It’s a standard, but it’s a very complex one...to the point where the ISO standard for imaging sensors is almost, “do whatever works.”
This is what you see in DxOMark ratings. They figure out what the true ISO rating is for a given ISO setting. The standard gives the manufacturers a range of acceptable sensitivities, and it gives the “okay” to correct discrepancies with software.
For example, my Nikon D90’s ISO 200 rating is really ISO 145. That’s about 0.46 stops less than ISO 200. If I take an image of a gray card and view the RAW file in Raw Therapee using the Neutral profile (demosaic only) I can see that the image is underexposed. That’s exactly what DxOMark saw. However, when I view the same NEF file in Nikon ViewNX, even with my custom “RAW” Picture Control (that removes all processing,) the image is properly exposed. Therefore, as a minimum level of processing, ViewNX is boosting the exposure of all images by half a stop (at least.)
Now here’s the question...is this some sort of farce? Have I been duped by Nikon? According to people who think the manufacturers are out to get them, the answer is yes. According to the ISO standard, the answer is no.
Personally, I don’t worry about these things anymore. After wasting way too much time investigating such things, I’m convinced that the manufacturers are NOT out to get me, and are actually giving me the best they could give me at the time of product design. For its class, my D90 takes great pictures. What more could I want?
