More specifically, by the term trajectory here it is meant a
complex combination of all the factors characterizing the viewing
experience of landscape. It includes the location of the camera as
it changes in time, the location of the camera target as it
changes in time, and the angle of view. The three variables can
also be summarized in an egocentric framework by means of
variables such as 3D location, pitch, yaw and
roll, but the former set of variables was used because it
follows closely the aspects of landscape implementation.
The variable of camera altitude is the primary differentiating
factor between trajectories. From a perceptual point of view,
altitude influences the viewing perspective of the landscape,
since the sizes of the textures change according to the distance
from the ground, and objects tend to be seen vertically from
above, thus exposing horizontal instead of vertical surfaces (for
example, the crown of a tree versus a tree trunk). Moreover, the
visual elements in a ground view appear larger than in a layout
view, and in the latter case there is a larger portion of
landscape being displayed at a given time.
There is converging evidence that camera altitude is an important
influencing factor in our experience of landscape. As seen in the
Literature Review, the variable of viewpoint height in maps was
found influential for the cognition of the represented terrain
surface as a whole. Moreover, panoramas from high observation
points always confirmed how the height at which the observer is
located dramatically influence the perception of, and the
experience in the environment.
However, camera altitude is not intended as an absolute variable
(for instance 100 versus 200 meters of altitude), because it would
not be a very meaningful measure given a varying topography. In
particular the absolute altitude of the camera is combined with
the relationship of the trajectory of the camera with the ground
surface. A trajectory that is very close to the ground, resembling
a low altitude, terrain-following flight (such as one of a
helicopter), is very distinguishable from a trajectory that is
high and uniform (such as the one of a small plane). The two are
certainly more qualitatively different from each other than two
uniform trajectories at two different absolute altitudes. In other
words, the variable of camera altitude is used in a more complex
and qualitative way than its quantitative meaning of "meters
above the ground", as it combines a component of absolute
altitude, a specification of a relationship with the ground
surface. The lack of a "pure" quantitative variable is
compensated by the gain in qualitative differences that otherwise
would not be captured by quantity alone. In practical terms, the
altitude of the camera in the low altitude condition will be set
to a constant value above the ground. The actual altitude will be
set to take into account the need for a reasonable altitude for a
"safe flight" avoiding collisions with the ground, and enough
proximity to the ground to fulfill the design requirements.
In the high altitude flight the altitude is less easily
determined. Each landscape stimulus (see Methods) presents a
different set of topographic characteristics, and the common
"cruising" altitude, although independent of the terrain, will
be determined on the basis of a safe flight over the highest
landform of the set (major landform case, see Methods). In the
plain, valley and minor landform cases the camera will have
exactly the same altitude as the major landform case, to ensure
consistency across stimuli.
Another source of dissimilarity between trajectories is the
elevation angle, or pitch, of the camera, that is, the angle
between the horizontal and the direction of viewing. The elevation
angle of the low altitude camera trajectory is approximately
horizontal and does not change with the variations of camera
motion, that is, when the observer moves upwards the camera is
still horizontal. This behavior resembles a helicopter maintaining
its attitude during flight and results in a visually more coherent
trajectory because the viewing parameters are not dependent on
vertical motion. The elevation angle of the high altitude
trajectory discussed above would be set in such a way as to convey
a more pronounced layout view of the landscape, which would
contrast with the ground level view of the first trajectory. In
general terms the elevation angle of a camera influences the
apparent depth, relief, slope and "blocking" (i.e. degree of
occlusion) of the landscape, and therefore might help create
different viewing conditions which are qualitatively different.
Thus, the camera of the high altitude trajectory will point
downwards, with an elevation angle in between the horizontal and
the vertical down to the ground. A camera looking vertically down
would show very little of the distant landscape beyond the
foreground terrain, thus excessively unbalancing the difference
between trajectories. On the other hand the same camera would
introduce the interesting comparison between a ground level
perspective and a layout map view. To counterbalance advantages
and disadvantages an intermediate solution was adopted. The camera
was set in such a way that it was able to see part of the distant
landscape but also offer the viewer a layout view. The optimal
elevation angle to look at a static landscape in a block diagram
was found to lie between 30 and 40 degrees [Sieber, 1996]. Within
the 10 degrees allowance, the value of 30 degrees is preferable
since part of the sky would still be present in the picture,
making the layout view still comparable with the ground view. On
the other hand the amount of sky present in the animation is not
controlled for in the current design. The fact that the ground
view has more sky in the picture, and that therefore might tend to
be preferred more for that characteristic alone, is considered a
distinguishing feature of that viewing condition. Instead of
controlling for the amount of sky it is preferred to maintain
meaningful qualitative differences across conditions.
Therefore we conclude that the comparison will be made between a
trajectory with elevation angle equal to zero (ground view) and a
trajectory with elevation angle equal to minus 30 degrees (aerial
layout view), on the basis of the qualitative difference between
ground perspective and optimal layout perspective, a difference
that indicates interestingly different modes of landscape
exploration.
The previous discussion on camera altitude and viewing parameters
referred to the variables accounting for differences in
trajectory. Now, the aspects that are equal across conditions are
considered in detail.
First, the velocity of the camera and the length of the trajectory
segment being flown need to be identical. Unfortunately, the two
requirements cannot be satisfied at the same time. In fact, if
each trajectory covered exactly the same segment on the ground,
then the horizontal velocity of the cameras in the two conditions
would be identical. However, in some hilly landscapes the actual
velocity would be necessarily different because the low altitude
trajectory has a vertical component, required to climb up or down
hilly terrain, that makes longer the trajectory distance actually
flown. The absolute amount of variation of these factors varies
from none in the plain landscape stimulus where there is
no vertical component of motion, to a maximum of about 30%
increase in the Ynez Peak (climbing landform) landscape
stimulus, where the camera climbs up the landform for half of the
length of the animation at an angle of 30-40 degrees. The
variation is not that great to change the self-report measures
substantially.
The duration in seconds of the animation is equal in all
conditions and for all stimuli. The animations are designed to
expose the viewer to very specific content (climbing a hill,
flying over a plain) that is not the same thing as offering a
general exposure to multiple aspects of landscape in the same
sequence. It was evaluated that 30 seconds for each landscape was
sufficient to have the required exposure to the stimuli. A
considerably greater duration was not really manageable for the
computational constraints on the generation of animations.
Amongst the other factors accounting for similarity across the two
conditions, the viewing angle (field of view) is a variable that
influences the perception of the scene in terms of the amount of
landscape contained in a view, and the relative size of the
textures. It is also related to photographic and cinematographic
aspects such as the use of zoom for enhancing distant detail. It
is proposed here that the viewing angle should not be manipulated
in this design because of the complexity of the variable and its
unpredictable effects on the overall landscape experience.
Another element of trajectory is the horizontal viewing direction
of the camera. This measure, called yaw, represents the direction
the camera is facing during forward motion. This is another
variable that is kept constant, and specifically, the camera
always points towards the direction of motion.
In summary the two trajectories being considered here are a low
altitude trajectory of a camera flying very close to, and
following, the ground, looking ahead and horizontally ("Low
Altitude-Terrain Following" or LA-TF), and a high altitude
trajectory of a camera flying horizontally, looking ahead, and
down at an angle of 30 degrees ("High Altitude-Uniform" or
HA-U).
4.2 Landscape categorization
The criteria of selection of three different landscape parameter
sets, resulting in the three different landscape designs, were in
part quantitative, such as for example the terrain characteristics
expressed by landscape profile. Although the general
geostatistical descriptors of terrain were not considered directly
in the design process, the landscape profiles were specifically
chosen in terms of visual complexity.
Landscape profile
The progression of landscape elevation from the start to the end
of the animation measured along the trajectory of movement (i.e.
the profile graph of the trajectory of the observer over the
landscape) does not allow us to have control of the
characteristics of the full three-dimensional structure of the
terrain that is visible during the fly-by.
However, the terrain profile is used in this study to simplify the
conceptualization of the terrain. After all, it is the most
significant piece of information in the entire terrain, because it
is the one directly involved in the interaction between observer
and terrain: in the terrain-following condition, the observer
"goes up" if the profile "goes up" as well, even if to the
right and to the left of the line of the profile deep cliffs might
suddenly open up.
In other words, this experiment design has a minimal, but
existing, degree of control on observer-terrain interaction, at
the expense of a degree of control on the actual information
available from the landscape, which is left open to randomness,
necessary for ecological experimentation.
However, an analysis of the terrain information available along
trajectories compared with the information of the rest of the
terrain, would probably give the result that the two aspects are
consistent with each other: "going up" in the profile results in
a "going up" in the entire surface (e.g., a trajectory along a
steep hill corresponds also to a steep hillside surface). This is
probably due to intrinsic properties of the real-world terrain
data used in this study.
In practice, two experiment conditions (Plain Cruise and
Ynez Peak, see below) were chosen so that they
respectively represented a stable, non-increasing profile, and a
gradual, monotonically-rising profile. They were identified in
real-world areas, respectively characterized by a flat plain and a
hill. Such mathematical abstraction of profile characteristics
will be useful afterwards in generalizing the results.
More precisely, the two trajectory types, flat and
rising, result from a qualitative attempt at categorizing
the observer's movements on the landscape into fundamental
categories, or primitives of camera movement on the
landscape, that in a previous experimental design comprised also
descending, overcoming minor obstacle and
overcoming major obstacle, concepts that lead to the idea
of visual complexity. Although it will not be
investigated here in detail, visual complexity represents
a measure of the aesthetic potential of the landscape. It is not a
geostatistical measure of terrain as a whole, but rather a measure
that uses as determining factors the visual arrangement (e.g.,
occlusions) of the landscape.
In practice besides selecting an area in Ynez Peak where
the desired monotonically-rising trajectory type could be applied,
it was also possible to include a special effect of final
reveal after the gradual hillside rising, where the
limited visual range could be extended indefinitely after the top
of the hill had been reached. Although the profiles are the
control on the generation of the trajectory from the source of
terrain form, special cases of visual complexity were
added ad-hoc in anticipation of future designs.
A series of three pilot studies were carried out before the final
experiment, based on a preliminary version of the following
Methods Section, and aimed at developing a better design strategy.
In the first pilot test, 14 undergraduate and graduate students
(11 males, 3 females) were selected. The experiment materials were
made of two animations (LA-TF/HA-U), consisting of a rectilinear 3
Km flight across a canyon system on central Santa Catalina Island
(CA). With respect to the following final experiment, the
animations were of lower quality and less distinguishable as
independent landscape experiences, motivating a more thorough
design work as a follow-up. The main objective of the pilot was to
test the overall procedure, and in particular the adequacy of the
computer animations in allowing the subjects to remember details
to be later reported in the questionnaires.
The results of the pilot indicated that the two experimental
conditions (LA-TF and HA-U) elicit different landscape experiences
at a significant (p < .1 and p < .05) level in at least
two close-ended questions, respectively suggesting that an LA-TF
stimulates less curiosity for newly revealed parts of the
landscape and a weaker feeling of possessing navigational
knowledge than HA-U. The latter result confirms the superiority of
high elevation views in conferring at least an impression of
layout knowledge, while the first result is in contrast with an
idea of landscape that elicits more curiosity because it is
revealed over time while advancing in a trajectory close to the
ground. One could infer that a lack of information does not
necessarily elicit curiosity, and the outcomes of the flow of
information in a landscape animation appeared complex.
Participants seemed in general to be able to encode a good level
of detail in their maps, even if the individual differences were
substantial and appeared stronger than the differences across
experimental conditions. The maps were not analyzed, but it was
possible to determine the variable extent to which the
participants detected the characteristics of the experimental
condition. In summary the pilot suggested that the approach was
feasible, and the procedure was able to extract useful information
producing complex results, although the analysis and
interpretation could be anticipated as difficult.
The second pilot study was entirely addressed to the verbal
component of the experiment, that is, the formulation and the
understanding of the questionnaire. A group of 10 undergraduate
students were individually introduced into an informal but
specific conversation aimed at evaluating their understanding of
the questions, after being exposed to the then already completed
animations of the main experiment. Besides gaining a better idea
of the general characteristics of the participants, and correcting
verbally confounded questions, the pilot was useful to fine tune
the wording of the required concepts to be investigated, by means
of a trade-off between the participant familiarity with the
textual format, and the linguistic precision in referring to the
entity being measured.
A third pilot study involved three undergraduate students. It was
aimed at checking whether the overall testing procedure,
consisting of three animations and three questionnaires to be
filled in, could be completed in the allotted test time. It also
allowed a complete execution of the procedure, inclusive of the
multimedia presentation and all the informational materials such
as informed consent and oral briefings and debriefings.
1 Design
This experimental design comprised two main variables, (1) viewer
(and helicopter) altitude and trajectory type, and (2) landscape
type. The first variable had two levels, namely (1) low altitude
and terrain-following, and (2) high altitude and uniform. The
second variable had three levels (landscapes), namely (1)
Plain Cruise, a uniform plain-based landscape, (2)
Silver Canyon, a narrow valley, and (3) Ynez
Peak, a mountain (see Figure 5.1).
|
|
|
|
|
| LowAltitude-Terrain Following |
|
|
|
|
|
|
|
|
|
|
Figure 1: Summary of experimental design
2 Participants
The number of participants in the final experiment was 36 (22
males and 14 females, distributed in roughly equal proportion in
the two between-subject conditions). The average age was 21 years
old. Participants were drawn from the undergraduate research pool
available at the UCSB Department of Geography. The pool was mainly
composed by students taking the introductory Human Geography
class, and also by a few junior and senior students taking more
advanced classes (Applications of GIS, and Geovisualization).
3 Materials
3.1 Landscape stimuli
3.1.1 Data sources
Figure 2: General location map
The landscapes used in this study were selected from a terrain
database composed of two datasets: the Conception Coast dataset,
and the Santa Catalina Island dataset. The Conception Coast
dataset was used to design two landscapes: Plain Cruise,
located in a large plain in the north-western area of the dataset
(indicated as the San Luis Obispo area), and Ynez Peak,
on the transverse mountain range close to the center of the
dataset. The Santa Catalina Island dataset, covering a much
smaller island area in front of the south-eastern coastline of the
Conception Coast dataset, was used only for Silver
Canyon, corresponding to an area on the south-eastern tip of
Santa Catalina Island.
Figure 5.2 shows the two entire
datasets, and
the inset maps represent the specific landscape areas being used.
Conception Coast dataset
The Conception Coast dataset is a 60 meters resolution Digital
Elevation Model (DEM) set, based on freely available USGS (United
States Geological Survey) DEMs, which was provided by the
environmental organization ConceptionCoast.org as a single large
DEM.
The original DEM was in the Albers projection, which had to be
re-projected in latitude-longitude coordinates (i.e., the
projection used by the 3DNature's visualization software World
Construction Set v.3 on which the visualizations were made). The
latter operation was carried out using the Grid re-projection
utility available in ESRI Arc/INFO Toolbox. The re-projected DEM
was then displayed using ESRI ArcView 3.1 and then saved in Arc
ASCII format for import into World Construction Set.
Using the import functions available in World Construction Set the
original large DEM was segmented in a regular grid of 135 distinct
but contiguous DEMs, approximately consisting of 300x300 cells,
and maintaining the 60 meters resolution of the original dataset.
Indicatively, just one of the final DEMs was sufficient to
completely cover one designed areas.
Santa Catalina Island dataset
The Santa Catalina Island dataset was not publicly available, and
was provided by Dr. Bill Bushing, formerly at the Catalina
Conservancy. It was originally produced from the digitization of
USGS topographic maps of the area. The dataset has a cell
resolution of 20 meters.
The original dataset came as a single ESRI Arc/INFO Grid file, which
was loaded in ESRI ArcView 3.1 and exported to Arc ASCII
format.
Before the import in World Construction Set, it was necessary to
modify the individual cell values for the non-land areas of the
dataset because they were initially set to a zero value. This
interfered with the realistic gradient rendering of the ocean,
therefore a text editor was used to set the bathimetry to a lower
value that produced in the visualization a deep blue color.
The preprocessed DEM was converted in World Construction Set DEM
format, resulting in 12 distinct but contiguous, and slightly
rectangular, DEMs, all correctly georeferenced. Additionally to
this land base, several additional DEMs were added in the western
zone by means of creating new oceanic DEMs (with constant negative
elevation value) that were re-georeferenced precisely at the
external boundary of the other ocean-defining DEMs. In so doing
the extent of the waters surrounding the island was enlarged. This
in turn allowed the animations not to show a sudden and incorrect
"end of the world" effect, which would have appeared in the
final animation if the ocean gave way to the sky gradient below
the horizon line.
Strategies for dataset enhancement
In order to keep the rendering times low, it was preferred not to
fractally sub-sample every individual DEM into a new set of DEMs
having four times the spatial resolution of the original DEM. To
increase resolution in the dataset, which in some cases determines
the overall realism of a scene, a less computationally
"expensive" solution was preferred. Fractal techniques added
random detail to the coarse terrain structure defined by the data
itself, process carried out at the final rendering step and not as
a data preparation procedure. Determining the ideal Fractal Depth
value consisted of a trial-and-error process that involved
considering a trade-off between rendering time and final image
quality.
On the trade-off an intermediate point was chosen, equal to a
Fractal Depth of 6 for Conception Coast, and 5 for Santa Catalina
Island, that leaned towards moderate rendering costs. This offered
an apparent ground resolution of sub-meter level, approximately
equal for both datasets.
This simplifying solution made the study feasible, considering the
almost prohibitive (in terms of time) rendering costs of six
animations, besides the other additional and numerous trial
versions. However, although some of the low altitude animations
showed occasional ground polygons, unavoidable without higher
fractal depths of subsamplings, the overall quality was considered
to be good, and adequate for the purposes of the study.
3.1.2 Landscape 1: Plain
Cruise
Figure 3: Plain Cruise site map
Figure 4: Plain Cruise profile graphs
Figure 5: Plain Cruise sample frames
Design objectives
In the first landscape, named Plain Cruise, the objective
of the design process was to provide the participant with stimuli
originated from a flat terrain, while the non-controlled variables
of land cover and general appearance were decided to give the
impression of a somewhat human-made and agricultural scenery,
although not specifically urban.
Terrain structure
The topography of Plain Cruise is based on the Conception
Coast dataset. The area that was selected presents an
approximately flat terrain form, although at a much closer
scrutiny it results to be sloping down very gently towards the
West. On the Easter horizon, the flat plain gives way to some
hilly terrain.
Plain Cruise topography was chosen to offer the least
visual complexity to the participant of the three
landscapes considered. There are no occlusions throughout the
animation, in both conditions, and no topographical mystery effect
is proposed since there are not any suitable topographic
configurations.
Landscape design elements
In this landscape, as in the other two for that matter, the
non-controlled design elements were not chosen to obtain a replica
of the ecological characteristics of the local environment
represented, but to define useful and distinguished stimuli to
show to the participants.
The landscape elements added on top of the topographic data
comprise also an artificial lake, created by digitization and
elevation manipulation. In the middle of the lake, two irregular
islands were added, by means of digitization and areal addition of
elevation.
The "universal" ground cover was set to low grass. On the back
and on the sides of the lake, several irregular "specific" land
cover regions were manually digitized. They were characterized by
different combinations of hardwoods, bushes, grasses, rocks, and
so on. In addition, a few rectangular fields, set to be filled
with corn-like vegetation, were arranged in two rows transversally
to the direction of motion, accompanied by sketched
buildings/greenhouses.
The animation was generated as if the local time was 3:00pm on
July 11th. The sun was producing intense illumination from an
almost vertical position, thus imposing a "flattening" effect on
all landscape features. For a general view of the area, see Figure
5.3.
Trajectories implementation
The trajectories designed for the Plain Cruise landscape
are shown in Figure 5.4. The HA-U condition
consists of a straight
trajectory at an average altitude of 761 meters from the ground.
In the LA-TF condition the trajectory is shown to be practically
flat (it is sloping upwards by a minimal percentage) and always
parallel to the ground, although it is not responsive of subtle
variations of topography existing in designed areas. Sample frames
of the two animations are shown in Figure 5.5.
3.1.3 Landscape 2: Silver
Canyon
Figure 6: Silver Canyon site map
Figure 7: Silver Canyon profile graphs
Figure 8: Silver Canyon sample frames
Design objectives
The Silver Canyon landscape was specifically designed to
address a particular aspect of visual landscape design mainly
related to the sense of closure offered by narrow valleys to
observers on the ground. Second, the topographical structure of a
valley, progressing downslope towards the ocean would have allowed
us to investigate the terrain profile condition of
landform descent. Third, the special site configuration
due to the closeness to the ocean, and moreover some more specific
definition of the non-controlled components of the landscape,
allowed us to have a "meaningful" landscape to show.
Terrain structure
After an extensive virtual exploration of the particularly rugged
topography of Santa Catalina Island, the Silver Canyon
area, in the southwestern part of the island, was selected.
Silver Canyon is a very narrow and long canyon, almost
perfectly straight, oriented NE-SW, and leading directly into the
ocean. On both sides of Silver Canyon, very rugged
topography can be found. The valley itself is characterized by a
thalweg that gradually descends down to the ocean in an almost
stepped fashion (probably due to the digital topographic source).
The visual complexity of this landscape is not due to
particular landscape configurations encountered along the
direction of motion, but to the steep slopes on both sides of the
valley, that are perceivable as such only in one of the two
trajectory conditions (see Figure 5.6).
Landscape design elements
Silver Canyon is probably the landscape where it is most
evident that the original characteristics of the source
environment (in this case, Santa Catalina Island), have not been
replicated, but only used and transformed according to design
choices.
The "universal" land cover was set to a conifer forest type,
reminiscent of Oregon or British Columbia landscapes, which gave
the landscape a distinct look with respect to the other two.
"Specific" land covers were digitized along the thalweg of the
valley, and in particular hardwood woodlands areas were defined to
suggest some sort of riparian corridor.
A stream was manually digitized along the line of maximum slope of
the thalweg of the valley. Together with the topography,
Silver Canyon offered also the presence of the ocean,
visible in the HA-U condition only, taking up a large part of the
last frames of the animation (there are also a few ocean pixels in
the last frame of LA-TF).
The general look of Silver Canyon was made to recreate a
gloomy morning (9:00am, May 28th) in some area of the Pacific
Northwest. The more contrasted shadows of the morning time tend to
exaggerate the topography, although a preliminary design set in
the evening was discarded for the too dramatic differentiation
between the opposite slopes of the valley.
An atmospheric effect, haze, was added to complete the ambience,
but also to hide the so-called "end-of-the-world" effect,
noticeable when a perspective view is set on a dataset that does
not reach in extent the horizon line.
Trajectories implementation
The trajectories designed for the Silver Canyon landscape
are shown in Figure 5.7. The HA-U condition
for Silver
Canyon is described by a trajectory at a constant altitude of 802
meters, that reflects the standard value of altitude above the
ground of 763 meters, that in this case was measured from the end
(and lowest part) of the terrain profile. The trajectory can be
considered straight, although it was minimally arcuated to follow
the direction of the valley.
The LA-TF condition is characterized by a reference distance from
the ground of 20 meters. The graph showing the 5-times vertical
exaggeration on that condition indicates how the trajectory so
defined was adapted to offer, at the same time of the
distance-from-ground constraint, a non-discontinuous and
reasonably gentle motion to the observer. A sample of the
animations' frames is in Figure 5.8.
3.1.4 Landscape 3 - Ynez
Peak
Figure 9: Ynez Peak site map
Figure 10: Ynez Peak profile graphs
Figure 11: Ynez Peak sample frames
Design objectives
The landscape of Ynez Peak (short version for
Santa Ynez Peak, the local highest point of Santa Ynez
Mountains, directly flown over in the animations) was selected to
be part of the final experiment design mainly for its distinct
topography and visual complexity. Again, the
haphazardly-placed landscape design elements did not aim at
replicating the local ecology, but to create a sufficient visual
variety.
Terrain structure
Topographically, Ynez Peak offered a good occasion to
examine a pure landform rising structure type of
landscape, which in other sites could not be found for multiple
reasons, including the presence of local irregularities which
denied the possibility of using a completely monotonic and
increasing surface profile.
Ynez Peak also offered the interesting landscape
configuration based on the arrangement of a perfectly transversal
mountain range, separating an internal valley from the ocean. Such
configuration allowed us to have the aesthetical effect of
"landscape reveal", whereby the short ranged-view of the
mountain while climbing it, gave suddenly away to the view of the
valley and the ocean below.
Finally, the fact that the animations took place on the opposite
side of the Santa Ynez Mountains with respect to the UCSB campus
reduced the chance for participants to be excessively familiar
with the area (if the campus was included as a landscape element,
it would have been visible in the very last frames of the
animation).
Landscape design elements
The "universal" land cover of Ynez Peak is subdivided
in a series of elevation-based vegetation bands that define a low
elevation (base of the mountain) cover of Oak Woodland, a
mid-elevation (center part of the hillside) cover of Shrubs, and a
high elevation (hilltop) cover of Grass.
Specific land covers were digitized and defined on the landscape,
as visible on the location map in Figure 5.9.
They comprise a
burnt forest area, a local concentration of oak woodland (outside
the "universal" bounds of the same land cover), and extended
areas covered with rocks and grass (imaginatively, some sort of
rockslide). The density of the vegetation was set very high to
obtain a "lush" effect.
A stream was digitized, from the base of the mountain in front of
the observer, to the western side of the mountain, close to the
top. The stream was modelled so that it cuts rather deeply the
mountainside, creating in the end a strong separation between the
two sides of the thick vegetation.
The Sun position was set to the early afternoon of the summer
solstice, thus characterized by very intense illumination, bright
colors, and limited shadows extent.
Trajectories implementation
The trajectories used in Ynez Peak are shown in Figure
5.10. The HA-U trajectory is characterized by a uniform flight at
the constant altitude of 1490 meters, which means an actual
distance from the ground corresponding to the one used in other
landscape conditions.
The LA-TF trajectory is constantly placed at 20 meters from the
ground, and follows the profile of the landscape, which is rather
smooth. There is no need to show a vertically exaggerated LA-TF
graph, since there are no small scale features characterizing the
landscape profile and the trajectory. Sample frames of the two
animations are
in Figure 5.11.
3.2 Instruments: questionnaire
The questionnaire utilized to test separately each landscape
experience was made of 43 questions, mostly written as Likert
scales, but also open ended, or based on graphical sketches (see
Appendix). They investigate several different
semantic areas of
landscape experience, whose categories can be broadly broken down
as follows:
- General preference and specific preference, in general in
the form of like or dislike of the landscape as a whole or in
terms of its constituting elements. This includes also the
trajectory as an item on which the participant can express a
preference (Q1-Q17).
- Spatial knowledge questions, including the ability to
remember the topographical layout of the landscape as altimetry
but also as the arrangement of the constituting elements. The same
consideration is extended to the participant's knowledge of the
trajectory along which the landscape was viewed, including
profiling and relationship with the ground (Q18-Q21).
- Aesthetics, and self-reported impressions referring to other
constructs such as presence and excitement. The term aesthetics
incorporates the direct measurement of the four Kaplan's
aesthetical factors (coherence, complexity, legibility,
mystery) (Q22-Q43).
The questionnaire aims at collecting data on the totality of the
participant's landscape experience, which, considering the
framework of this thesis, is largely mediated by constructs
related to aesthetics. The questionnaire is not the result of
design based on statistical analyses, ensuring independency and a
degree of control on the relationship between questions and
constructs. Such analysis was considered excessively demanding in
terms of the required testing procedure, which was only
preliminary to the core of this research.
The questionnaire is instead an extensive verbal articulation of
mainly aesthetical constructs all derived from the literature,
which however were extended conceptually to form viable questions
in ways that were aimed at non distorting the original idea. For
example, a way to ask about complexity was found in the Likert
scale statements "there were too many things in the landscape"
and, in the following question, "there was too little variety".
"Number of things" and "Variety" were both correlates of
complexity, although the relationship was intuitive and structured
qualitatively. In the end, questions were put down in common
English words, and organized in key statements, previously fine
tuned in wording by means of specific pilot tests (see Chapter
Pilot Experiments).
The questions and sketch maps for the spatial knowledge section of
the questionnaire for each participant and condition were coded by
means of grading schemes aimed at capturing the quantity of
landscape elements being drawn. Each landscape condition had a
different grading scheme: PC presented a total of 19 elements
(crop fields, lake, islands, etc.), SC was limited to a maximum of
12 elements (topographic features like valley sides, specific
ecosystems, etc.), and for YP the amount of elements was 15
(river, ecosystems, ocean, etc.). Since the analysis was conducted
across-trajectories and not across-landscape, the different amount
of elements was not a limitation of the procedure, but rather a
feature that gave adaptivity in the classification of each
landscape.
The second part of the data collected for spatial knowledge refers
to the accuracy of the depiction of landscape elements in the
sketch maps. This was not meant to be a drawing skill test, rather
it measured the precision with which the shapes were drawn, and
how accurate were boundaries and location of areal features like
topography and ecosystem extensions. Accuracy was measured for
each class of items included in the respective landscape design
template, assigning scores from 0 as "not drawn", to 3 as
"perfectly drawn".
4 Procedure
The experiments were carried out using an 800 Mhz laptop PC,
running Windows ME and equipped with the broadcast multimedia
presentation software SCALA iplaySTUDIO (by SCALA Inc.). The
computer had an LCD screen with a diagonal of 14.1 inches, and
displayed the animations in 24-bit true color. The location of the
experiment was the RUSCC lab, made available at the UCSB
Department of Geography. The laptop was set on top of a table and
the participant sat on a chair in front of it at normal operating
distance.
The computer presentation that led the participants, one by one,
through the experiment repeated first the oral and written
briefing instructions. The participant had to imagine being in the
cockpit of a helicopter flying over several landscapes, which were
declared to be taken from the real world, although generated on a
computer. It was suggested to the participant to be prepared to
report her experience with the landscapes being shown, whose
number, together with the number of questionnaires and the overall
sequence of the experiment, was anticipated. Before being exposed
to the first landscape, the participant was already informed of
the fact that she had to draw a map of the landscape. In fact,
although the order of the 3 landscapes was fully counterbalanced
across the entire pool of participants, there was an asymmetry in
instructions that made the participant unaware she would have to
draw the map in the first questionnaire, and instead allowed her
to know the details of the task just after the first
questionnaire, for the remaining two questionnaires.
The mouse click was used to progress to the next page, and led the
participant to view each of the three animations twice, and then
to complete the questionnaire on paper after each pair of
animations. After the third pair of animations was completed by
the participant, she had to fill in a questionnaire on their
personal background.
During the entire experimentation the system ran smoothly, and all
the animations were displayed in exactly the same fashion at a
full frame rate.
The results comprise the responses to the Likert scale questions,
concerning general preference, specific preference, aesthetics and
sense of place. The responses to questions are considered
individually, and are statistically analyzed by means of
two-tailed independent sample t-tests, entirely within
landscape conditions and across trajectory conditions. The results
also comprise the responses to the spatial knowledge questions
hinging upon the sketch maps drawn by participants in the median
section of the questionnaire. In particular, the spatial knowledge
questions consider the two fundamental elements of quantitative
landscape element detection, recognition and report, and the
average accuracy level in reporting them on the sketch maps.
The response of participants to specific designs and conditions
resulted extremely patterned. The general preference question (Q1)
was the one with the widest scope in the questionnaire, since it
summarized the totality of the aesthetical response to the
landscape stimuli. A significant result was not predicted before
the tests, since it incorporated the variance generated by many
contributing variables. The YP/LA-TF landscape condition was
reported significantly more liked than the HA-U condition
(p < .05), while the effect was not found in similar
across-trajectory comparisons for SC and PC.
The question addressing the development of a condition of mystery
(Q45) directs our attention to the aesthetical effect primarily
consequential to the presence of a high degree of visual
complexity in the landscape. Reflecting the pattern of Q1, the
YP/LA-TF condition appears to generate a significantly higher
effect of mystery than the corresponding HA-U (p < .01).
Question 39, concerning the level of excitement of the
participant, reached significance also in the YP/LA-TF condition,
more exciting than HA-U (p < .05).
Interestingly, question 29, asking about experiencing a surprise
effect at the end of the animation, did not result significant in
any of the conditions, not even in the YP condition were the
animation was constructed especially to generate surprise through
a specific landscape "reveal" effect.
The SC landscape was primarily designed to investigate the
influence of orographic features on landscape perception, and two
ad-hoc questions were added to the questionnaire concerning this
orographic perceptual effect. Interestingly, the responses allowed
us to find a significant difference between the SC/LA-TF and HA-U
in question 5 (p < .05), indicating a stronger preference for
the level of closeness in the LA-TF condition contrasted with
HA-U. However it must be noted that to several subjects did not
seem clear what was meant by a "closed" landscape, and the
minimal explanation in the testing phase might have in part
affected the results.
The "sheltering" pattern becomes even clearer in the specific
question 33 asking whether the participant felt sheltered by the
landscape. Sheltering in the SC/LA-TF condition, characterized by
high valley walls to the left and to the right of the moving
observer, was experienced significantly stronger than in the
corresponding HA-U condition (p < .01). This pattern was not
repeated in any of the other two landscape conditions (as
expected).
The direct questions generated from the four Kaplan's factors
(used as constructs) did not produce significant differences
across trajectory conditions, except for those measuring
complexity. In fact, the two partially overlapping questions on
the degree of variety in the scene, and on the quantity of
elements present, gave converging results, indicating that the
landscape in PC/LA-TF appears significantly more complex than the
same landscape seen in the HA-U condition (Q24 p < .01, Q25
p < .01)
While considering the across-trajectories differences of the
specific preference questions, the results seem to be the
consequence of the striking changes in appearance of landscape
when altitude is appropriately manipulated. For example, the
overall relief of the PC landscape was perceived to be
significantly higher in HA-U condition than in LA-TF (Q6,
p < .01). In light of the flatness of the plain and the
mountains on the horizon the result reflects more of the elements
being visible than a different evaluation of the elevation of the
same elements.
Along the same lines, the specific preference questions show how
the difference across trajectory conditions are due to the
specific visual structure of the landscape. Because of this reason
vegetation is more preferred in PC/HA-U than in LA-TF (Q10,
p < .05), and, conversely, preferred more in SC/LA-TF than in
HA-U (Q10, p < .01). Also, roughness is significantly
preferred in YP/HA-U, compared to LA-TF (Q9, p < .05).
The questions related to sense of place gave significant effects
in the "uniqueness" question (Q35), which indicates that, in the
HA-U condition, the artificially patterned PC landscape resulted
more convincingly unique than in the LA-TF condition
(p < .01).
Some questions referred to the concept of perceptual satisfaction
with the landscape experience being shown. The question addressing
the degree of satisfaction with the extension of the viewshed
available from the animations resulted in a significantly
different self-report. Specifically, the SC and YP conditions both
showed how the two LA-TF conditions generated dissatisfaction with
the amount of information available (Q41, SC p < .01, YP
p < .01). The related concept of desiring to know more of the
landscape (Q43) gives an inverted pattern, since LA-TF conditions
for SC (p < .05) and YP (p < .05) clearly indicate how the
lower altitudes elicit more curiosity for generic information for
the landscape, but also, in the case of SC (Q40, p < .05), for
further visual exploration. Interestingly, PC does not offer
significant results across condition in these respects.
As a general effect of the animated stimuli, the speed of the
animation (Q46), related to the concept of information rate, was
reported as consistently different across trajectory conditions in
all the three landscape conditions. In fact, LA-TF flights always
seemed less excessively slow than HA-U (PC p < .05, SC
p < .001, YP p < .01), although the actual flying speeds
were virtually identical. The confound constituted by the observer
moving at higher speed when climbing hills (see Methods) does not
interfere substantially with the result, also considering the
non-confounded conditions of PC and SC giving the same result as
the confounded YP.
In the case of the spatial knowledge questions, the analysis of
the number of elements consisted in running a t-test
comparing the total number of objects detected in the two
conditions (LA-TF and HA-U) by each participant, to detect
differences in the multiple conditions.
The main finding of this section is that the number of landscape
elements detected in PC/HA-U is significantly different than in
LA-TF (p < .001), indicating a better capacity in detecting
and remembering the elements of the landscape when flying high.
Conversely, the other landscape conditions did not show a similar
effect.
The accuracy scoring analysis was also ran by means of a
t-test that compared the average accuracy score of each
participant in the entire set of accuracy variables, across
trajectory conditions The PC/HA-U condition did not allowed
participants to score significantly higher in accuracy than in the
LA-TF condition, so that PC/HA-U is eliciting only the detection
of a higher number of elements. This supports in part the ability
of layout view to increase the chances of gaining a better spatial
knowledge.
The patterned results presented in the previous Chapter suggest a
solution to the problem of whether trajectory of approach is or is
not a factor influencing landscape experience. The direct,
between-subject comparison of trajectory conditions (LA-TF and
HA-U) indicates several interesting cases in which the reported
landscape experience is significantly different. This Section will
show how this behavior resulted coherent with the principles
reviewed in the literature review, and with the theoretical
considerations documented in the conceptual framework.
By means of considering that visual complexity underlies
the nature and degree of the aesthetical response to landscape, we
have empirically measured such influence by controlling the
characteristics of the terrain of the designed landscape being
filmed. In this context it was also suggested that the YP/LA-TF
condition was the most visually complex amongst the three,
specifically in terms of how the visibility of topography varies
dynamically during exploration in time. PC had instead a
topography which was the least visually complex. It must be
stressed that the other YP trajectory condition (HA-U) offers a
completely different complexity pattern, since the trajectory
comes in touch with very little of the topographical
visual complexity characterizing the landscape. In fact,
the helicopter flies very high, without ever encountering
temporary occlusions, reveal effects, and subtle changes in
terrain.
The empirical results provide converging evidence of the striking
difference existing between experiences done above the same
landscape but along different trajectories. The interesting fact
that, in the YP case, LA-TF is significantly liked more than HA-U,
combines with the lack of a clear effect in the other two
conditions. It is therefore suggested that even the most general
measure of preference (that is, like or dislike) might have a
different outcome in a given landscape according to the trajectory
being used. Along these lines, the same YP/LA-TF trajectory is
able to instill more excitement and to create more mystery than
the corresponding HA-U. The latter effect clearly indicates how
the higher visual complexity of YP was directly measured
in terms of the strongly correlated mystery factor. We suggest
that mystery perception is the natural experiential aspect of the
more implicit concept of visual complexity. In other
words, a high visual complexity creates a better
articulation of landscape features for mystery to be perceived and
to have a stronger psychological impact upon the observer.
However, the mystery effect is empirically detectable on a
landscape characterized by high visual complexity, but
only if the trajectory of the viewpoint crosses the spaces in
which such complexity is explicitly manifest and experienceable in
the form of partial or total occlusions, intriguing arrangements
and accessibility of distance information: in conclusion, in a
form useful for triggering survival-based activity.
Certainly it is true that terrain creates the highest degree of
complexity amongst all contributing factors to visual
complexity in a landscape. If a trajectory follows closely the
ground, such as the LA-TF trajectories, there is a higher chance
to be confronted with mystery, aesthetical pleasure, curiosity and
intense feelings. This consideration naturally stems from the fact
that a landscape animation does not imply necessarily a trajectory
of flight functionally related to the terrain (such as a TF
condition), but rather just a viewpoint moving above the terrain
in a terrain-independent fashion.
In a mathematical generalization, it is argued that if a
trajectory is a generic mathematical function of terrain form
it is easier for it to be closer to those aerial spaces appearing more
perceptually interesting, stemming from the accessibility to higher
terrain
complexity.
The experimental design does not offer a complete empirical
support to this idea, since the TF condition is also confounded by
the LA component. However it is expected that in a future study on
functional terrain trajectories, terrain-independent trajectories
will be found less able to obtain visual access to landscape
topographical complexity.
Visual complexity is also the source of other explicit
effects, such as the sheltering effect in the SC landscape. To our
knowledge there are not empirical investigations on the sheltering
effect, which is mostly an architectural and visual design concept
which is most often left in its qualitative, albeit useful, form.
The clear sheltering effect is detected in clear correlation with
the physical characteristics of the canyon landscape, that is,
present when the canyon walls were visually sheltering, and it is
not detectable in any other condition. This is another
confirmation of how the spectrum of perceptual experiences is
strongly correlated with the characteristics of the landscape, and
specifically terrain. In fact, the latter is a determining entity
in landscape experience.
In general the comparison between the effects on spatial knowledge
of the high layout view, versus the low altitude first person
horizontal perspective, gave the expected results. Almost as a
dual, or an opposite, of perception and aesthetics, spatial
knowledge is best gained when the visual complexity is
simplified or experienced from a more advantageous viewpoint, and
when consequentially there is not an involving dynamic experience
with the landscape. The fact that in both SC/LA-TF and YP/LA-TF
participants demanded for a higher vantage point for a future
better navigation on the ground is indicative of how altitude is a
fundamental correlate of the visual accessibility of landscape to
gain spatial knowledge for wayfinding and navigation.
It should be noticed how the grading scheme has in part determined
the statistical pattern, since PC was much more itemized (many
crop fields, vegetation patterns, island). In HA the participant
could simply tap in a more vast quantity of pick-up-able landscape
elements than the corresponding LA-TF condition. Also, the less
itemized forests of SC and the irregular ground patterns of YP
allowed less numerical counts. In other words, across landscape
condition, elements pick-up varies with the ability of landscape
to be subdivided in many constituents elements that can be singled
out (consider that, for example, many similar trees are seen as
one forest, as in the case of SC).
In other words, HA (with the additional component of the slightly
reclined pitch of the helicopter) produces in the observer less
involvement, excitement, immersion in the landscape, and instead
grants visual accessibility to what is not otherwise visible in
other conditions. However, the gain with HA-U trajectories reaches
"critical mass" only with certain landscape configurations, such
as the more artificial PC, since in other landscape configurations
is harder to distinguish enumerable features from confusing
vegetation aggregates and over-complex topographic arrays of
features.
In summary, the main finding of this study is that any landscape
is characterized by a degree and a type of visual
complexity, which in turn projects around it a space that
determines an accessibility to such complexity. Explorers
negotiate visual access with the complexity of the landscape and
their movement grants them access to certain aspects and not
others. Such accessibility space is highly patterned, and the
striking differences between HA and LA show that the approach to
landscape structures the experience of the approach itself. The
experiential differences stemming from the patterns of the
accessibility space comprise all component of human psychological
experience, including perceptions, spatial knowledge and feelings.
Once a trajectory on a landscape is decided, the landscape becomes
accessible in a certain way, which is defined by the implicit
complexity of the landscape and the trajectory negotiating with
it. The concept of filtering summarizes the previous
considerations, by stating how the selection of actual
trajectories from all the possible (and impossible) ones allows to
exploit different areas of this visual accessibility space
according to what we want to know about the landscape.
Landscape is the result of a process of visual selection of the
environment that contributed to the evolution and survival of the human
race.
The presence of evolutionary roots in landscape can be found in the
development of different expectations of survival in the variety of
environments accessed by human ancestors, and in the formulation of
different
predictions in the carrying capacity of certain landscape
configurations for
hunting. In general, landscape was of paramount importance in the
process of biological fine-tuning of the responses to landscape
features that
could mean the life or death of the ancient forefather.
We inherited the entire set of results of millennia of decision
making in the form of aesthetical conscience of what we like or
dislike of the landscape (or of anything else, for that matter). When
we take aesthetics into consideration, we are able to tap into the
depths of
that perceptual experience, which results in a unique measure of our
psychological relationship with the external environment. In fact,
aesthetics
can be considered the direct outcome of an enormous set of
instantaneous
measurements of the external environment.
Along the same lines of thought, the participants of this study
should be considered as measurers of the several landscape
conditions. As discussed in the previous two Sections, the final
experience of landscape significantly varied according to the
dynamics of approach to landscape, and to the landscape being
shown. In general, the patterned answers of the participants
showed how any landscape is not perceptually isotropic in relation
to the psychological variables being measured. Rather, it shows
preferred modes of approach to elicit stronger reponses, and areas
of approach where the visual complexity of the landscape
is completely expressed by means of multiple occlusions, thus
increasing the level of mystery. Instead, in other areas a strong
topographical visual effect is not detectable because the observer
is not in the corresponding observation "envelope" (or, "in the
right place to observe").
This study shows that having a selective and specific access to
landscape is feasible. Also, it is shown that by varying
trajectory also the dynamics of such selective access is varied.
Acknowledging this form of accessing landscape we can progress in
several research directions, such as, for example, the mapping of
visual accessibility spaces around landscape to classify how
psychological response varies. We can also measure the degree of
expression of visual complexity, and determine a
preferred set of trajectories for a better visual resource
exploitation. These new types of maps will be characterized by a
projection of the visual accessibility spaces as a function of
terrain form, since that resulted to be the determinant factor.
The process of filtering in this case would be dedicated to
rule-out the non-terrain-based possibilities of trajectory form.
The orientation of this study was fundamentally epistemological, that
is, an
investigation in the patterns of learning about and exploring
landscape.
Therefore, it might be opportune to conclude that
while we can endlessly discuss about how to better visualize and
represent
landscape, more importantly the issue would be to guarantee to
ourselves the
access to accurate knowledge while exploring landscape. Our ancestors
knew
that such truthful concept of the environment was a convenient
reference mark
to guarantee themselves good chances of survival.
At the same time, we want to develop knowledge to represent
planetary landscapes not to induce controlled (and thus, ethically
questionable) emotions, but to instill interest in what already
underpinned Flemish landscape painting five centuries ago:
landscape that first can help us in represent our own reality (and
desires) in
an externalized format; and secondly, landscape as a
placeholder, or a landmark, for a sense of mystery to be unfolded, able
to
stimulate the search for new accesses to what lies beyond.
APPENDIX 1 -QUESTIONNAIRE
Questionnaire
Landscape Appreciation
Administrator: Marco Ruocco
Circle a number from 1 to 7 to express
your degree of agreement or disagreement with each of the following
statements:
1 -
2 - 3
- 4 -
5 - 6
- 7
|
|
Strongly agree
Strongly disagree
1) I liked the landscape shown in the animation.
2) The landscape shown in the animation seemed natural.
3) I liked the level of naturalness that I found in the landscape.
4) The landscape shown in the animation seemed closed.
5) I liked the level of closeness that I found in the landscape.
6) In general, the difference in elevation between high and low areas
(e.g. between hills and plains) in the landscape seemed very high.
7) I liked the elevation differences present in the landscape.
8) The roughness of the landscape shown in the animation seemed very
high.
9) I liked the level of roughness of the landscape.
10) I liked the vegetation cover of the landscape.
11) I liked the sky above the landscape.
12) I liked the water bodies present in the landscape.
13) I liked the way the helicopter flew around above the landscape.
Answer the following questions in
the space provided:
14) What did you like about the landscape, if anything?
15) What did you dislike about the landscape, if anything?
16) What did you like about the way the helicopter flew around, if
anything?
17) What did you dislike about the way the helicopter flew around, if
anything?
Follow the instructions below, and write and draw in the spaces
provided.
18) In the space provided below, draw a map of the landscape that you
have been shown as if you were looking at it from above (i.e., a bird's
eye view). Try to provide information about the topography of the
landscape, indicating the location of features such as valleys, ridges,
peaks, etc. Put verbal labels on the map to define the objects that you
have drawn. Then, on the map, draw the line representing the trajectory
of the helicopter on the landscape, as if you were looking at it from
above.
[MAP DRAWING SPACE]
19) In the space provided below, describe in your own words all the
features you have drawn on your map.
20) In the space provided below, draw the line of the surface of the
landscape flown over by the helicopter, from the start to the end of
the animation, as if you were looking at it sideways from the ground
(i.e., draw the profile view, or cross-section, of the landscape).
Indicate variations in elevation such as those caused by valleys and
ridges. Then, on top of that, draw the
trajectory of the helicopter, as if you were looking at it sideways
from the ground (i.e., draw the profile view of the trajectory of the
helicopter), indicating variations, if any, in helicopter altitude
during the animation. Remember to label both profiles.
[PROFILE MAP DRAWING SPACE]
21) In the space provided below, describe in your own words all the
features you have drawn on your map.
Circle a number from 1 to 7 to
express your degree of
agreement or disagreement with each of the following statements:
1 -
2 - 3
- 4 -
5 - 6
- 7
|
|
Strongly agree
Strongly disagree
22) There were elements in the landscape that did not fit well with
each other.
23) The landscape made sense to me overall.
24) There was too little variety in the landscape.
25) There were many things in the landscape.
26) It would have been easy to find my way around if I was walking on
the
ground.
27) I felt it would be better to find my way around if the helicopter
flew closer to the ground.
28) I felt it would be better to find my way around if the helicopter
flew higher above the ground.
29) The part of landscape I saw at the end of the animation was
surprising.
30) During the animation I was curious about what I was going to see
next in the landscape.
31) I felt the landscape was a pleasant place to be in.
32) I could easily see what was going on around me.
33) I felt sheltered by the surrounding landscape.
34) If I imagine people standing on the landscape and looking upwards,
the helicopter would be very visible to them.
35) The landscape was unique.
36) The landscape had a specific character.
37) The landscape was easy to remember.
38) The scenery offered by the landscape was striking.
39) The way the helicopter flew around was exciting.
40) I would have liked to explore the landscape some more after the end
of the animation.
41) I felt I could see a large enough portion of the landscape at any
given time.
42) I felt myself present in the landscape.
43) I would like to know more about the landscape.
44) I feel there is not much more to see in the landscape beyond what
was shown in the animation.
45) Any slight movement of the helicopter offered me a different view
of the landscape.
46) The helicopter flew around too slowly.
47) The animation has shown a realistic landscape.
BIBLIOGRAPHY
Adams, A. J. (1994). Competing communities in the great bog of europe:
identity and seventeenth-century dutch landscape painting. In W.
Mitchell (Ed.), Landscape and power, pp. 35-76. Chicago and London: The
University of Chicago Press.
Appleton, J. (1996). The experience of landscape. John Wiley and Sons.
Bell, S. (1993). Elements of visual design in the landscape. London:
E&FN Spon.
Beneditk, M. and C. Burnham (1985). Perceiving architectural space:
from optic arrays to isovists. In S.R. Warren, W.H. (Ed.), Persistence
and change: Proceedings of the first international conference on events
perception, Hillsdale, NJ, pp. 103-114. Lawrence Erlbaum Associates.
Berleant, A. (1992). The Aesthetics of Environment, Philadelphia:
Temple University Press.
Berry, J., D. Buckley, and C. Ulbricht (1998, August). Visualize
realistic landscapes. GISWorld, 42-47.
Bishop, I. and R. Hull (1991). Integrating technologies for visual
resource management. Journal of Environmental Management 32, 295-312.
Burt, I. (1195). Clarity and sense of place in maps. The Cartographic
Journal 32.
Buttenfield, B. and Mackaness (1991). Visualization. In D. Maguire, M.
Goodchild, and D. Rhind (Eds.), Geographical Information Systems:
principles and applications. Harlow, Essex, England: Longman Scientific
and Technical.
Campbell, C. and S. Egbert (1990). Animated cartography: thirty years
of scratching the surfacve. Cartographica 27(2), 24-46.
Cosgrove, D. (1984). Social formation and symbolic landscape. Barnes
& Noble.
Cutting, J. and P. Vishton (1995). Perceiving layout and knowing
distances: the integration, relative potency and contextual use of
different information about depth. In W. Epstein and S. Rogers (Eds.),
Perception of space and motion. San Diego: Academic Press.
DiBiase, D., A. MacEachren, J. Krygierm and C. Reeves (1992), Animation
and the role of map design in scientific visualization. Cartography and
Geographic Information Systems 19(4), 201-214, 265-266.
Dubbini, R. (1994). Geografie dello sguardo: visione e paesaggio in
età moderna. Torino: Einaudi.
Eley, M. G. (1992). Component processing skills in the interpretation
of topographic maps. Cartographica, 35-51.
Gibson, J. (1979). The ecological approach to visual perception.
Boston: Houghton Mifflin Company.
Gillam, B. (1995). The perception of spatial layout from static optical
information. In W. Epstein and S. Rogers (Eds.), Perception of space
and motion. San Diego: Academic Press.
Graf, K. (1995). Realistic landscape rendering using remote sensing
images, digital terrain models and 3D objects. University of Zurich.
Graf, K., M. Suter, J. Hagger, E. Meier, P. Meuret and D. Nuesch
(1994). Perspective terrain visualizations - a fusion of remote
sensing, gis and computer graphics. Computer & Graphics 18 (6),
795-802.
Granö, J. (1929). Pure Geography. Baltimore and London: The John
Hopkins University Press.
Harrison, C. (1994). The effects of landscape. In W. Mitchell (Ed.),
Landscape and power, pp. 203-239. Chicago and London: The University of
Chicago Press.
Hartig, T. and G. Evans (1993). Psychological foundations of nature
experience. In T. Garling and R. Golledge (Eds.), Behavior and
Environment: Psychological and Geographical Approaches. Elsevier
Science Publishers B.V.
Heft, H. (1983). Wayfinding as the perception of information over time.
Population and environment: behavioral and social issues 6, 133-150.
Heft, H. (1996). The ecological approach to navigation: a gibsonian
perspective. In J. Portugali (Ed.), The construction of cognitive maps,
pp. 105-132. Dordrecht (The Netherlands): Kluwer Academic Publishers.
Helsinger, E. (1994). Turner and the representation of england. In
W.Mitchell (Ed.), Landscape and power, pp. 103-125. Chicago and London:
The University of Chicago Press.
Huss, R. and N. Silverstein (1968). The film experience. New York:
Harper & Row Publishers.
Jenks, G. and F. Caspall (1967). Vertical exxageration in
three-dimensional mapping. Technical Report NR 389-146, Nonr 583(15),
Geography Branch, Office of Naval Research, Department of Geography,
The University of Kansas, Lawrence, Kansas.
Johnson, L. (1974). Film, space, light, time and sound. New York: Holt,
Rinehart and Winston, Inc.
Kaplan, S. (1987). Aesthetics, affect, and cognition: environmental
preference from an evolutionary perspective. Environment and Behavior
19(1), 3-32.
Kraak, M. (1988). Computer-assisted cartographical three-dimensional
imaging techniques. PhD. thesis, Delft University Press.
Kraak, M. (1990). Theoretical aspects of three-dimensional cartography.
International Yearbook of Cartography XXX, 81-91.
Llobera, M. (2003). Extending gis-based visual analysis: the concept of
vidualscapes. International Journal of Geographical Information Science
17, 25-48.
Lorzing, H. (2001). The nature of landscape. Rotterdam: 010 Publishers.
Lowenthal, D. (1961). Geography, experience, and imagination: towards a
geographical epistemology. Annals of the Association of American
Geographers 51 (3).
Lowenthal, D. (1966). The american scene. Geographical Review 58 (1),
61-88.
MacEachren, A. (1992). Application of environmental learning theory to
spatial knowledge acquisition from maps. Annals of the Association of
American Geographers 82 (2), 245-274.
McGreevy, M. (1993). Virtual reality and planetary exploration. In A.
Wexelbat (Ed.), Virtual Reality applications and explorations, pp.
163-197. Academic Press Professional.
McGreevy, M. (1994). An ethnographic object-oriented analysis of
explorer presence in a volcanic terrain environment. Technical Report
TM-108823, NASA, Ames Research Center, Moffet Field, California.
McLaren, R. and T. Kennie (1989). Visualization of digital terrain
models: techniques and applications. In J. Raper (Ed.), Three
dimensional applications in Geographical Information Systems. Taylor
& Francis.
Mehrabian, A. and J. Russel (1974). An approach to environmental
psychology. Cambridge, Massachussets: The MIT Press.
Mitchell, W. (1994). Imperial landscape. In W.Mitchell (Ed.), Landscape
and power, pp. 5-34. Chicago and London:The University of Chicago Press.
Moellering, H. (1980). The real-time animation three-dimensional maps.
The American Cartographer 7(1), 67-75.
Muir, R. (1999). Approaches to landscape. Houndmills, Basingstoke,
Hampshire (UK): MacMillan Press LTD.
Rolfe, J. and J. Staples (1986). Flight simulation. Cambridge:
Cambridge University Press.
Shamai, S. (1991). Sense of place: an empirical measurement. Geoforum
22(3), 347-358.
Shepard, P. (1991). Man in the landscape. Texas A&M University
Press.
Sholl, M. (1996). From visual information to cognitive maps. In
J.Portugali (Ed.), The construction of cognitive maps, pp. 157-186. The
Netherlands: Kluwer Academic Publisher.
Sieber, R. (1996), Visuelle wahrnehmung dreidimensionaler
parametrisierter objekte und objektgruppen: eine empirische
untersuchung zur bestimmung eines optimalen betrachterstandortes.
Geoprocessing Series 26, Department of Geography, University of Zurich,
Zurich, Switzerland.
Sitney, P. (1993). Landscape in the cinema: the rhythms of the world
and the camera. In I.G.S. Kemal (Ed.), Landscape natural beauty and the
arts, pp. 103-126. Cambridge University Press.
Smith, R. and A. Brown (1996). Developing a sense of place. Journal of
Geography 95(2), 86-89.
Tayolor, R. (1984). Effects of map scale, complexity, and
generalization on terrain-map matching performance. Cartographica 21,
129-134.
Todd, J. (1995). The visual perceptiojn of three-dimensional structure
from motion. In W. Epstein and S. Rogers (Eds.), Perception of space
and motion. San Diego: Academic Press.
Tuan, Y.-F. (1975). Place: an experiential perspective. Geographical
Review 65, 151-165.
Tuan, Y.-F. (1979). Thought and landscape. In D. Meinig (Ed.), The
interpretation of ordinary landscapes. New York: Oxford University
Press.
UCGIS (2000). Emerging themes in giscience research: geographic
visualization. white papers. Technical report, University Consortium
from Geographic Information Science.
Warren, W. (1995). Self-motion: visual perception and visual control.
In W. Epstein and S. Rogers (Eds.), Perception of space and motion. San
Diego: Academic Press.
Zajonc, R. (1980). Feeling and thinking: preferences need no
inferences. American Psychologyst 35 (2), 151-175.
Zettl, H. (1990). Sight, sound, motion: applied media aesthetics.
Belmont, California: Wadsworth, Inc.
