> We're getting into the realms of psychophysics and perceived
> here (back to retinex algorithms, again). Suffice to say that
> you have to
> be extremely careful when comparing brightnesses over a
> changing background
> - the eye uses relative mechanisms for doing this, not
> absolute ones, and
> so what we see can be quite wrong, no matter how careful you are.
Somewhat OT, but there is a superb example of this effect here:
Even when you've proved to yourself that the squares are the same shade, it's impossible to see them as such when they're in context. It's rather disturbing!