> Mostly, a colored light means a tinted glass, which has a broad spectrum (except
> for LED).

It is incorrect to model such a light as srgb <1,0,0>. It should be 
modelled by something like <1,0.2,0.1> as you suggested. That isn't a 
fake or a hack, that's the sort of srgb value you would get from a broad 
spectrum light if you measured it with a colour meter. You'd only get 
something like <1,0,0> if you measured an LED or laser.

> And even if you do have such a light, the material you are
> illuminating will certainly not have a narrow reflective spetrum.

Same as above, you shouldn't be modelling surfaces with a colour of 
<0,0,1>, that corresponds to an extremely narrow reflective spectrum 
that doesn't normally occur IRL.

Post a reply to this message