Unsharp mask

[Cyclops]

Unsharp mask doesn't actually sharpen the image in the normal way. What it does is increase local contrast. That is,contrast between adjacent pixels so the image appears much sharper. Its an old darkroom method with a twist.

[Stanley]

Hi Cyclops,

Thank you for your reply.

Right, I have now learned the process behind sharpening. My question is, how much is too much? Sharpening does seem to make many photographs better, a little crisper. But you can also go too far and degrade the quality of the picture.

So I would like to know if there are guidelines as to how far to go without overdoing it.

Stanley

[DaveW]

Only things I can find are these ones giving examples:-

http://www.digi-graphics.com/pmain/howt ... image.html

http://www.imaging-resource.com/TIPS/SHARPEN/SHARP1.HTM

http://www.creativepro.com/article/out- ... afraid-ask

http://www.dpchallenge.com/tutorial.php?TUTORIAL_ID=4

http://www.bythom.com/sharpening.htm

Basically it seems to be a case of use other peoples settings as a starting point but then just sharpen to the extent that looks right for you, plus there is no overall correct settings since each image may require different settings to look right.

A highly sharpened macro shot may look superb, but not many women will thank you for sharpening their portraits to the same extent so that every pore shows!

To quote from the first link:-

"Knowing if it's Done 'Right'

It's easy to tell if an image has been over-sharpened, because you'll see telltale signs like halo's and/or objectionable artifacts, or pixelization in the on-screen image. The bottom line is this: if you can tell that the image has been sharpened... it's sharpened too much."

DaveW

[Cyclops]

Hi Cyclops,

Thank you for your reply.

Right, I have now learned the process behind sharpening. My question is, how much is too much? Sharpening does seem to make many photographs better, a little crisper. But you can also go too far and degrade the quality of the picture.

So I would like to know if there are guidelines as to how far to go without overdoing it.

Stanley

As a general guide I would suggest Amount,150-200
Radius 0.9
Threshold 2

[rjlittlefield]

As a general guide I would suggest Amount,150-200
Radius 0.9

In contrast, for my optics and workflow this would be way too much in most images. I generally use 30-50 at radius 0.7, applied after the image is resized for web presentation. High mag setups often require more, but very very seldom as much as 150-200 at any radius.

I don't think there's any way to get around the idea of sharpen-to-fit.

The only good way I know to learn this stuff is by practice, working with static specimens. Play around with the controls, including setting them so high that the artifacts cannot be missed. Make yourself a series at different settings, then go away for a day, come back, and see what they look like with fresh eyes.

I agree completely with that last quote that Dave lists. "If you can tell that the image has been sharpened...it's sharpened too much."

Only thing is, I'd go just a little farther.

I'd say that if it's been sharpened so much that it misrepresents the subject, then it's been sharpened too much even if nobody could tell just by looking at the picture.

This of course presumes that you're not intending to misrepresent the subject for some reason. There are lots of good reasons to alter a subject's appearance, including for example artistic impact or pedagogy -- highlighting subtle features that could otherwise be missed.

See for example http://www.photomacrography.net/forum/v ... php?t=2113. Compare images 1 and 3, and read the associated discussion about sharpening images of bones (a young otter's skull).

--Rik

[DaveW]

If there were only one set of perfect sharpening settings for digital the camera makers and software firms would by now only need to provide a "Sharpening" setting having no adjustments.

DaveW

[Stanley]

Hi everyone,

This has all been very useful information.

I will read it all carefully and do my best to absorb it.

Thanks to you all.

Stanley

[DaveW]

Just found this one too:-

http://www.photoclubalpha.com/2007/12/1 ... sharpness/

DaveW

[Stanley]

Rik,

Referring to your posting above, where you write the following:

In contrast, for my optics and workflow this would be way too much in most images. I generally use 30-50 at radius 0.7, applied after the image is resized for web presentation.

And with these settings, at what threshold? And I understand that you are just giving overall guidelines that may well be adjusted on a case-by-case basis.

Stanley

[rjlittlefield]

I'm a zero-threshold guy.

It's mainly a psychological thing.

Having written a bunch of image processing stuff, I'm conditioned to think that "threshold" means there's some magic value above which the operator does something, and below which it doesn't.

What that inevitably means is that in some places of the image it's going to do something significant (else why do it at all?), and right next to that place -- one pixel away -- it's going to do nothing, and those decisions are going to be made automatically all over the image, at more places than I can even think of checking by eye.

That concept makes me really nervous. So I don't use it.

Instead I just apply the sharpening operator everywhere -- threshold zero. If there are areas that I think would better treated some other way, then I'll make that decision myself and enforce it with something like a painted layer mask.

A lot of my images that are shot with compact digital are processed that way. I end up with essentially two layers. One of them contains significant subject that I have sharpened to taste; the other contains OOF background that I have blurred to taste. The transition between those two is controlled by a layer mask. It's a bit more trouble than trying to do something with thresholds and magic lassos, but I like the results better.

See http://www.photomacrography1.net/forum/ ... php?t=4989 for an example. Compare images 1 and 3. I can't think of any others I've posted that show before and after images, but I suppose I could gen up a pair if you need to see another example.

--Rik

[Stanley]

Hi Rik and DaveW,

First, Rik, I can see what you mean looking at the photo reference you gave in your last posting.

But your explanations are always so wonderfully instructive that I hate to turn down your offer to generate even another example.

So, by all means, if you have the time and the desire, please do produce another view. The more explanation you give, the better. But do only what time allows.

Thanks as always.

Now DaveW,

I really appreciate all those references.

This one,

http://www.bythom.com/sharpening.htm

was especially helpful. I created a Photoshop Elements file as specified in that article and did a few experiments, and that taught me a lot.

Thanks to you.

Stanley

[rjlittlefield]

So, by all means, if you have the time and the desire, please do produce another view. The more explanation you give, the better. But do only what time allows.

See "Sharper subject, blurrier background".

--Rik

[Cyclops]

As a general guide I would suggest Amount,150-200
Radius 0.9

In contrast, for my optics and workflow this would be way too much in most images. I generally use 30-50 at radius 0.7, applied after the image is resized for web presentation. High mag setups often require more, but very very seldom as much as 150-200 at any radius.

I don't think there's any way to get around the idea of sharpen-to-fit.

The only good way I know to learn this stuff is by practice, working with static specimens. Play around with the controls, including setting them so high that the artifacts cannot be missed. Make yourself a series at different settings, then go away for a day, come back, and see what they look like with fresh eyes.

I agree completely with that last quote that Dave lists. "If you can tell that the image has been sharpened...it's sharpened too much."

Only thing is, I'd go just a little farther.

I'd say that if it's been sharpened so much that it misrepresents the subject, then it's been sharpened too much even if nobody could tell just by looking at the picture.

This of course presumes that you're not intending to misrepresent the subject for some reason. There are lots of good reasons to alter a subject's appearance, including for example artistic impact or pedagogy -- highlighting subtle features that could otherwise be missed.

See for example http://www.photomacrography.net/forum/v ... php?t=2113. Compare images 1 and 3, and read the associated discussion about sharpening images of bones (a young otter's skull).

--Rik

Well i think it depends what size image you're working on. If its to be a web image,i.e. 800x600 then a lower amount is required,but if you're keeping a large image for printing you need the higher value

[rjlittlefield]

If its to be a web image,i.e. 800x600 then a lower amount is required,but if you're keeping a large image for printing you need the higher value

Good point, and another difference in workflow. It's true, sometimes I strongly sharpen an image just before sending it to a printer. But that's to overcome some peculiarity of the printer, so that what comes out on paper matches what would otherwise show on the screen or on the subject itself. The required value is different for each printer.

What I keep, in those cases, is the un-sharpened image, because that's the one that can be used for multiple purposes without accumulating artifacts. The strongly sharpened versions are just intermediate forms to drive the printer.

--Rik