Okay, so I read Adam's post about how the K-5 at 6400 is as good as the K-7 at 800. Then I actually take a look at Falk's posted picture (thanks Falk, I appreciate it!) and I think to myself - hmm... that doesn't look right.
When I first saw the dpreview samples from the A55, I thought the noise was blotchy and pretty bad, especially in some of the color channels (which made me wonder why people were saying it was good). Now that I see the Pentax K-5 image at 6400, and while it's definitely better than the K10D or K20D (at 6400), I don't know if it's as good as ISO800. Now granted, I don't have a K-7 anymore, I returned it after testing it for about 2 weeks - I didn't like the fact that it had greater amounts of shadow noise at lower ISO compared to the K20D. I didn't play around too much with the high ISO, but I continually hear people go on and on, defending the high ISO performance of the K-7 as a major improvement over the K20D.
So for those who are interested, I did a comparison of the K20D and Falk's K-5 image. I tried to follow Falk's processing settings (opened in CS3 with default color noise reduction (25%) and sharpening (25%, radius 1, detail 25 sharpening), saved as quality 10 jpgs, and then took samples from the JPGs at a light, medium, and dark area of the frame.
I think ISO1600 looks better on the K20D than the K-5. I think ISO3200 are similar, except there's more color noise in the K20D (which probably can be easily removed by an expert in noise reduction, which I am not). I think ISO6400, an extended setting on the K20D, is pretty ugly. I included ISO800, but I apologize in advance, I didn't think to use a clean sheet of paper instead of my wall - any texture you see in the light or medium blocks are from the texture on my wall. Clearly, much lower noise than the K-5 (at least in my opinion). So if high ISO performance is so much better on the K-7, then how is ISO6400 on the K-5 as good as ISO800 on the K-7?
Two options:
1) It isn't, and Adam should update the post on the front page of this website...
2) High ISO on the K-7 is worse than the K20D
Thanks again to Falk for his posting of images and answering of questions (even if we disagree, which I'm not sure we do, I appreciate him taking the time to post images).
(p.s. In the interests of disclosure, this is my first post here, and I have a thread on dpreview expressing my disappointed opinion on the K-5, which some people took offense to. So yes, I'm not too impressed with the K-5 so far, mostly due to the marginal improvements and the substantial price hike)
(p.p.s full sized JPGs here:
Dropbox - Photos - Online backup, file sync and sharing made easy.)