Originally Posted by dcstep
The DxO tests are a bit limited because they don't consider real world photography uses of those sensors; however, they do give a relatively accurate presentation of efficiency. It's not a "perfect" way to select a camera body. I like looking at side-by-side images, like DPReview and some others do.
I saw some side-by-side comparisons of the 5D3 and the D800 in-camera jpegs and the Canon blew away the Nikon. That's interesting, but since I shoot RAW, I'd expect anyone using either of those cameras to shoot RAW, I don't really see the relevance. Unprocessed RAW images are also meaningless because no one uses RAW images to print. I want to see Canon processed with DPP and Nikon processed with Nik for starters and then both processed by DxO (I use DxO, you might prefer LR) and make a decision.
this is where the internet is a let down.
untill a company makes a zip file for download with raw files from each camera for the general public to play with at their own convenience, none of these online tests are really any good. Who knows what kind of pixel manipulation goes on.