Noise is a result of how the digital sensor capture process deals with variables in light levels. The noise is caused by unwanted signal information on the camera’s sensor. Each sensor has a base level ISO and when the sensor needs to become “more sensitive” to light, the multiplied values of capture result in the introduction of artifacts seen as noise. The amount of noise generated depends on several factors including pixel size (larger pixels create less noise), exposure settings, and image content. Noise is also more pronounced in the darker parts of an image because the brighter parts of an image have a higher signal-to-noise ratio. 

For digital photography, most noise is combined into two elements: luminance and chroma. Luminance noise is grayscale. This type of noise has an impact on the brightness values of the pixels but has no real impact on the color. This type of noise is analogous to film grain. If you haven’t ever seen film grain, it sort of looks like old television static across the image or salt and pepper tossed across the entire image. 

Chroma noise shows up as various odd and random colored pixels scattered across the image. In many cases, the pixels appear as red/magenta and green. I often think of them as holiday lights. Chroma noise often looks worse and can render an image unusable, because of the color contamination. 

Because there are a number of factors that impact how visible noise might be in your image, you may need to use noise reduction software to remove the unwanted artifacts. In most cases, on modern cameras, images shot at 800, 1600, and higher ISOs will need some form of noise reduction. Your goal is to remove as much of the noise as possible while still preserving the details in the image and avoid any unwanted color shifts or artifacts. The amount of noise and the ratio of luminance and chroma can vary from camera to camera and setting to setting. 

Noise reduction is the result of slightly blurring areas of the image. In some cases, noise helps with the perceived sharpness within an image. There are some instances where you might introduce noise to increase perceived sharpness. It is important to keep both factors in mind as you start to work on your noise reduction strategy.  

When an image has significant levels of noise, it is almost impossible to remove all the noise and still preserve sufficient details or avoid a synthetic looking image. As good as modern software is at dealing with noise and leaving the image in a good working condition, in most cases you will not be able to remove all the noise without introducing some level of unwanted artifact (loss of detail being the most common).

As you develop your overall workflow, you should apply noise reduction before applying any type of sharpening to have the right balance between noise reduction and sharpening. Each image also will have the right balance of noise reduction, so while you can have a general starting point that is fairly consistent from image to image with a given ISO/camera combination, it is important to make minor adjustments to get the best image possible.

In printing, it is also important to note that the level of noise reduction will be different based on the size of the print, substrate, and the type of frequency data that dominates the file. An 8×10 will require different options than say 20×30. A high-frequency image will require different settings than a low-frequency image. As you begin to work with your noise reduction, make sure that you are properly tuning your software settings for the size and type of print you are making. 

Lightroom Noise Reduction

The noise reduction tools in Lightroom can be found under the Detail panel in the Develop module. Noise reduction is used to remove the two types of noise. 

When working on noise removal, I recommend that you view the image at a 1:1 ratio (100%) so that you can see the impact of the noise removal process. It is important not to be overly aggressive in the removal of the noise so that you can preserve an appropriate level of detail in the image. After you think you have the appropriate level, it is important to check over the other parts of your image to ensure that you didn’t inadvertently create a negative impact in other parts of the photograph.

To start your noise reduction, from the Navigation panel, zoom in on the image at 1:1 (100% magnification) and open the Detail panel in the Develop module.  

If you have any chroma noise, you will want to start by removing it first. Although it doesn’t necessarily matter which you remove first, chroma noise is often easier to remove and less destructive to the file. 

In the Details panel, move the Color slider to the right until you are at the point where the color noise is removed. If you need to refine your edit, use the Color Detail slider to set the threshold for noise levels. The higher the value, the more details that are protected, but you might see more speckled colors. The lower the number, the more colored speckles will be removed, but colors can cross-contaminate. Somewhere toward the middle is often a good place to start. The smoothness slider can be used to smooth out color shifts, particularly in low-frequency areas of an image. 

After you have tackled the chroma noise, you can move on to dealing with the luminance noise. Move the Luminance slider to the right to reduce the levels of noise in the image. You might often need to adjust the other two sliders under Luminance as well. Luminance Detail controls the detail preservation in the image. The higher the value, the more details get saved, but it also leaves in more noise. The lower the value, the fewer details are preserved, but the noise is more strongly reduced.  

Luminance Contrast is used to help preserve the overall contrast that might be lost from the reduction in noise. The higher the value, the more contrast is saved. But, you leave in more noise and increase the risk of producing artifacts. The lower the value, the less noise, and the smoother the transitions become, but the image might also suffer from a loss of detail and color. 

You can hold down the ALT (PC) or Option (Mac) modifier key when using the Detail and Contrast sliders, and the image will turn grayscale, allowing you to focus on just the impact of the noise reduction and not be distracted by the color information.

You can also use the adjustment brush and do selective noise reduction on a part of an image if you don’t feel that you need to apply noise reduction to the entire image.

Photoshop

I would recommend using the noise reduction in Camera RAW to deal with noise in a Photoshop document. The basic workflow from the Lightroom section above is directly applicable to working in Camera RAW. 

There is a Reduce Noise filter in Photoshop that can be used to help deal with noise, although in most cases it is not as effective as the Camera RAW filter. The basic method of reducing noise in is to use the Reduce Noise filter from the Filter Noise menu. When you open the filter you will see a Reduce Noise dialog box. In the dialog box, the first two sliders, Strength and Preserve Details are used to remove luminance noise. The Reduce Color Noise slider reduces color noise. Sharpen Details is used to apply sharpening during noise reduction. I would advise using a different sharpening method and not this option. Keep this filter focused on noise reduction. The final option is to Remove JPEG Artifact option which is used solely to reduce jpeg compression artifacts.

I would start with the Color Noise slider and set the value to 0%. Then slowly drag the slider to the right until you see the color noise disappear and stop. There is no benefit to moving the slider any higher than necessary. 

Once you have completed your work on the color noise it is time to move to Luminance noise. Set both the Strength and Preserve Details sliders to 0 and begin with the Strength slider. Move it to the left until you see most of the noise disappear from the image. You will not remove all the noise. Remember this filter is called Reduce noise so some will be left behind if you want a decent looking image. Once you have removed most of the noise, adjust the preserve detail slider to the right. This will begin to bring details back into the frame AND increase the noise. The goal here is to find a balance between Strength and Preserve Details that give you a good image with a minimal amount of noise. 

If you select the Advanced Option radial button, you will now see a tab Per Channel appear. This will allow you to see each individual channel of the image (Red, Green, and Blue) and adjust the channels individually for noise reduction. In most cases, there will be considerably more noise in one channel than the others. This will let you more aggressively deal with the noise in the bad channel and allow the other channels to help hold details and structures in the file. In most cases, the blue channel is the most problematic. 

Image stacking and median mode option

One of the best methods for reducing or eliminating noise in an image is to use Image stacking and the median stack mode. To do this method you will need at least two shots, but I recommend three-five, of the same scene. Noise is random in an image and when you have several shots of the same scene, you can use the data from the various files to eliminate the noise. 

  1. Open all your images into a single Photoshop file as layers.
  2. Align all the images;  using Edit/Align layers using the Auto option.
  3. Select all the layers and then merge to a Smart Object using Layer/Smart Objects/Convert to Smart Objects.
  4. From the Layer menu, Select/Smart Objects/Stack Mode/Median.

You should then be able to see the effects of noise reduction in the image from the median mode option. This is one of the most effective methods to reduce noise and if you are working with still life, landscape, or have an opportunity to make multiple exposures this is a great option. 

Third Party Tools

There are a number of third-party tools that do a great job with noise reduction. I recommend Noiseware, Topaz DenoiseAI or Dfine from DxOMark’s Nik suite.