124. Remove Specks and Spots
Before You Begin
122 About Removing Scratches, Specks, and Holes
123 Remove Scratches
125 Repair Minor Tears, Scratches, Spots, and Stains
126 Repair Large Holes, Tears, and Missing Portions of a Photo
143 Fix a Flash Thats Too Far Away
Dust, scratches, and other "age spots" aren't the only kinds of damage a picture can suffer. Digital cameras and cell phone cameras can add their own kinds of spottiness in the form of CCD noise. This kind of noise, which occurs in a digital camera when you take a long exposure or overextend the digital zoom, takes the form of graininess or colored specks in areas that should be fairly uniform. Cell phone cameras introduce a certain level of noise naturally, simply because of the low quality of their images. Noise can also appear in scans of halftone images printed in a newspaper or magazine, in still images captured from video, or in images recorded under low-light conditions.
Three noise filters can help eliminate these specks. The Median, Despeckle, and Reduce Noise filters can easily remove the random pattern of dots caused by noise, regardless of the cause. Because it tends to average out the tones in a selection (changing pixels to the same average level of brightness as their neighbors), the Median filter is often a good choice when the noise affects large areas that are supposed to be even in tone. The Despeckle filter blurs pixels to smooth out areas of low contrast, leaving areas of high contrast (which are typically the edges of objects in a photo) untouched. The Despeckle filter is lousy at removing dust and speck type spots, but is good at removing low-contrast noise in a large area while preserving your edges.
CCD noise Random distortions introduced into a photo by the CCD chip in a digital camera, cell phone camera, or scannerits principal light detector. In a digital camera or cell phone camera, CCD noise during long exposures, occurs or at high ISO settings (film speed), and is magnified when electronic zoom is used to simulate a close-up of a subject at a distance. In a scanner, CCD noise happens naturally, but especially at low scanning resolutions.
The Reduce Noise filter again looks for pixels that contrast with their neighbors, and in a manner similar to Median, reduces their contrast by averaging out their brightness. Unlike Median, the Reduce Noise filter also evens out color by averaging the hue of each pixel with that of its neighbors. And like Despeckle, the Reduce Noise filter preserves the contrast along edges in your photo. After removing general noise (tiny specks that occur in a random pattern over a wide area), you'll learn how to use the Spot Healing Brush to remove spots that occur in isolated areas.
Open an Image
Open an image in the Editor in Standard Edit mode and save it in Photoshop (*.psd) format. Then remove noise using any of the three Noise filters.
Remove Noise with the Despeckle Filter
To remove low-contrast noise from a photo, choose Filter, Noise, Despeckle. Although you could make a selection first, because Despeckle works only on low-contrast areas to remove noise, your edges will be preserved so there's typically no need to isolate the filter with a selection. The Despeckle filter works automatically, so you'll be able to judge its effectiveness right away.
This photo was taken to record the new paint job in our bedroom. Despite the sunshine coming through the window, the resulting photo had a lot of noise throughout. The Despeckle filter did a pretty good job of removing this low-contrast noise. I applied the filter multiple times to remove the noise completely, and yet the edges were preserved.
To remove small spots in a specific area, it's typically best to use the Dust & Scratches filter rather than the Median filter because the Median filter tends to remove not only spots but detail as well, as it averages out the tone in a selection. The Dust & Scratches filter removes small spots in an area without removing detail; however, in some situations, such as areas of even tone that contain specks or spots caused by age, the Median filter can do a better job.
Or Remove Noise with the Median Filter
To apply the Median filter to an area of similar tone, first select the portion of the image you want to change. If necessary, on the Layers palette, change to the layer that contains the data to change. Then use your favorite selection tool to select the area that contains the noise you want to remove. Because the sky in this example is nearly a uniform color, I selected it with a few clicks of the Magic Wand tool (see 76 Select Areas of Similar Color). Zoom in on the image so that you can see your changes clearly. Choose Filter, Noise, Median. The Median dialog box appears.
Adjust the Radius value to a setting that removes the noise without removing too much natural texture. The Radius defines the area in which neighboring pixels are examined, to calculate an average value for the central pixel. Typically, a low value such as 2 or 3 is sufficient. Click OK to apply the filter.
The Median filter averages out the brightness of neighboring pixels, so make sure that you do not include high-contrast edges in your selection because you might lose that detail after applying the filter.
Enable the Preview check box in the Median dialog box so that you can see how the changes you make to the filter's settings affect the selected area in the image window.
I used the Magic Wand to select the sky in this nighttime photo of the Washington Monument. The low-light conditions caused the noise, but it was easily removed using a low Radius setting in the Median dialog box.
Or Remove Noise with the Reduce Noise Filter
The Reduce Noise filter combines the best of the Despeckle and Median filters, so it's a good one to try in their place. Choose Filter, Noise, Reduce Noise. The Reduce Noise dialog box appears.
First, lower the Preserve Details value or you won't see any effect. This value controls how much contrast a pixel must have with its neighbor before its brightness is lowered to bring it more in line with the "neighborhood average." Increase this value to preserve your edges; lower it to reduce noise even more. Strength controls the Median effect on qualifying pixelsin other words, the amount that a pixel's brightness might be changed. To change pixel hue values in a manner similar to the way the Strength value adjusts pixel brightness, increase the Reduce Color Noise value. The effect of your selections on the image appears in the large preview window on the left. When you're satisfied, click OK to apply the filter.
Remove Large Spots with the Spot Healing Brush
The Noise filters are great, but they do nothing to large spots on a digital image. The Dust & Scratches filter does a wonderful job of removing such spots, especially when they are grouped together, but you must be careful to apply the filter to a small area or you'll lose detail. To remove spots that are isolated or larger than a small dot, use the Spot Healing Brush.
You don't have to make a selection first; the effect is controlled by the size of your brush tip. Zoom in so that you can see the spot you want to remove, and then select the Spot Healing Brush on the Toolbox. On the Options bar, select a brush tip and adjust its Size to something slightly larger than the size of the spot you want to repair. Set the Type option to Proximity Match. This option analyzes the pixels around the edges of the brush to create a patch for the repair. Click the spot to remove it. Here, I returned to the 1940s photo and removed several large-to-medium spots with quick clicks of the Spot Healing Brush.
If the area the spot is in has a definite texture, you can replicate that texture to a degree and create a more convincing patch than with the Proximity Match option. Just enable the Create Texture option on the Options bar instead, which analyzes all the pixels under the brush tip for both color and tone, and then uses that sampling to create a similar pattern.
When you're satisfied with the results of the Noise filters and the Spot Healing Brush, make any other changes you want and save the PSD file. Then resave the file in JPEG or TIFF format, leaving your PSD image with its layers (if any) intact so that you can return at a later time and make different adjustments if you want.