|OOOH!  Radioactive Ping-Pong balls!  Just dope each one with a
|binary-encoded brew of seven isotopes, and use a differential counter
|to tell which rad-bits are on.  128 unique codes possible.
|The balls would be cheap, but the detector could be a bit pricey.

That reminds me of another notion: since different colored LED's
output different wavelengths (rather than combinations of RGB), it
may be possible to illuminate the ball with different-colored LED's
and read back the reflectance values.  You'd have to experiment to
find what colored dyes absorbed/reflected what wavelengths, but you
could get a lot more colors than just RGB.

Of course, there'd still be plenty of other tricky issues but illumin-
ating the ball with colored LED's could give you more data points to
work with than just RGB.