Typical UnSharp Mask Attributes

keoj

Active Member
I know that many of you used scanners (as I have started to do) for negative and film digitization. The workflow that people use (usually with Photoshop) is generally pretty complicated with color correction but ultimately everyone that I know of uses unsharp mask in the final phases. Just curious, what do people use for the attributes typically.

I use Amount = 80 to 120, Radius = 3-4, Threshold = 8 to 10

Just curious, I'm self taught and I wnat to see if I'm in right zone or am I going overboard.

Joe
 

gwelland

Member
Take a look at the latest Scott Kelby book Photoshop for Digital Photographers - it has a great section on typical USM settings. He also covers quite a few other sharpening techniques too. I'd give you some of the settings here but I don't have my book with me at the moment.

Personally I use NikSharpenerPro and I find that it is outstanding for my needs. I used to use custom actions for USM/edge sharpening etc but I find that Nik does exactly what I want depending upon whether I'm printing or producing a web images. Highly, highly recommended.
 

gwelland

Member
Your settings will really depend upon the contents of the image and the size/media that you are going to print on.

One thing to be aware of is that you'll be surprised how much sharpening you can apply to an image to make it look correct on your printer. Don't worry about making it look great on screen, sharpen against the print itself - you may find that the screen image looks oversharpened when you do this.
 

swoolf

Well-Known Member
Their are lots of web-based tutorials on this , and , almost as many opinions on the best way to sharpen! Basically one size does not fit all with USM , as someone else mentioned , the parameters should fit the picture . Generally , the busier and more detailed the picture , the smaller the radius and the higher the amount , with little or no threshold . Conversely , for something like a portrait where you wouldnt want to see every pore on a models skin , you could use more radius and threshold but less "amount" . You could of course also just select , say , just the eyes and sharpen those leaving the rest as is , i.e. slightly "soft" . Sometimes , with B&W especially , I'll actually use a bit of Gaussian blur on the sky to lose some of the grain which can ocassionally be quite intrusive , and sharpen the rest of the pic . Like I said , horses for courses - hope that helps . Steve
 
>

Another trick is to not sharpen the blue channel. The blue in an RGB image or the cyan in a CMYK image generally has more grain and more nasty digital artifacts than the other channels - things you don't want to bring out by sharpening them.

You can sharpen each channel individually, allowing to to sharpen one channel with clothing or hair detail while not sharpening another channel where the skin texture is.

If you have a color portrait of a man, sharpening the blue will often over-emphasize beard stubble and make him look unshaven. Much of the detail in skin tones will be in the green channel - sharpening the green will bring out skin texture. A blonde's hair detail will mostly be in the blue, while her hair might look almost white and have very little detail in the red channel. I've always found it very interesting to look at each channel to see what's going on in the image.

If you're making a black and white from a color image, look at the channels instead of simply desaturating or converting to grayscale. If you look at the channels individually, you're seeing the image through a red, green and blue filter. (It's best to set your Photoshop preferences so that you see the channels in grayscale, not in their actual colors.) The sky will be very dark and have very bright clouds (but will not have much detail in the clouds) in the red channel, for instance, while the green channel will show lots of fine cloud detail. The blue channel will have the lightest sky and very little cloud detail - sometimes it may have an almost blank sky. This gives you a tremendous range of choices. You can go to each channel, copy it, then go back to RGB and paste one or more channels as new layers, then apply different amounts of transparency to them to blend them together.

Photos of women using just the red channel by itself can be extremely flattering - they make the skin look very smooth and clear, and can have a very sophisticated, fashion-photo look, especially if you increase the contrast slightly. Most blemishes in the skin will be almost completely invisible in the red. The green channel, on the other hand, will have darker skin tones and show lots of fine detail. A portrait of a man might use mostly the green channel.

I often use the red channel combined with just a tiny bit of the green for BW photos of women, and throw away the blue. Depending on hair color, you might need to use hair highlights from another channel. By copying the three channels (or any one or two of them) and pasting them as separate layers, you can combine them in different strengths to get exactly the image you want, or make alterations to each one. You can also make multiple copies of the same channel in different layers, and, for instance, multiply a channel onto itself to bring out highlight or shadow detail, which might be destroyed if you tried to do the same thing with curves.

If you keep your file in layers, you can easily fine-tune the tones in your image by changing the layer settings and opacities. You can also use layer masks to use just hair detail, for instance, from one channel and skin tones from another.

As far as color correction, the book PROFESSIONAL PHOTOSHOP by Dan Margulis (now in its 4th edition) is by far the best book I've used. If you read and absorb what's in there, you'll be able to not only color correct your photos, but to analyze them in a very sophisticated way. It's not a simple book, but it's well worth taking the time to understand what's in it. One of the most valuable things I've learned from it is the concept of finding unimportant tones in the image, and pushing them somewhere else to gain dynamic range for the important areas. (So, if you have a large range of middle tones in an unimportant area in the background, you could squeeze these out and gain more range to avoid blowing out highlights somewhere else.)

Margulis recommends a different approach from what I use for converting to BW; he simply manipulates the different channels in the color image before converting it to grayscale or desaturating, sometimes swapping channels around. Knowing how Photoshop creates a grayscale image (it uses more of the green, less red and very little blue) lets you tweak the color image to get the BW tones you want.
 

jsmisc

Well-Known Member
Dear Paul,
Thank you very much for that very helpful tutorial on sharpening. I am struggling with Photoshop and finding it to be very complicated. Your summary is very useful indeed.
John
 
Top