Fusion of Visual Channels - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 117

Flow Visualization often sees the need for multifield visualization, where a vector
field is depicted with other multivariate properties such as temperature and pres-
sure [ 13 , 16 , 19 ]. The primary vector fields are usually visualized using the orienta-
tion and size channels, which are commonly depicted using glyphs, lines, surfaces,
and textures. Different colors are commonly used for depicting one or two secondary
fields. In 3D flow visualization, it is common to visualize a vector field in conjunc-
tion with an iso-surface. In this case, any volumetric data that is not on the surface
is removed to reduce occlusion.
Uncertainty Visualization involves the depiction of the measurement of uncertainty
in conjunction with the primary data set. When both are in field representations, it
essentially becomes a problem of multifield visualization. The primary task is thus
association because the foremost requirement is usually the need to observe the
uncertainty measurement associated with some or all parts of the data field. The
visual channels used for depicting uncertainty include color, texture, transparency,
haziness, blurring, uncertainty glyph, and geometric transformation (for details, see
discussions in [ 11 , 18 ]).
One common dilemma in uncertainty visualization is that the underlying data
fields often require the use of several visual channels. For example, in surface and
volume rendering, many geometrical channels and optical channels are used explic-
itly or implicitly. The continuous spatial usage prevents substantial use of relational
and semantic channels. If the visual channels for depicting uncertainty were con-
fused with those for the primary data fields, it would introduce additional and unde-
sirable visual uncertainty. One effective means for addressing this dilemma is to use
repetitively-animated glyphs [ 15 ]. The dynamic nature of the uncertainty depiction
makes a clear distinction from the static depiction of the primary data fields.
12.4 Composition of Time-Varying Fields
Time-varying fields are often visualized as an animated sequence of images, each
of which depicts a single field or multiple fields at a particular time step. While this
form of visualization is intuitive and commonly deployed in practice, it has several
shortcomings. For example, viewing animations requires full attention, and is prone
to change blindness. Because of the limitation in short-term memory, an animation
is often watched over and over again in order to make comparison between different
time steps.
One alternative approach is to compose a static visualization to depict several time
steps. Hence, even when we consider only a signal field, F , at two different time steps
t 1 and t 2 , the composition of the two fields, F t 1 and F t 2 transforms the problem to
that of a conventional multifield visualization as discussed in the previous sections.
The requirements for such visualizations approaches often place a great emphasis
on spatial association, comparison and differentiation, posing a non-trivial challenge
to the selection of visual channels. On one hand visually disparate channels (e.g.,