ISO is NOT What You Think It Is – What is ISO, really?
ISO is the least understood and most misrepresented concept in photography. Let’s explain what ISO is and what it is not.
Because ISO is such a fundamental part of photography, I will be taking my time to explain it in the most accurate yet straightforward way possible. This blog article is lengthy, but using the menu below, you can navigate between sections:
Part I: The History of ISO
When the first commercial camera was invented in 1837, a device known as the Daguerreotype, the world became exposed (no pun intended) to one of mankind’s greatest discoveries: photography. In the coming years, the camera became society’s newest obsession, and quickly the rudimentary Daguerreotype system became superseded by more refined devices, using film canisters instead of photographic plates with mirrors, viewfinders, and focusable lenses. The age of the film camera was in its prime and all seemed fine and well for the photography industry, but one major flaw remained for photographers, the film. The first wave of cameras to use film had ratings corresponding to their light sensitivity that were based on the amount of light-sensitive coatings or crystals embedded in their structure. Because the industry was still blossoming, each film manufacturer would use their own film ratings or not include them at all, leaving photographers struggling to accurately expose their film, fueling the widespread adoption of photodiodes and light meters to aid in exposure calculation. Thankfully, relief came when systems of measurement were developed to regulate the sensitivity settings on film, and in the 1960s, DIN and ASA became the mainstream organizations used to standardize film sensitivities. The DIN standard adopted a system of logarithmic sensitivity settings while the ASA standard used a linear progression. It was common during the 1960s and 1970s to find film canisters listing both the ASA and DIN numbers as references to their sensitivities. But as film became ubiquitous, consumers and manufacturers wanted one system of standards to adopt instead of the two standards of ASA and DIN, so lo’ and behold, the International Organization for Standardization released the ISO standards in 1974, which became what is today, the international standard for measuring the speed of color negative film and “sensitivity” of digital sensors.
Part II: ISO in the Digital Era
When digital cameras entered the scene, the ISO standards developed in the days of film were adapted to suit new technologies. To put it simply, ISO in digital sensors is just an arbitrary scale that describes the amount of amplification applied to the analog signal from the sensor before it is converted into a digital number.
If that doesn’t make sense, let’s remind ourselves how a digital camera works.
First, a digital camera is composed of three main devices: the camera sensor, the gain circuit, and the analog-to-digital converter (ADC). When a picture is taken, the shutter curtains over the sensor open, allowing light to be collected. The length that the shutter is open, also known as shutter speed, and the rate at which light is collected, which is determined by the lens’ aperture and the light from the environment, are the two variables that will affect how much light ends up being collected by the camera. Photons hitting the sensor accumulate in tiny “wells” called photosites, and the energy of the photons is converted into voltages corresponding to each pixel on the sensor. These tiny voltages comprise the “analog signal” and directly after leaving the sensor, the analog signal is amplified by the gain circuit to be processed by the ADC. The ADC is a chip inside your digital camera that classifies the voltages of pixels into levels of brightness and assesses each level to a binary number, consisting of zeroes and ones. This is done through a process called “quantization” which effectively takes the analog signal and creates a sequence of numbers that best estimates the original input. All ADCs have a fixed range of voltages they can receive and quantize into a digital signal as well as a fixed resolution of possible output values. Many ADCs in cameras have a resolution of 8 bits meaning they can encode an analog input to one in 256 different levels (2^8=256). Now that we know a bit more about how a camera takes a photo and processes it from a light to a digital signal used to create an image, let’s go over how ISO affects all of this.
Part III: ISO Myths Debunked
First, unlike film, ISO does NOT change the sensitivity of the sensor in the literal sense. A sensor has a fixed quantum efficiency determined during production that cannot be modified due to any camera setting. The misconception that ISO is equivalent to the sensitivity of the image-sensing material is a myth derived from the days of film when such a declaration was true. ISO in digital cameras is exactly that, digital. ISO is amplification, and changing ISO changes how much the analog signal, the raw amount of light gathered, is boosted to produce an image that looks properly exposed.
The best way to explain this is with an analogy. The easiest one to conceptualize is that of a radio. In a radio, the radio box and antenna represent the camera sensor and lens while the radio waves the antenna receives are akin to photons entering the camera lens and falling onto the sensor, or radio box. The antenna receives radio waves and the radio box takes that raw analog signal and converts it to sound much like a lens receives light while the sensor and other electronics collect that light and convert it from an analog to a digital signal. Radios have a variety of buttons and knobs to control their functions, but all of them will have some sort of volume control. The volume control on a radio takes the signal being collected and simply amplifies it to make it louder or softer, but as you may have experienced, if the signal is weak, to begin with, meaning there is a lot of static and it is hard to hear, when the volume is increased the resulting sound is louder yet also has more static. On the contrary, if the signal is loud from the beginning, you won’t need to increase the volume as much, and the amount of static barely changes.
This is exactly like ISO in a digital camera: light is emitted randomly from objects and over time the randomness of the photons averages out to produce a clean signal. If the camera exposes the scene and gathers a large amount of light, the ratio of signal to noise (signal-to-noise ratio, or SNR), will be higher and the data will need no amplification. When the signal is not amplified, we call this the camera’s “native ISO” which for most cameras is anywhere from 50-200. If the light gathered by the camera is weak and not enough to produce a properly brightened image, the data will need to be amplified, and because light is composed of signal and noise, not only will the good signal amplified but the noise will also be amplified. The ratio of signal to noise then becomes lower and more noise is present in the final image. To recap, lower ISO values indicate less amplification of light gathered by a camera while higher ISO values indicate more amplification of light, thus increasing the brightness of the final image but not the exposure since exposure is determined by the physical amount of light captured before any amplification takes place.
Taking this notion further, another myth regarding ISO that does not transfer directly to digital is one stating “Increasing ISO increases noise.” The problem I and many other technical photographers have with this statement is that while it applies to scenarios where the ISO is increased with an equivalent exposure decrease to maintain a certain brightness, increasing ISO while maintaining the same exposure settings only amplifies the data, so while the noise will be amplified, the signal will be as well, and the SNR will not change. In fact, in ISO-variant cameras raising the ISO may decrease the visible noise. Let’s explore this concept.
PART IV: ISO-Variance and ISO-Invariance
Hot pixels are shown in the green box. Photon noise, dark current, and read noise all contribute to the general noise pattern shown in the red box.
In order to explore ISO-variance and ISO-invariance, we need to understand the sources of noise in digital images.
Dark current is noise caused by heat, and it is a relatively even distribution of noise that cannot be avoided without applying dedicated cooling systems to your camera. If you’re familiar with astrophotography, many Astrophotography cameras will attempt to mitigate dark current with advanced cooling mechanisms. Some high-end astrophotography cameras even use liquid nitrogen cooling!
Hot pixels are individual pixels that look much brighter than they should. They are also called “sparkles” or “dead pixels.” All cameras have a few hot pixels which will appear and fade with time, but with higher temperatures, they will become more numerous and visible. Hot pixels are caused by tiny current leaks in the wells of photosites on the camera sensor. When current leaks into a photosite, it causes it to have more voltage in the well and thus the analog signal will be converted into a larger digital signal, appearing brighter in the final photo.
Photon noise, or shot noise, is caused by the inherent particle nature of light. Photons from objects are emitted in random directions, and due to this, light is not distributed evenly across the sensor during an exposure. This causes some pixels to appear lighter or darker than others and is the main source of noise in most photos. The effects of photon noise, thankfully, can be mitigated with stacking and long exposures, which can average out the noise to produce a cleaner image.
Read noise is the last main source of noise in images. Read noise exists in all cameras and is produced from the electronics of the sensor, gain circuit, and ADC. Read noise is most apparent when low amounts of analog signal travel to the ADC, originating from downstream electronics, and in many cameras, read noise decreases as the analog signal is amplified. This is because if analog data is amplified in-camera, then the ratio of signal to read noise increases, since read noise is a fixed value.
This brings us back to the notion that ISO and noise are not directly coupled and the concept of ISO-variant and ISO-invariant cameras. In ISO-variant cameras, using low-ISO settings in combination with low light levels might actually be detrimental to noise levels, since the read noise is high and the signal is low, resulting in a low SNR, especially in shadows. Canon cameras are noticeably ISO-variant, and looking at this series of images from the Canon 6D, when exposure levels are raised in post-production rather than in camera, banding noise becomes clearly visible in ISOs of 100 to 1600. Beyond ISO 3200, the noise levels remain the same, indicating that the read noise has reached a level below the general noise floor. This is why in ISO-variant cameras, when shooting in low-light environments where not much light naturally exists, it is best to amplify the data in-camera instead of in RAW processors such as Lightroom.
ISO-invariant cameras have lower read noise levels such that in low-light shooting conditions, the SNR stays more constant as ISO settings change. Another name for these types of cameras is “ISO-less,” because, in theory, you can shoot at any ISO and boost the exposure in post-production. Even ISO-variant cameras will eventually act like ISO-invariant cameras because at a certain point the amplification of the analog signal will overcome the noisy electronics which produce read noise and any increase in ISO will show minimal to no difference in the noise present in the photo. In low-light environments, it is best to shoot at the native ISO for completely ISO-invariant cameras and amplify in post-production, while for ISO-variant cameras it’s best to shoot with high ISOs until the camera becomes ISO-less. Why would you want to shoot at a lower ISO if all ISO does is amplify signal and may even reduce noise? Because increasing ISO also decreases dynamic range.
Part V: Dynamic Range and ISO
The relationship between ISO and dynamic range is very technical, but all you really need to know is that as ISO increases dynamic range decreases. Because of this, ISO-invariant cameras allow users to shoot at low ISOs to preserve dynamic range, yet because of the low read noise, a sort of “fake” ISO in RAW processors can be applied by increasing the exposure slider, amplifying data to increase brightness. The reason why increasing ISO decreases dynamic range boils down to the fundamental nature of the ADC inside the camera. As I said earlier, the ADC converts an analog signal from the sensor into a digital signal that we can view and edit. Furthermore, the ADC can only accept a range of voltages to digitize, and it’s a restriction baked into its transistors, capacitors, and electronics; the range is fixed and cannot be altered. When ISO is increased, the analog voltage is amplified, and if, say, the maximum voltage the ADC can read out is 1.0 V yet changing from an ISO of 100 to 200 amplifies the brightest part of the image from 0.75 V to 1.5 V, some of the signal is lost because the ADC can simply not handle 1.5 V when its maximum readout voltage is 1.0 V. To compensate, the camera records a smaller range of light in order to stay within the ADCs constraints, and thus, dynamic range is decreased.
Recap so far:
So far, we have debunked two myths and introduced a little-known principle of ISO. The first myth we debunked was one that suggested ISO was linked to sensor sensitivity. Derived from the days of film, in the digital world sensors have one fixed efficiency determined by the manufacturer, and ISO merely serves to amplify already-recorded light signals. The second myth concerned ISO’s link to grain and noise in images, supporting the ancient idea that increasing ISO directly increases grain size. Another stain from the days of film, this mantra taken literally is false, although increasing ISO while decreasing exposure to maintain the same image brightness will increase visible noise, so if you only shoot in program modes, then changing ISO will affect the amount of noise in your photo. But because ISO is merely amplification, changing ISO doesn’t change the SNR in your image when exposure is held constant. After these two myths were debunked, we introduced the concept that increasing ISO decreases dynamic range. Indeed, this is true due to the nature of ADCs and how the digitization of light changes as it is amplified.
Part VI: Implementing ISO
A series of photos from underexposed (left) to normal exposure according to the camera computer and most people (middle), and a correctly-exposed ETTR image (right).
Credit for image: pocketlenses.com
In general, I recommend first exposing your scene with the right shutter speed and aperture to produce the image you desire. Remember that the quality of the final image you create is mostly dependent on the amount of light gathered in the first place, so try and maximize the signal your camera can gather. For example, if you are shooting a landscape, it is helpful to set your aperture first to optimize for lens sharpness and then set your shutter speed to produce an image with maximum SNR, leaving ISO, hopefully, at its native setting. This strategy produces the highest-quality image possible and I would encourage photographers to adopt this method. A method called expose-to-the-right (ETTR) is another name for this method. Essentially, ETTR suggests that in order to create an image with maximum SNR, one should push the histogram as far right as possible without blowing out any highlights by either adjusting shutter speed and/or aperture settings. ETTR should be attempted for all types of photography, although in some cases it will not be possible to create sufficiently exposed images without ISO amplification. In wildlife and sports photography, it is often the case where, due to the fast motion and long lenses involved, shutter speeds must be significantly shortened to freeze motion, and because many consumer telephoto and super-telephoto lenses do not have large apertures, the image is almost always under-exposed (and under-brightened). If your camera is ISO-invariant and if no shadows are being clipped, meaning no data is being lost, you can leave the ISO setting lower and boost the brightness in post. Most photographers, however, will choose to increase ISO in-camera to review photos easier or because they are using an ISO-variant camera. Scenarios like action, wildlife, low-light event, concert, street, or portraiture photography will often require ISOs to be manipulated, but remember that the more light you can gather, the cleaner your images will appear, no matter what.
Part VII: Editorial
ISO is a poorly understood concept in photography, and frankly, in order to grasp it fully, you need to be exposed to many scientific and technical terminology that may be confusing and a turn-off to learn more. While I believe you don’t need to fully understand how ISO works in order to produce stunning images, I believe we as people have a moral responsibility to absolute truth and its integrity. You can explain ISO at many levels, but what I believe many educators have done is use the film definition of ISO as a crutch and excuse to ignore the rich scientific nature of photography that influences every aspect of how photos are created. For a technical photographer like myself, it took many years to understand ISO due to the egregious levels of misinformation on the internet spanning articles, YouTube videos, and formal education. Out of three college photography courses I have taken at Century College and the University of North Dakota, I have been repeatedly told that ISO equates to sensor sensitivity, but upon challenging this notion, my professors would say it’s “not necessary for the general photographer to know ISO’s true meaning.” Frankly, this mentality is outright wrong and even insulting, because not only does ISO affect dynamic range, an extremely important principle of photography, but knowing how ISO behaves will influence the way you choose camera settings and will allow you to capture higher-quality images. As a photographer, beginner or advanced, I would never want information withheld from me under the assumption it would be too “complicated” or “technical” for my brain to handle. My mission is to try and boil down the complicated, technical aspects of photography into digestible chunks that can be understood by people without a scientific background. While this guide is a technical one, it’s important to realize that some concepts in photography and other fields require a technical explanation to be understood fully, and especially when explaining a controversial topic such as ISO, it’s important to back up claims with real science.
With that aside over, I would like to give you a message and responsibility as the reader to correct other photographers or educators when they misrepresent ISO and to only teach and preach truth to others. Because photography is a science as well as an art, basic integrity in the laws and knowledge that are put forth are required. Try to remember ISO, what it is and what it is not, and help others understand as well. If we all make an effort, we can begin to dispel the harmful myths and misconceptions surrounding ISO that have groomed a generation of misinformed photographers.
The TL;DR (Too Long; Didn’t Read)
ISO is a photography setting that dates back to the days of film. Originally a way to classify the sensitivities of film based on the amount or size of physical light-sensitive crystals, when digital cameras entered the photography scene, ISO became a way to categorize arbitrary amplification settings. What this means is that ISO is not the sensitivity of a digital camera sensor, nor does it directly contribute to added noise in images. Digital cameras capture light energy and transform it into a digital signal, and when the camera doesn’t have enough light to produce an adequately-exposed image, the signal AND NOISE is amplified, creating grainer images. Furthermore, because of the way digital cameras are structured, increasing ISO subsequently decreases available dynamic range. This leads to some photographers adopting the strategy of shooting dark images at low ISO’s and brightening those images in raw editors with the exposure slider. The exposure slider and the camera’s ISO work in the same way, and in cameras where this can be done effectively, we call them “ISO-less” or “ISO-invariant.” In some cameras boosting the ISO in camera actually reduces noise, and we call these cameras “ISO-variant.” So, really, ISO is much more than a setting that adjusts sensitivity and grain, and understanding what it is and what it isn’t will hopefully improve your photography.
Read my other blogs:
