What is it?

TileViewer is a distributed visualization framework, originally developed by Dr. Sung-Jin Kim as a part of the Ph.D. dissertation. The version 1.0 started around Fall 2004.

It worked on Windows XP, Mac OS X, and Linux (although for Linux, a couple of features were not available). Also different platforms could work together at the same time.

The main focus of TileViewer is to enable a highly interactive visualization on a display wall such as the HIPerWall. 'A highly interactive' meaning that user can display various data formats on the display wall and at the same time, the user can rearrange/zoom displayed data in real-time.

The following clip shows TileViewer in action. The user rearranges/zooms-in-out dozens of medical image data using a 3D mouse.

What various data formats?

TileViewer supports various data formats such as big images, NDVI data, MRI scans, SOAR 3D terrain, quicktime videos, and a webcam.

What big image?

The first data that TileVeiwer displayed were extremely high resolution images, aka 'Big images'. These big images have a resolution of around 200-600Mpixels. It is very difficult to display such images on a single machine not to mention a display wall.

For TileViewer, such images first should be preprocessed and converted into a tiled pyramidal TIFF format. Then the converted image is copied to all display nodes that consists the display wall. At run time, TileViewer control each display node to load and display only necessary parts of the image, enabling a interactive browsing of such image.

Since each display node has a reasonably powerful graphics card, TileViewer can use a Cg shader to docolor filtering on the displayed portion of the image.

N..D..V...what?

NDVI (Normalized Difference Vegetation Index) is a data format that contains vegetation index recorded by a satellite. Through working with Prof. Padhraic Smyth and his student, Lucas Scharenbroich, NDVI data sets were converted and could be displayed on TileViewer.

MRI scans

These are function MRI scans. This data set represents MRI scans of five different patients who travelled ten different hospitals in US. With the help of Prof. Padhraic Smyth and his student Seyoung Kim, the data were converted and displayed for TileViewer.

Each MRI data contains whole slice of brain scan and using TileViewer, the user can browse all slices of 50 data in real-time. Also the Cg shader is used to threshold the max/min value which isolates a certain section of the scan.

"I am a leaf on the wind. Watch how I SOAR."

SOAR (http://www.gvu.gatech.edu/people/peter.lindstrom/software/soar) is an out-of-core terrain visualization engine by Peter Lindstrom

The source code was borrowed and ported in order to show the flexibility of TileViewer.
No algorithmic change was made. Just wrapped the main function of SOAR engine as a C++ class. SOAR.h and SOAR.cpp are made by simply copying the functions in main.cpp of SOAR engine.

Videos ...of course...

What is the point of the big display, if we cannot watch a video on it.

*This work was supported in part by the National Science Foundation as part of the OptIPuter project (http://www.optiputer.net/) under grant No. SCI-0225642.