Then I can't decypher your
nothing they include in the file to instruct their own RAW converter how to apply NR and sharpening is applied by third party converters
"...third party converters profile files for DSLRs independently of the camera manufacturers, and Nikon has no more input than Canon does in this regard (nothing they include in the file to instruct their own RAW converter how to apply NR and sharpening is applied by third party converters)." That is self-explanatory; there are no instructions in the files directing third party converters (like ACR and Capture One) to apply NR.
Well, this is a good explanation for why the comparison was worthless.
Perhaps for the purposes of evaluating noise, what they do demonstrate though is that detail is not being sacrificed to attain lower noise -- that was really the point I was making with that comparison, and I believe it does show that. Ironically, many people looked at these very shots at DPR and based on different conversions and out of the camera JPEGs drew conclusions about the relative merits of the cameras. My own impression is close to yours in that regard -- mostly worthless. That said, nearly everything I have come across online comparing various cameras IQ in general and noise in particular has been similarly worthless; the best I have seen to date in that regard is the DIWA tests I offered earlier in this thread.
2. I'm afraid taking samples from the wall was a grave error. The wall does not look smooth. The face of the sculpture appears more suitable (and darker anyway, that's the point). There the D3 image looks much better than the 5D, which in turn looks much better than the D300.
Have you actually measured the differences? I just now took a series of measurements from the neck area of the sculpture and looking at the red, green, luminosity, and colors std. deviation there is negligible difference between the 5D and D300; the blue channel does show a noticeable difference in that area but in the area above the sculpture's right eye they are almost identical even in the blue channel.
Regarding my choice to measure the wall, one problem with the face of the sculpture is that it is hard to distinguish noise from tonal ramps that exist there. The wall on the other hand is relatively free of detail and has no significant tonal ramps where I measured it. Here is the D3 ISO 200 file with no changes and converted using NX (NR was turned off in the camera, and there appears to be no sharpening either): http://photos.imageevent.com/tonybeach/myp...//D3-ISO200.jpg
Any detail you see on the wall from the ISO 1600 files is false, and is in fact noise.
My reason for offering my time and effort to this thread is that the suggestion has been made that the lower noise in Nikon's D300 is accomplished by some sort of trickery (and the implication is that Canon is doing it without trickery); and that is unsubstantiated. Oh well, people will believe whatever makes them most comfortable. All I really know is that my D300 delivers very good high ISO noise performance and overall IQ, and I can adjust the NR in all the converters to taste (even in NX, although Capture One actually does a better job with its NR), detail is reasonably good up to ISO 1600 but starts to suffer after that.