What is noise? What is ISO?
All astrophotographers have one enemy in common…noise! Noise is created for many reasons, but the result is the same. It degrades your image quality. Through this post, I will try to describe what happens in your camera and debunk a few misconceptions about noise and ISO. I’ll try to present my statements simply but not in a too confusing way. Let’s start!
What is noise?
To begin with, noise exists in any type of signal and information. The sensor in your camera collects light, which is then translated into a data as a RAW image. There are 3 main sources of noise that add up to your final image:
- Shot noise: the noise that is added to the signal as it travels from the source to your sensor. It is mainly caused of particles that exist in the air, clouds or even Earth’s atmosphere work as an obstacle between the source and the camera sensor.
- On-chip/Read noise: the noise that is created when photons hit every pixel on your sensor. Many times the sensor fails to translate a photon into data, so noise is produced.
- Off-chip noise: while the data are transferred to the memory card to create a RAW file, the signal travels through the circuits of your camera. Due the law of thermodynamics, part of the signal is lost and converted into heat (e.g when your camera heats up while using it for a long time or when you shoot on a warm environment). This results in loss of information and therefore the production noise.
What is ISO?
ISO stands for International Organization of Standardization. Well, that’s not very informative…..
An amplifier exists on your camera that can boost the exposure of your image, allowing to compensate accordingly by increasing the ISO values. Furthermore, when you double the ISO, you get 1 full stop of exposure. For example, moving from ISO 400 to ISO 800, it’s like gaining 1 stop of exposure.
“So, ISO can increase the amount of light on your sensor? Does it make my sensor more light sensitive?”
Light sensitivity relies on your sensor's quantum effiency and can’t be changed. Some cameras have a more light-sensitive sensor than others, which means that they’ll perform better in low light conditions. Also, remember that ISO refers to the amplification that happens to the original signal. That means, that a good signal will be boosted to a good image and a bad signal into a bad one.
“But when I increase the ISO, I see more noise!”
That’s true because when you increase ISO, you increase the total brightness of your image, revealing the “noisy” pixels that exist underneath. It’s like when you increase the exposure slider on Lightroom and you see all that crappy noise on the shadows. I must insert a very basic math expression to justify my statements and that’s signal-to-noise ratio (SNR).
SNR = (POWER OF SIGNAL) / (POWER OF NOISE)
Higher SNR means a better quality image. By increasing SNR you make noise less visible. For a more detailed view on this matter, you can watch the fabulous video by Adrien Mauduit .
“So whatever ISO i put, the noise remains the same?”
This is partially true. I have to introduce another term here, ISO Invariance. The main concept is that the sensor's noise levels remain the same in some ISO value. E.g a camera can become ISO invariant from ISO 800 which means, that from that value and on the noise remains the same.
“If my camera is ISO invariant from ISO 1600 and on, why not always choose ISO 6400?”
Your camera may be ISO invariant, but ISO has a major drawback. It decreases the dynamic range of your image. You may have witnessed this when you put a high ISO and the distant lights are blown out. That's why you lose any details on the whites. With that in mind, you want to choose an ISO value on the "invariant zone" that does not blow up you highlights. A case that this can be extremely useful is at aurora photography, where aurorae are extremely bright or when you have city lights on your frame.
“What controls noise?”
At the end of the story, noise is mainly controlled by the amount of light that you introduce at your sensor which leads to aperture and shutter speed. By doing longer exposures or increasing the aperture, you give more signal to your sensor..so less noise and better quality.
I did a little experiment with my Canon EOS Ra where you can see the ISO invariance. From the results, Ra is ISO invariance from ISO 800 and above. However, you can clearly see that the details on highlights are lost at extreme high ISO due to low DR.
- I cropped the image to get a closer look at the noise levels.
- In every image I keep the shutter speed and aperture the same. The only thing that changes is the ISO.
- I only fixed the exposure, WB and removed any sharpening and noise reduction, in order to be close to RAW files.
- Note that ISO patterns are random, as photons hit the sensor in a random way too. So noise is random across the image.
The next two experiments below show that by using a small aperture and a high shutter speed, you decrease the light gathering on the sensor and therefore increase the total noise. Check the examples below that prove my point.
Changing the shutter speed (same ISO and aperture)
Changing the aperture (same ISO and shutter speed)
For astrophtography, try to take a few images with different full-stop ISO (400, 800, 1600, 3200, 6400, 12800 etc). Then, load them in Lightroom or any other post-processing software and fix the exposure to all images in order to have a relative same histogram. Finally, compare the images and try to find which ISO value keeps your DR and colors in a good level.
Ways to reduce noise levels
- Take images on a cool environment in order to reduce heat to your sensor and eventually noise on your images.
- Use a fast lens, f/2.8 or wider to capture more light and eliminate noise.
- Use longer exposures for the same reason. If you want to avoid star trails, you have to use a star tracker that rotates on an axis opposite to Earth's and lets you take minutes of exposure.
- Take 10-12 consecutive images with the same settings and stack them in Photoshop or Sequator (free) to reduce noise effectively.
- Use calibration frames. For a detailed information, check this article.
- All the above! Do you want the best result? You have to combine all the above methods and get minimal noise and the best signal on your images.