Technique: Noise Reduction Using… Noise?
- At April 07, 2014
- By Jay
- In Post-Processing, Technique, Tutorial
0

Digital camera and sensor manufacturers have made great strides over recent years in reducing image noise. Improved light gathering efficiency and processing has yielded dramatic improvements in high ISO performance to the point where one can use ISO 1600 and below with impunity, and generally ISO 3200 to 6400 with relatively little discomfort. In cases of the pro bodies and full frame sensors you can comfortably shoot in low light at insane levels from ISO 12,800 to 25,600 and expect good results (though you’ll pay a hefty price for this ability in the Nikon D4/D4s and Canon 1Dx). So why on earth would you want to add noise to an image in post? And, more paradoxically, use “artificial” noise to reduce image noise? Having been amazed by this technique for some time and recently working on new images/reprocessing old ones where it came into play it struck me as timely to post another quick tutorial.
Blue Noise
This is best to show by way of example. As good as these sensors are today, they tend to struggle in flat tonal areas and – for Nikon at least – blue skies in particular. Even properly exposed RAW images at base ISO will start to show blotchiness in the sky when pushed. For instance, photos like this are poster boys for noisy blue skies; there is absolutely nowhere to hide.
Here is a screen grab of a section in the upper left corner which represents the entire sky. This was shot at base ISO on a bright Cabo afternoon but with a polarizer to increase the contrast between blue sky and yellow architecture.
Lots of noise in an area that should be squeaky clean.
Noise Reduction
For my purposes I have found that Topaz DeNoise is my best tool for dealing with this type of noise. There are any number of plug-in or blurring solutions but here I’ll just focus on what a hit of DeNoise does to this sky. I have increased the contrast here to make the issue more visible online.
The noise has been nicely smoothed out. But a new problem has been created that is arguably worse: severe banding across the entire sky. All that blurring of the noise has also blurred tonal gradation; different shades of blue have been smushed together so now there are blocks/steps of tone rather than smooth transitions.
Adding Noise to Remove Banding
Here is where the incredible magic of adding noise comes in. For this I go to Photoshop’s Add Noise filter. Just by adding a minuscule amount of noise – the filter’s lowest setting and completely invisible on its own – those tonal gaps get filled and the banding is gone. Immediately. Invariably I’ll preview on and off several times as it still blows my mind how effective this is.
In most images it would be best at this point to mask out the noise reduction layer so only the blue sky is revealed and detail elsewhere is protected. Pick your selection poison. It’s worth mentioning for this type of image (clearly defined blue sky against contrasting yellow) creating a mask of the ‘b’ channel in Lab mode works brilliantly (short form: create temporary image duplicate for reference, change color mode to Lab, make a copy of ‘b’ channel, apply auto contrast, apply that channel as a mask back in original file). This is deserving of another tutorial but I would encourage anyone who is interested in channel masks, color correction, and the amazing power of the Lab mode to search out everything possible from Dan Margulis. He has forgotten more about this stuff than I could ever hope to know. Kelby Training (now KelbyOne) subscribers should watch his videos right now. But I digress. Here is a side-by-side review of the image states above…
High Contrast Black & White Noise
In color photographs you are relatively limited in how far you can push tones and saturation before things start looking gaudy. But in black & white interpretations all bets are off; there is much more creative latitude where you choose to place your tones. In fine art natural and urban landscape images this is often represented by very dark skies contrasted with bright clouds for separation and drama. The technique above can be an invaluable tool for fixing the damage in those areas caused by all of that tonal stretching.
A little goes a long way with the Add Noise filter. It won’t actually add visible noise, but will fill in those all-important tonal gaps to ensure clean gradations. Give it a whirl.