Originally posted by Ash A nice test.
I'm finding it difficult to see much in the way of noise from wither camera. I've never seen a K10D ISO 1600 image that clean before!
Either your exposure settings are spot on and the JPEG in camera settings are perfect for the scene, and perhaps the image has been cleaned up in PP - or your K10D has had a sensor transplant.
The images coming from the cameras are 6MP, so you are seeing reduced noise in both due to the downsampling (pixel averaging). I did this to produce a level playing field, to simulate what the output would look when printed a at fixed size.
I chose in-camera JPEGs to give the K-x the full benefit of it's more modern PRIME II image processor. This is meant as a results test, not a sensor test. The WB, contrast settings are slightly different, you'll have to factor that into the evaluation as best you can. It's also under fluorescent lighting, which doesn't help.
It's interesting that a number of you preferred the K10D output up to ISO numbers I personally tend to award to the K-x. Based on my own field evaluation, I consistently prefer the K10D at ISO100 and 200, but it's pretty much even at ISO400 and the K-x is better at ISO800 and above.
Even at low ISO it's mostly subjective as you'd be hard pressed to see a difference under controlled test conditions.
(Camera defaults, PEF converted in DCU4, DA35 macro. ISO200. F8. Tripod. Don't nag about focal plane, it's close enough and you can be sure to find something in the image that's in focus. Native resolution in both cases, so the K-x is at a slight disadvantage.)
Here's the K10, resize and 100% crop:
And here's the Kx, resize and 100% crop: