Originally posted by ManuH That may explain the strange noise pattern at ISO 3200 and above.
It might, or it might be that it would be even stranger without the NR applied. Whose to say?
Actually, I'm just taking you at your word about strange noise patterns. I don't have a K20D, but have seen some images where noise looks "normal", and others where there is a kind of cross-hatched pattern. Kind of like the infamous VPN many people saw with the K10D. *Occasionally* I'll see something like that when doing extreme push processing on the K200D, but mostly not. So I'm assuming that to some degree, this sort of thing *is* "normal".
Quote: According to DXO, the K200D also uses RAW NR at 800 and above.
OK; I can't prove or disprove that. I can only say that others who specifically looked for it didn't see it. Not that it matters to me one way or the other.
Quote: Maybe not much but enough that its score was equal to the K20D. That was strange as all other tests show the K20D performing better at high ISO.
I'm convinced that high ISO performance depends on so many variables that you can get a comparison to come out however you want depending on how you test.