Oh look ma, the pixel peepers are back. Serious photographers can leave now.
@tnargs: not pickle peepin at all, mate. as it is clear that less megapickels are crucial for high ISO IQ. look at my flickr and you know what I mean. That's among others a certain factor which puts the 1Dx above the 5DIII in high ISO IQ.
O please start to study the subject, less pixel are not good for anything, you are spreading a MYTH
more pixels are newer wrong, it is only a good thing
and regarding 1dx and 5dmk3, its more about the gain stage
The gain used is determined by the bit depth and native full well capacity of the pixel. The 1D X, quite literally and factually, has a higher FWC of around 90,000e- BECAUSE it has large pixels. If you compare an 18mp APS-C with the 18mp FF 1D X sensor, for identically framed shots (or to use Pi's very accurate term, "equivalent" shots), the 1D X will trounce the 18mp APS-C sensor. Doesn't matter who makes it, on an equivalent basis, for identical detail, the 1D X will have several stops less noise.
Conversely, in a focal length limited scenario, the APS-C will indeed capture more detail than the 1D X. At identical subject distances, crop the 1D X and downscale the APS-C to the same image dimensions, and the APS-C will have more, sharper detail. One could also use a higher resolution FF sensor and crop to achieve reach. All of these are different usage scenarios.
So it is not a myth. It depends on what your goals are. In reach-limited scenarios, higher pixel density is definitely important. One could also use a high density FF sensor and downscale it to the same image dimensions as the 1D X, and get sharper results with less noise (although not necessarily as low as the 1D X). It all depends on your goals, though. If you need to print large, it is unlikely you are going to downscale your FF image such that it has the same noise as the 1D X...so the argument that a high resolution sensor is no different than a low resolution sensor is just as much a "myth".
True, however aliasing in all its forms is not always a severe issue. Color moire, on the other hand, and moire in general, is a pretty severe problem with no real easy post-processing solution outside of manual correction on a per-instance basis...and even then, the solutions are not great. To take one of your own examples, the photo of the girl with a Foveon...I thought that photo was crisply sharp. It was aliased in a couple places, however it where aliasing occurred, it did not look any significantly worse than the kind of aliasing you get with the D800 in similar circumstances. A photo of a bird with moire, or a photo of a person with moire in the fabric of their clothing, etc. are pretty horrible problems that cannot really be fixed.
Aliasing cannot "be fixed", moire or not (it is a theorem). For example, the Sigma photo I posted is ruined. Who likes what is a matter of personal preference. Some like aliased images, some do not. But when you don't, you cannot "fix" them.
I'm guessing you have never used LR4's moire removal brush....
what ever you use you lose color information and resolution
Sure you do. Doesn't matter if you use an OLPF or a post-process moire removal brush. That wasn't my point, though...my point was that there ARE ways of dealing with aliasing and moire in post. It is not ideal, the results are not perfect, but neither is it impossible.
Again...depends on your goals. A D800E, or preferably something similar from Canon, would always be my choice for landscapes. I wouldn't use such a camera for birds or anything else prone to moire. I would take a 1D X any day for night sky photography, high speed action where I can get close, etc. I'd probably take a 5D III for everything else. It all depends, man.
HMMM long text again from you Jrista
FWC has nothing to do with results at high iso (FWC has only a meaning at base iso and DR, large FWC can also be contra productive, but that is another discussion)
The 5dmk3 has higher QE than 1dx 49% compared to 47%, the 5dmk3 has also higher resolution
QE is what counts when we are discussing high iso results and comparing 5dmk3 and 1dx, among many other factors as raw handling, electronic, read-out noise etc etc so there are no misconceptions
In plain words, the 5dmk3 sensor are superior to 1dx sensor regarding QE and resolution but it would be tough nut to crack for Canon to produce 1DX with 21MP and the number of frames / sec as now in 1dx
Q.E. is the ratio of photon strikes to electrons released. In other words, it could be considered the rate of conversion of photons to electronic charge. A higher rate does allow a lower gain...but you really need a much higher rate for it to be more meaningful than much larger pixel area. Yes, Q.E. does matter, but the amount of charge that can be held, at ANY ISO (since ISO is really just reducing maximum charge by a factor of two each stop), is fundamentally dependent upon the total surface area of each pixel.
According to sensorgen, at ISO 6400 the 1D X has a saturation charge of 1516e-, vs. the 5D III's 1079e-. That means the 1D X, DESPITE its lower Q.E., is quite literally more sensitive to light than the 5D III. As a matter of fact, it is approximately 40% more efficient. That clearly indicates that the difference in pixel area is a more important factor than a 2% difference in Q.E. You can't forget about the CFA strength either. Both the 1D X and 5D III employ a "weaker" CFA, which also has the effect of increasing sensitivity at high ISO.
Quibbling about a 2% difference in Q.E. is largely meaningless when you actually run the numbers. When the difference in Q.E. is 10%, 20%, 35%, then we can debate the merits of a higher Q.E. in a sensor with smaller pixels.
An interesting article on gain, Q.E., pixel area and ISO can be found at Roger Clark's site here
. Roger created the concept of "Unity Gain" as a way to compare the real-world sensitivity of cameras by factoring Q.E., pixel area and CFA transmission together, as all three affect real-world sensitivity to light.