I wrote about the sensor battle earlier this fall. I've gotten some new information and I've done a bit more analysis of the situation from Nikon's point of view.
Canon, as I've noted, seems to be coalescing around a single 24mp APS-C for compacts (GX), mirrorless (M), and low DSLRs (Rebels). We currently have one PowerShot (G1 X Mark III), three M models (M5, M6, M100), and five DSLRs (SL2, T6i/s, T7i, 77D, 80D) all using basically the same dual pixel APS-C sensor. We don't know how many more PowerShots are headed this way, but I'll bet several. So we could be at a dozen cameras using the same basic image sensor.
Now let's look at Nikon's lineup: Coolpix is still on sub-1" sensors, more than a generation behind where Canon is. There's really no mirrorless at present (though Nikon will sell you a two-year old J5 with a 1" sensor), and even if Nikon did produce an APS-C (DX) mirrorless, it would need a sensor that currently isn't in the mix. The D3400/D5600 share sensor tech, the D7500/D500 share a different sensor tech.
So, now you're the head of each company and you notice that demand is high for a couple of your models, not so high for a couple of your other models. How's that impact your sensor use?
For Canon, not really. They just keep making the same sensor at the fab and they just allocate the quantities differently among the product lines. Sensor costs don't really change.
For Nikon, it sort of depends upon which model went weak and which is strong.
Well, that would be D3400/D5600 going weak (DX sensor #1) and the D500 remaining strong (DX sensor #2). You can't shift the sensors you committed to for the low end products to another product, as there isn't another product to shift it to. Moreover, any DX mirrorless system Nikon might produce soon is going to require DX sensor #3, which will have PD-on-sensor and possibly more. Now, you could eventually roll that sensor to other models (e.g. D3500/D5700 and maybe even the other higher models), but right now you'd have five or six cameras using three different sensors against a competitor that could have a dozen cameras using one sensor. Who do you think has the advantage there?
So, on to the new news: Sensor shortage.
Let's just look at mirrorless for a moment, because you can see the problem very clearly there. The trailing 12-month numbers went down from 3.5m units in 2014 to 3m units in 2016, but back up to 4.1m units this year. Sensor demand just in that segment probably dropped last year by one-sixth due to the quake, but this year it's trying to increase by one third. Sensor demand is up pretty much across the board with dedicated cameras, and pretty much everyone is using some part of Sony Semiconductor at the moment.
Worse still, everyone is doing "custom jobs." Sony has their set of sensors, Olympus a different set with some different tech, Fujifilm still another set with different tech, and so on down the line. All coming out of essentially one vendor (other than Canon, which sources their own sensors other than for some of their PowerShots, and which I dealt with, above). New and different sensors always start with lower yields, and take time to work up to production levels that you can fully count on.
What I hear is that Sony wanted Nikon to use the 42mp sensor (or at least only a slight variation of it) for the D850. Nikon insisted on a new sensor and yields to date have been lower than expected.
So think about the FX lineup from Nikon: D610/D750 share a sensor, D850 has a different sensor, D5 has a different sensor, and the Df has yet another sensor. Five cameras, four sensors. And if Nikon makes a mirrorless FX system, guess what? Yep, another new sensor.
So Nikon alone is adding to the sensor shortage problem by trying to get new sensors to market.
Many of us noted something in a number of statements and presentations made in the past year by TowerJazz. TowerJazz is the company that initially started by taking over the old National Semiconductor fab (original makers of the Foveon sensor, I believe), and eventually formed a joint venture that bundled in Panasonic's sensor efforts and fabs, among others.
Several things that impact the above have been discussed openly by TowerJazz lately. The first of that I saw of was a report about a year ago that indicated that they wanted to increase their market share in CMOS image sensors for the photography market considerably. TowerJazz has been rumored to be the manufacturer of the sensor in the Leica Q, for instance, though that's never been confirmed to my knowledge. Pretty much any new client would increase their market share ;~).
Next up were announcements from TowerJazz about plans to build a new plant in China that could provide BSI for CMOS image sensors. The thing that stood out in that press release was "including large formats requiring stitching." That would be full frame, FX (also medium format and larger: TowerJazz can make a single sensor on an 8" or 12" wafer if you want).
That press release went on to say that Stacked sensors would be next in the portfolio of CMOS imaging technologies they have available to customers. State of the art CMOS (2.5 micron pixels with global shutters, 65 nm process, 100dB shutter efficiency, tighter stitching than currently available from others, better dark current numbers, and much more).
Most recently, at their quarterly financial statement release, TowerJazz was quoted in their conference call as saying they were "engaged with one of the [digital SLR market] leaders in the world in the development of their next generation sensors." Well, let's see, DSLR leaders? That would be Canon and Nikon. And Canon makes their own sensors. So it must be Nikon.
When the D850 came out there was a brief kerfuffle over whether TowerJazz was the sensor provider. It isn't. The technologies that the D850 has at the sensor aren't yet ready from TowerJazz: "provide our customers mass production [of BSI CMOS] starting in mid-2018."
And this all squares with what I'm hearing out of Tokyo. Sony fabs that produce the larger-than-smartphone sensors are basically running at full capacity, and every new sensor—yes, even for Sony Imaging—has to grapple with the increased demand for those sensors overall. Nikon has lost some key sensor staff to Sony itself recently, and it now appears that Nikon is going to try running with a different partner, at least for some of their products.
But this brings me back to where I started: the only reason to pursue the more costly "different sensor for every camera" strategy is if you can market the difference in image quality that produces. Canon, for instance, has somewhat worse dynamic range capabilities on that 24mp APS-C sensor than the Sony sensor cameras, including Nikon's. But where is Nikon calling that out? Are they expecting customers to figure that out for themselves?
No doubt we're in some very interesting times now.
