« Digital and its Discontents | Main | Six Comments on the Comments »

Sunday, 12 August 2007


Feed You can follow this conversation by subscribing to the comment feed for this post.

This technique (useful at that) has been around for a while. It's known under a number of guises as hiraloam (high radius, low amount), local contrast enhancement, haze reduction etc. The new Clarity slider in ACR 4.1 (and presumably Lightroom 1.1) is another implementation.

I do the same thing - I learned about it from the Luminous Landscape article a year or two ago. They just call it contrast enhancement and suggest 20 50 0 instead of 20 60 0 as a starting point. See: http://www.luminous-landscape.com/tutorials/contrast-enhancement.shtml

I should add that if you're using USM directly, it's best to limit the effect in the shadows and highlights with Blend If. The whole can be accomplished with an action.

One of my favorite "tricks"!! Also similar to the 'Clarity' slider in the CS3 version of ACR.

I do think, however, that the best printers (the operators, not the equipment) know how to "work the system" and are capable of amazing results. But that's only part of the process...

Wow, thanks for the tip Ctein! As far as image editing goes, I am mostly a dummy. I always wondered what the Low frequency sharpening in Neat Image was for. And I wasn't sharp enough to try and observe what it is doing :-) It apparently works in the same manner as your PS trick!

(Please forgive my ignorance, I can't figure out if Ctein has a first name. I couldn't find it anywhere! Or maybe it IS the first name?)

Thanks mate. Will try. What about the paper? There was a period where the manufacturers were boasting about ultra long life ink jet papers (I believe Canaon was claiming 25 years+). I notice that they seem to have dropped the boast now. Is this now so standard that they don't claim it, or did they have to cut it because of quickly fading prints? I had a few go downhill rapidly.

That technique is called Local Contrast Enhancement. It's pretty widely known:

Where Unsharp Mask is typically used to increase contrast at high resolutions, in this case it's being used to increase contrast at lower resolutions.

You're right ... this is a really great technique. I saw it first at Luminous Landscape



This technique is also called local contrast enhancement:


The vibrance control in ACR 4.1 and Lightroom 1.1 does much the same thing.

For a subtler, more controllable take on this technique, try Photoshop's 'High Pass' filter -- 'Filter' --> 'Other' --> 'High Pass' -- on a copy of the original, radii of 20-50 and a layer blending mode of 'Overlay', 'Soft Light' or 'Hard Light' (and attendant opacity) to vary the effect. Same basic effect, an increase in mid-tone contrast. Bruce Fraser lays this all out quite well in his 'Real World Image Sharpening'.

Ctein only has one name. It's pronounced "kuh-TINE."


P.S. He had to register on one site as "none Ctein" (first name: none [couldn't be left blank]; last name: Ctein) so sometimes I call him None Ctein...I don't think he appreciates it though.

Like any PS technique, the photographer applies it to the degree or percentage that enhances the resultant image. I just don't think that the characteristic of flatness is inherently bad and snap is good. Maybe this contrast enhancement compensates for some illusion in tonal separation deficiency in inkjet prints made from small megapixel originals. We've all been wooed by images as flat as Twiggy as well as those with crackle.

Dear Elauq,

Well, it ain't an illusion, and it's got nothing to do with the number of pixels. In fact, I first observed it in prints from files created with a 22 MP Phase One back. It's something I've seen across a wide variety of inkjet printers, ranging from cheap to pricey, on images ranging from a few million pixels to a hundred million. I won't swear it's universal, but it's damn near ubiquitous.

This isn't about artificially boosting the contrast, this is about getting the printer to render tonality that's there in the original. Think of it as a tweak that's needed to get tonality to print correctly, the same way that most printers require one apply a bit of sharpening to a file to get fine detail to render properly.

Before you pooh-pooh it, give it a try.

pax / Ctein

Dear Aussie,

Doesn't seem to be a property of the paper used. The tonality loss looks about the same in glossy, semi-glossy and matte prints, so far as I can tell. Not something I can measure quantitatively, though; so, it's just my impression.

Paper life does still get mentioned if you read the right promotional materials. Overall, though, everyone's numbers are so high that it's more about how much faith you have in our understanding of the degradation modes than relative life spans. These days, a projected life of 25 years is considered pretty lousy. This is only true of OEM papers + inks. Generic third party stuff is still mostly crap-- if the price is low, the quality usually is, too [sigh].

pax / Ctein

Thanks for trying to help us, Ctein. Before I even got to the end of the article, I just had a feeling that there would be several "I already knew that" posts in the comments section. So I scrolled down, and sure enough, there they were! Certainly some readers of TOP didn't already know about this, so thanks for sharing the info. Not all of us are omniscient.

Yeah, like everyone else, I've seen it before. For a typical image off digicam, I find 15px radius can be quite useful - on occasion I'll sharpen locally at something like (100, 1.0, 1) and then wider with (15, 15, 1), subject to some kind of `characteristic radius' of objects that should be sharp.

I remember a friend saying USM is basically introducing errors around the pixels, so it tends to be the *last* thing I do after the major levels/curve work. Works nicely to thicken shadows.

Alternatively, you could do Orton with the layers blended using the Multiply function for obvious reasons. Maybe even omit the gaussian blur stage and just multiply the image on top of itself to a given degree.

>>>P.S. He had to register on one site as "none Ctein" (first name: none [couldn't be left blank]; last name: Ctein) so sometimes I call him None Ctein...I don't think he appreciates it though.<<<

Don't call him "Mr. Ctein" either. He hates that. I was just trying to show him that I respected his talent and amazing dedication to photography. It doesn't matter though, he's good people.

Indeed, this technique is very similar (although not technically identical) to the "local contrast" boost now available as a "Clarity" adjustment in Lightroom 1.1. This technique, however, has the advantage of being maskable and therefore can be selectively applied.

Another similar simple, but generally overlooked/unknown/forgotten method for punching-up an image is to use a High-Pass filter overlay. The dance steps:

1. Duplicate the layer you want to punch-up, placing it on top of the master layer. Be sure to set its type to OVERLAY.

2. Select the layer, then: Filter -> Other -> High Pass...

3. Adjust the degree of punch by adjusting the radius. A setting of 5-12 usually works best for me.

I should note that this is a mild sharpening technique closely kindred to Ctein's suggestion. The "local contrast" (a.k.a. Clarity) technique is a selective color contrast technique.

The fun of Photoshop is that there are often so many ways to produce very similar results.

Duotones, as described by Steven Johnson in his new Digital Photography book, is another way. Here is Ken Rockwell describing the process:

Anyone like to comment on the difference between this technique as described, and doing the same thing but only in LAB color with only the Lightness channel used?


Ctein --

I asked the question that follows in a forum once
and got little response.
(Of course. People join forums only to argue.)

Decisions on how to vary the Amount
in this local contrast technique seem obvious to me
would we want to vary the strength of the Radius setting
based on the contents and/or resolution of the digital file?

For example, if the photo contains lots of fine landscape detail
such as branches, leaves and stems,
wouldn't we want to use a smaller Radius ?

This seems to be the case when applying this type
of local contrast enhancement on-screen.
But because I am a total printing slacker,
I don't have any real evidence there.

[...heh, heh...ermmmm....I just told Ctein, of all people, that I am a total printing slacker.....]

I've not seen that before, I tried it and it really seems to give photos a bit of a kick. What sells it for me is when you undo it, it looks like it adds a haze back over the photo.

Thanks! I just applied the 8%, 60, 0 settings to unsharp mask one of my rather low contrast images, and the improvement is amazing, i.e., without going over the top. The contrast is still not as high as on a sunny day, for example, but just right and looks natural, more realistic for an image taken under the shade.

I call it "digital haze" when I'm talking to myself about what's on the screen. It's as if a light veil were between me and the image. I like: amount=25 radius=8 or 15 or 30 or 60 threshold=0. Pick one of the settings and click the preview on/off to test it. I usually use the lowest setting that makes the image 'pop'. Portraits seem to like 8 and 15, but objects/landscapes like 30 or 60.

Dear Folks,

Indeed, no honorifics should be attached to "Ctein." (read into that what you will [grin]).

Occurs to me that I really should have laid out my general workflow.

1) Adjust tone and color to approximate correctness, so I can see what I'm working on. No pixel values below 20 or above 240, to avoid clipping; curves more linear than I'd want in a final print. But in the ballpark.

2) Do the fiddly bits. (That's pro talk for all the local adjustments, spotting, tweaking, etc.)

3) Make final tone and color adjustments.

4) Make a test print. If it's deficient in some respect, do more fiddly bits and tone/color refinement.

I do the local tonality enhancement between steps 1 and 2.

pax / Ctein

How do I create an adjustment layer, as suggested by Nuno de Matos Duarte for the Unsharp mask....I only see options like levels, curves etc, which will undoubtedly influence my adjustment layer?

"How do I create an adjustment layer, as suggested by Nuno de Matos Duarte for the Unsharp mask..."

He meant a duplicate layer. Technically the Blend If numbers used are dependent on the working space gamma.

If you want something that will do it all, download the "TLR Professional Sharpening Toolkit" from here:


The effect you want is "Haze Reduction" in the Creative Sharpening part. Because this is JavaScript, you can look at the logic for all the routines and build your own actions to do whatever you want. Most of the sharpening routines out there (Photokit etc) are just repackaged actions.

If you're starting from RAW however, you may be better off just using ACR 4.1. This does pretty well everything you want (other than final output sharpening) and you can revisit the settings over and over again if you open the RAW file as a Smart Object. Pretty slick!


I use this method all the time, and I apply it early in the process, soon after raw processing. The reason is that my judgments of brightness and contrast change after doing the local contrast enhancement, so I feel there's no point in doing such adjustments until after the LCM has been applied.

It's a good trick - it does remind me of the "clarity" tool in Lightroom, but I much prefer to be able to do it in Photoshop.

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007