January 5, 2014
Advanced sharpening techniques #3 ; Luminance Sharpening
Sharpening often produces unwanted colour shifts or artefacts, neither of which are desirable in a final print. Through a technique called luminance sharpening, these unwanted results of image sharpening can be avoided by targeting only luminance data.
A luminance sharpen will only apply itself to the black and white information in an image -- the detail; otherwise known as luminance. As a by product, there’s zero impact on colour; which reduces artifacts and weird colour shifts that other sharpening methods are prone too.
LAB vs RGB
Luminance sharpening can only happen in the LAB colour mode. LAB’s channel make-up has a critical difference to RGB, it separates out detail (lumiance) from color. LAB's channels break down as follows;
- L = luminance information
- A = contains the colour information for green and magenta
- B = provides the colour information for yellow and blue
By isolating luminance data in 1 channel alone, LAB allows us to target sharpening exclusively on black and white information. By contrast, RGB’s channels contain luminance and colour together in each channel, so it is more difficult to separate out.
Converting to LAB
To convert into LAB mode we simply go to Image -> Mode -> Lab Color
Now flick to the channels palette now, and you will see your new channels of L, A and B all selected.
Activating the Luminance Channel
Next we select the Lightness channel in order the sharpen directly onto the Channel, as opposed to the layer view as is common to the other sharpening methods covered so far like High Pass.
In your Channels Palette select the Luminance channel, your image will go greyscale -- revealing the ‘detail’ data.
Applying the Sharpen
Now we can actually start to apply a sharpen to the image. Go to Filter > Sharpen > Unsharp Mask to activate the Unsharp Mask dialogue (hereafter USM).
The options in Unsharp Mask look confusing, so if you are not familiar, here's a quick breakdown of what they all mean.
Amount = how light the lighter pixels get, and how dark the darker pixels get. When sharpening the image, the 'edge' or border between two colors USM makes light pixels lighter and dark pixels darker. Abuse amount and set it too high, your picture will look grainy and too contrasty.
Radius = the area that will be sharpened. A low radius means only the pixels right next to the edge will be sharpened. A high radius means a wider area will be sharpened. Setting the radius too high will porduce weird outlines or halos around your edges. Not good.
Threshold = how much contrast there needs to be between colors for them to be sharpened. A higher threshold means higher contrast areas will be sharpened, but low-contrast areas will not. Sharpening low-contrast areas (like a clouds) will introduce grain and speckle.
With your settings set to ‘taste’ hit OK. All done.
Now simply convert out of LAB by the same method as before in reverse.
Got to Image -> Mode -> RGB
For those who are uncomfortable with the LAB mode, or who see this as an extra workflow step, you can do an approximate method of this luminance sharpening in RGB. Personally, I do all my post-production in LAB anyway, so for those images where I use luminance sharpening it’s no issue.
Duplicate your final layer, then select the top layer. Apply the Unsharp mask filter as usual, intentionally over-sharpen a little.
Now select the sharpened layer and change the blending mode from Normal to Luminance. Then using the Opacity slider to reduce the sharpening layer to correct any undesired artifacts.