Good Sharpening

[Ari]

There was a post a while back on sharpening in PS.
I just wanted to share a video tutorial I found this week; I found it to be quite effective, depending on the subject.
It's a bit involved, but the results are impressive.

http://www.youtube.com/watch?v=r7JqcC2h5zk&feature=related

[C. D. Keth]

I am far from a PS expert (or PS fluent, even) but that's a very elegant solution.

[Peter De Smidt]

I've thought about doing something similar for a long time, but I've resorted to painting in sharpening when overall sharpening isn't good enough. But it's time to take the plunge to something more sophisticated. Thanks for the link!

[Ari]

My pleasure, Peter.
I find that while each step is important, the two "curves" steps make the most difference in the final output.
I have yet to play around with the Gaussian Blur to see what effect it has on the final output.

[ic-racer]

Computer graphics are great. When he was finished it did not look anything like a photograph taken with film and printed on photographic paper.

[Mike Anderson]

Very useful technique. Thanks.

[Mark Stahlke]

It looks like the edge masks from "Real World Sharpening".
By the way, you can invert the mask and do noise reduction without softening the edges.

[chuck94022]

Is this technique the same as the "Edge Sharpening" found in Aperture? Seems to produce nearly the same result anyway from my quick experiment.

I'm finding that there are a number of advanced features like this in PhotoShop that have worked their way into Aperture and LightRoom with much streamlined user interfaces targeted to photography. But it's nice to know about this one for use in PhotoShop directly.

UPDATE: After reading details on what Aperture's Edge Sharpening does, it is exactly this and a more. It also applies progressive sharpening (multi-pass sharpening that reduces ringing), it only sharpens the luminance channel (so that you don't get odd color artifacts), and it balances sharpening between the various tonal levels on a perceptual level (so that shadows appear to receive the same sharpening as mid-tones, when in fact they get different intensities - the visual system perceives them as the same).

I don't use LightRoom, but I presume it has a similar feature. Edge Sharpening appears to be everything that this advanced PS process is, and more.

[C. D. Keth]

Computer graphics are great. When he was finished it did not look anything like a photograph taken with film and printed on photographic paper.

It's a demo on youtube. If he stuck to amounts of sharpening a sane person would use on a photograph, it wouldn't look any different at bad video resolution.

[Ari]

To wrap this up, I want to add that the saved file behaves rather strangely, due to the new Alpha channel that was created.
For example, I opened up the sharpened file in Alien Skin filters, and got only a red layer to work with.
I don't think it will save as a jpg either.
It's impossible to merge the channels, also.

Oddly enough, I found that once you have finished the sharpening steps outlined in the video, you can dump the Alpha channel; there will be no effect on the image, the sharpening will still be there.
You can then work on other aspects of the photo, and save it any way you wish.

[Ken Lee]

Perhaps with good reason, some are wary of "canned" methods, because some photos are unique and (especially with Large Format) we usually prefer that they not look contrived after sharpening - as ic-racer has described so concisely.

The stated goal of the technique outlined in the video, is to sharpen selected parts of the tonal range - in order to avoid sharpening artifacts - but it's a rather circuitous approach.

You might like this simpler and faster method which also gives greater control: Sharpening the Dark and Light Layers Separately (http://www.kenleegallery.com/html/tech/index.php#Sharpening).

With respect to the Zone System, we might call it "Zone Sharpening" because we can sharpen any section of the tonal range to taste - separately - or not at all.

It's not a complete approach either: we often need to sharpen selected regions of the image (and ignore or even blur other areas), as Peter has explained. As with everything else in Photoshop, there are many ways to get there.

[Andrew O'Neill]

A much better solution than what I was always doing. Thank you fo sharing, Ari.

[Ari]

Ken, I always enjoy reading your tech page, and I like your method of sharpening as well.
As you said, there are many ways to get there.

Andrew, my pleasure; my knowledge of the "how" in PS is increasing, but my knowledge of the "why" has remained the same.
I never would have figured out this procedure on my own, and I don't know why the Alpha channel can be tossed afterwards, or why it's even there to begin with.
But, this sharpening method is certainly more refined than what I was doing previously (just the unsharp mask), and gives a subtler effect.
I had to find something since I gave up FF DSLRs, and sharpness on a $500 Lumix is a bit wanting.

[sanking]

Hard to talk about sharpening strategies where large format negative, FF DSLR, and micro four-thirds cameras are discussed together. Each format requires a different approach.

I personally find PhotoKit Sharpener2 a good piece of "canned" software that will usually give faster/better results that I can achieve on my own without a huge amount of time. This software has numerous types of sharpening routines, and most are accompanied by different layers that one can change by increasing or decreasing the effect. There are also different routines for capture, creative and output sharpening. With PK Sharpener, and other sharpening methods, it is important to evaluate results looking at the image file at 100%.

Image files of large format scanned negatives, and good medium format negatives as well, often need little sharpening, almost none if the scan is made with a high end flatbed or a drum scanner. With these type of image files I typically use PK Sharpener with a small amount of initial capture sharpening, and then no more sharpening until printing. Output sharpening is an unsharp mask with a low amount/high radius routine that often looks something like this.

Amount = 10-30%
Radius = 90 pixels
Threshold = 0 levels

Sandy

[Brian Ellis]

Hard to talk about sharpening strategies where large format negative, FF DSLR, and micro four-thirds cameras are discussed together. Each format requires a different approach.

I personally find PhotoKit Sharpener2 a good piece of "canned" software that will usually give faster/better results that I can achieve on my own without a huge amount of time. This software has numerous types of sharpening routines, and most are accompanied by different layers that one can change by increasing or decreasing the effect. There are also different routines for capture, creative and output sharpening. With PK Sharpener, and other sharpening methods, it is important to evaluate results looking at the image file at 100%.

Image files of large format scanned negatives, and good medium format negatives as well, often need little sharpening, almost none if the scan is made with a high end flatbed or a drum scanner. With these type of image files I typically use PK Sharpener with a small amount of initial capture sharpening, and then no more sharpening until printing. Output sharpening is an unsharp mask with a low amount/high radius routine that often looks something like this.

Amount = 10-30%
Radius = 90 pixels
Threshold = 0 levels

Sandy

I agree, I started using PKSharpen II a couple years ago. I always thought I was pretty good at sharpening manually, at least I put a lot of time and effort into it. But I think I get better results with less time and effort using PKSharpen II.

[Brian Ellis]

Computer graphics are great. When he was finished it did not look anything like a photograph taken with film and printed on photographic paper.

You're viewing it on a computer monitor so it's pretty hard to tell what the printed image looks like. There's usually a pretty big difference between viewing an image on a low-resolution device like a computer monitor and seeing it in print.

[Tyler Boley]

I'd suggest that vid as a way to spark ideas more then a strict workflow, there are refinements to this that are critical for me. The old Bruce Fraser Creative Pro article is a must as well. Just for starters, converting in and out of LAB is lossy... there are better ways.. if you do this on a separate layer, but use luminousity as your blend, same thing and helps eliminate added color noise. You also than have a lot of blending options.. that's one thing just for starters...

[bob carnie]

moving in and out of LAB is lossy?? could you elaborate.

I'd suggest that vid as a way to spark ideas more then a strict workflow, there are refinements to this that are critical for me. The old Bruce Fraser Creative Pro article is a must as well. Just for starters, converting in and out of LAB is lossy... there are better ways.. if you do this on a separate layer, but use luminousity as your blend, same thing and helps eliminate added color noise. You also than have a lot of blending options.. that's one thing just for starters...

[Ari]

In the video, he suggests using LAB, RGB or whatever, but switching to CMYK before printing.

[Ken Lee]

I too have read that LAB is a relic of earlier generations of Photoshop: for these purposes anyhow, there is no longer a need to convert in and out of it.

Conversion to and from LAB is a destructive or "lossy" operation. Unlike an operation performed on a layer, a conversion changes the original data or pixels (once you save the file).

How noticeable the image degradation will be, is another question. I presume that an instrument would detect the changes - because the actual data has been changed. A file "diff" tool would detect the differences too, since that's what we're talking about: re-writing the data, not appending a filter to it.

If we converted and image back and forth many times, a human eye would eventually see it too. It might show up on a big histogram. Monitors don't always show us the subtle details, so it might take a while before anything "visible" emerged.

[Peter De Smidt]

I would only move to LAB if needed for some other purpose. As Taylor said, using luminosity mode in RGB will avoid color artifacts, which is the main reasoning for sharpening the L channel. For inkjet printing, there's no need to convert to cmyk.

Moving in and out of LAB is one of those big controversies, with Mr. Margulies claiming that it'll never be a practical degredation. My take is that there are no worries, as long as you're working in 16 bit, and you minimize conversions.

Normally, I only use LAB for increasing color separation where needed. I do use an eyedropper LAB readout when in RGB to help with color correcting.

[Kirk Gittings]

On anything with strong grain I still prefer a High Pass Sharpen for output sharpening for prints. I still think I can get better sharpening of the details without accentuating the grain with a final HPS.

[Tyler Boley]

moving in and out of LAB is lossy?? could you elaborate.

it just is Bob, some would argue the relevance, but to and from lab losses levels.


In the video, he suggests using LAB, RGB or whatever, but switching to CMYK before printing.

This further makes me take the vid with some distance, there very very little reason for anyone these days to be converting to CMYK. Unless you are handing off your files for prepess work, and have the press guy's specific recommendation for WHICH cmyk to convert to... and know what to look for in that conversion process and specific cmyk optimization after the conversion.. it's completely irrelevant in today's digital work for the vast majority of digital workers.

Tyler

[bob carnie]

LAB is not destructive, I have done the tests that Mr Margulis says, on silver gelatin paper btw which is less forgiving than ink jet.

I would love to see someone here prove it on a print.. move from RGB - LAB - back to RGB then print. Verses a all RGB workflow. Show me the destruction on any size print you want. Now that would be eye opening.

We routinely use LAB for many purposes in colour and black and white with no issues. Black and white conversions with LAb moves and then RGB apply image is better than
any canned method I have seen.


LAB is a very powerful/potent tool in the right hands and a complete disaster in the hands of people who do not know how to work with it.

It is the largest colour space and every thing from paint chips , monitor calibration and profile making , as well as OBA are measured with LAB. I think
there is much misunderstanding about this Colour Mode.

[Tyler Boley]

Bob, why do I have to prove a known fact? I said some would say it's not relevant, and obviously it's not relevant in your work. It's safe to say you have a reputation of doing the highest quality work and deserve that reputation.
OK?
Tyler

[Peter De Smidt]

I would love to see someone here prove it on a print.. move from RGB - LAB - back to RGB then print. Verses a all RGB workflow. Show me the destruction on any size print you want. Now that would be eye opening.



Please, let's not get into this anymore. This topic has been beaten to death in many other places, and neither side has been convinced. We don't need another useless pissing match. It's very unlikely that a well thought out workflow, including a move from RGB to LAB if needed, will lead to noticeably lower quality than an equally well thought out workflow that stays in RGB the whole time.

[Tyler Boley]

I was merely pointing out a few reasons why I am less than 100% comfortable with the vid and it's recommendations, as I implied it's a source of ideas that can be looked into elsewhere, as others have also suggested. It rather casually brings up issues about spaces that require more than entry level use, CMYK as well as LAB. Bob himself just said the use of LAB is best done with knowledge and finesse, and he is clearly experienced with it.
Tyler

[Ken Lee]

I for one didn't know there was any controversy about LAB space, so I am grateful it has been mentioned and explained a bit.

There's always something to learn here, that's for sure.

[Kirk Gittings]

Over at LuLa mentioning LAB is like saying here Pyro X was better than Pyro Y here a few years ago. I went down the "LAB virtual rabbit hole" for awhile ( :) ) but since I don't print virtually any color I couldn't see any value to it.

[bob carnie]

This further makes me take the vid with some distance, there very very little reason for anyone these days to be converting to CMYK. Unless you are handing off your files for prepess work, and have the press guy's specific recommendation for WHICH cmyk to convert to... and know what to look for in that conversion process and specific cmyk optimization after the conversion.. it's completely irrelevant in today's digital work for the vast majority of digital workers.

Wow , we must be on different wavelengths.. much of the last two years my beta testing has been on making negative separations using PS.. RGB and Lab would be the last place I would work from.
CMYK is totally the place to do it.

Using RGB, CMYK and yes even LAB is totally relevant to alt printmaking.

But then again I never considered myself in the vast majority of digital workers.

[bob carnie]

Bob, why do I have to prove a known fact? I said some would say it's not relevant, and obviously it's not relevant in your work. It's safe to say you have a reputation of doing the highest quality work and deserve that reputation.
OK?
Tyler

You do not have to prove anything to me Tyler.. I was just curious as your findings do not match my findings in regard to LAB

[bob carnie]

I look at working in LAB very carefully as complete disaster is easily obtained.
Kind of like putting a finishing nail in with a sledgehammer.


I was merely pointing out a few reasons why I am less than 100% comfortable with the vid and it's recommendations, as I implied it's a source of ideas that can be looked into elsewhere, as others have also suggested. It rather casually brings up issues about spaces that require more than entry level use, CMYK as well as LAB. Bob himself just said the use of LAB is best done with knowledge and finesse, and he is clearly experienced with it.
Tyler

[bob carnie]

I think a lot of the controversy about LAB space is due to Dans videos where he works on really atrocious images and in all cases makes images that can only be
defined as Kitch , over sharpened and over saturated colour

but I spent four sessions with the man.. each session three days 12 hours long and what he has forgotten about PS I wish I knew.

I am a huge fan of the space and as Tyler correctly points out it works for me and one must be very careful when working in the space.
I was turned onto LAB right here on Large Format with threads by Chris Jordan who brought out controlling colour changes without affecting the neutrals.. this is one of the most powerful tools within LAB, and I was intriqued and followed the wormhole.( his reference at the time was getting rid of blue shadows on concrete in the open shade areas)

Most people relate to LAB about sharpening on the L , but once again as Tyler pointed out it is easily achieved in RGB on the luminosity mode.
Another area where I use it a lot is in BW conversion, where I will twist the AB channels to separate complimentary colours or better yet move them farther apart from each other on a layer paint in the effect from each other then immediately go to RGB and channel blend to make the best possible conversion.





I for one didn't know there was any controversy about LAB space, so I am grateful it has been mentioned and explained a bit.

There's always something to learn here, that's for sure.

[Tyler Boley]

This further makes me take the vid with some distance, there very very little reason for anyone these days to be converting to CMYK. Unless you are handing off your files for prepess work, and have the press guy's specific recommendation for WHICH cmyk to convert to... and know what to look for in that conversion process and specific cmyk optimization after the conversion.. it's completely irrelevant in today's digital work for the vast majority of digital workers.

Wow , we must be on different wavelengths.. much of the last two years my beta testing has been on making negative separations using PS.. RGB and Lab would be the last place I would work from.
CMYK is totally the place to do it.

Using RGB, CMYK and yes even LAB is totally relevant to alt printmaking.

But then again I never considered myself in the vast majority of digital workers.

man this stuff gets difficult sometimes, did I do something to piss you off?. I don't know what you are doing on CMYK or why you work there, I'm sure you have your reasons and they are good. Negative separations are a completely different issue than what more are doing here and I can imagine CMYK would be relevant. In general though, cmyk is an OUTPUT device profile, not a working space. In fact most people don't realize this, they think there is "a" CMYK space just like there are RGB spaces not knowing there are many CMYK profiles, each related to SPECIFIC press and paper conditions, and which they are converting to is determined in color settings most never look into. Which is selected, either by default or by user choice, will determine gamut, black channel build and limit, individual ink tints and limits, dot gain of each, total ink limit, etc etc. I convert to cmyk occasionally because I have a RIP and some CMYK ink/paper setups, and that way I can control the individual ink channels directly. All the previously mentioned issues are determined by me when I make the RIP setup for the paper and ink, and the build the profile. All this is off topic, again, I'm simply suggesting anyone interested in the kind of USM techniques suggested in the vid, look into it a bit more. Please.
Tyler

[patrickjames]

In the sharpening video the author states he is sharpening for print, as in printing press. That is a big difference than sharpening for inkjet. I thought the distinction should be stated because of some of the negativity above.

I edit color in LAB mode for scans. It took quite a while to learn, but I saw the advantages of it after I started doing it. It is not easy to wrap your head around though. It requires a different way of thinking about color. I imagine color now as a big 3 dimensional ball of muliti-colored play doh. The result, for me, is control of color I was never able to achieve before I started doing it. I agree with what Bob stated above. It is easy to make a mess out of it if you don't know what you are doing. I almost gave up at one point. I like the advantages though.

In the end, do whatever works for you. In this day and age, there really aren't any "better" ways, there are just different ways. Technology has afforded us that.

[bob carnie]

Tyler

no you have done nothing to piss me off, I think we have differing markets and our methods of getting to a final print may be vastly different, but obviously effective.

I am just now these past 5 years seeing the potential of CMYK for alternative printing, I hope some day to be a service provider of negatives for those who find it intimidating to make large
silver negatives for many if not all of the printing processes that require a darkroom.

I have very limited knowledge of the PRESS world as 100% of my work is dedicated to prints for hanging on walls, whether it be silver, pigment or ink prints.
I do not use CMYK for any reason other than making negs for pigment or alt prints and I must say this old dog is learning new tricks each day.
The combination of RGB and LAB is my world and for me it is second nature. Anyone who has printed on a colour enlarger will pick up LAB much faster than RGB in regards
to the info palette.
In fact I have the digital colour meter found on the Mac set to Lab and use it every day to take readings.... L channel 10 zones 100 points... 0 A or B means neutral way more easier to read than the RGB numbers.

I do use the L channel to sharpen if its a colour image which is appropriate to the OP's question
I also use Sandy's method for landscape work where there is a lot of flat surfaces
I use high amount low radius for sharp lines .
Most time I am painting in two different sharpenings depending on the subject.

I never, and I mean never use a plug in and for those who do , your heard it here,.. Using plugins is dumbing you down...

now I have pissed off a bunch of people and will take some heat.

Bob

[Tyler Boley]

HAH! I agree about plugins..
I don't do pre press either, all fine art prints here... that was just info on CMYK use. Things have become even more challenging here.. now it's CMYKOG!
Thanks Bob

[Ari]

In the sharpening video the author states he is sharpening for print, as in printing press. That is a big difference than sharpening for inkjet. I thought the distinction should be stated because of some of the negativity above.



That's what I wanted to say, thank you, Patrick.
The author seems to come from a printing background, he alludes to it once or twice.
Much of the vitriol here could have been avoided.