Hoping to clarify things a bit about what I meant when I stated "5 - 10 pixels per foot" for the contest...and hoping that it doesn't confuse things further.
Our (and I imagine most other) VTs will display images at a 1 image pixel = 1 screen pixel scale when the image first comes up. If course you can zoom in and out later, but that's the basic display mode. Therefore, if you have an image with 1280 X 1024 resolution it will exactly fill a 1280 X 1024 pixel screen. If you have a 1600 X 1200 screen, it will fill only a fraction of the screen etc. Dpi is meaningless as a standard measurement because different monitors with the same resolution can have widely varying screen sizes and therefore widely varying pixels per inch.
Suppose you want to represent a 20' wide street on the image. If you choose a 5 pixel = 1 foot scale, then the street would be represented as being 5 X 20 = 100 pixels wide. If you make the entire image at 1280 X 1024 pixel size, then it will cover an area equal to 1280/5
by 1024/5 = 256' X 204.8' on the imaginary area you are mapping.
The scale range was chosen because it allows tokens representing miniatures to be a reasonable size when displayed to scale.