Speck Removal Tool

The speck removal tool lets you remove minor imperfections in the image by interpolating image data from the surrounding area A common application is removing small dark circles from clear blue skies caused by bits of dust on the camera sensor.

You have the option of removing specks lighter or darker than their surrounding neighborhood, or both.

To define speck points, shift-click on the input image at the center of each speck. This displays an overlay on the input image consisting of a circle surrounding the speck as illustrated below.

The circle indicates the area that will be replaced. For best results, the speck should be just enclosed in the circle. You can adjust the speck radius using either the Speck Radius slider or using the mouse scroll wheel. For large images, it is more efficient to use the slider. The current speck is indicated by a cross displayed at its center -- all others have a small circle at the center instead.

You can drag the center of a speck point to adjust its location. To remove a speck point, Ctrl-click on it.

When you click near a speck point, the Speck Radius and Remove setting are updated to reflect the properties of the speck point. By the same token, adjusting the Speck Radius or Remove setting set the properties of the current speck point, if any.

Amount

The amount control lets you scale back the overall speck removal or you can create a mask to restrict speck removal to specific parts of the image.

Speck Radius

This slider lets you control the radius for the current speck in pixels. You can also vary the speck radius with the mouse scroll wheel.

Speck Strength

This slider lets you control the what percentage of the current speck is removed. The default value is 100%, but if, for example, you are using this tool to reduce reflections in a portrait, you can lower the strength so it leaves part of the highlight but makes it less obvious.

Speck Softness

This slider lets you control how the speck removal patch is blended with the input image around its edges. At 0%, the transition is abrupt -- at 100% the edges are heavily blended.

Remove

This control lets you select whether you remove the current speck if it is lighter than the background, darker than the background or both. If the speck you are removing is consistently darker or lighter than its background, using the corresponding remove mode will produce fewer artifacts.

Method

There are two different speck removal methods. Depending on the image, one may work better than the other.

Patch Match

This method is like cloning over the speck. It first searches around the edge of the speck for the adjacent part of the image that best matches the area around the speck. Having found the best match, it then copies the image data over the speck to remove it. The edges of the patch are blended smoothly to make it less obvious where it has been applied. This method usually works better when the background behind the speck is textured or when the speck is relatively large. In some cases, it can even be used to remove relatively small, unwanted objects from an image.

Directional Interpolation

This method interpolates image data across the speck in the direction that has the best match from one side of the patch to the other. This method usually works better when the background behind the speck is smooth or has a linear gradient and the speck is small.

Four Way Interpolation

This method interpolates image data across the speck both horizontally and vertically. This method usually works well against smooth backgrounds, but can produce artifacts if the radius is too large.

Shrink

This method warps the image around the speck to remove it. This method usually works well against smooth backgrounds, but can produce artifacts if the radius is too large.

Tips

The more tightly you can circle each speck, the better the result. While using a larger radius makes it easier to click on specks, a smaller radius produces fewer artifacts.

When using the Directional  or Four Way Interpolation method, the speck removal tool replaces the missing image data with a smooth patch whose color matches the average of the surrounding pixels. Thus, it works best when the background behind the speck is smooth. If the background behind the speck is highly textured, the Patch Match method will usually produce better results. Or, use the Clone tool, assuming you can find a matching patch of image to copy over the defect. It also helps to use the Remove setting so only the lighter or darker parts of the affected area are modified, as this helps preserve some of the background texture as well. Since specks don’t show up as well against textured backgrounds, this is usually not a serious problem.

The speck removal tool may not work as well at the edges of the image if it does not have image data around the speck to work with.

If you are using a split screen display, the results of setting speck removal points are shown on the output image. You can zoom in on both the input and output images if necessary to see more detail.