While I'm glad that DPreview decided to redo their samples, I'm finding that the K5 is actually the worst between the K5 and the D7000. While the K5 seems to have a little less chroma noise, the D7000 has more detail and less luminance noise. All the Chroma noise in the 50d makes it look the worst. I shoot RAW so really JPEG comparisons are useless for me, but I'll play along because everyone seems to want to compare JPEG to JPEG.
One thing I notice in this thread is everyone seems to be using the 3 1/2 year old D3 as a benchmark, and yes the K5 is getting close to that, even though it does not have as much detail. Why not compare it to the current model the D3s to see how it compares?
It looks way cleaner than the K5 sample, and all the others too, and by a wide margin. By about 2 stops actually.
Very telling, While both the D7000 and the K5 are class leaders for APSC cameras, if people want to compare and say it's as good as a full frame they should compare it to the current model, not a 3.5 year old one.
In a normal sized print very few people would notice the difference between the K5 and the D7000, and maybe even the D3s at moderately high ISO's, but 3200 and beyond even in an 8X10 I think you would notice a big difference between the APS-C and the full frame.
Again I shoot RAW 99% of the time so when looking at the RAW files, all the APS-C cameras look about the same. The 50d looks the worst, with the D7000 and the K5 being a essentially a tie. The K5 has a little more luminance, and the D7000 a little more chroma noise. The difference is slight both ways and I think both files will clean up the same in ACR.