In my GFX tests, I’ve found that the camera is about as sharp as you’d expect when tested with adapted exotic Zeiss glass, but it is surprisingly good with the presumably lower-quality Fuji 63 and 120 macro.
Why is that?
For the answer, I turned to a group of experts: the brilliant, unconventional folks at Lirpa Labs. Some of you may be familiar with the Labs. They are a division of the larger Sloof Lirpa optical conglomerate, and the yin and yang of Lirpadom have produced many breakthroughs over the years.
The Lirpalites are usually very busy this time of year, but they did take my question seriously. They disassembled my GFX so that they could reverse-engineer the tricks that the Fuji folk had put in there. They even reassembled and returned it to me, together with a handful of screws that were left over after the reassembly. Not satisfied with what they were able to find via that process, they took it upon themselves to befriend Fuji engineers at a technical conference, and take them out for drinks. Quite a few drinks, if the expense report that they turned in to me was any indication. By this social engineering, they were able to discover even more of what makes the GFX and its lenses a match made in heaven.
There are two pieces to the secret, and they work together.
The first is that the back element of the GFX lenses has a coating applied to it. Fuji has developed a special mix of chemicals that slowly breaks down, emitting nano-particles that enter the air space behind the lens and interact with the sensor. The Lirpa folks were able to learn the secret name for those particles; they’re referred to by those in the know at Fuji as floobydust. The Sony sensor in the GFX, like all such CMOS sensors, has back-biased photodiodes which trap electrons knocked loose by photons impinging on the sensor. Normally those electrons spread out more or less evenly over the capacitor formed by the back biased diode. The floobydust pins down the electrons so they can’t move around. That means that there are more electrons in the places in each pixel where more light fell, and fewer electrons where there was less light.
This means that the sensor can resolve details that are smaller than the pixel pitch. Pretty amazing, huh?
Now you know why the micro lenses in the GFX are smaller than usual. They have to be that way so that the floobydust can get around them. Understandably, Fuji doesn’t mention this in any of their product literature.
But floobydust isn’t enough. In order for the GFX to realize its full potential there had to be another breakthrough. In a normal CMOS sensor, there is a MOS transistor arranged in a circuit configuration called a source follower. The source follower buffers the voltage that is an indication of the number of electrons stored in the photodiode. A normal source follower would provide a voltage proportional to the total number of electrons stored in the photodiode. Thus all the sub-pixel detail would be lost. All the good things that the floobydust had done would be for naught.
The Fuji engineers have discovered a way to process the Sony sensors so that the source followers become unstable, and report the charge stored in only part of the sensor pixel. What part varies randomly over time. They call the modified source follower a Frammis Follower, after the legendary pioneer in the field of Marginal Engineering. Then they run the ADCs that look at the voltage fast so they can see the variation.
Pretty neat. But there’s a problem. Actually, there are two.
The first is that the sub-pixel readings come out in random order. Fuji solved that through extensive signal processing that looks at adjacent pixels and calculates what the order must have been for each sensor pixel for each exposure.
The second is more mundane. The floobydust only lasts about two years. After that the rear element of the Fuji lenses must be re-coated. Fuji solved that problem by engineering each lens to malfunction after two years. The user will send the lens in to be repaired, and will be none the wiser when they receive the repaired lens with a re-coated rear element.
So now you know the secret. Thanks to the Lirpa Labs experts for figuring it out, and thanks to Fuji for being so clever.
Andrea Blum says
Ok, I admit that for about a quarter of a second there I went for it because I was skim reading. That’ll teach me! “-)
Average DP Review User says
Another technical article way above my pay grade.
JimK says
Did you notice the date?
Michael Demeyer says
I remember first learning of the Lirpa Labs innovations in (I think) Audio magazine back in the 70’s. They are truly creative this time of year.
Nice to see they are still operating. Good to see some traditions preserved!
Michael
Doug Dolde says
I would find it unacceptable to have the lenses need repair after two years. Glad you pointed that out.
Steve Giles says
I have a GFX 50S and have recently been reading with great interest (and much gratitude!) your blog/articles on the same. This is the first time that I have been introduced to the wonderfully creative and inventive folks at the “Sloof Lirpa” labs. What a wonderful (and entertaining) resource to have available to you! I look forwards to their next innovation, presumably around the early part of Q2 next year!!
Bipin says
Brilliant article 🙂 cant wait for the yearly update