Unsharp Mask

[RichardtheSane]

Hi all,

I've been taking a few pictures with my 10d (as I've already posted) and now I am learning a lot about processing in PS7 but unsharp mask is a bit of a mystery to me. I have been using it sucessfully after reading a couple of tips online, but I am only following steps that someone else has suggested.
Can anyone explain what the the settings in unsharp mask are for, and what they do to the image.

Thanks in advance

Richard

[Roger_Cavanagh]

Richard,

It's not been updated recently, but this page has some info and links:

http://www.rogercavanagh.com/helpinfo/03_sharpening.htm

Regards,

[rdenney]

RichardtheSane wrote:
Hi all,

I've been taking a few pictures with my 10d (as I've already posted) and now I am learning a lot about processing in PS7 but unsharp mask is a bit of a mystery to me. I have been using it sucessfully after reading a couple of tips online, but I am only following steps that someone else has suggested.
Can anyone explain what the the settings in unsharp mask are for, and what they do to the image.

Thanks in advance

Richard

Unsharp masking is the digital equivalent of an old manual process. In that process, you'd expose black-and-white copy film as a slightly offset contact print from a slide. The offset renders all the lines a bit fuzzy--hence the "unsharp" part. It provides a gray line along all sharp edges. You then print through a sandwich of the transparency and the mask, and it brightens the bright side of edges and darkens the dark side of edges.

The digital version behaves similarly. It looks for an edge, and brightens the pixels on the bright side a bit, and darkens the pixels on the dark side a bit. This doesn't make the image sharper, but it makes it appear sharper by rendering the edge with more contrast.

You control how PS detects the edge with the USM controls. The threshold establishes the minimum difference (in average of RGB on the range of 0-255) between pixels on both sides of the edge before PS will consider it to be an edge. I usually use a value in the 8-10 range, but I choose it subjectively. You want a value that doesn't find edges in smooth gradients, or that doesn't create effects that look like posterization.

The edge itself can be fuzzy, and that's what pixel radius controls. If you want to sharpen edges that are fuzzy, choose a larger value. Mostly, though, I want to only sharpen edges that are already sharp but a bit too smoothly rendered (which is a Canon hallmark apparently, and I like it--it's easy to apply USM but it's really hard to undo it). Therefore, I usually choose a radius smaller than a pixel--around .6 or .8--so that only hard edges in the image will be sharpened. If the image is a bit out of focus, you can make it look a bit sharper by choosing a larger radius, but you can't get something for nothing, and this usually only works for lo-res images intended for web display.

Once you have defined how PS will find edges, then you can determine how far it will go with them. The "amount" setting controls how much the bright side is brightened and the dark side is darkened. I adjust it so that the brightness of the edge pixels are no brighter than the pixels away from the edge. That bright edge around everything is the hallmark of oversharpened images, in my view. I usually end up with a value of between 75 and 125, depending on the target (less for prints, more for web; less for higher resolutions, and more for images that I've downsampled).

Remember that USM is a targeting activity--it is done specifically to tailor the image for a specific output device. In my workflow, I don't do it at all, or perhaps only a tiny bit to expose dust, on the orignal image that I'm correcting to look good on the screen. When I prepare it for display, which I call targeting, I apply USM to the degree needed by the output device. Thus, in my workflow, I always correct color, curves, and saturation, then I do just a bit of sharpening to make dust appear (on scanned negatives--not a problem with the 10D) and then I clone out the dust. Then I store the corrected image. When I prepare it for printing or for web display, I resample it to the desired size, crop, correct color for that device using PS's preview, check for gamut errors, and then sharpen again at the final resolution.

Rick "who thinks a bit of practice with scanned negatives makes a 10D mighty appealing" Denney

[Conk]

Very informative Rick. Thanks very much for that info. Also Roger. I'm guilty of not visiting your site enough. Too much valuble information to miss.
I think as I get more time under my belt with digital editing the information I passed over yesterday is a definate must today.