LOL! How typical. I get one reply claiming that DxO is right, and another claiming that DxO is wrong and I'm wrong and there's a huge difference between FF and APS-C.
Sorry. You're both wrong.
DXO claim no such thing.
Yes they do, and it irritates me to no end that their defenders claim the opposite of what is staring everyone right in the face. Are you Baghdad Bob? Are you going to tell me U.S. Marines never set foot in Iraq? They assign an overall score to indicate which camera has overall better IQ. And there is something fundamentally wrong with their methodology when an APS-C sensor can out score a MFDB.
They post their measurements, which make it pretty clear that the 5DIII has substantially better performance at mid to high ISO. Low ISO performance (DR in articular) is a different matter.
DxO has never been able to measure DR. Sorry. They just don't have a clue how to do it properly.
The plots of the measurements are very informative, and don't appear to indicate any flaws in the testing.
Spoken by someone who has never once tried to replicate their results.
You must be joking, right? FF makes a big difference and I don't care what numbers are presented in DxOMark tests.
I don't care what numbers are presented in DxO tests either, which should have been clear from my post. What I do care about is whether or not human beings can reliably tell unlabeled, processed prints apart. They can't at low to mid ISO when the only difference is APS-C vs. FF. And I'm talking about 24" and 30" prints. None of the huge differences you claim actually exist. If we were speaking in person instead of over the Internet I would put you to the test fully confident you would fail just like everyone else.
Out of camera with no processing? FF shows some advantage. But it's not huge, which is why it can be post processed away.
Now at high ISO...yes, FF sensors show a distinct advantage that can be seen and identified in large unlabeled prints. No question about that, and no post processing it away.