Publishers of technology books, eBooks, and videos for creative people

Home > Articles

Noise and sharpening

  • Print
  • + Share This
This chapter is from the book

Work in the lab

Theoretically, the Lab color space contains every color the human eye can see and the entire gamut of RGB and CMYK. It's an excellent color space for editing digital images, as it doesn't have some of the color limitations of the other color spaces. Lab color comes in quite handy when targeting noise, as it breaks down the image into specific criteria.

The Lab color space is comprised of a luminosity channel and two chromatic ranges, called a and b. The luminosity channel represents the black and white information in your image, as well as the lightness. The a channel basically contains your greens through red, while the b channel ranges from blue to yellow. Since digital noise is contained in the chromatic channels, we can edit the a and b channels while keeping our black and white detail intact. Otherwise, we'd have to adjust either a single color channel (Red, Green, or Blue) or the entire image, making it harder to preserve image detail while reducing noise.

Convert to Lab mode

The conversion to and from Lab mode is very clean. While going from RGB to CMYK can cause muddy colors, transforming your color space to Lab is quick and painless.

  1. Open the image you wish to work with in Photoshop.

  2. Select Image » Mode » Lab Color.

  3. Open the Channels palette by selecting Windows » Channels and view your Lab channels. Here, you can see the Lightness, a, and b channels, as well as a Lab channel that produces the full color composite image.

Adjust the channels

The first step to fixing our noisy image is to view the specific Lab channel we're targeting.

  1. Return to the Channels palette and click on the a channel to make it active.

  2. Press the tilde key (~) to see the full color image while keeping the a channel active. Now we can see the entire image, but the adjustments we make are only applied to the a channel.

  3. Select Filter » Blur » Gaussian Blur to open the Gaussian Blur dialog box.

  4. Select your blur radius. The amount of blurring you apply to your image depends on the amount of noise and the level of detail you're willing to surrender. Take advantage of the preview window to test out different blur values. Also, be aware that we'll be blurring the noisy pixels in the b channel, so don't go overboard when working on the a channel. In our image, the artifacts on the a channel are moderate, so we selected a blur radius of 3.9 pixels.

  5. Click OK to apply the filter.

  6. Return to the Channels palette and switch to the b channel.

  7. Open the Gaussian Blur dialog box again and apply a blur radius of 1.5 pixels. Remember how we said that the b channel contains your blues through yellows. The higher radius is more aggressive toward our image, but since this is where the most noticeable noise is, this aggressive approach is warranted.

Look unsharp!

Now that we've blurred our a and b channels, it's a good idea to sharpen up the details in the Lightness channel to counteract the blur and keep the image looking crisp overall.

  1. Select the Lightness channel.

  2. Open the Unsharp Mask dialog box by selecting Filter » Sharpen » Unsharp Mask. The Unsharp Mask dialog box also offers adjustable controls and a preview window.

For our image, we set the Amount slider to 125%, the radius to 1.3, and the Threshold to 6 levels. Again, each image is different, so be flexible and try different combinations to get the best result.

Our Problem and Solution images show RGB versions of an enlarged portion of our image before and after we applied our filters. We've reduced the noise without overly blurring the image and losing detail. By taking the time to adjust the image one channel at a time, we can keep the image looking good while minimizing the detrimental effects of noise.

  • + Share This
  • 🔖 Save To Your Account