Sharpening in LAB Color

Intro | Multi-RAW Processing | Creating HDR Images by Hand [Part I] | Creating HDR Images by Hand [Part II] | Sharpening in LAB Color | Converting to Black and White | Using LAB Color Adjustments | Inverting Backgrounds with LAB | Intro to Compositing | HDR in Adobe Photoshop CS5 | Using Image Apply Image | Aging Photos Roundup | Making Colors Pop in Photoshop

Note about the intro image (click on the image to see it larger): For this image of Yosemite Valley to work, the trees in the foreground needed to look convincingly sharp in contrast to the majestic landscape in the background which I wanted to render with a slight golden glow. I used layers and the LAB color space to achieve this selective sharpening effect.

Why sharpen a digital photograph? If the purpose is to “fix” a photo that looks bad because the focus was unintentionally off or because the camera moved and the shutter speed was too slow, then sharpening is often an exercise in futility. The best bet is to go back and shoot it right.

Okay, okay, I know it’s not always possible to create sharp images in camera. And, yeah, I’m not talking about intentionally out-of-focus shots where the blur is intentionally creative, or where you’ve tossed the camera. But come on, if you want the photo to be sharp, do the right things: use a tripod, lock up your mirror, focus with precision, and so on.

However, there is a place for sharpening as part of a creative digital post-processing workflow in the Photoshop Darkroom.

First, it’s worth noting that some sharpening does take place as part of the default Adobe Camera RAW (ACR) conversion into Photoshop—or for that matter when you bring a RAW photo into Lightroom. Check out the Detail tab of the ACR window, shown in Figure 2, to see the default sharpening settings that are possibly being applied without you knowing about it. By the way, my recommendation is to keep these default settings in ACR—even if you are going to compositionally sharpen later in your workflow.

Figure 2: With the default settings you are sharpening an image every time you process it through Adobe Camera RAW.

In addition, there are sometimes reasons to sharpen images to given specifications for the purpose of printing. This kind of sharpening is generally referred to as output sharpening.

This column is concerned with yet another kind of sharpening, which is entirely unconcerned with fixing unsharp originals. The point of this kind of sharpening is compositional, meaning that by sharpening either the whole photo or a selective area you are improving the composition. When done correctly, compositional sharpening can subtly direct the eye of the viewer to certain parts of a composition; this emphasis can work on an unconscious level, and the viewer may not even be aware of what is taking place.

Figure 3: For this photo of a red camellia blossom to work, the petal edges near the interior needed to seem sharp. At the same time, the outer petals couldn’t lose their softness. Selective sharpening at two levels of intensity in LAB color delivered what was needed.

What Bad Sharpening Looks Like

We’ve all seen it. Bad sharpening is generally over-sharpening. Since all sharpening algorithms work by determining the boundaries of objects, and these boundaries are essentially lines, over-sharpening causes the objects in a photo to take on a brittle and almost cartoon-like appearance.

The Photoshop Smart Sharpen Filter is by no means the worst sharpening tool out there, but when applied to an RGB image it can easily be over done—particularly if the photo includes lots of thin shapes to begin with.

Consider the shot of trees following an ice storm in Yosemite Valley shown after processing from the RAW in Figure 4.

Figure 4: Here’s the image without any compositional sharpening in Photoshop.

Figure 5 shows the image with Smart Sharpen being applied at the default settings. This was the way I tried to sharpen before I learned about the technique I am about to show you. Sometimes it worked reasonably well, but other times I was dismayed by the crude results—particularly when I tried to boost the sharpening effect above the rather modest default values.

Figure 5: Smart Sharpen is probably better than nothing, but it often creates portions of images that are pretty unattractive.

Figure 6 is a blow up at 300% of the results of applying the Smart Sharpen filter to the RGB image. Please compare the image full size and blown up with sharpening using my LAB technique shown in Figure 6 and Figure 11.

Figure 6: Sharpening applied to the RGB image makes the branches look ugly and “too sharp” in an artificial kind of way if you look closely (shown here at 300%).Compare with comparable L-channel sharpening shown in Figure 11.

Let’s face it, this sharpening is ugly! There’s got to be a better way, there is a better way, and I’m about to show it to you.

Understanding LAB

Before I get to my sharpening technique, I need to digress and briefly explain LAB color. Don’t worry, there’s nothing very complicated here!

To construct the colors we see in the world and display them, a color model—sometimes called a color space—is used. Two of the most common color spaces are RGB and CMYK. The RGB—or Red, Green and Blue—color space is used to display photos on a monitor. The CMYK—or Cyan, Magenta, Yellow and Black—color space is primarily used for printing.
You may not know as much about the LAB color space, which is structured in three channels:

  • The L channel contains luminance information. Luminance is another way of referring to the black and white data in a photo.
  • The A channel provides information about greens and magentas.
  • The B channel provides information about yellows and blues.

The ins and outs of LAB color are beyond the scope of this article. I’ll explain more in an upcoming column Using LAB Color Adjustments.

For now, from the viewpoint of sharpening, the important point is that the black and white information is completely separate from the color information. The advantage of this is that you can sharpen the black and white pixels without any impact on the color pixels. This has two pluses:

  • Many shapes are outlined in black, so black is what you want to sharpen.
  • Color pixels tend to get unattractive and just plain nasty when sharpened—since sharpening the LAB luminance channel doesn’t touch the color channels this gnarly effect is avoided.

Since LAB separates black and white information from color information, keep in the back of your mind that it is possible to use a double-barrel approach in some cases: sharpening the luminance channel and adding blur to the color channels. This technique is beyond the scope of the relatively straightforward sharpening techniques explained in this column, but you might consider this giveth-and-taketh approach to sharpening as a potentially interesting special effect.

Note that output devices do not “play well” with LAB color. Before you can use an image that you’ve sharpened in LAB, you’ll need to convert it back to RGB or CMYK.

Converting to LAB Color

To convert to LAB color, choose Image > Mode > LAB Color (Figure 7).

Figure 7: It’s simple to convert a photo into the LAB color space.

Alternatively, you can also convert to LAB color by selecting Edit > Convert to Profile and choosing LAB from the Profile drop-down list in the Destination Space area (Figure 8).

Figure 8: It’s more formally correct, but functionally equivalent, to use the Convert to Profile dialog to convert to LAB.

Once you’ve converted an image to LAB mode, if you open the Channels palette as shown in Figure 9, you’ll see the three channels I’ve mentioned: Lightness, or L, containing the black and white data; and the A and B channels for the color information.

Figure 9: The three LAB channels are displayed as thumbnails in the Channels palette.

Using the Unsharp Mask

The Unsharp Mask Filter is one of the sharpening techniques in Photoshop that goes back a ways. At first blush, “Unsharp Mask” seems an odd name for a filter that in fact sharpens. The name comes from an old film darkroom technique that involved compositing an original photo with a blurred version of itself to bring out boundary lines—and the digital unsharp mask works much the same way as its analog namesake.

To use the Unsharp Mask on the L channel, open the Channels palette and select the Lightness channel (as shown in Figure 10). You should also make sure that all three LAB channels are visible—this is indicated by the “eyeball” icons shown in the left-hand column of the Channels Palette in Figure 10.

Figure 10: The Lightness channel is selected, and all three LAB channels are visible.
With the L channel selected as I’ve described, the next step is to open the Unsharp Mask filter (Filter > Sharpen > Unsharp Mask) as shown in Figure 11.

Figure 11: The Unsharp Mask Filter is shown applied to the L channel, with low intensity settings selected.

The Unsharp Mask window gives you three settings, Amount, Radius, and Threshold. The simple version is that setting Amount higher, and Radius higher, each give you more sharpening. Conversely, of course, lower Amount and lower Radius each give you less sharpening.

Threshold works in the opposite direction: the lower the Threshold the more sharpening, and the higher the Threshold the less sharpening.

Confusing, right? You’ll get used to it pretty quickly if you use these settings:

  • Leave the Threshold setting untouched at 9 levels.
  • Start with the Radius setting at 2.5 pixels. See what happens as you increase this setting gradually to 4.2 pixels.
  • Start with the Amount at 50% and gradually increase to 125%.
  • Magnify your image to at least 100%, If things look over-sharpened, scale down your Radius and Amount settings.

When you sharpen just the L channel, the effect may not be as pronounced as RGB sharpening depending on your settings. But is usually fairly pleasing (see Figure 12). You have to go way overboard with LAB sharpening to get lines as harsh as those that are often generated using sharpening in RGB.

Figure 12: The L channel sharpening looks attractive even when blown up to 300% (compare with RGB sharpening shown in Figure 6).

Of course, some photos look better when they are sharpened using different settings. More frustratingly, not all the areas within an individual image are created to be sharpened equally. For example, some parts of the Yosemite Ice Storm photo look better with slightly stronger settings (see Figure 13).

Figure 13: If you sharpen more extensively in LAB some areas look better than at lower settings.

The good news is that selective sharpening—strong where you want it, weak in other areas and maybe none at all for some parts of your photo—is easily possible using layers, layer masks and a layer stack.

Working on Layer Masks

I’ve made up a little rhyme to help you remember how to selectively sharpen with layers:
Sharpening Twice, Naughty and Nice.

In other words, the “nice” layer provides a bit of mild sharpening for a fairly wide area of the photo. The “naughty” layer does the hard stuff—a lot of heavy duty sharpening for the small areas that selectively need it.

Here’s how it works. Duplicate the background layer (Layer > Duplicate Layer). Name the new duplicate layer something reasonably descriptive, like Sharpen a bit.

Next, sharpen the L channel of the duplicate layer using moderate settings and the techniques I’ve explained earlier in the article.

Add a black Hide All layer mask to the Sharpen a bit layer (Layer > Layer Mask > Hide All. Paint in the areas you want moderately sharpened using the Brush Tool set to 100% opacity and 100% flow.

Next, go back to the background layer, and duplicate it again. Drag the new duplicate layer up to the top of the layer stack, and name it something like Sharpen a lot.

Apply the UnSharp Mask filter to the L channel of the Sharpen a lot layer with the Amount and Radius settings much higher.

Now, add a black Hide All layer mask to the Sharpen a Lot layer, and paint in the areas you want to be intensely sharpened.

You’ll end up with a stack of sharpening layers like the ones shown in the Layers palette in Figure 14. The thumbnails of the layer masks in the Layers palette give some idea of the relative areas that have been sharpened.

By the way, sometimes two sharpening layers are not enough. There’s nothing to stop you from adding additional layers for even more flexible sharpening if you’d like.

Since a layer stack works from the top downwards, the few areas that have been heavily sharpened are visible even if these areas have also been lightly sharpened. Areas where the layer masks are completely black for both the Sharpen a bit and the Sharpen a lot layers don’t get any sharpening at all.

Figure 14: Layers, layer masks and a layer stack are used to selectively sharpen different areas of an image at differing intensities.

Using LAB L-channel sharpening you can get pleasing results and it is harder to overdo sharpening. If you add the power of layers and masking to this technique, you get extreme flexibility in using compositional sharpening to create attractive imagery (see the finished Yosemite Ice Storm photo shown in Figure 15 for an example).

Figure 15: Selective LAB sharpening helps to give this photo contrast and a feeling of depth.

I often apply selective sharpening to very differentiated areas of images—for example, the edges of the petals of the Camellia blossom shown back in Figure 4. When I do this, the Layer Mask for the sharpening layer looks a bit like a reverse stencil of the parts of the image that have been sharpened (Figure 16). By the way, a layer mask is just another channel like R, G and B in RGB or L, A and B in LAB, so you can always take a layer mask out for a “spin” and close inspection via the Channels palette.

Figure 16: I wanted to sharpen just the petal edges of this Camellia (see Figure 4 for the final version) so I “painted” the petal edges on the black layer masked used to control the impact of the LAB sharpening layer.

After you are satisfied with your LAB sharpening, don’t forget to archive a layered copy of your photo. Then, flatten the image and convert it back to the RGB color space.


I use LAB L channel sharpening in almost all my images. In addition, it’s the rare photo of mine that doesn’t have some of this sharpening applied selectively using layers and masking. The great thing about this kind of sharpening is that it is like wearing snazzy underwear—you feel better, your photos look good, but nobody knows quite why. From a compositional viewpoint, you can draw attention to specific parts of a photo with selective LAB sharpening in an understated way.

This article has:

  • Explained the different reasons that photos are sharpened
  • Compared RGB with LAB sharpening
  • Discussed the point of compositional sharpening
  • Provided a practical overview of the LAB color space
  • Shown you how to convert an image to LAB color
  • Demonstrated using the Unsharp Mask Filter on the L Channel of a photo in the LAB color space
  • Shown you how to create and combine individual layers, each with different amounts of sharpening, with each layer applied to a selective area

If you need to know more about how to work with layers, layer masks, and layer stacks, you’ll find detailed step-by-step explanations of these concepts and techniques in The Photoshop Darkroom (Focal Press). The Photoshop Darkroom also explains how to work with LAB color generally, and how LAB sharpening specifically can fit within your Photoshop workflow.

Other Photoshop Tutorials

Creativity in the Photoshop Darkroom by Harold Davis: Multi-RAW Processing | Creating HDR Images by Hand [Part I] | Creating HDR Images by Hand [Part II] | Converting to Black and White

Advanced Photoshop Tutorials by Jay Kinghorn: Layer Masks | Smart Objects | Advanced Masking | Image Sharpening | Burning and Dodging


Harold Davis is a photographer and author. His photographs have been widely published, exhibited, and collected. Many of his fine art photography posters are well known. Harold’s images have won a Silver Award in the International Aperture Awards 2008 competition, and inclusion in the 2009 North American Nature Photography Association Expressions Showcase.

Harold is the author of The Photoshop Darkroom: Creative Digital Post-Processing (Focal), Creative Composition: Digital Photography Tips & Techniques (Wiley), Creative Night: Digital Photography Tips & Techniques (Wiley), Creative Close-Ups: Digital Photography Tips & Techniques (Wiley), Practical Artistry: Light & Exposure for Digital Photographers (O’Reilly Digital Media) and other books. Harold gives frequent digital photography workshops, many under the auspices of the Point Reyes National Seashore Association.

Text and photos ©2010 Harold Davis.

Sign in or Sign up to post response

    • Thank you for sharing your expertise. Much appreciated!
    • To post a reply Sign In
    • Same as Kelby's workshop, not too mention that by converting to LAB one loses tons of quality and then converts it back to sRGB makes it even worse. It might be the easiest of methods, but definitely not preferred or perfect. Anyway, thanks for sharing.
    • To post a reply Sign In
    • Didn't know that Kelby taught this technique, good for him! My tests don't show a loss of quality in converting to LAB and back, Photoshop uses LAB for its internal calulations no matter what space you are working in. And why on earth would you ever work in sRGB? Adobe RGB or another broader gamut RGB space is a better choice.
    • To post a reply Sign In
    • Fair enough. It does not matter what colour space one works in, when converting one loses information. Please see this image: Test image: Outcome: top: RGB; middle: converted to LAB; bottom: converted back to RGB This simple test has been done by our colleague at Canon Forum. A lot of information is irreversibly lost. I am not questioning that this method does not work I am saying that it loses quality, although probably not noticeable in print ;) Regards, Lukasz
    • To post a reply Sign In
    • If you start (as some would suggest you should), in high bit (more than 8 bits per color or what Photoshop calls 16-bit which is actually 15), the loss is moot. In 8-bit, its quite significant depending on the color space. Every time a conversion to LAB is produced, the rounding errors and severe gamut mismatch between the two spaces can account for data loss, known as quantization errors. The amount of data loss depends on the original gamut size and gamma of the working space. For example, if the working space is Adobe RGB, which has 256 values available, converting to 8-bit LAB reduces the data down to 234 values. The net result is a loss of 22 levels. Doing the same conversions from ProPhoto RGB reduces the data to only 225 values, producing a loss of 31 levels. Bruce Lindbloom, a well-respected color geek and scientist, has a very useful Levels Calculator, which allows you to enter values to determine the actual number of levels lost to quantization (see the “Calc page” at If you do decide to convert into and out of LAB, do so on a high-bit (16-bit per channel) document. There’s little reason to push the processors and do a mode change when you can use various forms of sharpening in RGB and just fade luminosity. Its not 100% the same as using the L* (Lstar) but its indirectly close and it solves the main advantage here, not altering the color data. Plus you have the opacity slider. And its all on an adjustment layer (less but not totally non destructive). The ground breaking work on sharpening was from the late Bruce Fraser who’s book (recently updated by Jeff Schewe) is the recipe book for the rounds of sharpening described in this original piece by Bruce: Capture and output sharpening in the Raw pipeline is available today based on Bruce’s work in Lightroom 2 and ACR 5. The only thing you really need to mess with is capture sharpening which is all creative in nature.
    • To post a reply Sign In
    • Thanks for all the very informative comments. I'd suppose it goes without saying to work in 16-bit (or higher) whenever possible.
    • To post a reply Sign In
    • >I'd suppose it goes without saying to work in 16-bit (or higher) whenever possible. And in the Raw processing pipeline whenever possible. That and (first) noise reduction.
    • To post a reply Sign In
    • Thanks, much appreciated knowledge to add.
    • To post a reply Sign In
    • Thanks for sharing.
    • To post a reply Sign In
    • Andrew, What do you mean by "just fade luminosity"?
    • To post a reply Sign In
    • >Andrew, What do you mean by "just fade luminosity"? Look under the Edit menu. You can fade any last applied effect and use blend modes and opacity as well.
    • To post a reply Sign In
    • Can hardly wait to try this. I've tried a variety of approaches to sharpening shadowed edges in order to emphasize texture, for example by creating a black and white conversion layer, increasing its contrast drastically and than layering it underneath a transparent colored layer. The results have been interesting, but not natural, with dramatic color changes and granulation of the image.

    • To post a reply Sign In
    • Excellent article and comments.  Just a small caveat:  your illustrations #6 and #11 are both labeled as 300% views, for comparison of smart-sharpen and L-channel sharpening, but they don't appear to actually be the same magnification.  For instance, if you identify the "Y" in the lower right of #6 (about 1/5 of the way from the bottom and a little more than 1/4 of the way from the right edge), and then look in #11, the same "Y" (just a little above and left of center) is clearly larger.  It would be nice to be able to make a truly same-magnification comparison.

      Thanks again for a very nice article!

    • To post a reply Sign In
    • hi Harold, I really liked your article!  I hope you don't mind that I quoted you and linked to your article on my own blog post about sharpening.  If you don't want to be linked/quoted, I can remove it; let me know. Here's my website. Thanks for the informative tutorial about LAB sharpening. :)

    • To post a reply Sign In

Sign in or Sign up to post response