Originally posted by falconeye Please have a look at this:
(click and view in size "O")
The rightmost column, rows 2 and 3, compare K-x and D5k at ISO 12800 (D5k by boosting ISO 6400). You can see how the K-x has a "dampened" noise floor but text looks just a bit smoother too. This is the same situation but for Sony's 12MP chip. Moreover, DxO confirms that raw NR is still applied for the K-r.
Would you see a similiar difference in your K-5/D7k comparison?
The "blotted" noise effect (it figuratively appears as though someone took a damp cloth and pressed it against what was likely a higher-frequency original noise) that is present on the K-x and D5k (though the D5k shows more occasional high-luminance artifacts, probably the algorithm was trying to preserve detail and ended up preserving random high-luminance dot-noise) in these shots are NOT present (or are so subdued as to be not readily apparent) in either RAW sample I looked at from the D7K or the K-5. A few minutes ago, I compared the RAW samples on IR for the D7K, D3100, D90, and 7D (all at 6400) and found the following:
*The 7D has the worst noise levels and there appears to be some attempt to blot out the noise (though it still left tons of noise and more prevalent random noise "spiking", like a hot pixel but too frequent to be so simply explained)
*The D90, IMO, was just ahead of the 7D, with the blotchy noise similar to the D5k sample you linked.
*The D3100 was just ahead of the D90, but had an even consistent noise grain to it, showing minimal NR, but with a prevalent deep-red noise pattern that creates a color cast that is unattractive.
*The D7000 was in the lead, with a similar noise pattern to the D3100 (very consistent, with no chroma smearing visible), though amazingly less per-pixel noise than the D3100. The strong red color cast present in the D3100 is not in the D7000, despite them both being same-generation Exmor sensors. That was the most surprising result.
I agree, the K-x NR is obvious...it looks unnatural, but I'm sure it subdues noise for the most part. I simply don't see the same effect on the K-5 samples (limited few I've seen) or any of my D7000 shots. If there is secondary NR, its VERY, VERY well hidden.
In looking at the D3100 and D7000 side-by-side, it makes me wonder if something was employed in the D7000 to mitigate the strong red noise to reduce color cast, but I have no proof one way or another.
To be fair, DXO should make a similar comment regarding the D5k/D90 in terms of smoothing. It's clear if you look at the SNR curve or any images that there is some obvious NR being applied at ISO3200+.
Last edited by AngryCorgi; 11-03-2010 at 08:11 AM.