PDA

View Full Version : Sharpening vs. Unsharp!?


downtrodden
04-02-2008, 02:22 PM
Ok guys so I have GIMP, and the help files won't work and there's something i've always wondered...

What the hell is the difference between Sharpening and Unsharp mask and why does something that's UNSHARP, make the picture SHARPER than Sharpening it!?!? :eek: Seriously...

And why do they call it Unsharp if it sharpens images... Shouldn't it add blur!!? :D:D:D

griptape
04-02-2008, 02:42 PM
It actually does add blur... and then from that blurry image, it detects edges, and sharpens them. It compares the blurry image it just made to the original image, and creates a mask with harder edges and tighter pixels to lay over that original image (the same way you would wear a literal mask on your face).

toriaj
04-02-2008, 10:58 PM
That's the best explanation I've ever heard, thanks :)

cdifoto
04-02-2008, 11:03 PM
Check this out:

http://www.luminous-landscape.com/tutorials/understanding-series/understanding-usm.shtml

as well as this:

http://www.luminous-landscape.com/tutorials/contrast-enhancement.shtml

downtrodden
04-03-2008, 12:55 AM
thanks Grip and thanks CDI for the articles, that definitely explains a lot!

downtrodden
04-03-2008, 01:00 AM
One question though, is there any particular time that one would be preferable over the other? I know when I use USM, the results look more.. authentic than using sharpening, but maybe i'm just getting lucky?

cdifoto
04-03-2008, 01:21 AM
Not for me. I always use USM.

Dread Pirate Roberts
04-03-2008, 01:46 AM
Simple rule of thumb really - Always use USM.

Try varying the AMOUNT and RADIUS when you use USM. If you use too much you'll get a prounounced halo around the edges though. Unfortunately it varies from image to image.

Oh one last thing the amount of USM also varies with image resolution (number of pixels) so do it on the final image when you know what size it will be. Eg a 10MPixel image on screen will tolerate a different amount of USM from a 1MP image on screen or a 10MP image printed as a poster.

Try and avoid messing with the USM threshold setting (leave it at 0) unless you have a lot of noise in the image.

downtrodden
04-03-2008, 03:26 AM
sweet... thanks all for the info!

AlexMonro
04-03-2008, 03:43 AM
Another way of using USM is for local contrast enhancement, which can give a flat picture taken on a dull day a bit more punch. Use USM with a small amount (maybe a tenth of usual for normal sharpening) and a large radius (try around 50 pixels). As usual with any USM, if it's overdone it can look hideous, but with a light touch it can bring a dead shot back to life, and without the risk of blowing highlights and shadows that can happen with a normal S-curve contrast boost.

Visual Reality
04-03-2008, 03:48 AM
I stick with 300-400, at .2, 0 threshold. Always one of my very last steps. You won't get the same results if you do it early in the process, especially before resizing.

erichlund
04-03-2008, 07:37 AM
I stick with 300-400, at .2, 0 threshold. Always one of my very last steps. You won't get the same results if you do it early in the process, especially before resizing.

This last, of course, is very software specific - for instance, the strength scale on CaptureNX only goes to 100, where Adobe products go to 500.

One thing I do find is that images destined for print can take more sharpening then images destined for the web.

Visual Reality
04-03-2008, 03:04 PM
Yeah and Photoshop lets you apply USM on only the lightness layer (in Lab Color Mode) which is optimal. With CNX you have to apply USM to the entire image.

Good point though - is 100 in CNX equal to 100 in PS?

erichlund
04-04-2008, 09:25 AM
No, they are proportionally about the same. NX lists it as a % and 100 in CS3 is about equal to 20 in NX. My point with the numbers was that other programs might not be the same, and you may be giving values that would mess with someone's mind if they don't understand your perspective.

Also, you can apply sharpening in NX to RGB, Red, Green, Blue, Yellow, Magenta or Cyan. Often, your shadows are mostly in the blue channel. This is not the same as the "lightness" layer, but it does give more flexibility than you might think.