viziblr

A few samples of how one could apply the ray-traced rendering goodness of the Persistence of Vision Raytracer (POV-Ray) to the task of displaying data. My goal was to create a "rich" feel to the charts using light and shadow, rather than explore alternatives to bar charts.

Below are several samples. All were hand-crafted .POV files except for the one with the 5 bars. This was generated by a Python script.

All POV files and the Python script are attached.

[Update 2007-05-14] Approximate Render times

with radiosity (as shown in samples above) 

• 1280x1024 = 32s
• 1024x768 = 23s
• 800x600 = 16s
• 320x200 = 4s

without radiosity (samples not shown)

• 1280x1024 = 5s
• 1024x768 = 3s
• 800x600 = 2s
• 320x200 = 1s

Thoughts about future explorations

"Data cube" concept: Use the faces of a cube to render more information.

"Data crystal" concept: Like the data cube except an external lightsource shines through the cube to produce an image on the table surface that represents a "view" or "slice" of the cube data. For example, one could use this to filter out kinds of data, or to scope the cube to a time span, or to identify only the "critical" items from the data set.