I have measured this with the step chart facility in Norman Koren's IMATEST program.
The technique used is to photograph a grayscale wedge such as can be obtained from Stouffer. It looks like this:-
Adjacent steps in this wedge differ by 1/3rd of an f-stop. The resulting JPEG file is then input to Imatest which sees how many steps are detected by the camera, measures the strength of the signal and the noise, and calculates the density response function of the camera, its gamma, and several different figures for dynamic range.
I have done this for the G2 using a variety of different exposures and camera to target distances. The average of these tests is given below. Four different figures are reported for the dynamic range, depending upon how much noise you are prepared to accept in the photograph. Norman Koren says that the "high quality" criterion is rather demanding, and the "low quality" criterion is an appropriate measure of "useable dynamic range" in most circumstances.
|Canon G2||Dynamic Range in f-stops|
|Gamma||Low quality||Medium quality||Medium-High quality||High quality|
So, to summarise: when taking JPEGs straight out of the camera, the useable dynamic range is about 10.5 stops and, for highest quality work, it is just short of 7 stops.
Details of the twelve tests averaged above can be seen here.
These results are slightly better than those I report for the Canon EOS 5D. The G2 cost me £618 in July 2002; the 5D body cost me £2,499 in September 2005 and two months later it was available for £1998. However, these figures were produced with different versions of Imatest; the G2 was tested at ISO 50 whereas the 5D at ISO 100 (camera default values); higher figures can be got from the 5D by using different Picture Styles and settings (some claim but I have not proved that yet) or RAW (as Mike Collins has shown); and, of course, the 5D has many other advantages. What the figures do show is that the PowerShot series is an excellent performer as regards dynamic range.
Camera was set at ISO 50, sRGB. Large fine JPG. Custom white balance. Manual exposure. Target illumination is specified to have a temperature of 5000 degrees K, but it may be closer to 5900.
The graphs below are output in PNG form by Imatest itself. The information above has been abstracted from twelve tests, of which one is reproduced below. You can see the relevant dynamic range figures at the right side of Figure 2 below. For further explanation of these figures, please see the Imatest website and look for step chart.
Peter Facey, Winchester, England