If this is the best the D800 can do...

[Tuggem]

http://www.fotosidan.se/forum/showpost.php?p=1792662&postcount=1457

I don't understand the language, but I'm assuming D800 samples reduced to 1600px. If this is the best this camera can do at ISO 1600+, the 5D MkII has nothing to fear let alone the Mk III!

I have the 5D2 and I think this is about 1/2 to 2/3 stop better than the 5D2 can do.
Hopefully the 5D3 has improved to at least this level to. I don't really expect it to be better since we are closing up to what is physically possible to do.
(a 100% efficiency FF sensor would probably be about 1 stop better than D800 so there is not much room for improvement)

Not sure what's wrong with your 5D2 then, because mine is at least 1/2 stop better than this at 1600+. That said the D800 is 36MP, so 1/2 stop worse than 5D2 is still an incredible achievement for Nikon. Just don't expect the D800 to be a low-light superstar, which is what I *AM* expecting the 5D3 to be. That - and video - will be the 5D3's trump cards in my opinion. As a professional wedding filmmaker and photographer, I know for fact my colleagues and I are going to lap up the Mk III if it improves on the Mk II in these aspects. Anything else is gravy. Nikon just can't compete when it comes to pro video on an SLR, and the early reviews of the D800's video capabilities (and quality) seem to pan this out.

Since you are talking about the number of pixels as they should have anything to do with low light capability it is obvious that you don't have any competence whatsoever in what you are talking about.
My guess is that the statement of your 5D2 is at the same lack of knowledge level.
If the 5D3 will have better or worse low light capability than D800 we don't know. We can hope for better but there is not any room to be a lot better. 0.5 stop would be amazing.
The only thing we know is that so far the D800 after D3s and D4 has the lowest high iso noise.

You've posted this before but on a pixel by pixel basis it does seem pretty clear that more megapixels = worse ISO performance on sensors of the same size and technology.

I personally do not buy the arguement that downrezing will always give equal or better results either, I'v seen more than a few NEX vs NEX 7 comparisons recently and the NEX 7 has looked significantly worst at higher ISO's even downrezed to 16 megapixels.

It would be interesting if you could provide links.
What I had found so far, dxo, dpreview, steve huff, and luminus landscape have all failed to show that NEX-5n would have less noise than NEX-7 at high iso.

I think you are right in your comment about downsizing. The result may vary somewhat depening on choosen method of downsampling.