# Why can't the ISO level on most digital cameras be set below 80?

Why is it that the ISO level on most digital cameras doesn't seem to go lower than ISO 80?

I understand (vaguely) that the sensors adjust the gain in order to achieve the equivalent of a high ISO film and I can see why there would be a limitation to this, but don't understand why it would be so difficult to make the sensor less sensitive to light?

• – mattdm Feb 26 '11 at 1:20

There are two ways to reduce the base of native ISO (the lowest setting possible on the camera). One is to make the sensor less efficient, so that the less of the incoming light is captured. It is easy to see how this reduces the sensitivity. However, as already stated in other answers, this would only be of use to the minority people who regularly find that they have too much light whereas the majority of photographers struggle with not enough light for the most part!

However there is another way to reduce the base ISO and that is to increase the capacity of each pixel (referred to as the well depth) so that it can store more charge (and hence more light) before becoming saturated. This method does have a genuine advantage at all sensitivities.

It is a little trickier to see how this affects the base ISO, however you have to remember that whatever readings the sensor takes eventually get mapped to the range 0-255 either by the camera or a person during raw conversion (or 0-65536 for 16 bit images). Imagine two sensors A and B, A has twice the capacity per pixel. When you shoot an image with A at 1/30s shutter speed the highlights are just about to clip. Now when you use B, you have to use 1/60s because of the decreased capacity - if you used 1/30s you would overexpose the image.

Whatever values the sensors record at the saturation point they still get mapped to 255 so will look the same. Given that with B you get the same image but with half the exposure time (all else being equal) this means B has higher sensitivity. If the base sensitivity of camera B is ISO100 then the base sensitivity of A is ISO50.

The point of all this is that if you can reduce the base ISO by increasing pixel capacity then this is a good thing as it allows you to have lower noise and greater dynamic range. To understand why this is the case, you first have to let go of the idea that increasing the ISO setting increases noise, it's simply not true, here's an answer I wrote on the subject: What is "ISO" on a digital camera?

The short version is that noise is mainly caused by lack of light (the less light you have the more random variations in photon emission show up in the image, see http://en.wikipedia.org/wiki/Shot_noise) it just happens that if you increase the ISO setting of your camera other settings have to change which result in you getting less light and more shot noise. It's the lack of light that produces the noise, not the act of amplifying the signal.

So even though we have to use amplification to get ISO100 out of camera A, we let in the same amount of light as camera B shooting at ISO100 and thus can expect the same level of noise! So camera A is no worse, however we do have the option of going to ISO50, letting in more light and thus getting less shot noise!

We also have the option of using ISO50 but with the same camera settings as B is using to shoot at ISO100, we let in the same amount of light so can expect the same level of noise, but if the scene suddenly gets brighter B with clip the signal whereas A will carry on with it's extra well depth, thus having higher dynamic range.

These examples are a simplification, there are other sources of noise, but in general having a "less sensitive" sensor can be a good thing. It is difficult to engineer deeper electron wells however, which is why you don't see cameras go below ISO50. You do see differences in well depth though, for example the lowest ISO available to the 35mm sensor Canon 5D is ISO50 whereas the same generation APS-C sensor in the Canon 30D the lowest ISO is 100 this is due to the extra capacity in the larger pixels of the 5D.

• +1 Fine answer (as always). Just to stave off possible misconceptions from future readers, though, please note that when you let in more light at ISO 50, you actually get more shot noise compared to the ISO 100 sensor. The point is that the typical amount of noise (in root mean square) relative to the amount of incident light goes down. (Quantitatively, dividing the native ISO by 4 reduces shot noise by one full stop relative to the signal.) – whuber Feb 27 '11 at 15:48
• Thanks! When I say less noise I'm almost always talking about the signal to noise ratio. It rarely makes sense to talk about absolute noise levels, and when you have more light, signal to noise ratio always improves (all else being equal) – Matt Grum Feb 27 '11 at 16:03
• I've been wondering so long about this exact question. A friend pointed me to this answer. Your simplified examples are so well explained. I'd vote twice if I could :-). – Nikos Alexandris Feb 8 '14 at 1:12

If a manufacturer wanted to do that, they could.

That being said, all sensors have a native sensitivity, usually ISO 100, 160 or 200. Gain is used to achieve higher sensitivities by multiplying the base signal. This introduces more noise, so if the base ISO was lower, them you would need to use more gain to achieve higher sensitivities and you would get more noise.

To get lower ISOs without changing the base setting, you basically divide the base signal. This causes a reduction in dynamic range, which is why you rarely see ISOs less than 50. If the base ISO would be divided by a larger factor, then too much dynamic range would be lost.

Finally, there is not much reason to develop lower sensitivity sensors since it is easy to use an ND filter to reduce incoming light without affecting dynamic range. Several cameras from Canon, Nikon and Fuji have a 2 or 3 stops ND filter built-in for that effect. On a DSLR, you can buy ND filters with variable strength which makes things very flexible.

• "if the base ISO was lower, them you would need to use more gain to achieve higher sensitivities and you would get more noise" this is not necessarily true - there are two ways to decrease the native sensitivity of a sensor, one is to reduce efficiency (so fewer photons get through, essentially the same as always having an ND filter) the other is to increase the capacity of each pixel, allowing more light to come in before the signal clips and overexposure happens. Comment continued below... – Matt Grum Feb 26 '11 at 7:52
• ...continued Both ways of reducing base ISO will require amplification to get to say ISO100, yes, but increasing ISO doesn't introduce noise (except a tiny amount in the amplification process). Noise mostly results from lack of light causing non-uniformity of the incoming photons. Increasing ISO can only reveal noise already present. So if you have an ISO25 sensor with a super high capacity per pixel you record the same number of incident photons per pixel as an ISO100 sensor, you have to amplify to get the same image but you wont have any extra noise in pushing the ISO25 sensor to ISO100! – Matt Grum Feb 26 '11 at 7:57

but don't understand why it would be so difficult to make the sensor less sensitive to light?

That's almost counter-intuitive to the concept of photography which, after all, is about the capture of light. Having said that, the CCD/CMOS is going to have a natural floor for the light capture so any less than that is going to require the deliberate discarding of information which, in general, is probably not going to be done to the satisfaction of photographers. Software, which is what would do this, is always going to be a compromise on various factors.

In any case, as Itai noted, there are other ways to reduce light. You can use ND filters, smaller apertures, faster shutter speeds, etc. If you shoot raw, then you can manipulate exposure even more. Net effect, going below ISO 80 is simply not likely to be worth it to manufacturers, they are much more interested about the other end of the range.

There is nothing stopping the firmware programmer from implementing lower "ISO" values by averaging multiple frames. For example if the base sensitivity was determined to be "80 ISO". "40 ISO" can be implemented by averaging two frames.