D7 vs. D7i Noise tests:
This page will compare the noise levels of the D7 against the improved D7i. A recent review of the D7i on the excellent dpreview.com site claims that the D7i is noisier than the D7. I have tried to shoot some very controlled shots to test this, since dpreview did not actually have a D7 to compare side by side with the D7i.
(Version #3, Last revised on June 25, 2002)
The test shots:
Here are two identically exposed shots containing Kodak gray cards taken at ISO 100. These shots were taken in manual mode at 1/15 second at F4.5. You can click on these links to see the original images, they are too large to display here. Take a look at these and convince yourself that these shots have very close exposures.
As you can see, the noise levels are actually very similar.
For example, compare these two areas...
You can see that the D7 is actually slightly noisier than the D7i, not more noisy.
Here is a lighter area...
I think that the D7i is actually less noisy in this area also.
Now compare a very dark area...
Well, OK that is pretty dark, but I think that you will agree that there isn't much if any difference (slight mottling is due to re-compression of the JPEGs that I used for the site)
Now for some sections of a different shot with color...
Maybe, maybe there is a little more noise in the D7i shot in the darker areas.
Now for a higher ISO shot, these are crops of an ISO 400 shot
Now this is interesting, the D7i shot does look maybe a little noisier, but the D7i shot also has much better contrast and color. All of the D7 high ISO shots have a bit of a milky look, and they seem to have addressed that in the D7i. I think that I prefer the D7i approach, even if it does result in the noise being a little more visible.
Gray patch tests:
I took identical exposures (not shown) of Kodak white and gray cards using manual white balance with both cameras at ISO 100. Then I measured the standard deviation of the brightness identical patches on the cards, measuring the combined RGB images and each color channel. Here are the results.
D7 White Patch D7i White Patch
RGB 180.93 +/- 2.43 182.14 +/- 1.91
Red 181.12 +/- 2.45 182.62 +/- 2.61
Green 181.96 +/- 2.54 182.43 +/- 2.16
Blue 178.80 +/- 2.47 183.20 +/- 2.29
D7 Gray Patch D7i Gray Patch
RGB 57.28 +/- 2.88 58.81 +/- 2.84
Red 57.54 +/- 4.04 59.87 +/- 3.99
Green 59.22 +/- 2.94 60.02 +/- 3.02
Blue 50.56 +/- 3.24 53.56 +/- 3.03
As you can see, there is very little difference in the standard deviation of the noise in either the bright white patch or the darker gray patch. You can see that the darker patch is noisier, as is to be expected. In fact if there is any difference, the D7i may be just slightly less noisy, but the difference is probably not significant.
Effect of using the DiMAGE viewer utility:
I recently discovered something, using the DIVU 2 utility on D7i shots and only doing color space conversion on them, can apparently reduce the noise level on D7i shots. Here are two example areas...
D7i straight from camera D7i converted to sRGB
Note the shadow area under the arm, it is actually less noisy in the sRGB converted shot.
Here is another example... (the following shots were enlarged)
D7i original shot D7i DIVU processed
Again, (although it is more evident on the original shots) the DIVU processed shot looks less noisy. My theory about this is that it has to do with the color correction that DIVU applies in converting to sRGB color space; it appears to remove a little Red, and there is a lot of Red channel noise in the shadow areas. The effect is very noticable in shadow areas that contain Red, especially skin tone areas. In the end though, the DIVU processed shots are a little bit less noisy
The D7i does not appear significantly more noisy than the D7. In fact, in the well exposed areas of the shots it is actually less noisy. In the shadow areas the noise levels are comparable. At high ISOs the D7i does appear slightly noisier, but it does have much better tone and contrast (which is probably why it looks a little noisier). There is certainly not enough noise difference between these cameras to warrant the "I heard that the D7i is noisy" posts that are appearing; the D7i images are virtually identical to those of the D7.
One possiblity regarding dpreview was that since they didn't have a D7 for direct comparision that they were comparing shots that had been processed with the DIVU program to the unprocessed shots from the D7i. The DIVU program applies excessive contrast by default, and this tends to hide noise in the shadows.