Judging from my In Box, some of your are as confused as the camera makers. Capture size is a very tricky aspect of photography, as it has both cost and performance implications. Both those things are moving targets, too, which means that you can bet wrong if you're not careful. Additional factors come into play, too, which I'll get to in the detail.
Let's take a look at where I think the various sensors are and are headed:
- Medium Format — This is a speciality market with very little volume. Some think that volume is as little as 10,000 units a year, but whatever the actual number is, it is low; very low. This immediately puts into play a volume penalty. Medium format sensors are really expensive to produce due to the area of a silicon wafer they consume and the stitching necessary to create them, but when you're only making a small number of them, it just increases those costs. It's difficult to see how medium format ever breaks out of it's problem: it will remain a low volume, high cost solution for as long as there's a viable audience. We're already down to only a small number of producers, too. Medium format is already more a labor of love than of commerce. It will only become more so.
- Full frame (FX) — The usual response I get is that full frame sensors will come down in price, thus full frame is the future. To some degree, that's true: full frame is the primary large sensor imaging capability both now and in the future. But like medium format, full frame sensors consume a lot of that costly silicon wafer and they are usually made via stitching, as well, as the optical gates on the steppers making them are usually smaller than 24x36mm. If you can drive the costs of FX down, you can drive the costs of the smaller sensors down proportionally, too. In an FX body, the imaging sensor is by far and away the most costly component, so even minimizing other body costs doesn't necessarily give you a lower cost to produce (at least as long as we make DSLRs the way we do, with thousands of parts on traditional assembly lines). What FX has going for it is lenses. Not just the number of available lenses both new and legacy, but we have lenses that provide things like subject isolation that we don't really have in the small formats. One reason why FX stays around for a long time is that it's been around for a long time, ironically. There's little you'd want to do that you can't do with the Canon and Nikon full frame cameras. From that standpoint alone they're safe (at least as long as cameras are being sold). But there are trends that are fighting against them: cost, size, weight, technical progress. Is FX the sweet spot? No. The cost is too high, too many aging SLR/DSLR users want smaller/lighter rigs, and technical progress in smaller sensors is pushing the level of what those can do into territory beyond what the majority of customers want or need.
- APS (DX) — The 1.5x to 1.6x crop basically gives up a bit more than a stop in DOF isolation and in noise handling to the full frame cameras. But it does so with a huge drop in cost, at least 75% and probably more. Technically, you should be able to produce near equivalent crop sensor product to full frame at far lower costs. The question of whether this is the sweet spot is determined by how "near" the equivalence is and how "far lower" the costs are. You should also be able to drop size and weight proportionally with the APS/DX sensor cameras. As most of you are aware, I've been decrying the camera maker's poor job at filling out their APS/DX lines. Canon and Nikon, in particular, seem to be much more interested in getting you to opt for the more expensive, larger, heavier full frame systems. There's probably a spreadsheet somewhere in both companies that shows that the marginal profit gain is slightly higher by doing that, even if volume is lower overall. In fact, overall volume is one of those additional factors I mentioned: every household is not going to have two or three DSLRs, like they do TVs or cell phones. About two-thirds of households will have one DSLR. Thus, I suspect that Canon and Nikon are afraid to fill those households with only lower cost APS/DX cameras. My point, however, is that you can push APS/DX pretty much right up to the boundary of marginal returns (24mp, basically) and deliver very high quality product and images to everyone at a low cost and high profit. Would some eventually move to FX? Yes. Not everyone lives in the sweet spot forever. If you really want to excel and stand out, you need to move out of the crowd and find small incremental things that can put you ahead of them. So even if APS/DX had been fully pursued by the camera companies, I'd bet they'd still be selling full frame systems. Maybe not quite as many of them, but I can make the case that someone who maxes out a 24mp DX DSLR and wants more will buy a 36mp FX DSLR as a next step. Note the price range in APS/DX: US$600 to US$1500. That by itself is a sweet spot. Go below that and you're in the mass market and have to efficiently sell quite a few more product to bring the same ROI. Go above that and the sales volume drops dramatically.
- m4/3 — I mentioned price and elasticity of demand just a moment ago in the APS/DX commentary. It definitely is an issue as we get to the 2x sensor size. Notice what the OM-D E-M5 and GH3 are selling for? Yep, as much as a D7000. So what are you getting for the same price? Smaller physical size, slightly less performance (we're now two stops away from full frame). Wait. I'd pay the same price for those two things? Well, some people will. I've been clear that I stopped carrying my D7000 on long hikes because my OM-D E-M5 takes a great deal of bulk and weight out of my pack. Now that I'm in my sixties, I don't really have the stamina or desire to hike 12 miles at a time with a 40 pound pack. So I'll take the slight performance tradeoff for the size tradeoff at the same price. But is that a sweet spot? No, it isn't. In theory, the reduced sensor size again provides a strong cost reduction to the camera maker. But we're not seeing it from Olympus and Panasonic in these bodies, probably because they don't have the volume of sales necessary to take out the R&D costs quickly. Sometimes the sweet spot can only be achieved by an out and out gamble (e.g. price far lower to get the volume needed). Neither Olympus nor Panasonic are exactly in a financial position to take huge gambles. But the fact that we see the lower end previous generation bodies on closeout for so little money (GF5 with X lens is currently US$400) shows you that the sensor costs are enough lower that the price can go lower than APS/DX product. The other aspect of sweet spot, though, is performance. While the sensor is certainly there, the real question is whether the lenses are. We hit diffraction very quickly in m4/3, and we're two stops removed from full frame in DOF isolation. Is that the sweet spot, two stops away? I'd argue that m4/3 is the bottom of the sweet spot and APS/DX is the top. Because Canon and Nikon are doing so poorly protecting the APS/DX DSLR line, they're leaving the m4/3 crowd a chance to step in and grab volume. Let that run too long and you'll have a reversal of fortunes in the heart of the market.
- CX (1") — Moving another stop away full frame we have Nikon's mirrorless entry. Again, every time we reduce the size of the sensor we have some immediate gains: cost, size, and weight. If this is where the sweet spot was, Nikon screwed up big time. Note that at the US$400 closeout costs on the Nikon 1, I believe Nikon is still making a substantive profit. The Nikon 1 models have less than 300 parts in them (compared to over 2000 for most DSLRs), they have a very low cost sensor in them, they're made very efficiently in a state-of-the-art plant (except for the hand painting of the logo—what's with that?). Nikon even managed to get those small sensors to perform above expectations for their size (e.g. they were absolutely state of the art pixel-wise). Then they priced the whole thing higher than two of their APS/DX DSLRs. Remember, we're giving up a couple of stops of performance and DOF isolation from APS/DX. In fact, it's tough to get useful DOF isolation with the CX sensors except with fast telephotos. We didn't get the full benefit of the smaller sensor size (the cameras and lenses are bigger than they have to be for some reason, probably because any smaller is tougher to hold, which is another factor in finding the sweet spot).
So let me take a different approach to where the sweet spot might be:
- Final product cost to customer between US$500 and US$1000.
- High ISO performance capability equivalent to at least current 16mp APS/DX.
- Pixel count high as possible without putting diffraction impact in at all apertures.
- Clear step up from any and all compacts.
- Clear upgrade path to full frame should user eventually desire and be willing to pay for those benefits.
CX (1") has issues with #2, #3, and #5. While the Nikon 1 fits into #1, it's at the wrong end given the rest of its traits.
m4/3 has issues with #3 if we push pixel counts up much further, and they have problems with #5. Currently, the m4/3 vendors are pushing the boundaries on #1, too.
APS/DX doesn't have any issues. Sweet spot found.
That's why I've been saying for some time that Canon and Nikon are absolutely crazy to be lagging so much with filling out their APS/DX offerings. Yes, the D400 and 7D Mark II fall above the #1 price range, but that's actually part of solving #5!
