Why bother changing base ISO?

Started Jul 18, 2007 | Discussions thread
Bill Janes Senior Member • Posts: 1,848
Re: You achieve better amplitude resolution

Marianne Oelund wrote:

Ideally, we would have converters which are matched to the sensor's
amplitude resolution, which is typically 30,000 to 80,000 levels (the
electron count at full well capacity). This would require 15- to
16-bit converters. If we had that, your point would be completely
valid and there would be little value in using ISO settings above
However, our converters (Nikon) are limited to 12 bits (4096 levels),
so they fall short of the resolution contained in the sensor output
by about 8:1 up to about 12:1. Thus when you are not using the full
well capacity of the sensor, it makes sense to increase the analog
gain upstream of the converter. Failing to do so simply deprives you
of the amplitude (tonal) resolution you can obtain with the higher
ISO settings. Depending on the subject nature, you may or may not
notice this loss of tonal resolution.


Your analysis is lucid, but I must take issue on one point and I will present data that expand on your analysis. The full well of a camera may be 30,000 electrons, but the camera can not resolve 30,000 discrete levels because of noise. The following table shows the noise characteristics of a Nikon D200 camera according to the data on Roger Clark's web site.

The top line of the chart represents the data for full well exposure at base ISO and the remaining lines show the values for exposure at ISO 1600. The main components are the shot noise (photon counting statistics) and read noise and these are shown along with the total noise and the signal to noise ratio.

At base ISO the camera can use the full well capacity of 33,000 electrons. However, since the standard deviation of the total noise is 181 electrons, the camera can not resolve 33,000 discrete levels and it is not necessary for the camera to count every last electron.

At ISO 1600, the full well of the sensor is not used, and the camera gain is such that 2043 electrons represents full scale on the ADC. The number of electrons for f/stops below full scale are shown along with the noise characteristics. Six stops down, we collect only 32 electrons and the standard deviation is 9.3 electrons. The camera gain is about 0.5 electrons/ADU (analog to digital units or raw pixel values), and noise rather than ADU resolution will limit effective resolution.

If we left the camera ISO setting at 100, the gain would be 7.98 electrons per ADU and this is the effective resolution; to quantify 32 electrons in the shadow area, we obviously need better resolution. However, the noise of 9.3 electrons will limit effective resolution in the shadows. As you noted, there is little benefit of using an ISO above unity gain, which is about 800 for the D200.

Noise limits the dynamic range at high ISO, and the effective photographic dynamic range depends on how much noise you are able to tolerate. According to Norman Koren's Imatest, the D200 has only 3 stops dynamic range at ISO 1600 for high quality results, but nearly 7 stops of DR for low quality results. This limitation is shown on the table for the limiting shadow value.

-- hide signature --

Bill Janes

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow