We’ve all been there. Wandering around a city at night, and we come across a great composition. Unfortunately, although you’ve still got your camera with you, the tripod is still sitting in your hotel room as you weren’t planning on any serious shooting.
You turn on the camera and set up your shot, but to get a shake-free image even with your aperture wide open, your ISO is ramping up to 6400 or more. So, you’ll get the shot, but it’ll be horribly noisy. What to do? Take a look at the image below (it’s some 100% crops). The right hand side is virtually noise-free, compared to the right. What if I then told you that both of these images are at ISO 5000 handheld, with no noise-reduction applied in post?
How is this result achieved? The trick is to understand how noise works in an image. At higher ISOs, the noise forms a bigger proportion of the signal from the sensor than at low ISOs. This is why you get the grainy effect. Noise reduction software works by looking at the image and averaging out the high frequency variations in values which noise exhibits itself at- but this has the side effect of also getting rid of detail in the image if you turn it up too much, and why heavily noise-reduced images look soft.
Noise has one helpful feature though- it’s random in nature. So, by taking several images of the same scene at a high ISO and then combining them, we can increase the proportion of the image signal that is the thing we’re photographing, and the noise will average out. And that’s the technique used above.
How do we do this in practice?
This technique can’t be used in all instances – if the scene has people moving, trees blowing around or traffic speeding along, then these will obviously be in a different place in each frame. The effect may look quite interesting in some cases, so by all means give it a try, but you won’t end up with a static low-noise scene
The technique described here is commonly used in astrophotography to reduce noise, but as you can see from the above, can also be applied to other scenes.
We’ve all been there. Wandering around a city at night, and we come across a great composition. Unfortunately, although you’ve still got your camera...
This is a very interesting solution to an all to common problem. I will have to give this a try next time I’m faced with this situation. I have also been doing a bunch of reading on Expose To The Right, where you push your exposure up, intentionally over exposing your image, and then turn down the associated parameters later in post-production. You’d push the histogram plot to the right but not too far. You just want to get the most amount of pixel readings on the right half of your histogram without any major clipping of data. Noise primarily exists in the dark shadow areas of an image, so the lights can be much easier to deal with in post-production.
(note: I stumbled across your website because of your “review” of the 28mm f/2 fe. Great post!)
Thanks Paul, and sorry it’s taken over a year to reply! ETTR is a great solution for reducing noise if you can get a long enough exposure without blurring, just watch the highlights 🙂