Showing Shape with Texture – Two Directions are Better than One
Author(s): Sunghee Kim, Haleh Hagh-Shenas, Victoria Interrante.
Proceedings:
2002.
[BibTeX] 
Abstract:
If we could design the perfect texture pattern to apply to any
smooth surface in order to enable observers to more accurately
perceive the surface’s shape, what would the characteristics of
that texture pattern be? The answers to this question have
important potential impact across a wide range of visualization
applications, from molecular modeling to radiation therapy
treatment planning, in which scientists need to attain an accurate,
intuitive understanding of the shapes of complicated, smoothly
curving surfaces in their data. Over the past several years,
researchers in our lab have carried out a series of experiments
intended to investigate the impact on shape perception of various
characteristics of surface texture patterns. In this paper we report
the results of our most recent study, in which we compare
performance on a surface attitude probe adjustment task under
three distinct conditions of principal direction pattern orientation
and a control condition in which no texture was present. The
three texture conditions were: a doubly-oriented texture in which
approximately evenly-spaced lines follow both of the principal
directions, a singly-oriented line texture which follows only the
first principal direction, and a singly-oriented line integral
convolution texture, from which information about texture
compression in the direction of the texture flow may be indirectly
accessible. Over a series of 200 trials (4 texture conditions x 10
surface/probe locations x 5 repeated measures), a total of five
naïve participants were asked to adjust a circular probe, randomly
located on an arbitrary doubly curved surface, so that its base
appeared to lie in the displayed surface and its perpendicular
extension appeared to be oriented in the direction of the surface
normal. An analysis of the results showed that performance was
best in the two-directional texture condition, closely followed by
the line integral convolution condition. Performance was further
decreased in the one-directional and no texture conditions (in that
order).
The paper is organized as follows. In section 1 we describe
the motivation for our work. In section 2 we very briefly discuss
previous and related work in the field of vision research, and
briefly recap the findings of our first two experiments [9, 10]
which focused on the effects of pattern anisotropy and the impact
of texture orientation. In section 3 we describe our experimental
methods, including a brief summary of the process of the stimuli
preparation, and we present a detailed statistical analysis of our
experimental results. In section 4 we discuss the implications of
our findings, and in section 5 we outline plans for future work.