Discussion in 'Photographic Discussions' started by Alpha, Jun 2, 2008.
Is the true EI of a digital sensor the same as its stated ISO equivalent?
Though your meaning is clear enough, strictly speaking there isn't really such a thing as a 'true EI', because the EI is simply calculated from the light intensity at the plane of the sensor and the exposure time. It doesn't matter what the result of the exposure is. The ISO speed is equal to the EI (rounded to the nearest third-stop in the standard series) when certain conditions are met in the recorded image. For a digital camera there are two main ways of arriving at the ISO speed: saturation-based and noise-based. If you like, you could call the saturation-based speed the true ISO speed.
The saturation-based speed is calculated from the exposure that would reach the upper limit of the capacity of the sensor or circuitry (ie on the verge of being blown out), and it is generally the method that is used for the highest quality.
It is fixed so that in effect an 18% grey will give an exposure that is 12.8% of the exposure that would result in saturation, or to put it another way, a 100% diffuse-reflective surface will give an exposure that is 71% of the exposure that would result in saturation.
So, on another note. It's common in film for manufacturers to up-rate their film speed, depending on development. That is, box speed is often faster than the film actually seems to be. Could the same apply to sensors?
Lets compare to Ilford 3200 which is in fact ISO1000. They aren't lying about anything, the datasheet clearly says ISO1000. That said it also clearly needs a vastly different developing times for ISO1000 exposure and ISO3600. Regardless of if it's pushed in development the ISO rating is still right. It is still ISO3200 when pushed to 3200 it's just grainier.
The same does apply to sensors. Lets look at the D200. It has a rating from ISO100-1600. Oh wait, there's HI1.0 which gives it 3200. This is a software push, just like pushing the exposure in film it's done after the light has been recorded. Again this isn't hidden from consumers or lied about. It's called HI1.0 for a reason. (which some people would say means HI noise lol).
The rest of the numbers are ISO equivalents. There is no ISO in a photo detector. There is only internal quantum efficiency which is the sum total of every other efficiency rating of a photo detector put together. The ISO comes from running this photo detector with a certain bias, boosting it with a certain gain through a MOSFET (CMOS) or BJT (CCD) embedded under the photodetector, and finally doing the Analogue to Digital conversion which again has a gain factor that contributes to the overall sensitivity of that pixel.
Separate names with a comma.