In a fascinating video interview with Lau Norgaard, VP of R&D at Phase One, several contentious forum topics were simultaneously settled at the OEM level. I've marked these at their various time stamps:
Lots of good stuff there but also a fair amount of chaff and marketing hand waving.
True, Kevin Raber (the interviewer) does wave his hands a lot. ;-)
Based on my limited playing around with a similar file from both Backs I would say that an oversimplified way to understand the difference in color produced between the two CFAs is to metaphorically consider the difference between Adobe and ProPhoto RGB: Adobe RGB with its denser space is represented by the Trichromatic while ProPhoto by the Standard Back. For the reasoning behind that metaphor you can refer to the articles linked to by an earlier post.
Thanks for writing up your analysis, Jack. It was an interesting read with nice links to additional material on color. I'll respond below to the limited conclusions you reached.
Personally, I don't think your Adobe RGB vs. ProPhoto RGB metaphor fits well for describing the difference between a trichromatic and standard sensor. A better metaphor would be polarized versus un-polarized sunglasses. The polarization cuts through haze allowing the wearer to see more pure detail much like a denser trichromatic sensor reveals purer colors uncontaminated by the color haze of a weak CFA.
I've debated all of these points on DPR over the last several years citing industry references, CFA spectrograms, and image examples. You will see many of these same points raised in my posts on the problem with weakened CFA's, CCD vs. CMOS, and the generally poorer color accuracy (SMI) of modern cameras.
Except that the CFA changes implemented in the Trichromatic appear to directly contradict those conclusions: in my limited investigation, the Trichromatic's CFA may be 'denser' or 'purer' but the Standard Back's linear color is distinctly more 'accurate' out of the box. Compromises, compromises...
I would be hesitant to draw any firm conclusions given the limitations of what we had to work with. Let's face it, there was only a third party's test scene that just happened to have had a color checker dropped into it. Along with that came the imperfect "lighting, different ISO and exposure, position of cc24 (probably resulting in contamination of some of the patches by reflectance from items in the scene), absence of direct spectral measurements, etc." which you cited.
Further complicating things, you built a custom profile from this very uncontrolled test scene with only a color checker and without the sort of profile tweaks that Phase One would have surely added to their camera profile. Lau Norgaard specifically mentions in the video that Phase One consider color checkers a starting point for camera profiling, not an endpoint.
If their marketing material is to be believed some have the feeling that some of the changes they brought were actually corrections to mistakes made with color setup in the Standard Back (e.g. was its hot mirror too 'weak', as has been reported?). On other tweaks though I think they moved in the direction of optimizing color for a specific application, trying to answer questions like this: what changes would we make if this camera were always/predominantly used outdoors, indoor performance be darned? The opposite question could be asked as well and different compromises would be reached. Then there is the much larger target market that wants their camera to be setup so that it works well both indoor and outdoors...
Perhaps it coincidentally resolved a weak hot mirror issue for them. But the core benefits of pure color still remain. These presumably include clearer hue separation of browns from reds, purer blue skies, and more naturally vibrant saturated colors when they are photographed alongside weak colors.
I do see Mark Scott Abeln's point about the potential downside of a more sensitive red channel when it comes to indoor lighting, but let's see how Phase One handle this when they release some indoor photos. In general they are touting better noise performance in their trichromatic sensor which may by itself resolve Mark's concerns.
So once the compromises are understood I am not sure the average photographer would want a specialized 'sensor' set up like in the Trichromatic for their full frame camera.
IMHO it's ok to be unsure. Drawing firm conclusions from compromised test data is another matter.
fPrime
