We now have four partially-stacked sensor cameras in the middle of the camera market, two of them from Nikon (ZR and Z6III), one from Panasonic (S1II), and one from Sony (A7 Mark V). I believe Canon was the first to pioneer DGO in the C300 Mark III cinema camera, and the technique has now expanded into multiple models.
The Panasonic and Sony cameras use something called Dual Gain Output (DGO), but the Nikon ones don't.
So what is DGO?
Let's start with dual gain (DG), a technique built into image sensors for quite some time, and at the heart of most Nikon Z System cameras. In a DG sensor, each column of data coming to the analog-to-digital converter (ADC) is amplified at one level for lower ISO values, and at a higher level for high ISO levels. In lots of light (low ISO) the values don't need as much amplification to be optimized. In low levels of light (high ISO) the values need more amplification.
What DGO does is to combine those. It does this by using two ADCs, one adjusted for "bright" data and one adjusted for "dark" data. Basically, there's a blending step that then happens after the ADCs to create the final digital number, but that number is now the best possible data from the highlights and the best possible data from the shadows. The net effect is that dynamic range below the second gain point is increased. You can see this clearly in Bill Claff's dynamic range plotting:
Thing is, the Panasonic S1 II and the Nikon Z6III shown in the above chart use the same image sensor. Why is there that clear difference between base ISO and the second gain point at ISO 800?
My hypothesis has to do with rolling shutter. At the same capture specifics, for instance, the Panasonic has a rolling shutter of 27.5ms, while the Nikon is at 9.33ms. This almost certainly has to do with that extra blending step needing to be done on sensor. Weirdly, Panasonic seems to have optimized for dynamic range while Nikon has optimized for rolling shutter. Given the strong video emphasis by Panasonic and the strong still emphasis by Nikon, I would have guessed the opposite. (With the ZR and R3D raw, Nikon has opted for yet another approach in how they present data to the ADC and record that.)
More curious is that neither company seems to have given the user a choice. Yes, that would be yet another setting that would need technical explanation, but I believe it would have been possible (and desirable from a user standpoint).
While I'm not one of them, there's been a serious outcry for the past several years with serious users demanding more dynamic range. But there's no free lunch. If you increase one parameter of today's image sensors you get consequences in other areas. The good news is that increased bandwidth lowers the consequences (here, rolling shutter). And Moore's Law tells us that more bandwidth is still coming in the future. Indeed, basically guaranteed.
Nikon juggles two key design decisions these days that impact final image quality. The first comes with balancing decisions at the image sensor, including but not limited to the dynamic range/rolling shutter one described above. The second comes in lens design, where you can do the Chinese menu thing and pick one from column A, and one from column B, but you can't pick everything from every column.
Do I wish for better? Yes. Remember, I'm an optimalist at heart (optimal data capture, optimal data processing). Do I find it difficult to make great images from a Z6III? Nope, just the opposite. Will the eventual Z6IV be better? You bet. Will it be better enough to trade in my Z6III? My Magic 8 Ball says "Ask again later."
