David Hayden Photography
ISO need to explain this!
This is our third discussion of the Exposure triangle and will deal ISO, what it is and how it gives us another tool for getting the exposure and effect we want. I recently went down a deep rabbit hole on the topic of ISO, sensor sensitivity, the relationship between ISO and digital noise and so on.
From an engineering perspective, this topic is very complex. Fortunately for our purposes there are a few things we need to know.
What is ISO and why do we have it anyway?
Referring to our exposure triangle / or pie charts as I presented the topic in an earlier article, we see that exposure is a combination of volume of light (number of photons) and sensitivity of the photographic material.
We can control the quantity of light that actually hits the sensor by controlling the size of the aperture to let in more or less light. And, we can control the amount of light on the sensor by changing the amount of time the light has to expose the sensor.
The level of exposure of the image is also dependent upon the sensitivity of the sensor. ISO is the current standard for establishing that sensitivity. Imagine trying to get a proper exposure without knowing how sensitive your film or sensor is to light. Back in the 1930s the Germans developed the DIN standards photographers could use to now how reactive their film was to light.
Later in the 1940’s the USA developed the ASA standard for film sensitivity. The ASA standard is based on a scale where each doubling of sensitivity amounts to one “stop” of exposure. Therefore, if you had a set up where your proper exposure was F4 at ASA 100, to Get the same exposure using ASA 200 film, you would have to close your aperture to F5.6. ASA 400 film would require a narrower aperture of F8 to get the same exposure and so on.
Different standards in different countries made things difficult. To create a universal standard that could be used internationally, the International Organization of Standards was established and in 1974 they issued ISO standard 2240 as a measure of light sensitivity. Because of its ease of use, the organization adopted the ASA standard. The current version of the standard, ISO 2240:2019, was published in 2019.
Here is a comparison of ISO, ASA and DIN standards for our purposes. Obviously, you would not always a combination of F/4, 1/200s at ISO 200. The purpose of these charts is to show how, when ISO doubles, the exposure gets brighter or darker by one stop. By changing aperture OR shutter speed, you can maintain desired exposure.
Imagine you have the depth of field you need for your portrait at F/2.8 but you need a faster shutter speed to freeze some motion. In that case you could increase your ISO to 400, set the shutter speed at 1/400.
Or, perhaps you need more depth of field, you can stop down the aperture to F/8, leave the shutter speed alone and just push the ISO to 800.
Getting the most out of higher ISO settings
Digital sensors turn photons into electrons that can be processed by computer devices into the images we see. This is not a straightforward process. For lower levels of light the sensor / processor needs to amplify the incoming signal to render an image. The problem is every signal also has within it a certain amount of noise. While the noise is fairly constant, the light signal is not.
In low light situations the signal is low compared to the background noise, so when the sensor amplifies things, the noise is amplified too. This results in more noise in the shadows compared with the brighter areas of the photograph where there is enough signal to stand out against the background noise.
Imagine you walking down the street having a conversation with a friend. As you are walking a firetruck pulls up with its siren blaring. Now you have so much noise, you need to amplify your voice for your friend to hear it. That is what I mean by signal to noise ratio.
If you were to use a microphone at that moment it would also amplify the siren noise. Your voice would still stand out a little more, but high levels of noise would still be there. To overcome this you speak a lot louder into the microphone, raising your signal well above the background noise.
Let’s compare a couple of images, same subject but with different exposures.
These first set of images were exposed so the white gauze has the best detail. The brightest white point in the gauze is 248 out of the max 256. All of the black velvet has about 0 brightness. What if we wanted details in the black areas?
By pulling up the black point with out adjusting the whites we can now see some detail in the blacks like the black netting and some folds in the velvet. But that comes at the cost of some noise. You will hear this referred to as “noise in the shadows.”
The image below is the same subject but exposed by raising the ISO from 1600 to 3200 to see detail in the dark areas. You can see the black netting on the velvet. But there is very little detail in the white gauze.
But as you will see by bringing back the whites and lowering the contrast in postprocessing, we can reveal the texture in the gauze without adding additional noise.
The point of this is not that you should raise your ISO, because doing so always introduces more noise.
The point is, if you are going to use higher ISO, set your exposure to have the desired detail in the shadows and you will have less noise. That said, if there is a wide range between the dark and light areas you will lose detail in the highlights.
Shooting brighter exposures without losing all details in the whites is referred to as exposing to the right (ETTL). In the case above, using ETTL, I can even raise the blacks a bit to pull out the texture of the velvet with very little introduction of noise.
Digital noise and cropping
Another contributing factor is cropping. If you look at an image it may be perfectly acceptable, but cropping in reveals the digital noise. Therefore, one way to reduce noise is to shoot in tight so you don’t need to enlarge the pictures to clearly see the subject.
Below is an example of some dancers I photographed a few days ago. I cropped in on the both the back and front dancer’s faces. If you look closely you will see there is a good bit more noise in the face that has the most enlargement.
Technical stuff you may not care about - Not all sensors are created equal
Digital sensors in cameras have a rating that describes the amount of noise they contribute to an image. This rating is known as the input-referred read noise (IRRN). IRRN is a measure of the noise that is introduced into an image by the sensor itself.
The lower the IRRN value, the less noise there will be in the image. The higher the ISO setting, the higher the IRRN value will be. This is because the camera must amplify the signal from the sensor more in order to capture more light at high ISO settings.
To interpret this information, it is helpful to compare it to the signal level in the image. The signal level is the brightness of the image, and it is typically measured in digital counts (DN). A typical signal level for a well-exposed image is around 100 DN.
If the IRRN value is much lower than the signal level, then the noise will be negligible. However, if the IRRN value is close to or higher than the signal level, then the noise will be visible in the image.
What this means is that the amount of noise in your images dependent on the sensors capability and how far you push the ISO.
Learning high ISO Photography
Here is a lesson or experiment you may want to try.
Make a simple set up with some different cloths with various colors but including a black and a white cloth.
Start taking pictures using manual exposure and manual ISO, then By changing the ISO setting only.
Expose to the left (ETTL) so the brightest cloth has good texture
Expose to the right so the black cloth has discernable texture
Expose in the middle so the mid-range items have the best exposure
Open your images in your preferred software and push the exposure, light & dark (white and black) sliders around until you can predict the level of noise you are likely to have to deal with in your images.
Here is a practical example that worked for me. When I was shooting a Nutcracker ballet for Sangre de Cristo Arts Center’s School of Dance, I wanted to keep my ISO as low as possible and still catch action. I started at ISO 6400, thinking I would have lower noise than if I used a higher ISO. The highlights were slightly underexposed, I then pushed the images in post processing to achieve a proper exposure.
The highlights had some noise, but the shadows had a lot of noise.
I changed my approach. For the next showing, I raised my ISO to 10,000 in order to have more detail in the shadows. I was surprised by the fact that there was actually less noise in the set of photographs with the higher ISO because, by exposing to the right the camera had more light to work with in the dark areas. Definitely some bright areas such as white costumes or bright lights on faces caused them to be over exposed, but I was able to tone those down and get respectable images.
This high ISO (6400) image of a dancer at the top of this article actually came out so well, it was able to be used for this wall-wrap in the school of dance.
Coming next week:
The first article in a series about the properties of light and how they apply to photography
May the images in your mind become your inspired creations!