Why would using higher ISO and faster shutter speed yield more noise than using lower ISO and slower shutter speed?

Question

I'm trying to see how this answer correlates with practice. That answer basically says I need to let enough light into the camera and then use the highest ISO value and I'll get the lowest noise then.

So I make two shots with the same camera. Camera has aperture set to a specific value, fully manual mode, camera is on a tripod and focused onto a specific faraway object. Shooting is done around midnight, so it's rather dark. The scene is a large industrial building standing faraway across a field from the camera. The field has no light sources, all lights are on the industrial building.

I dunno if it's important but images are written as JPEG inside camera - no shooting to raw and external post-processing. Maybe that's why I see my unexpected result.

Camera highest ISO value is 3200 (not using extended ISO values).

So I first shoot with ISO 1600 and shutter speed set to 1/125 second and then I shoot with ISO 3200 and shutter speed set to 1/250 second. The amount of light should be identical and indeed both shots look properly exposed and exposed the same way.

When I open the pictures in an image viewer and zoom in to one hundred percent I see that ISO 1600 shot is much cleaner than one with ISO 3200. I checked that several times and I'm sure - it's not the other way around.

Why is this? Is this because of post-processing difference?

Answer 1

So I first shoot with ISO 1600 and shutter speed set to 1/125 second and then I shoot with ISO 3200 and shutter speed set to 1/250 second. The amount of light should be identical and indeed both shots look properly exposed and exposed the same way.

The amount of light is not identical. You let twice as much light into the camera at 1/125 second than at 1/250 second. You then amplified the image with half as much light to match the brightness of the other image. In so doing, you amplified the noise in the image by a factor of two as well. To let the same amount of light into the camera at 1/250 second, you would need to also open up the camera's aperture by one stop compared to where it was set at 1/125 second.

When we say noise in the context of an image, what we often really mean is the Signal-to-Noise Ratio (SNR).

In the context of digital photography, the 'signal level' is determined by the amount of light in the scene and how much of that light we allow to enter the camera by selecting a particular aperture size and shutter time.

The amount of actual noise is divided into two main types: Read noise and photon shot noise. Read noise is fairly constant for a given shooting condition. It can be affected by the camera's internal temperature, but it won't usually change from one shot to the next under the same general environmental conditions. Photon noise, often called 'shot' noise or 'Poisson distribution' noise, is variable and increases with the amount of light, but increases at a lower rate than the increase in light.

Adjusting ISO affects how much the signal and noise combined are amplified. Increasing the ISO setting multiplies both the signal and noise equally, so the SNR remains the same. However, as the image is amplified more it becomes brighter and we may be able to see more of the noise in the image that was previously masked in the darker shadows of the image.

The end result is that the more light that is allowed into the camera, the less influence the noise will have on an image because the SNR is increased by adding more additional signal (light) than additional noise.

The advice in the answer referenced at the beginning of the question is to be understood as "... let enough light into the camera and then use the highest ISO value that doesn't result in blown highlights. If enough light is let into the camera, that ideal ISO could well be the camera's minimum ISO setting. Only when we are limited by the amount of available light, the camera's maximum aperture width or the aperture setting needed to get a desired depth of field, or the shutter time needed to prevent blurring a moving subject does the advice to increase ISO to the highest setting that does not overexpose the highlights become applicable to an ISO setting higher than the camera's baseline ISO.

Answer 2

Increasing ISO does not give you more light. It only tells the camera to amplify the signal coming from the sensor to "simulate" additional light. However, this simulated light is not like the real thing, because when the camera amplifies the signal, it also amplifies the noise.

Hence, it is always preferred to use a slower shutter speed or a wider aperture if possible, because those really give you more light. In your case, since the aperture is constant, 1/125 will give you twice as much light as 1/250, which is why the 1/250 shot needs to be amplified twice as much to obtain similarly exposed results.

Answer 3

Getting the exposure correct is the goal. We are talking about camera settings and adjustments that control the amount of light energy that will play on the image sensor. The objective; make an image containing the desired intensities and colors.

Three quantities make up the exposure.

1. Image brightness 2. Length of exposure 3. Sensitivity of image sensor

All three of the components of the exposure can be manipulated. Likely, there will be a range of exposures that produce acceptable results. Exposure has an equation E=IT. This equation is known as the “law of reciprocity”. Verbally stated – Exposure = Intensity multiplied by Time.

We can adjust the sensitivity up or down. We are talking about the ISO of the imaging chip. The best quality is obtained when the ISO is set low. This is because when we up the ISO, we are upping the amplification required to obtain our goal.

Now the CMOS chip contains millions of photosites, each with an amplifier. The efficiencies of each amplifier are slightly different. This difference induces “fixed-pattern noise. Noise is the counterpart of static in an audio system. All circuits contain “good” signals intermingled with “corrupt” signals. This is called “signal-to-noise-ratio”. As we up the amplification, both parts are intensified. The bad news is, the ratio of the corrupt part escalates thus we get more noise then we want. Noise is seen as a lack of uniformity. In other words, a granularity akin to grain seen in many film based images.

The bottom line is: There are multiple settings that yield a correct exposure. You should likely choose to set the ISO as low as you can to achieve the desire effect. The good news is, photo engineering marches on, todays chips and software yield less noticeable noise over a wider range of elevated ISO settings. Tomorrow’s chips and software keep this a moving target.

Answer 4

The referenced answer is dealing with a very specific situation. As other answers have noted, to reduce noise in the final image, you need to get more light (signal) into the camera. The referenced answer is based on the assumption that getting more light into the camera isn't possible; you can't open the aperture or switch to a longer shutter speed, but the image is still underexposed. In that specific situation, it is better to increase ISO than to brighten in post.

Without going into all the technical details, the signal-to-noise ratio at the sensor will be the same either way; increasing ISO won't reduce noise. But the values from the sensor get rounded when converted to an image file. Increasing the ISO multiplies the unrounded analog values, while brightening in post increases the rounded digital values. Also, the sensor output (including the effect of ISO) is linear, but image files (whether raw, JPEG, or whatever) follow a nonlinear response that matches that of the human eye. The combined effect is that brightening in post posterizes the darker values in the image. That makes the noise appear more prominent than it would if the brightness were increased through the ISO.

As far as I know, the extended ISO actually multiplies after the conversion to nonlinear digital data, so there's no benefit there over brightening in post.

Answer 5

The main sources of noise in a camera sensor are read and and shot noise. As you increase the amount of light captured (increase exposure time) both these sources of noise decrease relatively. One extra stop of light will increase the signal to noise ratio by about 41% simply due to the noise statistics. Thus you will always do better by capturing more light (longer shutter times) until you reach the saturation limit of the electronics.

Increasing ISO increases the gain used in the camera - this amplifies both noise and signal (it also adds in some more noise but that is usually not an issues in modern cameras). For the same incident light increasing ISO will mean reducing shutter duration, reducing the amount of light captured and this an increase in noise and reduction in signal to noise ratio.

Reducing ISO and increasing the amount of captured light is therefore better from a noise point of view. However once you get to the base (unamplified) ISO of a camera, further reductions reduce the dynamic range available which may have a different detrimental affect.

Answer 6

It does not. You get less noise by shooting at the lowest ISO possible which is why high-end cameras offer ISO 64, 50 and even 32. When shooting at a lower ISO, exposure must be longer or the aperture must be wider. In both cases, you would get more light.

More light increases the signal relative to noise and hence you get cleaner output.

Keep in mind that some low ISO are Expanded. This still reduces image noise since scaling down the data scales down the noise too but it can result in lower dynamic-range but it depends on how reading off the sensor is done.