Bob Mc
05-18-2005, 12:59 PM
I wish I could take credit for uncovering the following information about the Unsharp Mask feature - but all I actually uncovered was an article I cut out and saved - within one of a pile of other articles I had save for further review. It's appearance was timely as I did my quarterly re-shuffling of the piles ... as I was using the feature ...and getting lousy results.
Full credit goes to Ed Meyers who wrote "Unmasking the Unsharp Mask" published in Digital Photo Pro magazine - November/December 2003 issue.
I don't want to rehash the whole article - nor should I. Mr Meyers included several illustrated examples and a lot of detail. I'll just try to summarize the major points I got from it.
The Unsharp Mask - now to be known from here on as USM.
What you are looking for is a crisp sharpening - high contrast and clean edges - without a halo
USM looks for contrast in your image by comparing each pixel to its neighboring pixels. It has 3 controls it uses to analyze and adjust the image. Amount, Radius and Threshold. It's important to know that these controls interact with each other to generate the adjustment.
AMOUNT control - higher % values mean a greater contrast is induced along the edges.
RADIUS control - effects how far out - in pixels - the USM effect will be applied.
THRESHOLD control - this is the value difference ( contrast/levels ) between adjacent pixels above which an adjustment to each pixel is made.
Example 1 - Threshold is set at 10. 2 adjacent pixels have values of 79 and 85 Since the difference between 79 and 85 is not 10 or more - the pixels are not changed.
Example 2 - Same Threshold setting (10). 2 other adjacent pixels have values of 85 and 140. Obviously this is a much bigger difference and certainly greater than the Threshold value of 10 so USM makes the light pixel lighter and the dark pixel darker - thereby accentuating the edge.
-------
"Haloing" effects seem to be caused by using too high a value for radius (how far out the effect is applied). You can often use high (e.g. 200% or more) for the Amount control if you keep the radius to a small value (e.g. 1 or even less)
As usual, there are many caveats to the process. It depends on the image, size of the image, the output characteristics (rgb printer, cmyk press, watercolor effect, etc)
Meyers makes several suggestions.
a. test, test and test some more - on the specific image
b. view your image at the pixel level - to judge the effect (double-click the magnifying glass in the tool bar to get there and double-click the hand to get back to the previous size.
c. after applying the USM (and probably test printing ) if the image is not right, go back to the unsharpened image and re-apply an adjusted USM - resist the temptation to sharpen an already sharpened image.
d. If you use USM in the full RGB mode (all 3 channels) color shifting and aliasing can occur. If you switch to the LAB mode - all the color information is in the A and B channels and NOT the Lightness/Luminance channel. USM applied to the Lightness channel alone reduces the aliasing and should eliminate color shifting (no color info in that channel).
d-1 A By-the-way - Meyers indicated that if your digital camera introduces a bit of color fringing, a slight Gaussian blur to the A or B color channels can reduce that fringing ( Filters > Blur > Gaussian Blur).
d -2 B Another BTW - Going back-and-forth often between Modes/Color Spaces may degrade the image so keep those Mode changes to a minimum.
e. Reducing the size of an image after applying USM is not as "good" as applying USM after the reduction. Reduction averages (in some way) pixels together, the USM effect will be averaged too. USM after image reduction - you get what you asked for (control wise)
----
Hope I've made USM a little clearer for everyone and to Ed Meyers - I hope I summarized the key points of you article correctly.
Remember - Test, Test, Test
Full credit goes to Ed Meyers who wrote "Unmasking the Unsharp Mask" published in Digital Photo Pro magazine - November/December 2003 issue.
I don't want to rehash the whole article - nor should I. Mr Meyers included several illustrated examples and a lot of detail. I'll just try to summarize the major points I got from it.
The Unsharp Mask - now to be known from here on as USM.
What you are looking for is a crisp sharpening - high contrast and clean edges - without a halo
USM looks for contrast in your image by comparing each pixel to its neighboring pixels. It has 3 controls it uses to analyze and adjust the image. Amount, Radius and Threshold. It's important to know that these controls interact with each other to generate the adjustment.
AMOUNT control - higher % values mean a greater contrast is induced along the edges.
RADIUS control - effects how far out - in pixels - the USM effect will be applied.
THRESHOLD control - this is the value difference ( contrast/levels ) between adjacent pixels above which an adjustment to each pixel is made.
Example 1 - Threshold is set at 10. 2 adjacent pixels have values of 79 and 85 Since the difference between 79 and 85 is not 10 or more - the pixels are not changed.
Example 2 - Same Threshold setting (10). 2 other adjacent pixels have values of 85 and 140. Obviously this is a much bigger difference and certainly greater than the Threshold value of 10 so USM makes the light pixel lighter and the dark pixel darker - thereby accentuating the edge.
-------
"Haloing" effects seem to be caused by using too high a value for radius (how far out the effect is applied). You can often use high (e.g. 200% or more) for the Amount control if you keep the radius to a small value (e.g. 1 or even less)
As usual, there are many caveats to the process. It depends on the image, size of the image, the output characteristics (rgb printer, cmyk press, watercolor effect, etc)
Meyers makes several suggestions.
a. test, test and test some more - on the specific image
b. view your image at the pixel level - to judge the effect (double-click the magnifying glass in the tool bar to get there and double-click the hand to get back to the previous size.
c. after applying the USM (and probably test printing ) if the image is not right, go back to the unsharpened image and re-apply an adjusted USM - resist the temptation to sharpen an already sharpened image.
d. If you use USM in the full RGB mode (all 3 channels) color shifting and aliasing can occur. If you switch to the LAB mode - all the color information is in the A and B channels and NOT the Lightness/Luminance channel. USM applied to the Lightness channel alone reduces the aliasing and should eliminate color shifting (no color info in that channel).
d-1 A By-the-way - Meyers indicated that if your digital camera introduces a bit of color fringing, a slight Gaussian blur to the A or B color channels can reduce that fringing ( Filters > Blur > Gaussian Blur).
d -2 B Another BTW - Going back-and-forth often between Modes/Color Spaces may degrade the image so keep those Mode changes to a minimum.
e. Reducing the size of an image after applying USM is not as "good" as applying USM after the reduction. Reduction averages (in some way) pixels together, the USM effect will be averaged too. USM after image reduction - you get what you asked for (control wise)
----
Hope I've made USM a little clearer for everyone and to Ed Meyers - I hope I summarized the key points of you article correctly.
Remember - Test, Test, Test