This appendix describes the RendAsunder demonstration program that is supplied with Array 3.0 software.
RendAsunder is an interactive parallel software volume renderer. During the process of volume rendering, a three-dimensional array of data elements (each assigned a corresponding color and transparency) is rendered to a two-dimensional image, from a user-defined viewpoint.
A 375 MB volumetric data set from the National Library of Medicine's Visible Human Project is provided with RendAsunder. The data set consists of a 584 x 1878 x 341 element volume of eight-bit samples of a cryosectioned human male.
 | Note: Use of this demonstration implies agreement with the terms of the contract described in the file /usr/array/gifts/RendAsunder/Readme. Read this file before executing the demonstration. |
Install the component RendAsunder.sw.base from the array CD on any node of the array. Execute /usr/sbin/RendAsunder on that same node to begin execution of RendAsunder on the entire array.
RendAsunder keeps a configuration file of its own to describe the array. When you launch /usr/sbin/RendAsunder, it uses Array Services to generate a configuration file in /usr/array/gifts/RendAsunder/config/curr.config. It then runs the executable /usr/array/gifts/RendAsunder/start to launch the demo.
Once you have started the program this way, you can later use /usr/array/gifts/RendAsunder/start directly to start without regenerating the configuration.
The graphics appear as determined by the standard X Windows environment variable DISPLAY. For best results, output should be displayed on a graphics device directly attached to a node. DISPLAY should be set to select a graphics-equipped node, for example
setenv DISPLAY bitblaster:0 |
If no node is graphics-equipped, you can display graphics on a remote host, although at some cost in performance. To do so, set DISPLAY to select the remote host.
When started, RendAsunder opens a control panel and a graphics window, as depicted in Figure A-1.
The Graphics Window contains a rendering of the data from the current viewpoint, overlaid by rectangles that delineate the areas of the screen rendered by each processor. Different colored lines delineate areas that are rendered by different nodes. During execution, RendAsunder dynamically adjusts the regions of the image rendered by each node and processor to balance the workload.
Figure A-2 shows the Controls menu.
The Controls Window contains the following menus:
File menu
Home View option, which takes you back to the default view, in case you get lost
five Save and Restore options, which allow you to save and restore five viewpoints
The viewpoints are not saved between sessions. If you restore a particular viewpoint before you have saved to it, RendAsunder displays one of the five default viewpoints.
You may find it interesting to look at these five default positions before you save over them. If you have saved over them and want to reset them to the defaults, use the Reset Default Stored Positions option.
Record Movie and Play Movie options, which allow you to record a sequence of viewpoints, and play them back.
Quit option, the best way to exit cleanly
Class menu, which lets you select classes of data to be displayed by changing the color map in use. Select All Classes to render all data visibly. The other Class settings omit one or more types of data to produce different displays.
Help menu, which displays information about the program (later versions will contain more extensive help)
The Controls Window contains a number of sliders that let you navigate around the data. Click on a slider to highlight it, and then use the left and right arrow keys to move the slider.
The sliders provide a convenient way to get smooth animated motion. Select one slider, scroll it with repeated key taps to the position you want, and then select another slider. For big jumps, drag the slider with the mouse.
Refresh the display by clicking on the currently highlighted slider.
The demo uses a geographical metaphor to navigate around the data. From any position you choose, you are always looking towards a center focus point in the middle of the data.
These two sliders allow you to move across lines of latitude in the north and south directions until you reach the poles, or move east and west across lines of longitude. These controls allow you to reach any point on the surface of a sphere; recall that from the surface of this sphere you are always looking in the direction of the center focus point.
The center focus point is initially in the middle of the body. Use the elevation of center point control to raise and lower the center focus point.
For example, if you raise it to about 2.8, you will be looking at the head from the surface of a sphere centered about the head. If you lower it to about -3.0, you are centered about the feet. Any movement about the sphere orbits around this center point.
Use the radius from center point control to zoom in and out from the center focus point. This control is often used in conjunction with the Near Clipping Plane control, which determines the distance of the near clipping plane from your eye.
If the value of the near clipping plane is much larger than the radius from center point, you might not be able to see anything because the entire volume will be clipped.