Ed,
You don't say anything about the light in your office when the shots were taken, or the exposures. It's been my experience that noise is not a simple inevitable and consistent result of a given iso, but rather, it's a result of some combination of iso and settings and available light that I've never been quite able to turn into a calculation. What I do know is, if the light is good - in other words, if I don't absolutely HAVE to use iso 1600, say - I get a better end result than if the light is bad to start with and I have no choice but to shoot at a high ISO and push the exposure up in post-processing.
Here's a quick link to a gallery of test shots I just took with my K20D:
high iso tests with K20D
My messy desk at this particular moment (6-16-2008 7:03PM) is a pretty good high iso test, I think. My desk is right next to a big west-facing window. Sun is going down here in Dallas and the light is waning, so this should would be impossible at iso 400 (say). I took three shots: at iso 3200, 6400 and 1600 respectively. The gallery shows the shots without processing, with processing, and also a screenshot.
What would be most useful would be to export these photos at the basic size needed for, say, 4"x6" prints and then compare the prints. However, I will note the following:
- iso 3200 and 6400 ARE indeed noisy. I am not saying that this is surprising. But from some of the stuff I've read about the K20D, I've gotten the impression that some people think it produces nearly noise-free images even at 3200.
- the iso 3200 and 6400 images ARE usable - if you're pressed.
- I think iso 1600 is better, even when the exposure is pushed up a bit in post-processing (in Bibble Pro).
- I do think the output from the K20D at iso 1600 is better than the output from the K10D at the same iso. Sorry I didn't have the K10D handy or I would have taken a quick shot at 1600 with it, too.
Anyway, while I think the K20D's high-iso performance is very good, I very much doubt that at, say, 3200, it can really compete with the Nikon D3. It's not even a fair fight, pitting the K20D against a full frame sensor. I'm happy with the K20D and seldom dream about the D3 since I can't afford one. And perhaps it's fair to note that the D3 might not be FIVE times better (in other words, as much better as the difference in price).
Will