Petrludvik,
I noticed that in your node setting you gave a higher IOR to the red component, but it should be the opposite, the blue component having a higher IOR then red, and the green in between, because for the Snell’s Law, the IOR increase with the frequency of radiation (for normal transparent materials).
Apart from laziness, the right deltas between IORs at the R-G-B frequencies, on per material base, should be easy to know, so to get quite correct glass, diamond and so on.
In reproducing your concept by myself, I mixed the output of the above calculations with a basic glass (same parameters), and with the ‘Is Transmission Ray’ lightpath as fac; it should give a more physically correct result, bacause actually the reflected rays should not be dispersed.
paolo
