LCC of 23HR, 35HR and 40HR on IQ250

[voidshatter]

Courtesy to Steve at Teamwork Photo for providing the digital backs, and to Paula at Linhof Studio for providing the lenses - great dealers to offer a test drive!

This test was supposed to investigate the color cast and crosstalk issues of current Rodenstock wide angle lenses on the Sony CMOS sensor (IQ250, IQ150, Credo 50 etc), and to help predict / mimic whether these lenses can hold up value if Sony makes a 645 fullframe CMOS sensor (around 54 x 40mm in size).

Test setup:

ALPA 12 SWA
23HR with center filter
35HR
40HR
IQ250
IQ260 (as a positive control)

One obvious flaw is that I forgot to test the horizontal shifts. The popular 32HR was not tested, because the center filter is too big to accommodate ordinary filter holder and we were not interested. Hopefully someone else would be able to contribute on the 32HR.

Below shows how the 23HR with center filter performs on IQ250 after LCC correction (with IQ260 as a positive control):

Below shows how the 40HR performs on IQ250 after LCC correction (with IQ260 as a positive control):

I have attached the LCC files of these 3 lenses (23HR /w CF, 35HR, 40HR) on the IQ250: https://www.dropbox.com/sh/ze3cx44vp1xjpz9/AADQ8TT5uCQGQ3kL1ECfXWZFa?dl=0

Early conclusions:

a) The "pink-banded" Rodenstock lenses (e.g. 23HR and 35HR) may hold up better against the Sony CMOS sensors.

b) The "blue-banded" Rodenstock lenses (e.g. 40HR) may have slight desaturation of color after LCC correction if shifted to the extreme - see the blue sky.

c) It is very likely that the IQ250 has the microlens offset design (same as the Sony A7R) - see the sudden occurrence of color cast in the lower part of the IQ250 LCC shots when shifted to the extreme.

 

[Paul2660]

Thanks for the work. I would have loved to see the same equipment shifted. Love the 1st photo.

Strange that the 70mm image circle lenses seem to have less issues.

Paul

 

[voidshatter]

Thanks for the work. I would have loved to see the same equipment shifted. Love the 1st photo.

Strange that the 70mm image circle lenses seem to have less issues.

Paul

Actually the 90mm image circle lens (i.e. the 40HR) had more issues because the amount of movement was more (i.e. 25mm vs 17.5mm).

This is perhaps something to keep in mind - the Sony CMOS sensor will not be so friendly to shift movement - at least for the current lens lineup. Expect to have reduced value of the current Rodenstock lenses when Sony releases a 645 fullframe CMOS sensor.

 

[torger]

The Sony MF CMOS has offset microlenses yes, which make it perform best with centered lens. Note that it actually has less color cast than the IQ260 for a centered lens.

The wide yellow blob you see in the LCC shots is about where it the crosstalk starts to become really bad. I've seen quite many IQ250 LCCs by now, and the color pattern is the same, on the longer lenses you see the same color pattern but zoomed in.

It has also asymmetric behavior so if you turn your back upside down you get a different color, and shifting to the sides also is a bit different, but on the whole it seems to be quite symmetric in terms of the gravity of crosstalk, ie you can tolerate about the same amount of shifts in all directions. I'm a bit uncertain about horizontal shifts though due to lack of test files.

If you're really picky about color fidelity and thus want near-zero crosstalk you can't shift much at all with this sensor. If you want to sanity check where you have crosstalk you can open up an LCC shot in RawDigger and move around the mouse pointer and compare G and G2 (green 1 and green 2), when they start to separate you have crosstalk (as one gets most leaks from the red and the other from the blue they separate). With the IQ260 you'll find that you have at most like 5% difference, but with the IQ250 there's about 20-30% when you reach the yellow blob.

If you're more pragmatic about color and can compromise some and we interpolate from Guy's Credo 50 tests where he got good real-world results with hard-to-detect color issues with as much as 15mm shift on the HR40, ie quite much of the yellow blob visible, it seems the 23HR should have good results up to 7.5mm and the 35HR up to 12mm. I think this is great news, that means that for some tech cam shooters these backs will be flexible enough to be useful for wide angle shooting.

Full frame with the same pixel design like this will look much much worse. The 44x33 look pretty good despite the circumstances because the smaller size makes the shifts it can do larger relative to the size. Sony and others can suppress crosstalk much better than this if they want to but this time around it was not in the design. This is only happens for tech wides, and tech wide angle shooters is likely not a focus group when designing sensors.

Samsung has shown some interesting sensor technology lately, such as the first APS-C-sized BSI sensor... when/if BSI comes to MF I will be happy, then we'll probably have crosstalk suppression in the same class as the Kodak 50 megapixel sensor which excel in this metric (due to incorporating light shields in the design).