The Usual Suspects

Microsoft tells me that I've received several hundred thousand emails from you readers in recent times. While I haven't statistically analyzed them all, I can tell you that a large percentage of them boil down into questions or comments about only a few topics. Recent new cameras (Z9) and usual end-of-year reports have re-surfaced two Usual Suspects in my In Box: noise and noise about market share.

Let's start with noise (Usual Suspect #1).

The operative question I keep getting is this: how good is the Z9 versus the [fill-in-camera-name-here]? Some ask about dynamic range. Some ask about noise. Some ask about image quality. But all are essentially asking the same thing while not understanding a basic fundamental of where we are in digital image sensors at the moment: any state-of-the-art BSI image sensor of the same size is going to produce about the same result, with only some small hiccups that can mostly be ignored.

Let me put it another way: modern cameras are recording the randomness of photons quite well, and randomness of photons is by far the primary source of noise in an image when you look closely. 

The invention of 100% pixel view has been one of the worst inventions of the digital age so far, because people believe that what they see at 100% view is "the truth." The actual truth in photography revolves around your output, not your input. If you're printing at 300 dpi at 19" on the long axis, pixel view tells you nearly nothing. 

The truth is this: collect light over the same area (sensor size) and output the result to the same area (print size) and the differences between two cameras (almost completely) disappear. A 45mp Z7 II will look identically to a 24mp Z6 II at a 19" print size. And yes, a 45mp Z9 will look identically to a 20mp D6 at about that size, too. 

Now, if we get out our microscopes and detailed measurement tools, yes, we can find small differences between those cameras. The Z9 has fewer hot pixels and somewhat less pattern noise than the Z7 II, but most of you didn't notice that your Z7 II had either. 

This morning, someone wrote to me that they didn't buy a D850 because it was far worse than their D500. Really? The D500 is 20mp and 10 fps, the D850 at DX crop is 19mp+ and 9 fps (with grip). One reason why the D500 stopped selling so well was that the D850 came out and effectively gave you both a better D810 and a near equal D500 in one camera. 

What's limiting image sensors today is essentially quantum efficiency. Sticking color filters on top of the photo diode steals light. We've essentially hit a ceiling with how much light we can get into the photo diode and convert to charge, and the silicon specialists have pretty much got how well they do that down to a science. 

Sure, there are some tweaks that are used to maximize something in the eventual digital numbers that are produced. The whole dual gain thing is just one of those. And the choices for those tweaks introduces some small differences when we get out our sophisticated measuring equipment. But for any given image sensor size (APS-C, full frame, medium format, etc.) you'll get remarkably similar results in your normalized output size. 

So, to all that keep asking, here's my initial take on the Z9:

  • Really good at long exposures due to lower hot pixel production and a few other silicon-based improvements.
  • Nearly as good as a D850 at base ISO; good enough so as to ignore.
  • Some small weakness compared to the D850 at ISO 100 to 400 due to the way the low ISO gain was set.
  • Nearly identical from ISO 500 and up, with the Z9 having less tendency towards magenta at the insane ISO values.

But in terms of 19" prints? I'd defy you to see any of the above other than perhaps hot pixels on 15 minute exposures.

Of course, the comparison that some of you keep trying to make is with the D6. Again, if we're looking at 19" prints from the full frame of each, I'd defy you to see a difference. If anything, the Z9 might produce a slightly higher perceptual acuity. But otherwise? Not enough difference to worry about. 

Of course, you start pixel peeping, and then you see a difference. Of course you do, because a D6 pixel is bigger than a Z9 pixel, and the randomness of photons you'll see is dependent upon capture area when viewed that way. 

Aside: get ready for some more arguments involving photons soon. Eric Fossum's QIS (quanta image sensor) idea is now in prototype form and producing `16mp images. That sensor accurately records the randomness of photons (when each arrived, and where from). Initial results are impressive. Of course, randomness is still randomness, so you still have the issue of what most of you call noise. Which is where computational photography starts to play a role.

Bottom line for Usual Suspect #1: sensors got excellent and thus equivalence is now a real thing. At least for traditional processing viewed at conventional sizes.

Suspect #2 gets triggered early every year by BCN's yearly report on (some) sales in Japan. This year's results:

  • Compacts: #1 Canon, #2 Sony, #3 Fujifilm
  • DSLR: #1 Canon, #2 Nikon, #3 Pentax
  • Mirrorless: #1 Sony, #2 Canon, #3 OM Digital Solutions
  • Action: #1 GoPro, #2 Sony, #3 DJI
  • Lenses: #1 Canon, #2 Sony, #3 Sigma

What happens after BCN reports their results is that Web sites around the world overly interpret those results and come to generally erroneous conclusions. That has to do with: (1) breakdown of who BCN collects data from; (2) Japanese market only; (3) unit sales are reported and that favors low cost product. There's some more nuanced issues, as well, but projecting worldwide success based upon #1, #2, and #3 is guaranteed to lead you in wrong directions.

Nikon, for instance, has for some reason decided to put minimal effort into their home market. I'm not sure why that is, but you can see it even in the consumer DSLR sales in the Japanese market: Nikon backed off. A lot of that has to do with #3. The Japanese market is notoriously price sensitive, and BCN exaggerates that by collecting much of their data from mass retailers (Japanese equivalent to Big Box stores). Nikon has backed away from Japan so much that some recently introduced products are still not available in Japanese stores: Nikon prioritized first shipments to North America and Europe. Moreover, when Nikon management green-lit the Z System, they also pivoted away from consumer cameras, preferring to sell fewer, higher-priced cameras than running the Thailand factory to capacity with consumer products.

Meanwhile, if you examine what put OM Digital Solutions into the #3 mirrorless spot, it's tons of small consumer models (E-P7 and older E-M10 models). Ditto Canon and Sony in the top two spots: it's mostly the APS-C cameras that produced the volume that BCN counted. 

Thing is, the CIPA "value" numbers tell the real story these days: higher-priced products are what the entire Japanese camera industry is mostly pursuing now, and what is keeping the overall gross receipts from collapsing. You see it in the mirrorless body and full frame lens categories, where value is growing at high rates. 

Another problem with BCN's numbers is that they simply don't tell you anything about volume year-to-year. So, did #1 Canon sell more compact cameras in 2021 than 2019? (I'm skipping 2020 because it was a pandemic-impacted year all around and has highly distorted numbers.) I'm pretty sure the answer is no. I'd bet that all the categories are down in volume, and have been for a decade. So wouldn't a better tracking method from BCN be year-to-year volume tracking, not intrayear? Oh, right, that doesn't create winners, does it? ;~)

Ricoh Japan's recent statements about how they were going to change how they sold cameras in their home market got a lot of attention, too. Many we-don't-actually-do-news-but-scrape-it photography sites seemed to interpret the initial reports as meaning Ricoh/Pentax was making a step towards getting out of cameras. That the business was closing down. 

Nope, just getting smaller in overall volume, and adjusting how they do business. Pretty much the same thing every Japanese camera company is doing right now. At least Ricoh and Nikon have been fairly upfront about it. Essentially, both have said "we no longer mass marketers, we're specialty marketers." 

Bottom line for Usual Suspect #2: Every Japanese camera company has a downsizing plan and is executing it. Technically, only Canon and Sony really seem to care about market share any more, and I'm not sure how much they actually care; financial results will be more important to them. So higher profitability on lower volume is what everyone is seeking.

 Looking for gear-specific information? Check out our other Web sites:
DSLRS: | mirrorless: | Z System: | film SLR: all text and original images © 2024 Thom Hogan
portions Copyright 1999-2023 Thom Hogan
All Rights Reserved — the contents of this site, including but not limited to its text, illustrations, and concepts,
may not be utilized, directly or indirectly, to inform, train, or improve any artificial intelligence program or system.