Untitled Document

Greetings to the group from Helsinki, where I've just finished a class with a dauntless group of Finnish photographers. It was a most enjoyable experience that I'll discuss more when back in the U.S. in two weeks.

Meanwhile, I wanted to throw out a sharpening idea that I've been having success with recently. I am intending to show this in my advanced classes later this year but haven't tested it as well as I should have with enough of a variety of images. So, to those interested in testing it , suggestions or comments would be welcome, although I don't know whether I will personally see them for the next two weeks.

For the moment, this idea pertains only to conventional sharpening, meaning with a high Amount and a relatively low Radius. I have had only mixed success with hiraloam. Although it is mildly applicable in CMYK and somewhat more in LAB, its real power would be for those forced to sharpen conventionally in RGB, which is technically the weakest sharpening space for several reasons.

We all know that conventional sharpening is highly effective at adding focus, but also prone to adding undesirable detritus, particularly in lighter areas. Consequently, there have been many ideas about sharpening through a mask. John Ruttenberg has suggested using Blend If sliders to limit the amount of sharpening that can take place in lighter areas. I have suggested using Command-Option-1, then Command-I, to load an inverted luminosity mask and sharpening through that. And many people have mentioned some type of edge mask that tries to select only the areas of transition, and sharpening through that. I have never been a big fan of this method because it seems to me to yield unnatural results. The first two methods OTOH are relatively safe but don't always allow as much sharpening as I would like.

I wrote about why darker areas accept sharpening better than lighter ones as early as 1994. Since then, I've realized a corollary--in addition to lightness, a strong color is an argument against sharpening. We don't like to oversharpen skies, the petals of flowers, and human skin, for example. Inverted luminosity masks don't take account of this effect. A deep red rose is a fairly dark object, so an inverted luminosity mask would allow a lot of sharpening, which we probably don't want. More neutral objects OTOH accept heavy conventional sharpening much more gracefully, and not just on their edges.

My suggestion is a mask that caters to both--that allows more sharpening where the image is darker but also restricts it where the image is colorful. While it is possible to make a convoluted Action that generates such a mask by a series of blends of the RGB channels, there's a faster way--make a false separation, and use an inverted Heavy GCR black as the mask for the RGB sharpening.

Here's the procedure, which of course should be reduced to an Action to save having to do it over and over.

1) Copy the RGB image.

2) With the copy, Convert to Profile>Custom CMYK.

3) Fill in: Heavy GCR, 70% black ink limit, 340% total ink. Dot gain is basically not relevant as you can always lighten or darken the mask after applying it, but I just use the default 20%.

4) Click OK twice to generate the false separation.

5) Command-4 to expose the black channel, and Mode: Grayscale to discard the CMY channels.

6) Invert the channel with Command-I, yielding a negative image.

7) Auto Levels.

8) Gaussian Blur, radius 2.0 pixels to eliminate noise and make for a softer sharpen.

9) Return to the RGB image and create a duplicate layer. Sharpen conventionally with a very heavy hand--500%, 1.2 pixel Radius, 3 Threshold might be a good starting point for most images.

10) Add a layer mask. To it, load the artificial black channel that was made in steps 1-8. This should confine the sharpening to the desired areas.

11) If you feel the image is not sharp enough, apply a curve to the mask to lighten its midpoint. If you find the image to be too sharp, darken the mask in the same way.

Dan Margulis

___________________________________________________________________________

Good one, Dan! You should travel to Finland more often.

Could you please post a couple of images that have shown the superiority of the Black channel mask over a transparency mask? I've tested a variety of images, including the Woman from your earlier edition of Professional Photoshop. Your sharpening technique gives superior results on everything I've tested, but I have yet to see an instance in which the Black channel mask gives superior results.

Come home soon. It's too quiet around here with you gone. Only a couple of new posts have appeared since you left.

Howard Smith

___________________________________________________________________________

First a point of clarification: conventional sharpening does not add focus. Focus concerns the size of circles of confusion caused by the optical properties of the image capture. Conventional sharpening adds acutance. Focus and acutance are two well-recognized different aspects of image appearance. An out-of-focus image cannot be sharpened with a conventional unsharp mask. . Focus Magic, for example, is an application (I'm not using) which claims to bring out-of-focus into focus by correcting several kinds of blur. Photoshop CS3 has apparently adopted some of that technology as well in the new Smart Sharpen tool, but there is more to Adobe's new tools as well.

This brings me to the heart of the matter, which is the existence of these new sharpening tools in Camera Raw, Lightroom or Photoshop (e.g. Highlights, Shadows, Mask, Fade), aiming to achieve the objectives underlying the interesting 11 step procedure which Dan lays out. The new tools in these three applications only require the use of several sliders and no colour space conversions; hence it would be of considerable interest to compare Dan's procedure with the new tools in Camera Raw 4.x (or Lightroom) and Photoshop CS3. I wonder whether Dan has done so himself already and found the new CR/PS tools lacking relative to his new procedure. I shall not be testing this - no time just now, and with the flexibility of the third-party application I'm using, no need. But I shall retain Dan's procedure on file "just in case". Thanks Dan.

Mark Segal

___________________________________________________________________________

Hi Mark, Don't kid yourself. That statement is akin to the classic example on page 190 of PP #6 by Dan, machines and some people will tell you all day the two greens are the same,of course they are not just look at them.If an image appears clearer,sharper,more contrast than it was before you started 98% of the world will call it more "in focus" adding focus or what ever term you want to use is irrelevant. If it looks more in "focus" it is,why would anyone care if "focus" is added? Jim Donovan

___________________________________________________________________________

Mark,

Focus (noun): "the centre of interest or activity" (Oxford English Dictionary)

Having taken both of Dan's courses, I never was under the impression that Dan was suggesting some "un-doing" of lack of photographic technique that resulted in the inability focus the lens. I don't think PP suggests anything different, although I must admit I am behind on reading the 5th Ed. When he mentions "focus", it might be helpful to think of it as enhancing the area of importance in the image.

Since the Smart Sharpen filter appeared in CS2, I would assume that you have had an opportunity to use it on a variety of images. While it is suitable for some classes of images, don't you find that the lack of a "threshold" slider is an issue? (I'm guessing that is the filter you're referring to when you mention the "Highlights" and "Shadows" sliders.)

Darren Bernaerdt

___________________________________________________________________________

Hello Dan,

it's a pleausre to be back on this forum after several months and my impression is the same as the first day I subscribed it: oh, damn I've been missing so many things before this time...!

The Mask sharpening is an interesting technique, I've tested on few images but in the most it worked very well.

The black channel and its many ways of create it, is a great element that helps in a lot of color corrections, masking techniques and even photoretouchs.

In addition to what you've shown, in some images I tried also to change the values in the "blend if" grayscale option and this added further control to the sharpening allowing a more accurate final result. In other cases I made a gaussian blur on the mask and the result was also interesting.

I would be happen to know if anyone has other suggestions regarding this.

But I've found that the result is not so good on critical images (underexposed and grainy images) and even after changing the luminosity of the mask (the black channel) the image looks not so good due to its grain. I know that images with a lot of noise are the most diffcult to sharpen and I thought that this technique could help more than others but maybe there are some further steps to focus on.

Ciao!

Alessandro Benardi

___________________________________________________________________________

Howard Smith wrote:

That was so Howard, currently, the new USM mask method post by Dan and the replies are going as expected - while the following earlier post has remained ignored, except for one reply:

http: //tech.groups.yahoo.com/group/colortheory/message/18612

Ric brings up some very good questions on converting to CMYK, where the thread could easily head into half a dozen separate discussions of great relevance and importance to many on this list (unless Ric and myself are the only two members still interested in discussing CMYK, I presume that Ric is still waiting for more replies before responding).

Stephen Marsh.

___________________________________________________________________________

Hi Jim,

I won't kid myself - promise. And I don't want to make a big deal of this point, but I guess it's a compliment to Dan that he isn't 98% of the world - he's one of the top professionals in his peer group, so naturally I get surprised when I don't see accurate technical terminology being used on a technical forum - but OK, let's carry on and assume Dan means acutance and the picture "looks sharper" with greater acutance than without it, provided it isn't out of focus to begin with :-). So with that out of the way of course an important question is how this procedure compares with the new sharpening algorithms in CS3. I'd be interested in anyone's observations about that. One of these days I'd like to test it comparatively, but doing this correctly and thoroughly is a big job - needs lots of time - and focus!

Mark Segal

___________________________________________________________________________

Moderation Notice - Re: A New Sharpening Mask Idea

Dear members, the moderators have made a decision to manage this topic and new posts regarding what specific words should be used in describing sharpening will no longer be accepted. The moderators will not be accepting ACR/ALR vs Photoshop sharpening discussions at this time (as per Dan's message closing the last ACR thread). Posts discussing the pros/cons of the method are valid for discussion.

We currently have 11 messages in this thread, eight previously approved and three waiting for approval (with two being rejected that never made it to the list for a total of 13 messages so far).

I am appending the final 3 posts currently awaiting approval to the foot of this message.

The break up of the 13 messages is as below:

1. Original post by Dan Margulis. (21st September 2007).

2. On topic reply from Howard Smith (22nd September 2007).

3. Reply from Mark Segal. This post was considered worthy of a negative reply by 5 members, that all had similar things to say to Mark in regards to his OP. Three were allowed to comment as is (the other two were asked to reword their rejected post (22nd September 2007).

4. Jim Donovan's reply to Mark Segal. (23rd September 2007).

5. Reply from Stephen Marsh to Howard Smith reminding the list of another earlier topic on Perceptual CMYK conversions that could do with more discussion. (23rd September 2007).

6. Reply from Darren Bernaerdt to Mark Segal (23rd September 2007).

7. Reply from Alessandro Bernardi to Dan Margulis. (23rd September 2007).

8. Mark Segal's reply to Jim Donovan (23rd September 2007).

9. Mark Segal's reply to Darren Bernaerdt (23rd September 2007, see below).

10. Richard Wagner's post (23rd September 2007, see below).

11. Jonathan Clymer's reply to Mark Segal's OP. (23rd September 2007, see below).

In summary, we have had five negative replies to Mark's criticism of Dan's post. Mark has replied to both Jim and Darren. The moderators feel that all parties concerned (apart from Dan who is away) have stated their position and feelings on the matter and that there is no benefit of further on-list discussion of these matters, please take the discussion off-list if you wish to continue it.

The moderators look forward to approving on topic posts discussing the applied theory of sharpening through a K channel mask, or with other comparable methods in Photoshop.

Respectfully,

Stephen Marsh.

List Moderator.

===

From: "Mark Segal"

Date: Sun Sep 23, 2007 1:09 pm

Subject: Re: [colortheory] Re: A New Sharpening Mask

Darren,

You've hit on an important distinction in your first para below the Oxford

Dictionary. Well, you're two up on me because I haven't taken either

of Dan's

courses. I just read the books. "Enhancing the area of importance in

the image"

as you put it - maybe you are thinking of local contrast enhancement -

high

radius low amount? That does what it says - enhances local contrast,

using the

unsharp mask as the tool alright - but it isn't "sharpening" if one

means by

that improving the overall appearance of edge sharpness by gently

enhancing

acutance without seeing much change in the overall contrast of the

image. I

assume we understand HEIRLOAM creates a more contrasted image

appearance than

"conventional sharpening" does.

Re your last para, I haven't made much use of Smart Sharpen because I

use PK

Sharpener Pro, designed for flexible treatment of a wide range of

capture types,

output types and image conditions, using combinations of tools (and

settings)

all within Photoshop. For clarity, when I referred to Smart Sharpen I was

thinking of the presentation in CS3 (which appears to be the same as

that in

CS2), where if you select the Advanced radio button, you get three

tabs, one of

which is highlights and another shadows; and they allow one to shape the

tonality of the light and dark contours, with the user determining the

split

between the two. Perhaps Adobe figured that between these more refined

controls

and the Fade slider which controls the extent of the sharpening they

didn't need

Threshold as well. However, Ben Willmore, who provides a very lucid

discussion

of Smart Sharpen in his "Up to Speed Photoshop CS2", also regrets not

having a

Threshold control. But Bruce Fraser, in his Image Sharpening book is

very much

less enthusiastic about the usefulness of the Threshold control (page

112) and

by the time you get to page 122, he wasn't too satisfied with the

amount of

control achievable with Smart Sharpen either. I, like many of us I

think, have

tried numerous approaches and plugins. I settled on what I am using

now, with

much satisfaction. It is good that Dan has developed a new procedure,

and in

light of the existing literature and toolsets, I think it deserves

comparative

testing, which is a very considerable task when you get down to

thinking about

what it would involve.

Mark Segal

===

From: Richard Wagner <Rich@WildNaturePhotos.com>

Date: Sun Sep 23, 2007 3:38 pm

Subject: Re: [colortheory] A New Sharpening Mask Idea

On Friday, September 21, 2007 5:05 PM, Jim Donovan wrote:

> If an

> image appears clearer,sharper,more contrast than it was before you

> started

> 98% of the world will call it more "in focus" adding focus or what

> ever term

> you want to use is irrelevant. If it looks more in "focus" it

> is,why would

> anyone care if "focus" is added? Jim Donovan

Well, presumably the audience here is not "98% of the world," but

rather imaging professionals. If we all start misusing terminology

there will be utter confusion. Adding contrast or edge sharpening is

not the same as "adding focus." There is a long, historic misuse of

standard imaging terminology on this List. That is unfortunate for

those just starting out, who don't recognize the problem. It is

hardly "irrelevant."

--Rich Wagner

===

From: "new_news" <new_news@patmedia.net>

Date: Sun Sep 23, 2007 11:00 pm

Subject: Re: A New Sharpening Mask Idea

>>>First a point of clarification: conventional sharpening does not

add focus.

Focus concerns the size of circles of confusion caused by the optical

properties

of the image capture. Conventional sharpening adds acutance.

Oh give me a BREAK! Is there anyone on this list that didn't know the

difference

and didn't know what Dan meant? And he is not the only expert who uses

the term

"focus" this way (I can cite examples) so it's not unique to him.

Perhaps you should visit all the printers in the US to tell the

pressmen to stop

using the term RED when they are actually using magenta ink!

Jonathan Clymer

___________________________________________________________________________

Dear All,

Here is an action attempting to implement the procedure Dan describes.

http: //www.pochtar.com/DanMargulis_RGB_Sharpen_K_mask.zip

I found USM Radius settings [0.7..1.5] to be useful depending on the image. On the sharpened layer I used Blend If sliders as usual for me to exclude extreme highlights and shadows from sharpening. Depending on the amount of noise in the image it seems useful to adjust shadow limiting sliders further.

Iliah Borg

___________________________________________________________________________

Dear All,

There is a lot of talk over the internet what is the place of the sharpening method Dan suggested: is it "capture" sharpening, "creative" sharpening, or output-oriented sharpening. I think it is good to mention that "capture" sharpening applied after the demosaicing is a crunch (not only my opinion, at least three other teams coding raw converters will tell you about the same, this includes Bibble and Nikon Capture Teams).

Demosaicing of unsharp and un-deconvoluted images "burns into the image" the effect of anti-aliasing filter and other smearing factors present in the optical and electronic paths in cameras; and leads to misdirection of demosaicing gradients resulting in artefacts, one of the types being worm-like structures (digital reticulation) you can see here (ACR conversion, view at 200%):

http://www.pbase.com/image/85092804.jpg

The above is far from being the worst case because the area in question is bright and close to neutral; and the image is taken in the daylight; that is the exposure of all channels is more or less uniform. That is, demosaicing can find a lot of information in each pixel under the colour filter array. This is not always the case.

Sharpening applied to such structures after the demosaicing makes them worse and visible on a normal-size print.

If I need to apply pixel-level sharpening after demosaicing (that is what some call "capture sharpening") it is 0.3-0.4 radius, 350-450%, zero threshold, L channel extracted to a separate document and sharpening applied in 32-bit mode (floating point in Photoshop). After applying sharpening I compensate gamma via "Exposure" adjustment and place the sharpened channel as a separate layer into the original document.

It is often that I need to direct sharpening into darker areas of an image where the noise appears blotchy. Pixel-level sharpening helps dissolving noise into finer structures that appear more like film grain ("hiding leaves in a forest"). Radius for such sharpening usually can be 0.4 to 0.6.

--

Best regards,

Iliah Borg

___________________________________________________________________________

Iliah,

I use a Canon 1Ds, real good camera but sensor technology a bit old by the digital standards of today, I process my raw files in Adobe Camera Raw 4.x without sharpening and apply PK Capture Sharpen after rendering the image into Photoshop. Then I make any other necessary adjustments under soft-proof and apply Output Sharpen before printing. I have been satisfied with the image quality from these procedures and I am receiving very positive feedback from critical/expert people who see what I print. However, coming from you, and given what you have written below, I got a bit scared, asked myself whether this could really be happening right under my eyes, so I opened some raw files I hadn't adjusted yet, converted them doing NOTHING in the raw converter, pumped them up to 200% and looked closely for these worm-like artifacts (or other such animals) on my colour-managed LaCie 321 display, in both dark and lighter areas of the images, and frankly I couldn't find any. That was a relief. Then I applied PK Capture Sharpen to them in default mode to see whether this sharpening, viewed again at 200%, would bring out such artifacts; much to my relief, it didn't. By the way, a 200% zoom of a Canon 1Ds file is equivalent to a mighty big print. Bottom line, I'm pleased to report I can't replicate your findings in my workflow.

Mark Segal

___________________________________________________________________________

Thanks, Iliah, I've been wondering exactly what that was. It's sort of organic looking, which obviously can't be right, though it depends on how analog the digital artefacts are, if you see what I mean.

Cheers,

/andrew webb

___________________________________________________________________________

Howard writes,

I do not know what you mean by "transparency mask". What I was suggesting was an *adjunct* to current sharpening methods, not a replacement. As I indicated, it's speculative at this point--I'm pretty sure that the *concept* is correct but the execution that I suggested can probably be improved upon. Because testing it fully would take a great deal of time, and because the improvements would be visible but not spectacular, I don't foresee being able to take this much further myself, so I threw it out to the group for consideration.

I have observed that the more colorful the object, the less amenable it is to accepting sharpening. The most obvious example I can think of recently: I've just been through the Utah national parks. In Bryce Canyon and Grand Canyon, there are areas where the rocks go from nearly neutral to strongly colored. There's no difference in darkness or in texture, just color. And even when light halos are properly lessened, the colored areas don't take the sharpening as well as the more neutral ones.

Here's why this might be true.

I believe that almost everyone now accepts the basic theory of color correction that I have been laying out since the early 1990s--that the objective of manipulating the file is to produce something more like what the viewer recalls seeing and less like what a camera or other artificial instrument saw. That's the whole theory behind balancing highlights and shadows (compensating for the human phenomenon of chromatic adaptation, which a camera lacks). It's the whole reason that LAB correction is often successful where other methods are not (introduction of color variation that mimics the human phenomenon of simultaneous contrast). It's why we retouch nasty reflections out of faces (imitation of the human phenomenon of filtering out bright flashes of light). We desaturate backgrounds (imitation of the human phenomenon of losing color perception in areas of little interest). And on and on.

Unfortunately for this philosophy, the human visual system is astonishingly complex. It processes a huge amount of information in ways that would overwhelm the most powerful of today's computers, even if we fully understood it, which we are far from doing. Scientists have identified more than a dozen ways in which human perception significantly varies from that of an artificial instrument. Catering to these is one of the keys to successful color reproduction.

Almost all of these effects bear on survival. We (or at least our prehistoric ancestors who were around when the human visual system evolved) need to be able to identify things that are potentially good to eat. More importantly, we need to identify other animals that may think that WE are potentially a good thing to eat, and we need to do so very, very quickly.

All of the phenomena listed above bear on the need to make rapid evaluation and identification of the objects that may confront us. When we recognize the color of something edible, we don't want to be distracted by lighting conditions, hence chromatic adaptation. When something is camouflaged in the forest among similar colors, we want to be able to see it, hence simultaneous contrast. If a predator jumps at us out of the sun, we want to be able to see it, hence our rejection of bright light.

Since the time of Chevreul in the 1830s, we have known that the human visual system intensifies certain colors because of the phenomenon of simultaneous contrast, thus aiding us in identifying immediately what is before us. Being able to identify things by color makes it less necessary to identify them by detail. If the human visual system can in effect say "this is a flower, we don't need full definition of its detail, or, this is a butterfly, it is not a threat to us and we don't need to study it closely" it can devote more of its resources to resolving detail in more neutral objects faster. A complicated scenario? Yes. But hardly the most amazing thing that the human visual system does. Dangerous animals don't like to advertise their presence with bright colors. It's reasonable, if not proven, to think that a way of subconsciously emphasizing focus in tones where these animals are likely to be found may have evolved.

That speculation, if correct, accounts for why hiraloam sharpening doesn't seem to be as strongly favored by this method as conventional USM is: we are programmed (so I imagine) to ignore fine detail in colored objects to some extent, but we can still see large variations in shape, such as hiraloam halos. It also explains why whenever in literature a character encounters a potentially dangerous animal, there is likely to be a description of how sharply focused the scene becomes and how everything besides the animal is ignored.

This has an impact on color correction. We are always trying to get the maximum sharpen that doesn't cause the image to fall apart. Unfortunately, that forces us to sharpen for the weakest link--for the area that breaks first, which is commonly a colored area. It means that we can't sharpen the more neutral areas as much as we'd like. That's why it's so often better to sharpen the two weak channels in CMYK, but we don't have that move available in RGB. If we have a way of slightly protecting the colored areas, without excluding them entirely, then we should be able to be more accurate in the neutrals.

I've tested this as follows: on a top layer, create a sharpened version of the original (remember, conventional narrow-radius sharpening only) using whatever method you like. Leave it somewhat oversharpened so that ordinarily you might want to reduce the opacity of the layer to 90% or so.

Instead, from a copy of the bottom layer, generate a blurred saturation mask and load it as the layer mask on the top layer of the original file. That cuts down on the sharpening of the colored areas and may make it possible to retain 100% opacity on the top layer.

From my limited testing, this always makes for a better result--if you look hard enough. The problem is that sometimes the gain is so small that it's almost invisible. I do not currently think that such an approach would be worth the bother.

My suggestion of using a layer mask based on a heavy black channel is, I think, more promising because of the mask's luminosity component, which allows us to target the desired sharpening area very precisely. Because the mask has been blurred, there is no harm in applying curves to it in a random shape. In principle, we want to sharpen more as the image gets darker; usually the main target of sharpening falls in the midtone- to three-quartertone range, occasionally as light as a quartertone. But if for some reason we want to sharpen the quartertones as much as the three-quartertones--or even * more* than the three-quartertones, there's no sweat with the mask--you just apply a curve to it that flattens the range between them, or even apply a curve in the shape of the letter N if you want more sharpening in the quartertone than in the three-quartertone.

Again, this doesn't *replace* other techniques. Generally when we sharpen overall in RGB we want to restrict the lightening effect throughout; the method I recommend ordinarily is to do a heavy sharpen on the top layer, which is set to Luminosity mode then apply the bottom layer to the top at 50% opacity, Darken or Darker Color mode. With this method the darkening isn't needed as much but it's still needed--I usually cut it to about 35%. Since everything is by Action, it takes seconds to get to this point and the result is going to be passable.

The final step (if there's time) is to see what areas, if any, we want to emphasize or deemphasize. We click into the mask curve in any points that we want to emphasize, and lighten them, and darken any that we want to de- emphasize.

Iliah's Action, I note, also includes Blend If sliders to restrict the sharpening in the lightest and darkest areas of the image. There's certainly no law against doing that if it makes the image look better, but I question whether they're really necessary. Because the mask is based on inverted luminosity, sharpening of the lightest areas is already seriously restricted. As for the darkest areas, sometimes they can be sharpened effectively but sometimes we encounter excessive noise. If that's the case, applying a U-shaped curve to the layer mask is likely at least as effective as Blend If.

The suggested mask is useful because it is prejudiced against allowing sharpening of colored areas, which makes it better IMHO than a mask based solely on inverted luminosity. But, somebody can probably work out a mask that would be even better. It's a work in progress, so I'd be interested in hearing from those who have given it a workout.

Dan Margulis

___________________________________________________________________________

Dear All,

The second version of the action with Blend If sliders adjusted based on more images put through the method; some comments also added:

http: //www.pochtar.com/DanMargulis_RGB_Sharpen_K_mask_v2.zip

Total feedback - 117 e-mails. Most used the method for output sharpening while softproofing. Radius variations mostly are in the range of 0.9 to 1.4. Typical successful applications - landscapes, some shots in a zoo, portraits. Most users asked to thank Dan for the suggestion. Few are asking if it is possible to improve the mask by combining it with "true" saturation mask. Several times I was asked how Luminance differs from L (or why not to use RGB sharpening and Fade to Luminosity). The answer obviously is "it depends" :) IMHO understanding of the difference may be beneficial when choosing how to sharpen.

In both images L is to the right:

http://www.pochtar.com/L2Luma.jpg Command-Option-1/Control-Alt-1

http://www.pochtar.com/L2Luma2.jpg Command-Option-~/Control-Alt-~ (~ is "tilde")

Histograms for comparison

http://www.pochtar.com/L.vs.LumaHistograms.jpg

Some negative comments (9 in total) are along the lines - why do we need yet another sharpening method when we already have PK Sharpener, FocusMagic, FocalBlade, and SmartSharpen.

More then 45000 downloads as of the time of writing.

--

Best regards,

Iliah Borg

___________________________________________________________________________

It has been a long while since I have responded to this group. It seems being over-worked (aka “burn-out”), death in family, etc., can make a person view color-corrections as much less important than was previously thought. Never-the-less, as I come back around to my love for good color, I took note of the comment below;

The above programs are valuable (some more than others), but I enjoy knowing the “whys” and “hows” of sharpening (maybe that is why I enjoy Dan’s in-depth reasons for what he does). That way I can construct my own way of sharpening the photo based on what I understand (instead of putting all my trust in a program that will sharpen the photo for me - somehow). The more I understand the better the end results. That is why I welcome any new sharpening method, and the methodology behind it.

If my own “ultimate workflow” contains a combination of several sharpening tweaks, so be it - bring on more! Personally, I try to think of photos as paintings (as others has suggested we do) on which I can build a corrected 3-D image out of the supplied 2-D image. For me, the various sharpening methods help me do this.

Andy Adams

___________________________________________________________________________

Andy,

You've summarized my own reasons for not using third party plug-ins. As useful as they might be, they don't exactly contribute to creative thinking. All the effort spent trying to come up with methods that are at least almost as good as the plug-ins has been well worth the time for me. And of course there is also the obvious beneift that you mentioned about how it's helpful to know what is going on when you use any tool. Of course anyone working on a tight schedule with the need for fast results will value plug-ins if for no other reasons than economic.

There is probably some new tool or technique yet to be developed that will make Photoshop accessible to all of us, with the result that even beginners will be able to do professional work with very little training. Such a comment on my part is almost sure to elicit a barrage of angry retorts, but-let's face it-progess is inevitable. Things we couldn't envision 10 years ago are now being utilized by young students who simply take all of the technology for granted.

The only thing about such a new tool or technique that bothers me is the possibility that it will stifle creativity instead of supporting it. But then that kind of fear is common in all fields of advancing knowledge, even though the reality is that the progress almost always benefits more people than it hurts. Every new development that eliminates jobs somehow manages to create even more in another direction.

Howard Smith

___________________________________________________________________________

Dan,

My post may have sounded like a challenge (we get those too frequently), but it was only a request for clarification. You have pointed out on several occasions that little things can make a big difference. I thought there might be some images where the Black channel mask might be superior.

What I referred to as a transparency mask, which may well be incorrect terminology, is a layer mask prepared from a selection made from the composite channel.

Apparently there are variety of masking techniques that can be a great help in sharpening among other things. It's something we all need to consider more thoroughly. Your original post contained a considerable amount of material for exploration. Thanks for taking the time to share all your ideas.

Howard Smith

___________________________________________________________________________

Howard,

New technology that solves mechanical problems in newer and better ways and works more efficiently liberates us to spend more time being creative. I'm a fim believer in the theory of comparatived advantage. It has served the progress of human-kind very well.

Mark Segal

___________________________________________________________________________

Dan wrote;

My suggestion is a mask that caters to both--that allows more sharpening where the image is darker but also restricts it where the image is colorful. While it is possible to make a convoluted Action that generates such a mask by a series of blends of the RGB channels, there’s a faster way--make a false separation, and use an inverted Heavy GCR black as the mask for the RGB sharpening.

Based on the principles above for RGB files, I have done the following in Lab and feel it produces better results. I would appreciate any comments.

1) With an Lab file I follow the above steps 1-8.

2) In Step 9, I am (of course) only sharpening the “L” channel.

3) I finish steps 10 and 11. The comment in the paragraph after step 11 is accomplished by lowering the opacity of the sharpened layer to 65%-70%.

Dan wrote, “a strong color is an argument against sharpening”. But what about hiraloam sharpening? So, I do the following;

4) I option>merge to create a 3rd layer and add a layer mask filled with same mask as the 2nd layer - but inverted, blurred another 2 pixels, and opened up with curves (50>35) in order to include some of the “less colorful” areas in the transition.

5) Apply hiraloam sharpening to the “L” channel and adjust opacity to taste.

I accomplish the above with a series of actions with pauses.

Andy Adams

___________________________________________________________________________

Mark:

Something is gained, and something lost. Which is more important?

Speed dial is great, but what about the time when you really need to know a number that you use often, but you don't have that crutch?

Just for the record, I'm firmly against progress.

Ron Kelly

___________________________________________________________________________

Ron,

Starting from the message of Howard's to which I responded, I think this thread started to drift outside the bounds that were set for it, and I know there is limited tolerance for discussion of plug-ins on this List, so I'll be very brief in answering your questions and leave this tangent thereafter. My basic philosophy is that high quality plugins qualify every bit as much - and often more so - as viable solutions to the imaging factors they were designed to address. They aren't a crutch, they aren't a substitute for knowing what you are doing, they are part of the arsenal, and they are comparators of equivalent value to any other usage of Photoshop for achieving similar objectives.

Mark Segal

___________________________________________________________________________

I was going over some old notes about ghosting images.

If a customer asks us to cutback an image down to 20%, my notes say to cutback (using curves) the black to 0% and the CMY to 30% (10% above what is asked), instead of just moving all four colors to 20%. But my notes also refer to a better way, something about re-seperating the file so that when you cutback the file (again cutback all black and CMY to 30%) it will be more colorful.

The problem is, my notes are vague about the re-seperation settings. Any thoughts?

Andy Adams

___________________________________________________________________________

Howard writes,

I was not familiar with this term. I'm more familiar with "inverted luminosity mask" to describe what you're suggesting. By loading the composite channel, you're forcing Photoshop to convert to grayscale, and then loading that grayscale image, which you arethen presumably inverting and blurring slightly.

Provided you are comfortable with editing the layer mask, this is a powerful way to sharpen and shares most of the functionality of using an inverted black channel. Either is useful for quick and dirty sharpening because they both avoid excessive sharpening of light areas.

The main differences between the two are:

1) Dark, yet colorful, areas, such as the blue and red parts of the flags of Australia, Canada, the UK, and the US, get a lot of sharpening through an inverted luminosity mask but little with an inverted black mask.

2) Relatively pure pastel colors, such as skies and light flowers, have sharpening reduced by an inverted luminosity mask but get almost no sharpening at all through an inverted black mask.

These two differences are pretty minor, IMHO--in most cases easily controllable through editing the layer mask. I would have a mild preference for the black mask, but either can work well.

Dan Margulis

___________________________________________________________________________

Andy Adams writes,

Sharpening this way in LAB can indeed be "better" than RGB/Luminosity, but it's the kind of better where if a person holds a gun to your head and tells you to choose the better one, you can do it if you look hard enough, but otherwise you're unlikely to notice.

A LAB sharpen will produce less noise in the three-quartertone and shadow areas, but it seems easy to control through the layer mask.

When sharpening without a mask, I've stated in my books that around half the time one can't even tell the difference between sharpening LAB and RGB/ Luminosity without looking at magnified channels. Most of the rest of the time the LAB advantage is small. About the only time it becomes significant is when the lighter of two edges is of a light pastel color, which can be lost in RGB/Luminosity mode but is preserved in LAB.

For these reasons, my general sharpening recommendation is (assuming not doing channel-sharpening in CMYK) do it in LAB if convenient, but it isn't worth the bother to make a special trip to LAB for sharpening. I believe that recommendation holds for this mask also.

This brings up the philosophical question of whether hiraloam and conventional sharpening are friends or foes. The above method, and also PP5E, suggests that where conventional sharpening is strong we should be careful about adding too much hiraloam. This implies use of a layer mask to blend the two methods, as above. OTOH, I've also gotten good results by just ignoring the conventional sharpen and applying hiraloam maskless.

One way or the other, though, I'd add steps to the action as follows:

First, AFAIK it's never useful to sharpen the AB channels conventionally--they're way too noisy, and don't help define edges in any case. But in hiraloam, as I discovered *after* PP5E was published, sharpening the AB can help a good deal. In certain cases I sharpen the AB more than the L.

Second, I thought that because hiraloam doesn't make ugly white halos the way that careless conventional sharpening can, it isn't necessary to restrict the lightening function as we usually do in conventional USM. Aesthetically, I now know, the darkening effect of hiraloam is just about invariably more pleasing than the lightening effect.

So, I'd add the following steps to any hiraloam action regardless of whether it uses a layer mask:

First, rather than sharpen the L only, I'd sharpen the LAB, on a separate layer of course. I'd use 100% Amount, which is definitely too high, intending to reduce to taste later.

Second, I'd blend the L channel from the lower layer into the L of the upper layer, Darken mode, 50% opacity, to reduce the lightening effect of the hiraloam sharpen.

Third, I'd set the top layer to Luminosity mode and adjust opacity to taste.

Fourth, I'd duplicate the top layer, reset it to Color mode, and see whether I'd like to reduce or increase the opacity chosen in the third step.

Dan Margulis

___________________________________________________________________________

Dan,

I appreciate the improved approach. Thanks for taking the time.

Andy Adams