D2H noise test and conclusion (large post)

Started Mar 4, 2004 | Discussions thread
Flat view
Rory Veteran Member • Posts: 3,239
D2H noise test and conclusion (large post)

There has been so much written about D2H noise issues, and I have found it problematic, so I thought I would do my own testing. I came to the following conclusions after conducting a small battery of tests with my D2H (SN 2009427). I welcome any suggestions where I may have gone wrong, but this is what I have found.

D2H noise test conclusions (Factors in order of impact):

1. ISO has the largest impact on noise, becoming quite discernable at 800 and above.

2. Under exposure has the next largest impact, with noise becoming quite obvious with adjustments of .66EV at ISO 400 and 1.33EV at ISO 200.

3. Setting tone to auto results in underexposure of between 0.12 and 0.22 EV, which in turn results in more noise.

4. Getting the WB right appears to have the lowest impact on noise.

5. In-camera sharpening has no impact on noise if you shoot NEF and set sharpening off when processing in NC. This allows you to evaluate the sharpened out-of-camera JPGs without sacrificing quality.

D2H noise test methodology

This is the base test shot, with white balance (gray card measured in camera) corrected using gray point sampling in NC4.0. The arrow points to the 200 pixel crop that is compared in the composite further below.

I mounted my camera on a tripod and made a base exposure of the scene using incident metering from a Sekonic L-358. The subject was lighted by one ceiling incandescent light. The luminance was 3.2EV (100 ISO). The resulting histogram looked like this:

I proceeded to take a sequence of 28 photos, (shooting NEF and basic JPG) with changes to camera settings summarized in the following table. Highlighted cells indicate where various settings were adjusted.

  • When tone was set to auto EV adjustment was required in NC4.0

Each NEF was loaded into NC4.0, where the WB was corrected using Gray point sampling, the sharpening was set to none and USM set to 65,4,5. Exposure was normalized in each photo by bringing the RGB luminance on the gray card to 87. (See the NC EV Corr column in the table) Each NEF was saved as a JPG (Good Quality).

Next, the 200x200 pixel crop was copied from each JPG to a composite using PS CS, as can be seen here (the numbers correspond to the table above):

Three WB settings were set in-camera. The first was set by measuring a gray card, and resulted in the closest temperature:

The WB set to Incandescent was also very close

The WB set to auto was too warm

And for comparison again, here is the WB corrected in NC4.0


Flat view
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow