In previous tutorials I’ve
covered noise removal through various methods. Complete noise removal isn't always possible, though. Sometime we just have to accept some level of noise in our image. What level of noise is acceptable? What kinds of noise are acceptable? The following techniques won't rid your image of noise but they will change the way noise looks, and hopefully they'll give you an
overall more pleasing picture.
1. Noise Reduction on the RBG Channels
Here is the image we’ll use in our examples:
It’s incredibly noisy, and
genuinely so: I accidentally had the ISO cranked up on a bright day!
Assess the Image: Evaluate Noise on the RGB Channels
Most digital images are a composite of information held in Red, Green, and Blue channels. You can use the RGB channels to help you to see where exactly the noise is causing the most trouble in your picture. If you don’t have the channels window open in Photoshop, click Window and Channels.
You’ll see your picture
broken down into three channels: Red, Green and Blue. You want to evaluate each channel to look for noise. Here is each channel for my
The red channel is incredibly noisy, pretty much across the whole image.
Green isn’t so bad apart from the trees and riverbank.
Blue isn’t bad either; mostly the noise is in the water but you can’t see that as much due to the nature of the image.
Apply Noise Reduction on the RGB Channels
You can apply a filter on a particular channel. Taking my Red channel, which was the worst for visible noise, I selected the sky and applied a Surface Blur from the Filters list.
Surface blur will take larger areas of the same colour and blur it without ruining your edges too much, so you can try applying it to the whole picture. If it’s a particularly noisy image, like mine, applying this strength of blur to the whole image would soften the details considerably, despite only being on one channel.
So, rather than applying the blur filter to the particular channel, let's use a noise reduction filter instead. Hit Filter > Noise > Reduce Noise:
Then click Advanced and press the Per Channel tab:
You’ll see you can check
each channel for noise here too. The dialogue will allow you to click on the actual
image to see a particular section in the preview. You also have the option to Preserve Details, so you can experiment with moving the sliders to see which works
First, select the channel with the most noise. Now reduce noise using this process:
- Move the Strength slider to the right, past the level where noise reduction starts to detrimentally effect the image. The image will start to look smudged.
- Then move the slider slowly to the left, or use the up and down arrow keys, to gently reduce the filter Strength until you return to an acceptable image quality.
- Repeat this process with the Preserve Details slider to fine tune the filter.
Then do the same steps for the remaining two channels.
2. Reduce Noise in Lab Colour Mode
Let's process our example in another way. This method will tackle colour noise more directly while preserving as much resolution in the image as possible.
Hit Edit > Convert to Profile and choose Lab Colour:
You’ll notice in your Channels window that instead of RGB you now have Lightness, a and b (L, A, B… Lab!).
The different between Lab and RGB is basically (very basically!) that Lab is how we see colour as people, as opposed to how a machine or device might interpret colour, like your monitor.
The Lightness channel, as it suggests, contains the information about the luminance, or black and white information, of your image. The a channel holds green and magenta information and b yellow and blue.
This method is very simple: select the a or b channel and apply a blur filter. Because the a and b channels hold colour information only, not detail, you can tackle colour noise without taking away any of the detail in the image (held in the Lightness layer).
Often when we are disturbed by noise in an image what we are actually bothered by is colour noise. When you remove the colour noise you'll likely find that the luminance noise that remains is actually rather pleasing!
3. Add Noise to Hide Noise
When we think of noise we usually think of random noise. This is a very common kind of noise but it is not the only kind.
High ISO images suffer from all sorts of image degradation. Depending on the camera and the situation you might encounter loss of resolution, harsh and unnatural tonal jumps between bright and dark areas, chromatic aberration, colour shifts, banding, or reciprocity failure. Noise is any irregularity in the picture that is part of the capture but detracts or diminishes the image quality.
Of course there are methods we can use to improve these defects, but sometimes adding random noise is the fastest (and sometimes best) way to improve an image.
Plus, noise isn’t always a bad thing: it can add character and interest to a photograph. If your photo is already a little noisy why not add more grain to make it look purposeful?
This is a picture I took of
the Prague Funfair Orchestra and converted to black and white.
Assess the Image
definitely gave it more character and I think some grain would really add to it
too. This a fairly high contrast image, with deep blacks in the shadows of their pants (and the gate) and bright whites on the hats, but it still has lots of nice texture.
I want to keep that nice texture and the high dynamic range without crushing tones or losing detail. As I adjust contrast with this image I'm in danger of creating solids blocks of black or white. I particularly want to keep the smooth gradation in the hats, which have some parts that are very nearly completely white. A touch of noise will help avoid creating areas of solid tone and keep the smooth gradation that creates the three-dimensional look of the hats.
The easy and ‘quick fix’ way to add noise in Photoshop is to simply hit Filter > Noise > Add Noise.
Which, at 10%, will give this:
Alternative Method: Add a Noise Layer as Smart Object
If you’d like more control over what you’re adding though, try this method.
First, create a new layer:
I’ve named mine ‘Noise’. Choose Overlay or Soft-Light as the Mode and then tick Fill with Overlay-neutral colour which will say 50% grey for those modes.
Hit OK, then right-click your new layer and convert it to a Smart Object:
Then go to Filter > Noise
> Add Noise like before. Have monochromatic ticked if you’re using a colour
image and select Gaussian for more ‘random’ noise or Uniform for more ordered
I used 10% again so we can compare it to before:
Looks about equal, right? The effect is the same as hitting Add Noise but with the added
bonus that I can now adjust the Opacity if I wish, change the Blending Mode or
even add a mask and use a soft brush to erase certain parts of the added grain.
There are so many ways to reduce noise in Photoshop that it’s hard to know which to choose or which is right for your image. If you’re looking to reduce noise without applying something to your image as a whole, then using channels is definitely worth a try, particularly on images which are heavy in colour noise. Channel methods are more subtle and you may not find they’re what you’re looking for where images are very noisy on the whole.
If your image is really grainy and any kind of channel noise reduction is too subtle for you, then I’d recommend a bespoke piece of software or plugin for Photoshop such as Noiseware.
Remember though, not all noise is bad! Sometimes it can help fix other issues. And it can add depth and character to your photo, so don’t dismiss an image just because it’s grainy. You can always add more grain to give it that ‘film’ look.
Try a few of the methods to
see which works best and which you prefer, but if you find yourself getting
frustrated with the picture put it to one side and come back to it
later. Sometimes it’s hard to see the potential in something when we’ve worked
on it for too long. Better to use fresh eyes and decide if you’re going to
remove as much of the grain as possible, or add some more to make it something new
and different from what you imagined.
Envato Tuts+ tutorials are translated into other languages by our community members—you can be involved too!Translate this post