I've taken a quick look on the tech HR40 raws. Thanks for sharing.
I think it looks quite good. There is crosstalk going on, as one can measure in green channel separation in the LCC shots (if there is no crosstalk green1 and green2 are the same). This separation can also be seen as demosaicing mazing when using more sensitive demosaicers than the one in C1. Haven't yet checked in C1 but I know that mosaicer is robust so I would be suprised if I find any mazing artifacts at this level.
One can see the asymmetric behavior of the sensor, the left shift LCC has a different color scale than the corresponding right shift. If symmetric behavior is desired one can turn the back upside down on the left shift for example. It might be the case that the right shift has a bit less issues, ie a tiny little better result could have been had if turning the back upside down on the left shift.
So how much is color affected? A color checker test would be needed to make really sure, but from the paint on the buildings in the examples and grass etc I'd say that there is desaturation but a very low amount. It would probably not be acceptable for reproduction photography, but for any normal landscape or architectural photography I think most users would be satisfied. Of the sky I'm not sure if it's a shift or if it's natural. I suspect that the fitness center magnenta left could have been different (ie bluer) if the back was turned upside down when shooting that segment.
Can the sensor be shifted more than 15mm? From Doug's library test we know that at 30mm there is total breakdown, but I have not seen any test files examining the range between 15 and 30mm. My guess is that 15mm is as far as you'd want to go.
How will the 32HR perform? Need to test to make sure, probably a little bit less good but not by much. Speculating now - perhaps 12mm shift would yield a similar result as 15mm on the HR40.
Do we have the full picture? No, to really document color stability of this system with asymmetrical behavior you'd need to make a lab test. Personally I'd want to do that before investing in this type of system so I get a full understanding of its color performance as I'm quite picky about color and lack of other crosstalk-related artifacts, but I understand that not everyone is that.
Would I recommend using the Credo 50 with the HR40? No, not really, but at the same time I'd say that many will be satisfied so I think some people will go this way and I think that is an okay decision, real-world post-processed result will be fine almost always. If buying into it with open eyes I cannot say it would be a bad decision, color and tonality is just one of many measures.
Note that some of my skepticism is that I'm an engineer, and as such I just don't think it's sane design for a system to push the sensor into operation in an area which it's not designed for and patching up with a robust demosaicer, it's also obvious that the great technology advances in both fine tonality and DR is compromised in the shifted areas, moreso than we're used to. I think having to use LCC is bad enough, and at some point I think is pushed too far than is sane, and this sensor lens combination is such a case. I can't however deny that thanks to the huge DR of this sensor and the stable C1 demosaicer the real world results you can have look pretty good.
I think it looks quite good. There is crosstalk going on, as one can measure in green channel separation in the LCC shots (if there is no crosstalk green1 and green2 are the same). This separation can also be seen as demosaicing mazing when using more sensitive demosaicers than the one in C1. Haven't yet checked in C1 but I know that mosaicer is robust so I would be suprised if I find any mazing artifacts at this level.
One can see the asymmetric behavior of the sensor, the left shift LCC has a different color scale than the corresponding right shift. If symmetric behavior is desired one can turn the back upside down on the left shift for example. It might be the case that the right shift has a bit less issues, ie a tiny little better result could have been had if turning the back upside down on the left shift.
So how much is color affected? A color checker test would be needed to make really sure, but from the paint on the buildings in the examples and grass etc I'd say that there is desaturation but a very low amount. It would probably not be acceptable for reproduction photography, but for any normal landscape or architectural photography I think most users would be satisfied. Of the sky I'm not sure if it's a shift or if it's natural. I suspect that the fitness center magnenta left could have been different (ie bluer) if the back was turned upside down when shooting that segment.
Can the sensor be shifted more than 15mm? From Doug's library test we know that at 30mm there is total breakdown, but I have not seen any test files examining the range between 15 and 30mm. My guess is that 15mm is as far as you'd want to go.
How will the 32HR perform? Need to test to make sure, probably a little bit less good but not by much. Speculating now - perhaps 12mm shift would yield a similar result as 15mm on the HR40.
Do we have the full picture? No, to really document color stability of this system with asymmetrical behavior you'd need to make a lab test. Personally I'd want to do that before investing in this type of system so I get a full understanding of its color performance as I'm quite picky about color and lack of other crosstalk-related artifacts, but I understand that not everyone is that.
Would I recommend using the Credo 50 with the HR40? No, not really, but at the same time I'd say that many will be satisfied so I think some people will go this way and I think that is an okay decision, real-world post-processed result will be fine almost always. If buying into it with open eyes I cannot say it would be a bad decision, color and tonality is just one of many measures.
Note that some of my skepticism is that I'm an engineer, and as such I just don't think it's sane design for a system to push the sensor into operation in an area which it's not designed for and patching up with a robust demosaicer, it's also obvious that the great technology advances in both fine tonality and DR is compromised in the shifted areas, moreso than we're used to. I think having to use LCC is bad enough, and at some point I think is pushed too far than is sane, and this sensor lens combination is such a case. I can't however deny that thanks to the huge DR of this sensor and the stable C1 demosaicer the real world results you can have look pretty good.