Photoshop has a dizzying array of tools and techniques available to sharpen and enhance detail in your images. That causes lots of confusion, and gives rise to certain mythologies that not only become persistent, but also tend more toward dogma than wisdom. My own mythology is based on the idea that sharpness is a perceptual element, rather than technical. While we can talk about image processing and math, and compare notes on untold ways to sharpen, in the end, the image is the important thing. Sharpening should not only present the image in a pleasing way, it should also support the intent of the image.
So how much should you sharpen? Exactly enough. But not too much. As you try various tools, keep in mind the creative aspects. Rather than worry about which tool is best or what “perfect” sharpening looks like, think about how the effect changes how you see your work. It should be a conscious part of your work, not a compulsory move or afterthought. Learning about the various methods gives you choice, and choice is power.
Sharpening Tools in Photoshop
Looking under the Filter>Sharpen menu shows a handful of different ways to sharpen your images. Each generally does the same thing (except Shake Reduction and the Motion Blur option in Smart Sharpen), which is to increase contrast on a very small scale. Each tool uses some variation on a special kind of equation called a kernel, which is specifically meant to operate on an array—in this case, a two-dimensional grid of pixels. A kernel operation looks at each pixel and compares it to its neighbors, then writes a new value for that pixel and moves to the next one, repeating the process for every pixel in the image.
The basic Sharpen filter uses a simple kernel with no variables for you to control. When you apply it, you get what you get. If you want more effect, you repeat the filter. If you want less, you’ll have to apply it to a copy of your image and lower the Opacity of the copy. Sharpen More behaves the same way. Except, you know, with more. Sharpen Edges makes a selection from the Find Edges filter and only applies the effect there.
Unsharp Mask (USM)
Unsharp Mask gets fancy, letting you adjust a couple of variables called Amount, Radius, and Threshold. In the kernel above left, each box represents a pixel, and the numbers are values that get multiplied to the brightness value of each pixel. In this example, the kernel involves nine total pixels, so is called a 3×3 kernel. The next size up would be a 5×5 kernel, etc. The Radius slider controls how many pixels are involved in the operation. Note that the Radius slider allows changes of 1/10 of a pixel, so you can get really small changes.
The Amount slider is effectively a scale for the basic results. Think of it like a power control. It acts to increase the contrast between pixels, whereas the Radius slider determines how many pixels are involved in the kernel. They work together and can balance each other.
Threshold is like a cutoff filter that limits the effect to areas where the pixels are at least that number of brightness levels apart. The higher the number, the bigger the difference needs to be for the effect to be applied. Any areas that have less difference get ignored.
So, the Amount slider reads out power, the Radius slider reads out numbers of pixels, and the Threshold slider reads out levels of contrast. These are the basic building blocks of the more advanced sharpening tools, whether they’re built in to Photoshop, come from a plug-in, or are applied with layer-based techniques. Understanding these basic principles is important to helping you explore and choose different sharpening methods, as well as developing your own variations.
When applying USM, start by setting both Radius and Threshold to 0, then crank up the Amount to around 200%. Ease up the Radius slider until you start getting what I call “ringing” halos—the kind where the edge contrast is high enough that you just barely start seeing those white edges around some features. Then roll back the Radius a tiny bit, and start bringing down the Amount slider.
Look around the image to see if there’s noise in the darker areas, including along edges. If so, nudge up the Threshold slider. The values you actually use are going to be relative to both the size of your image and the kind of features you want sharpened, so you’ll have to tinker.
High Pass Sharpening
Unsharp Mask does have something fun going on under the hood: It uses the same principles as the legendary High Pass sharpening technique. The High Pass (HP) filter also uses the same basic kernel idea, but it shows only the output result of the kernel without blending it back into the original image. In other words, the HP filter is only doing half of the work; the result still needs to be blended back into the main image. We do this using one of the Contrast blending modes: Overlay, Soft Light, Hard Light, Linear Light, and Vivid Light.
In fact, all of the Sharpen tools use the same method of filtering and blending, it’s just that Sharpen, Sharpen More, and Sharpen Edges don’t give you any controls. Unsharp Mask does the blending for you (you don’t get to choose the blend mode), but lets you “fiddle with the knobs.”
Let’s take a closer look at HP sharpening. Zoom in to 100%, create a copy of your photo layer to be sharpened, and then go to Filter>Other>High Pass. The only slider is Radius and it does exactly what you’d expect. The reason HP is a filter is because it only shows you pixels that are changed by the kernel, turning everything else to 50% gray.
Remember what I said earlier about sharpness being a perceptual element? Here’s what I mean: Sharpness in a digital image is really about contrast between small features. The Radius slider is controlling the size of the features that are included, and that’s what you see onscreen. This is also the start of Frequency Separation; the size of the features is considered a frequency. The Radius slider is choosing which frequency to show you.
Small Radius values will apply to only the tiniest details, while large values will affect bigger features, and will often start to cause a halo effect, especially in areas such as clear skies against a contrasting landscape. To start with, choose something smallish, just to pick out fine details. Being able to see the HP filter in action helps you visualize the Radius slider’s behavior in other sharpening tools. For overall sharpening at output, choose a higher Radius value, but mind the potential for halos and other artifacts.
Now we need to use that result to change microcontrast. Choose the Overlay or Soft Light blending mode for the HP layer, or Linear Light for even more effect. Each Contrast blending mode gives slightly different results. You can further refine the effect by lowering the HP layer’s Opacity, and by opening up the Layer Style dialog and working with the Blend If sliders. [For more on the Blend If sliders, see “Blend If Masking” in the May 2018 issue of Photoshop User, or Lesson 6 in Dave Cross’s Mastering Layers: Advanced Techniques course.—Ed.]
To get the next level of control, add a Curves adjustment layer clipped to the HP layer, and then set a pin at the 128/128 point in the middle of the Curve. This locks in the gray values so only the correct features get affected. Tiny changes to the Curve can have dramatic effects on the sharpening, and you also get to choose whether you’re changing only the dark edges, light edges, or both.
If you’ve ever oversharpened an image and ended up with ugly halos, hard edges, and other artifacts (and we all have), then this technique should have obvious merit—you can tweak the light side or the dark side mostly independently.
A good application of HP is when you want to see the sharpening mask itself, and perhaps do some direct correction or spot manipulation with the Sharpen tool, which I’ll talk about later.
Next up is Smart Sharpen. It does all the same stuff as USM, with most of the advanced HP technique, but it does it in one convenient spot. Smart Sharpen should be your starting point for most sharpening work in Photoshop. Yes, I really committed to that statement, so let me explain.
Within Smart Sharpen, you get the familiar Amount and Radius sliders, but Threshold has been renamed “Reduce Noise,” and it behaves a little differently. Instead of choosing a cutoff contrast value, it smooths out the sharpening effect.
The Blend If technique we mentioned with HP is handled by the Shadows and Highlights sliders. That is, these sliders provide for a transition range so that the applied effect doesn’t have a harsh cutoff. The Fade Amount sets how much of the effect is applied to the region: 0% means everything in the shadows or highlights gets sharpened, while 100% means nothing in those areas is sharpened.
Tonal Width is like a band-pass filter that defines the range of the shadow or highlight region. Higher values include more tonal range, while smaller values restrict the effect to only the brightest or darkest regions. If the darkest areas of your image are noisy, set the Fade Amount to 50% and bring up the Tonal Width slider just to the point that those areas aren’t included in the sharpening.
Radius in this case has to do with the transition from sharpened to unsharpened areas, measured in pixels. A larger Radius value here means a smoother transition from the unsharpened areas to the sharpened.
Finally, there’s the Remove drop-down menu, which offers your choice of Gaussian Blur, Lens Blur, or Motion Blur. I generally like to use Lens Blur because it smooths tiny, specular highlights. Gaussian Blur is a much more general algorithm that assumes blur is uniform across the entire image, and it tends to oversharpen tiny highlights. The difference can be very subtle, but try it out on some smooth skin with modest pores and specular highlights. You should be able to pick out the slightly harsher effect from Gaussian Blur.
Motion Blur attempts to apply the sharpening filter in a specific direction, but between you and me, skip it; use Shake Reduction if you have movement issues.
Smart Sharpen does most of what you can do with the HP technique. The Amount slider has an extended range, and when combined with the Shadows and Highlights sections, it mostly replaces the Curve and Blend If features. You still don’t get a choice of blending modes and you can’t easily visualize things such as Radius and Amount but, given the other features, that may be a moot point.
Use Smart Sharpen the same way as USM, but realize the controls have slightly different sensitivity. If you find there’s too much noise, or you get sharpening artifacts (squiggles) in smooth areas, use the Shadows and Highlights sliders.
Camera Raw Details Tab
For several versions now, Photoshop has allowed Camera Raw (CR) as a filter (Filter>Camera Raw Filter), and that gives you access to the top-tier sharpening capability, the Detail controls. In the Detail panel, you get the usual suspects of Amount and Radius, but Threshold is gone altogether. Instead, there are Detail and Masking, along with quite a few Noise Reduction sliders.
Detail increases the edge contrast without increasing the radius or the power overall. It’s meant to bring out fine textures without introducing artifacts. On many images, the effect is very subtle, but for fine-grained textures, such as sand and cloth, it can be pretty dramatic.
Masking is an edge-finding filter that restricts the effect. At 0, sharpening is applied everywhere, but at 100, only the highest contrast boundaries of the smallest features get sharpened.
The coolest part about CR’s Detail tab, however, is that it lets you visualize the effect even better than HP! The trick is to hold Option (PC: Alt) while adjusting the sliders. When holding the modifier key, Amount shows a desaturated image so your eye isn’t fooled by color variations and only sees luminosity. For Radius, you see something similar to what you get with HP, but again desaturated. With Detail, the result is the brightness of only those tiny edges. And Masking shows you white where sharpening is applied, and black where it isn’t.
Noise Reduction does what it’s supposed to, but is split into Luminance and Color controls so you can adjust each independently. Do not underestimate the power of these sliders! Pay special attention to dark areas and regions with high saturation but low detail; these sliders can help prevent the weird little artifacts introduced by the Sharpening section, and do a little correction on general noise from the camera itself.
The algorithm behind the CR sharpening is a little more advanced, resulting in less chance of the obvious artifacts you can easily get with the previous techniques; however, it isn’t as strong overall, reflecting the needs of photographers who are simply trying to remove the softness inherent in digital capture.
After all is said and done, this is my favorite tool for overall sharpening because it’s fast, accurate, and easy to control.
If sharpening is about enhancement, Shake Reduction is about correction. Shake Reduction uses a Deconvolution filter to try and analyze the movement of the camera during a capture, and then works backward to recover what should be the original image. Opening Shake Reduction begins the analysis, and produces an estimate of the blur path if possible. Ideally, you should choose an area that was the intended focal point by moving the Blur Estimation Region box and reducing its size to just what’s needed to determine the shape of the blur, called a Blur Trace. Photoshop will attempt to draw a shape that estimates the movement, apply that shape to the Deconvolution filter, and then sharpen the results.
The Blur Trace Bounds let you tweak the result of the shape a little to correct for “almost, but not quite” results, including jagged edges and halos, or effects that look like the Emboss filter. Generally, it should be left alone unless you absolutely must adjust it. Even then, you’re more likely to get good results from moving the Blur Estimation Region around a bit.
Source Noise, Smoothing, and Artifact Suppression all work to try and prevent jagged, noisy elements from being augmented by the process. Again, Source Noise should be left on Auto for most operations, while Smoothing can be adjusted for general noise. Artifact Suppression deals with larger artifact issues that tend to look like compression more than noise.
Shake Reduction works best on images with very small amounts of motion blur, especially when there’s a tiny specular highlight that clearly shows the movement.
The Sharpen Tool
No, I didn’t forget this tool; I saved it to introduce the more philosophical elements associated with sharpening. In particular, the best ways to approach sharpening an image, and what you should be looking for. More and more photographers consider sharpening as a creative element just as much, if not more so than technical. That means they’re sharpening at different stages than the traditional wisdom of “at the very end,” and they’re applying different amounts of sharpening to different regions of their images.
In fact, many pro shooters prefer to use the term “microcontrast” to move away from the idea of sharpening as detail enhancement. Microcontrast pays homage to the perceptual aspects of these techniques, and implies a more organic, smooth result. This is where the Sharpen tool comes in. (You can find the Sharpen tool under the Blur tool in the Toolbar.)
This tool does the same thing as the Sharpen filter, but it only applies it where you brush. In real time, the tool samples the canvas, applies a level of sharpening, blends the result with the original pixels, and paints the output back to the canvas. The advantage is that you don’t have to attempt to sharpen an entire 36M layer to get results in only a few small areas. That’s to say, you can easily target your sharpening effects.
The tradeoff, of course, is that you don’t have the fancy sliders and other tools available to you. But don’t give up on this spiffy tool, just yet. With the tool active, check out the Options Bar: You get a handful of blend modes that get applied as you brush; there’s a Strength slider; and, best of all, you can paint the result to a blank layer by turning on Sample All Layers. The last option, Protect Details, is to reduce pixelization, which does work, so I recommend leaving it on.
Enabling the option to stamp to a new layer gives you all kinds of flexibility. First, it means you don’t have to create a flattened copy of everything before applying sharpening (more on this in a bit). Second, you can choose exactly where to apply sharpening and in what strengths. Using a digitizing tablet such as a Wacom Intuos, you can slowly build up sharpening just like painting. Third, you can treat the output like any other layer content, including applying layer styles, additional blending and opacity, and filters. (Look for a detailed article on the Sharpen tool in an upcoming “Photoshop Proving Ground” article!)
Before I leave you to experiment, let’s talk briefly about workflow and intent. The usual wisdom is to sharpen for output, and do it at the very end. But every image needs at least a little sharpening straight out of the camera. Sometimes the tool you use to get photos onto your computer does this, and sometimes it’s done when you’re about to open the image in Photoshop (or Lightroom). This is called “ingest sharpening,” and should really be treated like preparation to get a neutral starting point for later editing. Tread lightly and err on the side of softness to leave some room. Sharpening too much at ingestion is likely to create artifacts as you progress.
Next is “in-process sharpening,” which is where microcontrast comes into play. At this stage, the goal is creative application, and is usually where you start to apply finishing touches to local features. Mostly, this comes after heavier retouching such as frequency separation, healing, and big contrast adjustments. The image is being worked to look its best in Photoshop, before output considerations are folded in.
In-process sharpening should have a goal, such as drawing the eye through the image, or calling attention to individual features. Many artists go back and forth through iterations of overall and detail enhancement at this point, because the two techniques affect each other. Again, restraint and balance is key.
Finally, the classic output sharpening! This happens when you’re done with everything else in the image, and you need to make the picture look great for different viewing situations; whether it’s social media, being projected on a screen, or printed on any number of different substrates. For enlargements, you may even need to go back and reduce some of the creative sharpening to prevent artifacts from being exaggerated. Generally, however, for print and screen, you’ll need to make a smaller version of your image, and then enhance sharpening specifically for how the image will be viewed. Social media especially will require additional sharpening to make up for the compression and reduction in size.
This means working nondestructively to the extent possible. Keeping separate layer groups (or even separate files) for sharpening allows you to be selective in making later adjustments.
So how do you know which tool to use? Start with what you know best; but also practice other techniques. There are subtle differences in most of the methods, and your perception (there it is again) of how much control each gives you is one of the more important aspects. They’re your tools after all. Make them work for you.
With that in mind, for Smart Sharpen, Unsharp Mask (no!), and the Camera Raw Detail panel, I recommend the general method of setting everything to 0, cranking up the Amount, then easing into the effect by raising Radius, and bringing Amount back down. Then, look to the noise and artifacts. Also keep in mind that you can lower the Opacity of the sharpened layer, and when sharpening, zoom in to 100% in the area of interest—don’t worry about areas you’ll sharpen differently, or will mask out. At the end, look to smoother areas to make sure you’ve not added any artifacts. If so, mask them out, or apply a tiny bit of smoothing with the Blur tool.
Remember, too, that you don’t have to stick to one method or tool, even in the same document. Use global adjustments such as Camera Raw or Smart Sharpen for ingesting your starting image. Layer-based techniques or brushing with the Sharpen tool are great while working in-process. And finish with Smart Sharpen for output.
There is some broad philosophy to go with this as well. Remember the question about how much to sharpen? It turns out that usually less is more. Viewers are more likely to notice and be distracted by oversharpening. Very few will be bothered by slight softness, especially if the image itself is otherwise strong. In the end, however, you’re the artist and you get to decide how much and what to sharpen.
In preparing to write this article, I had some great conversations with a number of amazing professionals (see below). The general theme was to move away from sharpening as detail enhancement, and toward microcontrast as expressive control. Think less technical, more creative. Images from newer cameras, especially medium-format and some of the higher-resolution 35mm systems, may need very little help once in Photoshop. Conversely, the iris image above was shot with a phone, so consider the entire process of capturing through processing if you feel you want more detail and sharpness. Oh, and a few anonymous Photoshop engineers want me to send you this message: Stop using Unsharp Mask! They put a lot of effort into Smart Sharpen and the Camera Raw Detail tab, and it breaks their hearts when they aren’t used.
Remember: Never let your tools get in the way of your art!
I’d like to thank the following for their wisdom, time, and generosity:
- Robert Clark
- Karen Hutton
- Vincent Versace
- Russell Williams Adobe Chief Architect, Photoshop
- Noel Carboni Adobe Senior QE Developer, Photoshop
- Eric Cheng Adobe Senior Principal Scientist
- Meredith Stotzner Adobe Product Manager, Photoshop
This article originally published in the June/July 2019 issue of Photoshop User magazine.