High-Performance Computing Applications for Visualization of Large Microscopy Images - High-Throughput Image Reconstruction and Analysis - page 309

solution for solving the need for large pixel real estates [7]. Figure 14.4 shows one
such display being used by researchers for visualizing high-resolution content. The
collective resolution of these displays can run up to a few hundred megapixels.
They are usually run by a cluster where a single node drives one or more tiles and
all the nodes are interconnected by gigabit Ethernet or equivalent hardware. The
software responsible for managing the displays is inherently distributed in nature
and can synchronize the visualization across tiles. Ideally we would like to use the
entire display as one giant desktop where multiple applications can simultaneously
reside. Datasets can be placed adjacent to each other for visual comparison using
these applications. These windowed applications can either run locally on one or
more local nodes or remotely across the network on a cluster.
Tile-displays at times span entire room lengths and can be cumbersome to in-
teract with because of the large physical space they cover. Because of their large
size, the paradigm of a keyboard and a 2D mouse on a fixed workstation does
not apply very well. Users sometimes want to analyze the high-resolution content
up close and sometimes want to step back to do a visual comparison of multi-
ple datasets. Multiple users want to interact with different sections of the display
simultaneously and want their own mouse pointer and keyboard. This is differ-
ent from the typical one-desktop-one-user paradigm used by desktop operating
systems. Thus, the displays bring with them a host of unique human-computer
interface problems that require unique solutions. Computer scientists have exper-
imented with alternate input devices such as wireless 3D mice and wireless tablet
PCs which can be held by the user as they walk around the room and help provide
user input to the displays. Groups are also working on camera-based face, gaze,
and hand-tracking systems which will one day help do away with the need for
carrying any physical input device on the person.
Figure 14.4 Researchers using NCMIR's 40-megapixel display wall for conducting collaborative
data exploration experiments. The display is managed by Scalable Adaptive Graphics Environment
(SAGE), which allows multiple graphical applications to use the tile-display like a shared desktop
space.