Originally posted by ZombieArmy I don't understand this purism in a system where all pictures are an interpretation of light as 1s and 0s that hit a sensor.
This "interpretation" is based on solid reconstruction techniques.
Yes, the Bayer CFA bandwidth characteristics do not coincide with that of a human standard observer which is why cameras have trouble getting certain hues right.
Yes, only one colour is captured per pixel position and the full RGB image must be interpolated through a process called demosaicing which can be performed in many different ways, with every approach yielding a slightly different reconstruction. The list goes on.
However, engineers know what to do in order to obtain a faithful as possible reconstruction because they know what information has been lost in which ways.
The fact that a perfect reconstruction is not possible does not imply that interpretations become arbitrary.
Of course no manufacturer is obliged to provide the most faithful reconstruction and many manufacturers deliberately do not do that in order to give you nicer "memory colours", sunsets, etc. However, again, this does not detract from the fact there is an objective notion of "image quality" and another, subjective one, that would be better referred to as "image appeal".
All I wanted to achieve is to point out the difference between these two notions.