We test the Nikon D80 digital SLR camera for noise during completely dark frame shots to expose the basic noise of the camera.
A dark frame image is one taken with the lenscap on. The purpose of a dark frame image is to show the noise that the sensor and camera combination will produce during an exposure of the given duration. The RAW files for all these test images is available on the DIMi camera test page.
To attempt to show you the noise in a way that works on the web, I’ve been experimenting with a number of different ways of doing this. While I am sure to collect flak from some people on the Net no matter what method I use, I believe this is a workable solution that almost everyone can reproduce for themselves to do their own checking.
Two series of dark frames are taken, at 1/250th of a second and at 30 seconds, each series covering every ISO value. These represent a typical daytime and nighttime exposure. To be able to show you some of the noise characteristics of this camera I am going to use three features of Adobe Camera RAW: the histogram, shadow clipping warning and the Blacks slider. All images are taken with noise reduction turned off on the camera.
The Blacks slider in ACR allows you to set the pixel brightness value that gets converted to a pure black. By default in the current version of ACR, for the D80, any pixel value of 5 or less gets converted to pure black. To show what is actually happening I reset this setting to zero, as well as all the other sliders to try to produce as true a version of the RAW file as possible. Shadow Clipping Warning turns any pixel that is a true black value to a blue so you can see where your shadows may have blocked up. The histogram shows us the distribution of pixel values in the image.
So by setting the Blacks slider to zero and grabbing the histogram images of each ISO image, you can see the spread of noise values.
1/250th of a second exposures
30 second exposures
The next step combines the Blacks slider with the Shadow Clipping Warning. As I said before, the Shadow Clipping Warning turns any pixel blue that is set to black. Now as you slide the Blacks slider above zero a larger range of pixel values are being turned to black, and so will show as blue of the preview screen. You can use this to examine both where the noise occurs in the image and what range of values it covers.
100ISO image with Blacks slider set to zero shows that there are no pixels at all in the image with a zero value
With the slider set to 1 we can see that most of the image is covered with pixels with a brightness value of 0 or 1 but that there is a scattering of pixels throughout the image and especially at the top of the sensor (where major heat sources must be located) where there is more noise
With the slider at 2 effectively all pixels are covered, but still not all, especially at the top of the image.
Taking the 400ISO 1/250th of a second exposure as an example, we can see that as we increase the slider from zero to 8 we gradually include all the noise in the image. This shows that there are some patterns to the noise and also that, at 400ISO, all the noise is below a brightness value of 8 (on the Blacks scale).