My question is based on ISO and regarding grain. So when taking a photo with high iso normally you would get grain. To combat this you would normally just simply lower the iso, but I've seen and heard that having iso as low as 100 can be negative to the picture.

I wondered is this true? and if so, why? If the picture is perfectly bright enough then why should iso need to be high? Additionally, do other settings also affect grain, such as aperture and shutter speed, or is it purely iso?

In a nutshell is it bad to have a very low iso, almost the lowest the camera can do, if the picture is bright enough?


7 Answers 7


I've seen and heard that having iso as low as 100 can be negative to the picture. I wondered is this true? and if so, why?

This can be the case with cameras where the "true" base iso setting is 200, so that iso 100 is done with in-camera processing, rather than adjusting amplification on the sensor itself. The sensor cannot physically go lower than iso 200, so the image at iso 200 is simply taken at +1EV overexposure, then processed to bring the brightness of the image down -1EV to simulate ISO 100.

However, this reduces the dynamic range of the image. And it's why using an "extended" ISO setting at either the high or the low end of the ISO range is something you only want to do if the tradeoff is worth it. Using a very low ISO can be similar to having a -1EV neutral density (ND) filter built-in, and could still be worth it to get a slower shutter speed, or to use a bigger aperture than you could if you were limited to ISO 200.

At the higher end of the ISO range, the reverse happens. If, say your camera is actually limited to ISO 3200, but you have "extended" settings for 6400 and 12800, those settings are achieved by the camera using ISO 3200, and underexposing, then "pushing" the exposure brighter in post-processing. This will probably not only enhance the noise than exists, it will also reduce dynamic range, just at at the low end.

There are a lot of current model cameras that have "ISO-invariant" sensors where theoretically shooting at a lower ISO and boosting it should not create additional noise. But most of these sensors may only be invariant over part of the range, and may not be perfectly invariant. And, regardless, if you use an extended ISO setting at either end, you're reducing dynamic range.

If the picture is perfectly bright enough then why should iso need to be high?

If you are not using extended ISO settings, but are within the native range of the sensor amplification settings, there isn't much reason, unless you need it to get good exposure with a faster shutter speed or smaller aperture setting.

Additionally, do other settings also affect grain, such as aperture and shutter speed, or is it purely iso?

Actually, ISO has nothing to do with "grain." Grain is a film term that related to how on a film stock, the ISO was determined by the size of the silver crystal grains on the film emulsion. The faster the film was, the bigger the grains are, and the grainer the images will be.

But with digital, we don't have any silver crystal grains or grain size. What we have is noise. Noise can come from the photons (not) hitting the sensor (a smaller amount means lower signal and possibly higher noise; i.e., darker areas tend to be noisier); from all sort of sensor electronic issues, like "dark current" heat, residual signal from the sensor not being completely cleared, or analog amplification.

But in general, to get grain with a digital image, you have to post-process it in as an effect. Grain is not uniform in size and far more random than noise typically will be in an image.

Basic beginner instructional material will often be wrong and confusing by calling digital ISO "grain" or "sensitivity". But those are how ISO worked with film. With digital, we have "noise" and (mostly) "amplification" instead. Some of the amplification is analog and adds noise, some of it is digital, which may or may not. So just increasing the ISO may not actually add more noise, depending on which methods are used under what conditions. It's not always as simple as lowest ISO => least noise; highest ISO => most noise.

The other exposure triangle settings typically do not add noise to an image, unless they require you to increase the ISO to get a good exposure.

In a nutshell is it bad to have a very low iso, almost the lowest the camera can do, if the picture is bright enough?

Depends on whether you're using an extended ISO setting.

  • \$\begingroup\$ AFAIK, using an extended low ISO does reduce the resulting dynamic range, but it does not increase read noise nor recorded image noise (shot noise); because it does not reduce the amount of light actually recorded and the resulting SNR. At least I have never seen any test results that demonstrate that it does/can; and I cannot come up with a reason why it would. \$\endgroup\$ Commented Feb 7 at 21:48
  • \$\begingroup\$ At low light levels and relatively short exposures (which are what necessitates using higher ISO settings) the primary source of noise is usually Poisson distribution noise due to the random nature of light itself. Photons are not perfectly distributed evenly within a light field of a given strength. Over time the randomness averages out, which is why shorter exposures show more randomness (and thus "shot" noise) than longer exposures do. \$\endgroup\$
    – Michael C
    Commented Feb 8 at 11:25

There's a common misconception about cause and effect when it comes to ISO and noise.

Noise is caused by too little light hitting the sensor. In very simple terms, photography is (in a way) the measurement of light, and if there is not enough light hitting the sensor, the measurements become inaccurate, and that is what noise is.

Using a larger aperture or longer exposure time is a way to get a brighter image by getting more light on the sensor. This also makes the photo less noisy.

Using higher ISO is a way to get a brighter image without getting more light to hit the sensor. As no more light is hitting the sensor than with low ISO, this will result in a bright but noisy image. This is where the misconception that high ISO causes noise comes from. It doesn't. It is more accurate to say that both high ISO and noise are caused by too little light hitting the sensor.

This means that you will not get a less noisy image by lowering your ISO, but leaving everything else the same; you will merely get a darker exposure.

In fact, depending on the camera, you might even get a noisier image if you do that. The best keywords for further research are "isoless sensors" and "ISO invariance". For complex reasons that I do not claim to understand, on sensors that are not isoless (ISO-invariant), lower ISO images that are later brightened will be noisier than ones that were shot with the right ISO in the first place.


In a nutshell is it bad to have very low iso, almost lowest the camera can do, if the picture is bright enough ?

No, using a low ISO is generally beneficial, and in no case is it bad as long as the image is recorded bright enough. Even extended low ISO's do not cause an increase in noise.

On a camera that is not ISO invariant, using too low of an ISO and having to increase the brightness on post is worse... with such a camera using a higher/more appropriate ISO actually reduces noise (grain). On a camera that is ISO invariant it makes no real difference.

With modern cameras the primary cause of image noise is photon shot noise. Basically that is the strength of the signal being recorded, and how much of the signal that is recorded. I.e. a dark scene doesn't contain a lot of light; that is a weak signal with a poor signal to noise ratio (SNR). And because it is a weak signal, you have to use a higher ISO.

Anything that causes you to record less light causes you to record less signal, and more noise results. So smaller apertures and faster SS's also result in more noise. The higher ISO is just a byproduct and only makes the amount of recorded noise more or less visible.

Similarly, bright light is a stronger signal with a better SNR. You can use a higher ISO in bright light and still record the same or less noise than a lower ISO would result in in low light; you just don't normally need to. This is why a recorded image with a range of brightness will tend to have a range of noise levels across it; with darker areas of the image showing more noise.

With digital ISO isn't actually exposure; it simply compensates for the third factor of exposure, which is light intensity/scene luminance. And it isn't actually a variable (except dual gain sensors have two levels). ISO is more akin to the brightness setting on your monitor... if you have recorded a noisy image, then making it more visible (brighter) will make the noise more apparent. Similarly, higher resolution (more MP/smaller pixels) only allows you to enlarge/magnify the image more, and that will make the recorded noise level more apparent.

Edit to add:

I think I understand the root of the question...

Currently there is a lot of talk about how dual gain/dual ISO sensors have lower read noise at higher ISO's than it does at low ISO's. This is due to the smaller capacity single capacitor state having a higher conversion gain and less switching noise.

However read noise is not image noise, it is just one contributor. And the people getting wrapped about higher read noise at lower ISO's are making a fundamental mistake/misinterpreting the data. At lower ISO's you collect more light; which results in a much greater SNR, and the signal drowns out the noise component (read and shot noise contributions).

The lower read noise at higher ISO's is significant at higher ISO's where the SNR becomes weaker. And at very high ISO's, where SNR is very low, read noise could be the primary noise contributor to the resulting image noise. But that isn't really relevant to the low ISO high SNR scenario.

This is NOT image noise. Note that read noise at base ISO (and extended low) is still less than the higher ISO's after the gains sate switch.

enter image description here

  • \$\begingroup\$ There's no such thing as an ISO invariant sensor. There are some that are much less variant than others, perhaps enough so that they can be used as if they were functionally invariant. But the claim that a silicon based image sensor can be perfectly invariant is a myth. \$\endgroup\$
    – Michael C
    Commented Feb 8 at 10:41
  • \$\begingroup\$ Actually, the original Nikon D1 is essentially completely ISO invariant, as is the D5500 and D7000 (slightly less so); they just are not particularly good performers. LINK \$\endgroup\$ Commented Feb 8 at 13:58
  • \$\begingroup\$ Those aren't flat lines. There appears to be some pushing/pulling going on at different ISOs. \$\endgroup\$
    – Michael C
    Commented Feb 12 at 23:30
  • \$\begingroup\$ @MichaelC; I don't think there is any manipulation going on... less than 1/10th of an EV is probably close to sample error. It is certainly far below visibility levels. \$\endgroup\$ Commented Feb 13 at 12:52

When speaking of ISO as an isolated parameter: you definitely get less noise at lower ISO values at the cost of image being dark. I guess it's not your intent. You can then bring the brightness back in the editor but you get the same (or greater) amount of noise back.

In general cameras have lower shadow noise (in terms of how well image is preserved) at higher ISO values because of how electronic image sensors work. All older digital cameras exhibit this behaviour strongly and it's less pronounced in later models.


In a nutshell is it bad to have a very low iso, almost the lowest the camera can do, if the picture is bright enough?

You don't change ISO in a vacuum. At a given brightness, lowering ISO will be offset by longer exposure time or wider aperture or a darker end result.

If you assume that you will adjust neither exposure time or aperture to make up for lowering the ISO, you will get a darker end result that you have to brighten in postprocessing.

Brightening in postprocessing means that you have not made use of the full dynamic range available to the sensor, and that means that the contribution of quantization noise to the end result will be larger than necessary.

Now some modern sensors have an "ISO-invariant" range where the effect of the ISO setting does not actually affect the raw image data but just at which brightness level this already digital data is converted to JPEG. When the sensor is not ISO invariant (or in the part of the range that is ISO invariant), then the effect of the ISO setting will be on the analog pathways, increasing sensor sensitivity (at the cost of adding base noise) or analog amplification before digitization.

But using lowest ISO can even be problematic when recording raw pictures and being in the sensor's ISO invariant range: the camera then meters in a manner where it will utilize the full sensor sensitivity range. Without specific high dynamic range processing (like strong shadow lifting), this will not deliver visible improvements while it makes the image comparatively susceptible to blown highlights. If you feel you want to dial down exposure in postprocessing, those blown highlights may limit the amount to which you can achieve good results.

So if you expect your image to possibly blow highlights on a modern sensor, using slightly raised ISO may give you reserves to dial down overall or just the highest brightness levels without resulting in flat color patches.

  • \$\begingroup\$ It's impossible to increase the sensitivity of the same sensor hardware. It's all analog amplification if the analog signal is increased for the same amount of light entering the camera due to changing the ISO setting. \$\endgroup\$
    – Michael C
    Commented Feb 8 at 10:44

ISO is not part of exposure and noise is not increased as a result of high ISO

This is contrary to what we typically read -- which is that ISO is part of the "exposure triangle". But ISO is not actually part of exposure.

The second part ... that noise is not increased as a result of high ISO may also sound confusing since most of us definitely notice more noise at high ISO. But it helps to understand what noise and exposure actually are -- and what causes noise -- to understand why this statement is true.

The short version is that noise is more noticeable in under-exposed images and that ISO is not part of exposure.

The sensitivity of your camera sensor is whatever it is ... ISO doesn't change that nor does it increase the noise. This is because the sensitivity is fixed. The camera opens the shutter for some duration in time and allows some flow of light (based on the focal ratio) and then the shutter closes. At that point the image has been captured -- the exposure is complete.

But ... BEFORE the camera writes the image to the memory card, it will apply "gain" ... and that's where the ISO comes in. This involves either amplifying the analog signal or multiplying the digital value (sometimes called ADU's for Analog-Digital Units) of the pixels ... and then writing the results to a file on the memory card. But it is important to note that this does not happen until after the image is exposed -- which is why ISO is not part of exposure.

So why do you notice noise in some photos (typically of high ISO) and not in other photos (typically of low ISO).

The ratio of signal (the light you are trying to collect) vs. the noise (the randomness inherent to the sensor) -- or Signal to Noise Ration (SNR) affects whether you may notice the noise. If you collect enough light the SNR is dominant with signal compared to noise -- the noise is still there but it is overwhelmed by the much stronger signal. This also means you wont need to increase the gain (amplifying or multiplying the pixel values) -- which also means the noise present in every image doesn't get amplified.

If, on the other hand, you do not capture enough "signal" then you may need to apply gain ... which also amplifies the visible noise. And the more gain you need, the more you'll notice the noise.

We can do experiments to test this. If you leave the lens cap on -- such that no actual light can be collected -- and take a exposure, you can inspect the value of each individual pixel and not only will you notice that each pixel has a non-zero value (despite not allowing the camera to collect any light), you'll also find that the pixel values will all be different. Given that you have a lot of random pixel values ... once you amplify those values you will find significant differences in brightness rather than subtle values that you hadn't noticed. The "noise" was technically there in every shot ... but was too subtle to be seen.

What is noise?

There are numerous things that contribute to noise. The biggest factor is something called "read noise", but there is also pattern noise, stuck pixels, amp-glow, and even quantum noise (although that's really subtle). Read noise is usually the most dominant factor and it is also affected by heat. A hot sensor generates more read noise -- which is why it tends to be more noticeable in very long exposures (and some cameras include a feature called "long image noise reduction" to try to combat the problem.)

But the main takeaway is that EVERY IMAGE has this noise... and the SNR determines whether or not you are likely to notice it.

The way you get a higher ratio of signal to noise ... is to take longer exposures (and this is why I say the ability to see the noise is a result of under-exposure.)

What's the downside to low ISO?

High-ISO means you are cheating the exposure. For example maybe you take an exposure that is only half of what is needed ... and then rely on gain to make-up for the under-exposure. While you can do this, the true image was under-exposed and that means you have to amplify everything (the good signal and the bad noise) -- making noise more noticeable.

But if your subject is moving and you don't want a blurred subject (or you have to hand-hold the camera) then you might want to force a higher shutter speed ... and that might not be possible at low ISO. This is a compelling reason to move to a higher ISO (just recognize there is no free lunch ... you'll also increase the visible noise.)


Assuming you mean noise on a digital camera (rather than grain on film) - no it's not really true in the real world, I'm sure there may be a technical reason based on how digital sensors work but to the normal eye the lower the ISO the better.

Why shoot higher ISO in bright light? To increase your shutter speed (say for longer focal lengths) or to use a higher f-value aperture for a greater depth of field.

You will have a sharper image at say a 300mm focal length with a s/speed of 1/500th at 1S0-800 than at 1/125th at ISO 200.

It depends what you do with your final image anyway to be honest, if you pixel peep on a PC screen you'll see the difference between ISO100 and ISO800, but if you print at a reasonable size on art paper from a normal viewing distance - then not really noticeable at all.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.