What is Camera Noise?
Camera noise, as wedding photographers it can be the bane of our existence. There is an argument about the value of noise, the character it adds to a photo, almost a nostalgic throw-back to the days of film grain. Sometimes it's applied to counter other issues like banding, or to enhance consistency between images taken from different sources or design elements. Many programs like Lightroom, Photoshop, and Capture One Pro even feature filters by which to add noise. But what is it? It's funny how many professional photographers themselves don't know. I've always found myself intrigued by it, and so I tried to learn as much as I could to counteract it, and today I'd like to take a minute to clear this up and set some records straight.
Let's start this one of by being clear about what it ISN'T. Digital camera noise IS NOT the same as film grain. That being said, though they're quite different, they're not exactly dissimilar. They share many commonalities such as a correlation in the strength of the light in the photo, so let's consider them cousins. Now I could get into a detailed description about the film development process and how the solutions interacted with the density of the silver grains... but that's long and boring, and for general purposes it's not really useful for modern digital photographers.
So to clear it up let's just clarify with two very distinct ways of telling the two apart so you can clearly see that they are distinctly different, and move on:
1. Film sensitivity directly affects the size of film grain and can present in different sized spots, whereas digital noise will always be limited to a single pixel in size despite digital sensor sensitivity (ISO).
2. Film grain reflects luminance, whereas digital noise reflects both luminance AND colour (Think: red, green, and particularly blue spots).
Now that we've established that the two are different, let's talk about what digital noise actually IS.
It's the visible representation of random fluctuations which exist in ALL electronic signals as a result of many variables, relative to the strength of the signal.
Longer (BUT STIll relatively short) Answer:
Modern digital cameras typically rely upon CMOS sensors to receive and interpret light, which it sends as a series of electronic signals to encode into data format as a file. The thing about electronic signals is that they're subject to noise which, summed up, is a random fluctuation in electronic signals. It doesn't just apply to cameras though, it's pretty much anything with an electronic circuit. For example. if you apply that to audio you'd see it represented in thermal noise, or to put it in layman terms, "white noise."
However, it should be noted that if you're going to make that connection that there's a HUGE difference between distortion as it applies to audio and distortion as it applies to photography (they're not cousins – they don't even share the same species). In the same line there are various different types of noise in the various fields, both in how they present, as well as in their causes. For instance in photography one often sees the effects of electronic noise visibly reflected as in Random Noise, Fixed Pattern Noise (a.k.a. "Hot Pixel"), and Banding, whereas with audio you would hear Band Noise, and Correlated/Uncorrelated Noise in various forms. There are two categories that apply to noise in most fields, but that's not to say that they're entirely mutually exclusive, those types are external and internal. Examples of external would include atmospheric, or man-made noise, and examples of internal would include thermal, shot, and transit-time, etc.
So, noise is present at some level in all electronic signals, meaning all images have noise to some degree. Which is honestly fine for almost everyone ... until it isn't. There is obviously a threshold for what some people deem an acceptable amount of noise, and what others do, so we often rely upon an SNR (signal-to-noise ratio – also frequently abbreviated as “S/N”) to establish and quantify the results. Think of it as a scale whereby on paper we can more accurately assess the strength of a signal relative to the level of background noise and apply it in contexts that work for us. The signal, in photography contexts, is your camera's photosensitive cell responses to the light reflected or transmitted from your subject hitting your sensor. So, if the signal that is sent to your sensor is bright and strong, then your sensor's signal likely remains stronger than the noise and the signal more often wins-out. However in turn that means when the signal is weaker (think low-light situations) then noise tends to show up.
Thankfully as cameras and programs are pretty well developed and they compensate for noise, they tend to be better and better at resolving noise problems from generation to generation. To that end, certain types of noise are easier to counter than others. Take for instance Fixed Pattern Noise, as it is repeating it is easier for camera profiles to be developed that counteract the patterns. If Fixed Pattern were the only type that applied to photography then noise would likely be for the most-part a thing of the past, however it isn't.
It's a bit of a stretched simile I'm about to use, but bear with me here. On my old guitar pedalboard, I used a noise gate called the ISP Decimator to counter the electronic noise issues I was having with my audio signal as a result of ground loop (another cause of audio noise). This pedal has one single dial that adjusts the threshold, and one button that toggles the unit off and on. When the unit is active the threshold acts like a dam, blocking ALL signal from passing through like water gets stopped by a dam. Once the signal surges stronger than the threshold however, the gate is opened and everything flows through. Everything. Noise included. You can adjust the threshold so that when you're not playing guitar it mutes the signal, killing any buzz/hiss you might hear, but when you strum everything comes through the gate. The thing is though, assuming the SNR is adequate, the signal is strong enough to drown out or overpower the noise. If however there is too much noise, the signal gets muddied and lost in it, and it sounds horrible.
Now to bring it all back to photography terms, think of your ISO as a knob that controls the amount of noise your image signal has to contend with: The higher the ISO, the higher the noise. In photography the dam can't really be a dam because there's no value to blocking all the signal as that would be completely black, so by its nature it has to be more of a filter/strainer with adaptable openings. This filter in this metaphor represents the programming the engineers have integrated into your camera's processing. They design your camera to be actively adaptive to noise, filtering out things like fixed pattern noise, and doing their best to accommodate the other noise factors at play. But despite years of progress the signals in many areas such as the shadows/dark areas are very weak, and sometimes the intended signal is so low that the noise is stronger; with the filter openings wide open they let more through than you, as the operator, intended: noise. When the light in your environment is low, and you open the filters up wider by raising your ISO hoping to increase your signal by letting more light through, you also inadvertently let through noise. In the regions of the photo where the light is strong you won't see as much noise, but where there's less light... red, green, and particularly blue pixels will abound. Blue more so than the others because the blue channel is where it tends to manifest most often.
If your goal is to minimize the amount of camera noise, then I definitely recommend Lightroom's noise reduction filters. They're INSANE. But I should warn you to take it slow, and keep an eye out for how using the luminance filter can REALLY degrade your sharpness. Try watching this video for some feedback as to how best to navigate noise in Lightroom. In Photoshop you should take a glance at this article, it's very well written and functional (or watch this short video). If you're a Capture One Pro user, then check this one out!
Anyways, for an overview, I have to leave it somewhere, and I think that this is a pretty decent introduction to the subject. Keep in mind that to say that this is just the tip of the iceberg would be an insane understatement. Camera manufacturers employ teams of engineers to develop effective responses to digital noise, this blog post is more of a toe-in-the-water to the electronic field (pun intended). If you're looking to research the subject further may I suggest the book Noise in Electrical Circuits by F.N.H. Robinson, you can buy a copy from Amazon HERE.