Thanks, Jon Rista for trying to put this back on track.
Apparently some don't get why this matters. If physics really does limit how well an APS-C sensor can perform at higher ISOs (and I have no reason to doubt you on that), then the direction Canon decides to go with the 7DII sensor will tell us much about what the company thinks about the future of high-end crop sensor DSLRs.
If Canon were to release a 16mp 7DII, they are saying something quite different than if they release a 22-24 mp 7DII.
The relevant question for Canon is most likely to be – which one will generate more demand in the marketplace?
Many would say the 16mp sensor, which should have better high ISO performance than the current 18mp sensor. In effect, Canon would be following the same path with its flagship APS-C body that both they and Nikon have followed with their flagship full-frame bodies.
But, what really would be the demand for such a body – a good, maybe even great all-purpose crop body, but still not as good as the almost identically priced 6D in terms of high ISO performance. The 6D would be a better all-purpose body; and would there be sufficient differentiation between the two in the marketplace?
Or Canon could go the other way and release a 24mp crop sensor body -- essentially conceding the high ISO niche to full frame. Would this camera find a bigger market?
While the relative advantages of a crop sensor for reach have been much debated, almost everyone concedes that in cases where the shooter is distance limited and significant cropping is required, pixel density does matter. You will always reach some point where there simply aren't enough pixels to give you a usable image.
So, why the reference to the 1D IV? Because that was the point at which Canon abandoned the sensor that many argue passionately was the ideal compromise between size and reach. We can't assess or intelligently speculate without first knowing what the constraints are.
If the ISO performance of the APS-H sensor can never be achieved with the smaller APS-C sensor, then Canon must decide which path to go down.
So, unlike the many, many threads where individuals focus solely on what they want and assign human motives to a large corporation (Canon doesn't care...Canon doesn't listen...Canon had better do this...) this is simply an effort to explore what the reasonable expectations may be, so that it gives us a better idea of what choices Canon is facing and, when they announce their decision, we have a better idea of where the market is going.
Thanks for the detailed thoughts. I understand where your coming from now, so maybe I can provide a clearer answer.
For all intents and purposes, all else being equal, the only thing that actually matters when discussing IQ is the size of the sensor. The size of the pixels does not actually matter. You could have an FF, APS-H and APS-C with the same size pixels, to eliminate pixel size as the ultimate determining factor here. Assuming equivalence (same framing, same depth of field), per-pixel noise will be the same, however the FF will always be better than the APS-H which in turn will always be better than the APS-C. ALWAYS. Note the factors here: Same framing, same depth of field.
So, why? Why is it that the IQ edge will always
be FF > APS-H > APS-C > 4/3 > Small Form Factor? If we dive into what is necessary to achieve identical framing AND identical DOF, you'll understand why. Identical Framing
To achieve identical framing, you either use a longer lens, or get closer. It's pretty much as simple as that. However in achieving identical framing, you are changing the total amount of light, for your subject, that falls on the sensor. It really doesn't matter if the sensors have the same pixel pitch, or the same total pixel count, either way, the larger sensor is going to be gathering more light in total
when you normalize framing.Identical DoF
Now comes the part that everyone usually assumes is the caveat that somehow lets a smaller sensor achieve better results than a larger sensor. To achieve the same depth of field with a larger sensor, you need to stop down the aperture. By stopping down the aperture, your negating the benefit of gathering more total light. The caveat stops there, however. At worst, you'll simply normalize the results...the FF sensor will have the same amount of noise as a smaller sensor. The Identical DOF factor is not a magic bullet that can ever allow a smaller sensor to perform better than a larger sensor.
Let's take two sensors that have the same number of pixels. The 1D X and 7D are pretty ideal examples, since they both have exactly the same pixel count. Take the same shot with both cameras, and in not a single case will the 7D image ever be better than the 1D X image. The larger pixels trounce the 7D, at all ISO settings.
Assuming your aiming for the same depth of field, then the 1D X image will at worst look the same as the 7D images, however there are actually technological improvements that make 1D X images always look better than the 7D, even when using significantly narrower apertures.
Everyone always talks about the "equivalent" cases, however in my experience, in practical scenarios equivalence is rarely ever actually desired. One of the big reasons for buying a camera with a larger sensor is to get a THINNER DOF. The moment you open the aperture up on a larger sensor, all hope for the smaller sensor is gone. Not only are you gathering more total light simply by virtue of greater sensor area, but now your allowing more light through the lens. As is also always the case these days, the pixels of FF sensors are larger than the pixels of APS-C sensors...so you have more total light with
larger pixels along with
the use of a wide aperture (at least as wide as the one you were using with the APS-C).
There is also often another benefit with larger sensors. They usually have more pixels than smaller sensors. For example, if you fill the frame with your subject with both a 5D III and a 7D, the 5D III is not only putting larger pixels on the subject, it is not only gathering more total light...it is ALSO putting more
pixels on the subject. If you then downsample your 5D III images to the same dimensions as native 7D images, the extra pixels of the 5D III provide more source data, such that when interpolated, the 7D-normal outcome is even better
, sharper, less noisy, more accurate.
This is not always going to be true anymore...especially if Canon moves to a 24mp 7D II. However that does not negate all of the other advantages of using a larger sensor. The more total light is always going to be the case...larger sensor, more total light...more total light, higher SNR, less noise.
Finally, there is the one caveat that actually does give smaller sensors the edge. Or rather, to be more accurate, the caveat that gives smaller pixels
the edge. Reach. The much-vaunted reach factor. The only case where a smaller sensor can give you a performance edge is when you are literally reach limited. You cannot use a longer lens, and you cannot physically move closer. You are either blocked by some active barrier, at the shore of a body of water, or moving closer would scare away your subject.
In this case, and pretty much only in this case, we are actually NOT talking about an equivalent set of circumstances. Equivalence requires identical framing...however the larger the sensor, the less total area of the frame your subject is going to take. We now have identical subject size. Assuming that your FF sensor has larger pixels than the APS-H, and that in turn has larger pixels than the APS-C, then the APS-C is actually going to perform better. Some would argue that the larger sensor is still performing better...after all, it still has larger pixels. If one is willing to completely ignore the level of detail being resolved, then indeed, the larger pixels will still be less noisy. But were talking about reach here...the level of detail being resolved is exactly what matters. In this case, as the facts of physics would have it, your actual subject is getting the same amount of light in both systems. It's a matter of area, and the absolute area of the sensor is the same, even though relative area (to the frame) is different. The key difference is the amount of detail...larger pixels resolve less detail, and blur the edges of your subject with nearby background (and/or foreground) detail.
In reach-limited scenarios, with identical subject size at the sensor, smaller pixels perform better.
There is a lot of value in smaller sensors with smaller pixels. For the less skilled, it means you can get some detail and ultimately end up with a good composition without having to have other skills to get close, or compose right in camera, etc. With more pixels, you can always crop to enlarge and compose better.
If you have skill, and tend to photograph things that need reach...wildlife, birds, airplanes, even sports, reach is a very valuable tool.
Even more so, it is a valuable tool to those with limited budgets. To achieve equivalence with a larger sensor, you need bigger lenses. To justify the cost of full frame, especially a high end full frame, you need to be able to produce better
images, so you need bigger and
better lenses. A crop sensor with lots of pixels means you can get an order of magnitude more bang for your buck, and often in a smaller, more portable and manageable package. More pixels doesn't really hurt you, either, as in a reach-limited scenario, your subject is still covering the same absolute area of the sensor...more pixels simply means more detail...and you can always downsample.