Optics meets sensor architecture -- a GFX mystery
1 week ago
1
I'm either closing in on a peculiar GFX mystery, or I'm barking up the wrong tree. At this point, I need the insight of people with deeper knowledge of optics and digital sensor architecture than mine.
In a nutshell, I have now encountered two, and possibly three, cases where a lens that is properly calibrated for film and other digital sensors does not properly on GFX without being "improperly" calibrated. These are all symmetrical or near-symmetrical lens designs where the rear end of the lens is very close to the sensor.
Case 1: Mamiya N 43mm f/4.5 L on GFX. This is a rangefinder lens with a rear end that is inside the GFX mount at infinity. When Mamiya technician Bill Rogers disabled the shutter so I could adapt it to my digital view camera, he checked and confirmed that the lens cells were at the correct spacing for film. However, when I put lens on my GFX-based digital view camera, image quality at the edges was unusable. On a hunch, I removed the correct shim, and replaced it with a thinner one that brought the rear cell closer to the front cell. Image quality is now excellent at f/4.5. Therefore, the cells have to be closer together on GFX cameras (GFX 50R and GFX 100S) than is "correct" for the optical design.
Case 2: Schneider-Kreuznach APO-Digitar 35mm f/5.6 XL on GFX. This lens came in a factory Copal 0 shutter. It was unusable because the edges were terrible. There were no shims under the front and rear cells, so they were as close together as they could be. I returned it because adjustment was not possible.
Case 3: Schneider-Kreuznach APO-Digitar 35mm f/5.6 L on GFX. I bought this one in a Schneider-Kreuznach electronic shutter for the Plaubel digital view camera system. To use it on my outfit, I moved the cells into a donor Compur 0 shutter. In the original shutter, there were three shims under the front cell with thickness totalling 0.21mm. It's common for #0 shutters to not be exactly 20mm between the lens mounting surfaces, so I was not surprised to see shims. With the three shims, image quality was unacceptable. However, using only the 0.09mm shim, image quality was as good as can be expected at f/5.6. This seemed normal. However, I noticed that the shutter in the Compur 0 no longer closed because it hits the front cell. That means the front cell is much closer to the shutter than it is supposed to be. Once again, the correct spacing for the cells is wrong for my GFX 50R and 100S.
Why is this an optics meets sensor architecture question?
People I know who are using the Schneider-Kreuznach APO-Digitar 35mm f/5.6 XL on Hasselblad CFV 100C and Phase One IQ4 150 cameras are using unmodified lenses, with the "correct" spacing. They don't have this problem.
This suggests to me that the way Fuji implemented the sensor in the 100S (and it seems the 50R, which has a different architecture than the 100S) is the source of the problem. Could it be the thickness of the cover glass in GFX cameras?
What else could explain the fact that the lens cells in these two lenses need to be closer together on GFX than they do on film or other digital sensors?
