... a waste of time! (IMO!)
I always had my suspicions about these DxOMark tests and have read various people talking them up (recently Nikon fan-boys about the D800 score) and also people saying they were rubbish. (Note this is NOT a brand war thread!)
Well today I've gained some experience to make my mind up. I normally shoot with a 7D but recently I forgot to take it to a herding event and borrowed a Nikon D5100 to take some shots. (the best camera you have is the one in your hand and at least I got some shots!!) I've just edited those photos and I'm blown away by how washed out and poor the photos were. They all required almost a 1 stop bump in exposure and a strong contrast tone curve to be applied and then further pushing with highlight and shadow sliders. I really struggled with getting a decent looking end result without breaking those pixels. I am amazed at just how poor the files were out of the camera. Examples are below, sadly I don't have a 7D photo from the same conditions to compare, but I have shot there with my 7D on different days and been happy with the results. Of course the lens will have a significant bearing on this as well, but again, my experience with my old 450D and kit lens was far better that these. (Again I'm not trying to bag the 5100, just trying to illustrate the point about the test scores).
So out of interest I jumped onto the DxO scores for the 7D and the D5100 and was amazed to see that the D5100 scores massively higher than the 7D. I can honestly say that these results are total bollocks! The results should be the exact opposite in my opinion. These results may give them a number in a lab but in a real life situation, it couldn't be further to the truth. To take it a step further I compared the D5100 to the new Canon 5D3. OMG the D5100 scores just one point lower!! Now this of course is a very unscientific test, and the glass will make a difference, but the comparison of these scores is simply staggering.
Comparison at DxOMark here.
I'm blown away by this. I was looking at these results for the 5D3 a couple of weeks ago trying to get a gauge of what sort of difference I would see from my 7D to the 5D3 I'm thinking of upgrading to. After seeing this today, I will be deleting the DxOMark from my bookmarks. It is useless. The real world is where it matters.
This is one example from the day, all the photos had very similar setting changes in Lightroom. This thread could easily double as a "why you MUST post process your photographs" thread!
Original jpg straight out of the camera cropped mildly to improve composition, shot in Aperture priority, ISO400 f5.6 1/800s, pattern metering. The histogram is bunched up in the middle and no-where near the blacks or whites.
Lightroom edited, Strong Tone curve applied, Exposure +0.9, Contrast -5, Highlights -78, Shadows -54, whites +2, blacks -71, clarity +25, vibrance +5, saturation -15, and some sharpening. I'm still not happy with the hue of the red, but I think the result is reasonable. Overall I think I've managed to push this image as far as I'd dare, I'm not real impressed with some of the banding in some areas that has resulted.