Most photographers avoid using high ISOs in certain situations—when shooting macro images, or landscapes, for example. But if you ask the average photographer why she avoids those higher ISOs, you’ll probably get a fairly limited reply. “Noise,” or more generally speaking, “quality” are the reasons most people will give for avoiding high ISOs, but did you know that there’s more to quality than just noise?
[ Top image Loves His Mommy, Grainy B&W Edition by Flickr user Rich Renomeron]
It is true that the most noticeable quality issue you’ll see in high ISO photographs is digital noise, which is that sort of sandy or gritty look that images will take on, especially when the light is low. Noise can, in some photos, have an almost artistic quality to it, but you don’t usually want to see it crop up in images where your goal is to showcase detail.
What is ISO?
Before we start to talk about some of the problems associated with higher ISOs, it’s helpful to understand what ISO is and how it fits into the exposure triangle. The term “exposure triangle” refers to the three settings that factor into exposure: shutter speed, aperture and ISO. Shutter speed and aperture control how much light reaches your camera’s image sensor, while ISO controls your camera’s sensitivity to that light. At low ISOs, your camera is less sensitive to light; at high ISOs, it’s more sensitive to light. In other words, brighter conditions require lower ISO numbers, and darker conditions require higher ISO numbers. But the problem with using higher ISOs is that they can have an impact on image quality. At very high ISOs you may start to see digital noise, which can muddy details and make your image look soft.
What causes digital noise?
Digital noise looks similar to the grain you see in a photograph that was shot with high ISO film. Both grain and noise have that sandy or gritty appearance, but film grain is actually a physical thing. Film is made from light-sensitive silver halide crystals—millions of them, and in low ISO films they are small enough that they can’t be seen with the naked eye. But as the ISO increases, those crystals get larger, and in very high ISO film they start to become really obvious.
With digital images the visible result of high ISOs is similar, but the cause is completely different. Digital noise is feedback, which is actually a lot more similar to the static you hear when you turn the sound up on an amplifier than it is similar to those large silver halide crystals. And its appearance is somewhat different, too. Film grain gets larger as the ISO increases, but digital noise is always exactly one pixel. Instead of an increase in size, you get an increase in quantity. And while film grain has a neutral color, digital noise can appear in different colors, although it is typically most visible in the blue channel.
Cameras increase ISO by amplifying the signal, which has the effect of also amplifying digital noise. This is called the “signal-to-noise” ratio, and it’s what causes visible noise in a photo. An image shot at a higher ISO will have a lower signal-to-noise ratio, so there will be more visible noise in that image. An image shot at a lower ISO will have a higher signal-to-noise ratio, which means there will be less visible noise.
Digital noise is caused by imperfections in your camera’s image sensor. An image sensor is made up of millions of individual pixel sensor regions, which work by converting photons (light particles) into electrons, which are negatively charged subatomic particles that are then measured to determine the amount of light that they were exposed to. But the problem lies in human limitations—modern manufacturing techniques are actually incredibly accurate, but they’re not perfect, especially when you consider the complexities of trying to measure something as small as a photon. We haven’t figured out how to build pixels that are exactly the same size and shape across an entire image sensor, so there are always going to be differences at the molecular level, which means that two pixel sensors that appear identical and receive the same number of photons during an exposure may have slight structural differences that will cause inconsistencies in the number of electrons recorded. And the camera’s circuitry isn’t going to read the charge on each pixel with 100% accuracy, either. There may be electrical interference or thermal variance (temperature-related variables) that contribute to slight inaccuracies in each reading.
These problems exist in both good light and in poor light, but the difference lies in the number of photons that each pixel can “see” in any given lighting situation. When the light is bright, for example, a pixel might see roughly 400,000 photons, but when the light is poor it might only see 12,000 or so. If it’s only seeing 12,000 photons, then that number has to be amplified, which is a way to artificially increase the value so the information will be useful (that’s what happens when you boost the ISO). But what happens if there were a few errors in the reading? At lower ISOs it’s going to be a pretty small number compared to the total—there isn’t much difference between 400,000 and, say, 399,950. But if you’re amplifying the signal 32 times (which is what happens when you turn the ISO up from 100 to 3200), now you’ve got a difference of 1,600 photons instead of just 50, which is going to have a visible impact on quality.
Digital noise is the best-known artifact of high ISO photographs, because it’s the easiest to see. But it’s not the only side effect of shooting at higher ISOs. At very high ISOs you may also lose tonal range and color range, so you’ll have fewer, muddier-looking colors. You’ll also get a sort of flattening effect that can hide detail and make an image look two-dimensional. These are much more subtle problems than noise, which is obvious to pretty much everyone who sees it. To really start to notice that loss of quality, you need to look at the details, and compare your high ISO photos to similar low ISO images.
So what causes those less-obvious quality issues in high ISO photos?
Besides noise, the most obvious ISO-related quality problem you might notice in an image has to do with dynamic range. “Dynamic range” is a way of measuring the difference between the darkest part of a scene and the brightest part of a scene. An image with good dynamic range will have both a true black and a true white with a broad range of middle tones in between, while an image with poor dynamic range may only have a limited range of middle tones without any true blacks or whites. The amount of dynamic range you get with any individual camera is related to its ability to pick detail out of the shadows, and the amount of detail that a camera can pull out of the shadows is limited by image noise. So, the more noise you have in the shadows, the less dynamic range you’ll get, which is why higher ISOs tend to produce photos with decreased tonal and color range. So it follows that a camera that produces less noise when it amplifies the signal will have better dynamic range than a camera that has poorer high ISO performance.
If dynamic range is important to you (and it should be), it’s also worth noting that there are some very low ISO settings that can actually decrease dynamic range as well as decreasing digital noise. At ISO 50, your camera is actually just overexposing your photos by one stop, and then truncating the highlights after the fact. The end result is an increase in the light signal, or an overall decrease in dynamic range.
Now this is all complicated by variations between camera models. There are really two different types of noise—the type of noise that exists at the pixel level, or would be there regardless of which ISO you choose, and the noise that happens as a result of signal amplification (when you select that higher ISO). Pixel-level noise won’t really start to become visible until you raise the ISO, so at low to midrange ISOs you may only see noise in the shadows (depending on how well your camera performs). But as you increase the ISO even more, any noise that exists at the pixel level gets amplified right along with the signal, so you’re going to start to notice it outside of the shadows, in the brighter areas of the scene as well as the darker areas.
_DSC0734bw by Flickr user mathei
Most modern cameras don’t add a lot of noise at the pixel level, so at lower ISOs the differences in quality really have more to do with the ISO performance of the individual camera. A camera with good high ISO performance will not show noise in the shadows at low ISOs, and will show only limited noise in the shadows at higher ISOs. What this means is that you can underexpose an image at a lower ISO and then make post-processing adjustments to improve your exposure without having to worry about increasing the amount of visible noise in the photo. If you have a poor ISO performer, on the other hand, you’re going to see some limited noise in the shadows even at low ISOs, which means if you underexpose a photo and then try to correct the exposure in post-processing, you’re actually going to make the noise problem worse. So if you have a camera that has poorer ISO performance, you’re actually better off shooting at high ISOs than you are turning down the ISO and then attempting to brighten the image after the fact in post-processing. In other words, it’s often better to just err on the side of overexposing at higher ISOs than underexposing at lower ISOs.
How do you know?
You can do a simple test with your DSLR (or any other camera) to see where it falls on the spectrum of ISO performance. Place your subject in a fairly low-light situation (if you are shooting in bright light you may not be able to test the full range of your ISO without overexposing the last group of images). Mount your camera on a tripod and take a series of photographs at increasing ISOs, changing only the shutter speed, not the aperture (when you change the aperture, you’ll get differences in depth of field which can make it harder to pinpoint any quality issues). Now open each photo up on your computer screen and look at the details (use 100 percent magnification). Which images start to show visible noise? Can you also see a difference in tonal and color range? Which images show only a slight decrease in dynamic range, and in which images does it become obvious?
Light and shadows by Flickr user Paco CT
The results of this experiment will determine how you use your ISO setting. If your camera has poor high ISO performance, you should avoid underexposing your photos and then attempting to lighten them in post-processing, as this can exacerbate the noise that your camera naturally produces. A camera with better high ISO performance will give you greater flexibility.
Ultimately, it all comes back to noise—the amount of noise that your camera produces directly impacts the amount of dynamic range you’ll get at higher ISOs. Of course, this all has to be balanced out with your goals for the photos you shoot. Not every image has to have broad dynamic range or be completely noise-free, because the subject matter may not demand a lot of detail and color range. A black and white image is an obvious example of a photo where color range just doesn’t matter, and in certain genres even tonal range isn’t that important—street photography, for example, is more about people, “decisive moments” and that urban jungle than it is about capturing detail and tonal range. We tend to overlook noise and those less-obvious quality issues that are related to noise when the subject matter is more emotional or story-telling than it is visually beautiful, so you always have to keep your goals in mind when you decide to turn your ISO up or keep it low.
Most people think this post is Awesome. What do you think?