The author is correct that "image quality" (as interpreted in the article) has not changed much since 2017. That is because:
- Read noise has been fairly constant at around 1e- once you hit the high conversion gain mode
- It's kinda hard to improve on 90% quantum efficiency (peak value--there is some variation depending on frequency)
- Full well capacity has not changed for a given pixel pitch
The end result is that the dominant form of noise in many/most photographs taken with full frame sensor cameras is shot noise rather than sensor noise. Camera's can't do much about shot noise--it is inherent in the light. Thus, few improvements since 2017 in "image quality" aside from resolution.
However, the author is flat wrong about a few things, and misleading on others. First, he suggests that R&D efforts on new sensors essentially stopped in 2017 due to the adoption of cell phones as primary cameras. That's demonstrably not true. Megapixel counts may not be climbing at present, but stacked sensors have allowed dramatically better autofocus, the adoption of electronic shutters, etc.. Depending on your photographic genre, sensor features like these have a bigger impact on "image quality" than do dynamic range or megapixel counts.
Next, the author implies that cell phones are very nearly as good as larger cameras in terms of image quality. That's simply not true. Cell phones do an amazing job considering their limited size, but the photographic window in which they can perform as well as larger camera systems is quite narrow. They don't have the same dynamic range. They don't have the same signal to noise ratio. They don't have the same resolution, regardless of megapixel count. They don't have the same ability to control depth of field (and, no, computational photography and AI are not yet substitutes). They don't work with well with strobes. Are they great for capturing casual outdoor and travel shots for instagram? Absolutely. Convenient? Absolutely. Easy to use? Definitely. Can they be used even for professional work? Sure, within a certain range of requirements. Would I ever choose one for product photography? Portraiture (aside from candids and street)? Sports? Wildlife? Astrophotography? Not a chance.
Next, the author tells us that the SNR on the A7RIII is higher than newer cameras because it has larger pixels. Generally, measuring SNR at the pixel level is really silly unless you are comparing sensors with the same pixel pitch. Why? Because we don't view our images at 100%! We look at them, whether printed or on screen, at totally different magnifications, and we won't look at them at the same magnification with a 42mp camera as we would with a 102mp camera, so the per pixel SNR isn't a useful measure of, frankly, anything.
My biggest problem with the article is the implication that resolution, dynamic range, and SNR are the only components in image quality. AF accuracy matters. AF speed matters. Lens quality matters. Ease of use matters. Focus tracking matters. Frame rate matters. Lots of other factors matter, and there continue to be advances from one generation to the next in digital cameras.
Keep in mind, in the days of film one could claim there were NO improvements in image quality due to cameras--ever--if using the metrics the author seems to have adopted. After all, film cameras had nothing to do with image quality as long as they didn't have light leaks, right? So a Kodak Instamatic was every bit as good as a Nikon F5 if loaded with the same Kodacolor film, right? After all, the film was capable of the same lp/mm, had the same grain structure, the same sensitivity, the same dynamic range, right? So, really, the same image quality? Forget the advantages of larger negatives. Forget the advantages of better lenses. Forget the advantages of higher frame rates. Forget the benefits of a wider range of focal lengths. It's really the same "sensor" in an Instamatic as an F5, so no change in image quality. That seems to be what the author is implying.
