> What gets recorded in the "blue" channel is a kind of weighted
> average of the shifts over all of the different "blue" wavelengths
> that were present in the illumination.
Ah, it is sort of like a quantification error. Then I just got Michel wrong.
This does sound logical to me. I can imagine colours leaning towards infrared or
ultraviolet could have a substantial influence.
> In summary, it is the detailed shape of the spectrum -- where are the
> peaks and valleys? -- that cause TCA to be different for different
> light sources.
I believe ignited-gas lamps usually have weird peaked spectrums, so I can imagine
them causing these weird aliasing type of effects.
> The phenomenon called "metamerism" is due to a similar cause. Much
> info about that can be found by googling.
Ah, I knew the effect, but not the name.
Thanks for the great explanation (as usual)!