Accurate on-screen sharpening?

[Dan Altick]

Deke McClelland put out a new article on sharpening in the latest (July/August) edition of PhotoshopUser Magazine where he explains how to use on-screen sharpening to accurately predict output print sharpening by viewing the sharpening on your display at screen resolution.  Jeff Schewe stated in the "Resolution" video from the "From Camera to Print" series that you can't do that accurately.  I would like to know some opinions on this article.  Thanks.

Dan Altick

[stamper]

Deke McClelland put out a new article on sharpening in the latest (July/August) edition of PhotoshopUser Magazine where he explains how to use on-screen sharpening to accurately predict output print sharpening by viewing the sharpening on your display at screen resolution.  Jeff Schewe stated in the "Resolution" video from the "From Camera to Print" series that you can't do that accurately.  I would like to know some opinions on this article.  Thanks.

Dan Altick
[a href=\"index.php?act=findpost&pid=211381\"][{POST_SNAPBACK}][/a]

I haven't seen the article but Bruce Fraser in his book "image sharpening" states the view should be between 25% and 33% At the end of the day it boils down to opinion and taste There isn't a one size fits all solution

[digitaldog]

Deke McClelland put out a new article on sharpening in the latest (July/August) edition of PhotoshopUser Magazine where he explains how to use on-screen sharpening to accurately predict output print sharpening by viewing the sharpening on your display at screen resolution.  Jeff Schewe stated in the "Resolution" video from the "From Camera to Print" series that you can't do that accurately.  I would like to know some opinions on this article.  Thanks.

Dan Altick
[a href=\"index.php?act=findpost&pid=211381\"][{POST_SNAPBACK}][/a]

Deke should be reading Bruce's book (well based on a NAPP article, he sort of does) and those of Jeffs.

Jeff is correct. Deke took Bruce's tip about figuring out the display resolution to get a ROUGH idea of proper zoom ratio but nothing will change the fact that the display is a very low resolution device and isn't close to being effective for sharpening since its not WYSIWYG.

Where Deke is off is the bit about "accurately" viewing the sharpening but the term accurate is thrown around so much these days, its lost its meaning.

[Dan Altick]

I think I may have to run some tests of my own on this.  There might be some validity to what Deke is saying.  Note that he called it soft-proofing the sharpening; which means it's not an exact science, but it might be the best approximation we can hope for on a monitor.  At a minimum, it might be a good place to start (outside of using a third party plug-in) in order to minimize the number of prints necessary to get it right.

[digitaldog]

I think I may have to run some tests of my own on this.  There might be some validity to what Deke is saying.  Note that he called it soft-proofing the sharpening; which means it's not an exact science, but it might be the best approximation we can hope for on a monitor.  At a minimum, it might be a good place to start (outside of using a third party plug-in) in order to minimize the number of prints necessary to get it right.
[a href=\"index.php?act=findpost&pid=211893\"][{POST_SNAPBACK}][/a]

The question is, can you use this soft proof to visually decide what sharpening to use?

The "other" soft proof itself as you point out, has limitations such as the gamut of the display and the output media. But short of that, and the fact that you're comparing and emissive and reflective media, the two match pretty closely.

How can a 100 dpi or so output device provide a meaningful preview of an output device like an Epson that has something like a 480 dpi output? At what zoom ratio? If less than full image, how can you see the image that's not visible due to the zoom setting? This makes the out of gamut soft proof seem pretty severe.

[Dan Altick]

And yes, it may very well be too severe to use.  Deke never mentioned what zoom ratio he was using, but I would assume 100%.  That would only make sense here I believe.  And yes, even at screen resolution, for a large image, you wouldn't see the whole image.  I personally don't have enough experience with this without testing to say whether 100ppi (in my case 120ppi) is sufficient to reliably apply a multiplication factor to the sharpening radius; and I really don't even know if it's safe to assume a 50% increase in the sharpening amount to account for the ink softening.  I would imagine though that that percentage can be determined with some accuracy for a given configuraton (i.e. printer, ink, paper, dpi).  My guess is it probably is somewhat independent of the image content.

[Schewe]

I personally don't have enough experience with this without testing to say whether 100ppi (in my case 120ppi) is sufficient to reliably apply a multiplication factor to the sharpening radius; and I really don't even know if it's safe to assume a 50% increase in the sharpening amount to account for the ink softening.
[a href=\"index.php?act=findpost&pid=211929\"][{POST_SNAPBACK}][/a]

I like Deke...I've known him for a loooong time (I've been on stage with him several times speaking together) but he doesn't ALWAYS know what he's talking about...sometimes he postulates (but doesn't really do the followthrough).

But I will tell you this, in working with Bruce Fraser in designing and developing PhotoKit Sharpener and more recently working with the Lightroom engineers to put PK Sharpener into Lightroom's output sharpening, a computer display is useless for determining the amount or type of sharpening to do to an image for optimal output...

If you are shooting higher rez images and are printing out at 300-480PPI, the BEST you can do is use about a 25% zoom to sorta/kinda see what's happening. But an image at 480PPI being displayed at 25% even with a display of 120DPI which is semi-simulate the display, you are looking at the image on a device with 1/4 the resolution of the image. So, trying to judge anything at 25% zoom at 1/4 the resolution will be a terrible predictor and will lead you to miss-sharpen your images...

The ONLY way to accurately judge the sharpening required for an image is to make the friggin' print. Look at it, and make adjustments and make another print. And you gotta keep doing that over and over until you find the sweet spot, for that image content, image resolution and paper type.

I'll try to find the article and when I run into Deke at Photoshop World in Sept, I'll let him know what I think...but on the surface, based on my experience, it's a fool's errand...

[digitaldog]

I'll try to find the article and when I run into Deke at Photoshop World in Sept, I'll let him know what I think...but on the surface, based on my experience, it's a fool's errand...
[a href=\"index.php?act=findpost&pid=211938\"][{POST_SNAPBACK}][/a]

I have a copy I'll send you. I don't think you (nor Bruce) will be happy with it.

[Kevin M]

I have never been able to accurately judge print sharpening by viewing on screen - either at actual size, print size or indeed any other size . I think the problem arises because print sharpening may look disconcertingly over-sharpened on screen at 100% view but disappears almost entirely at print-size view.

In line with what Jeff stated above, I have always found the best option is make a test print.  In the cause of saving paper and ink I just take a small crop (postcard) of some relevant detail to check the amount of sharpening.

[Dan Altick]

Jeff,

Thanks for your comments.  You've saved me allot of time, paper, and ink.

If you find anything different though after reading the article and/or speaking with Deke, please post back.  Thanks.

[One Frame at a Time]

The ONLY way to accurately judge the sharpening required for an image is to make the friggin' print. Look at it, and make adjustments and make another print. And you gotta keep doing that over and over until you find the sweet spot, for that image content, image resolution and paper type.

[a href=\"index.php?act=findpost&pid=211938\"][{POST_SNAPBACK}][/a]

If this is the case (I am not doubting) are the sharpening programs (NIK, etc) of value?
(Not trying to start a war but seems a valid question in light of the posts here)
Can they make the right adjustments without a bunch of proof prints?  Also, how do these programs relate to the Smart Sharpen tool?  I have only recently gone from using USM to Smart Sharpen (in advanced, More Accurate mode or whatever its called).  Doing a two pass work flow for most of my printing.  Thanks!

Paul

[bjanes]

Deke should be reading Bruce's book (well based on a NAPP article, he sort of does) and those of Jeffs.

Jeff is correct. Deke took Bruce's tip about figuring out the display resolution to get a ROUGH idea of proper zoom ratio but nothing will change the fact that the display is a very low resolution device and isn't close to being effective for sharpening since its not WYSIWYG.

Where Deke is off is the bit about "accurately" viewing the sharpening but the term accurate is thrown around so much these days, its lost its meaning.
[a href=\"index.php?act=findpost&pid=211732\"][{POST_SNAPBACK}][/a]

Readers should refer to Real World Sharpening with Adobe Photoshop PSCS2 by Bruce Fraser, page 218, for guidance. Since the material is copyright, I will paraphrase the passage.

Optimizing for content and creative sharpening require skill and human judgement, but sharpening for output is a deterministic process and creativity at this stage is unnecessary and should be discouraged.

When you have accurately determined the output parameters such as final output resolution, media type, and output device (with PKSharpener or by other means), it may be nice to have an onscreen preview, but such a preview is not really necessary since optimal settings have already been determined by the mechanistic process and one does not need visual feedback to make further adjustment.

For capture sharpening, judgment, skill, and artistic intent are needed and visual feedback is helpful. The current advice seems to be that one should preview at 100% or more; in fact, ACR does not allow previews at less than 100%.

Bill

[Dan Altick]

I still don't have a clear definition of input sharpening.  The sharpening that Deke does on the image before computing the output sharpening, would that be considered input sharpening, or would that be too much for input sharpening?

[digitaldog]

I still don't have a clear definition of input sharpening.  The sharpening that Deke does on the image before computing the output sharpening, would that be considered input sharpening, or would that be too much for input sharpening?
[a href=\"index.php?act=findpost&pid=212600\"][{POST_SNAPBACK}][/a]

Hard to say. Its "supposed" to be capture (what you're calling input) sharpening. Is he doing this correctly based on the capture device? Who knows but Deke. He's got the prints to examine.

[joeholmes]

A lot of these posts seem to start "I haven't read Deke's article, but..."

Well I'm printing thirty 17 x 25" prints this week on my Epson 3800 (for a fall solo show in Seattle), and sending another five files off for 20 x 30" prints, and so I've been doing a bit of sharpening.

Last year I read Bruce's book, cover to cover twice, and absorbed everything he advises.

I also read Deke's article when it came out, and I followed it precisely on a batch of prints this week.

And in my opinion, Deke's method works 9 times out of 10 exactly as he describes. It does an excellent job. The sharpening you see on screen is very, very close to what you get in your final print.

That said, of course, you'll want to tweak by eyeball, and I've been doing just that -- I'll increase or decrease the opacity of the sharpening layer depending on how things look. I doubt Deke would disagree with that method. But by following Deke's method, you'll get very close to what you want in your final print. My tweaking is truly minor.

Following Deke's method, I've run about 15 prints so far, and I'll only need to re-run one of them to improve the sharpening. The rest look better than anything I've ever printed before.

At the very least, Deke's method gives you a very good starting point for sharpening. In many cases, I'm finding it gives you a terrific final sharpening.

Pick up that back issue of Photoshop User and study Deke's method and try it yourself.

-=-Joe

[teddillard]

Deke McClelland put out a new article on sharpening in the latest (July/August) edition of PhotoshopUser Magazine where he explains how to use on-screen sharpening to accurately predict output print sharpening by viewing the sharpening on your display at screen resolution.  Jeff Schewe stated in the "Resolution" video from the "From Camera to Print" series that you can't do that accurately.  I would like to know some opinions on this article.  Thanks.

Dan Altick
[{POST_SNAPBACK}][/a]

Not to stir up the soup, but...

FWIW, I totally agree that the "wysiwyg" approach to on-screen previewing is, like screen-to-print color matching, almost more of a distraction than a help.  In both, consistency is the key.  I echo the advice, "make a frikkin print".  heh...  

That said, I like to view at slightly higher zoom than "Print Size".  I just activate the rulers and zoom in so an inch on screen is a little bigger than an actual inch.  The (fairly simple-minded) notion is that I want to see the print at about what it's going to look like coming out of the printer.

I have a post on my own blog with details of how I approach Unsharp Mask...  [a href=\"http://www.teddillard.com/2008/08/unsharp-mask-settings.html]http://www.teddillard.com/2008/08/unsharp-mask-settings.html[/url]

The most interesting thing, (to me anyway), is that my application of Unsharp Mask in the last year or so has been to use it in a very localized way...  I have completely stopped doing "global" sharpening and, in a Smart Object/Smart Filter workflow, now mask and selectively apply Unsharp Mask almost every time.  Selective sharpening gives you the nice impression of a crisp image where you want it, the eyes, the branches, etc...  but avoids sharpening noise and backgrounds.

 ...of course, consistency in "visualization", for this approach, is immensely helpful.

[The View]

Ted, I looked at your article.

Your screenshot shows, that you are viewing the document you are about to sharpen at 66.7% zoom ratio.

So far, I have always avoided to look at an image at 33.3 or 66.7%, because of the interpolations Photoshop makes when displaying at this zoom ratio.

Why are you using 66.7%?

[teddillard]

Ted, I looked at your article.

Your screenshot shows, that you are viewing the document you are about to sharpen at 66.7% zoom ratio.

So far, I have always avoided to look at an image at 33.3 or 66.7%, because of the interpolations Photoshop makes when displaying at this zoom ratio.

Why are you using 66.7%?
[a href=\"index.php?act=findpost&pid=218377\"][{POST_SNAPBACK}][/a]

As I mentioned above, I'm simply trying to look at the image at the physical size that it will print.  The actual numbers of the zoom don't really matter to me, neither Photoshop nor the screen can display what the printer will do, as others have noted above...  (The printer's not even printing the same pixels, right?  It's printing little droplets.   Much the same issues as the wysiwyg color conundrum.)  

I have tried almost everything and found that the simplest, easiest way to consistently predict the effect of sharpening is to view the image at a little higher zoom (one or two "clicks") than "Print Size".

The actual zoom numbers will depend on your true screen resolution...  but I quite literally hold a ruler up to the screen.  

To tell the honest truth, I just don't understand how you can evaluate sharpening without viewing the thing at roughly the size it will print... I realize the screen has limits and Photoshop does wacky things too, at harmonic resolutions, but I find that seeing it at "Print Size" gives me everything I need to see.

[Schewe]

To tell the honest truth, I just don't understand how you can evaluate sharpening without viewing the thing at roughly the size it will print...
[a href=\"index.php?act=findpost&pid=218426\"][{POST_SNAPBACK}][/a]

And even that is of dubious use since you are trying to look at something on a display that only has 1/4-1/2 the resolution of the print. Look, trying to divine ANYTHING about the detail in an image and how it will relate to the print is useless. You may THINK you are seeing something useful but you're not. And yes, I suppose if you've spent the time doing the trial & error of relating the poor display to many prints you might be able to train your eyes to interpolate the display results to the print, but you would be far better off making a friggin' test print.

And yes, I HAVE now read Deke's article and I'll be sure to have a word with him at Photoshop World next week.

[teddillard]

And even that is of dubious use since you are trying to look at something on a display that only has 1/4-1/2 the resolution of the print. Look, trying to divine ANYTHING about the detail in an image and how it will relate to the print is useless. You may THINK you are seeing something useful but you're not. And yes, I suppose if you've spent the time doing the trial & error of relating the poor display to many prints you might be able to train your eyes to interpolate the display results to the print, but you would be far better off making a friggin' test print.

And yes, I HAVE now read Deke's article and I'll be sure to have a word with him at Photoshop World next week.
[a href=\"index.php?act=findpost&pid=218525\"][{POST_SNAPBACK}][/a]

heh...  I imagine you will...  

Just wanted to mention the Smart Object drum I've been beating a lot of lately...  
 one of the wonderful things about them is the use of USM on a RAW Smart Object.  It's totally "amendable", because it becomes a "Smart Filter".  Make your adjustments however you like to view the thing, make your prints, re-do your adjustments, salt to taste.