On the Referred Term

As per the previous page discussing what I believe are tremendous fault lines in "radiometric" usage, I am equally rather disappointed with the notion of "referred" as it has grown in usage. I do not believe that open domain tristimulus represents anything to do with a "scene" in any way. Of the regrets I have, propagating terminology largely derived from Giorgianni and Madden's Digital Color Management: Encoding Solutions book. While I have tremendous respect for Giorgianni, I find this entire discussion rather nuanced, containing a large degree of depth, and ultimately entirely faulty.

My current belief is that it is nonsensical to refer to pseudo-observer based metrics as "scene". What matters are the amplitudes of the values, which are relative to specific RGB observer based colourimetry, not foolish notions of "radiance". A gaussian blur applied in one RGB colourimetric space will yield different results to another, even when the underlying colourimetry is "properly" transformed between them, for example.

We may do ourselves well to interrogate the surface of our assumptions here, and peel apart incorrect inferences.