Sharpening digitally captured images in Photoshop, or any other editing software is unavoidable because of the softness introduced in the digitizing process, no matter how good the capture device may be. When we digitally photograph the continuous tone real world, we force the view to be broken into pixels. This brings with it the potential to introduce stepped diagonals, interference patterns, and other digitizing artifacts.
To combat this problem, all capture devices use a process called aliasing through which suitable intermediate tones are added to where there are “edges” in the photograph. This alone introduces some softness to the image. Additionally, most cameras use moiré filters to minimize the interference patterns that could be distracting in the finished image. So, we start with a slight handicap of softer than real captures. Depending on the in-camera settings, some “sharpening” may be applied by the camera, or in the case of RAW image captures, nothing is added. Therefore, we need to inject some level of sharpness into the images that are lost in the process of photographing.
Different software handles this in different ways. But the fundamental principle of sharpening is as follows:
- There is an “edge” when a lighter tone abruptly changes to a darker tone
- The “sharpening” process lightens the lighter side and darkens the darker side
- Thus, increasing the edge contrast creating the illusion of sharpness
Sharpening is a carefully controlled workflow that consists of capture sharpening, creative sharpening, and output sharpening. Although each stage may need different strengths of sharpening applied to various parts of the image, their fundamental principle remains the same as explained above. If applied carelessly, sharpening can introduce unwanted halos around the edges detracting from the overall quality of the photograph. So, the process should be such that we introduce halos everybody sees but nobody notices. In other words, the illusion remains enchanting but elusive.
The main sharpening tool in Photoshop has been the Unsharp Mask for many generations in the software. It has improved over time, but the main parameters remained the same. USM is applied by Radius, Amount, Threshold. They collectively control how wide the halos will be, how strong the contrast between the dark and the light sides of the edge will become, and how much difference between the two will trigger the contrast enhancement. As I mentioned earlier, the light halos visible at the junction of lighter tones with darker tones are far more noticeable than the dark halos, which seem to magically disappear into the shadows, literally!
In a straightforward application of the USM filter, reducing the light halo will need one or more of the following:
- Reduce the radius
- Reduce the amount
- Increase threshold
All of these, alone or in combination will reduce the effect of sharpening, perhaps to the point of defeating its purpose. The method I will explain in this post will allow controlling the white halos separately from the dark halos. I call this “Split Sharpening”.
Look at the original and the sharpened images below. Click to enlarge and use the arrows to see one image after the other.
The white halos along the edge of the wind turbine wing, the building roofline, and the wall are certainly visible. I applied USM with a radius of 1 and an amount of 70. I could lower the radius and/or the amount which will reduce the white halos but will not eliminate them.
Split Sharpening
To combat this problem, I will separate the lightening function from the darkening function into two separate layers. First, I will explain how to do this manually then I will provide a download link to a Photoshop action which will do this for you automatically.
My layers palette contains two layers, the original background layer, and a sharpened image layer titled USM, and shows the parameters of sharpening. Follow the simple steps below to enter the comfort zone of Split Sharpening:
- Target the USM layer and change its name to USM R=1 A=70 Lighten
- Change the blend mode of this layer to Lighten
- Press Ctrl-J to make a copy of this layer and change its name to USM R=1 A=70 Darken
- Change the blend mode of this layer to Darken
Now, you can do the following to control the white halos:
- Lower the Lighten layer opacity
- Add a layer mask to the Lighten layer and mask the areas to eliminate halos
- Do both
Compare the sharpened image above to the one below. It is the same sharpened layer simply split into “lighten” and “darken” layers and some areas masked as shown in the mask image.
And the layer stack now looks like the image on the left.
Now, let’s talk about the Photoshop action to split a sharpened layer into lighten and darken components. It is nothing more than the steps above captured in the action which plays the steps back in the same sequence. Download the Ekin Sharpen Action and save it to a location on your computer (right-click on the link and choose to save it).
Then follow the steps below:
- UnZIP the file
- Copy it to where you keep the Photoshop actions. This, on a Windows system, is
C:\Users\Your Name\App Data\Roaming\Adobe\Adobe Photoshop CS6\Presets\Actions
But the uncompressed action file can be anywhere so long as you know where it is - Start Photoshop, open the Actions palette if not open
- Click on the panel menu icon on the top right corner of the panel and choose Load Actions
- Navigate where you saved the Split Sharpen action and click on OK
- From now on, you can make a layer to sharpen, apply an unsharp mask the way you normally do, and then select this action and click on the play icon at the bottom of the Actions palette.
- It will create a group with two layers with lighten and darken blend modes and adjust their opacity to 50% which you can change to suit the image needs and mask either layer to hide the effects of sharpening.
As the title implies, this is part 1, and there will be at least part 2, but there may even be more posts and actions coming on this subject. If you do not normally receive new post alerts, you may want to do so using the Subscribe for New Posts.
Stay tuned!
Here are the full-size images showing the original, sharpened, split sharpened, and the lighten layer mask hiding the halo-producing edges.
J S Turner
Awesome my friend. Also, the Vimeo of the RI Coast is as beautiful as the music.
A. Cemal Ekin
Glad you liked them both Jim.
Cemal
Don Park
I am greatly impressed by the concept of Split Sharpening. I’ve read Part 1 through Part 3, and tried to play Ekin Sharpen Action linked to the Part 3 article.
However, I have difficulties with the following two things:
1. I have no idea how to create a layer mask showing the halos in Lighten layer. (Of course, I know I can add a layer mask, either Hide All or Reveal All, to the image thumbnail.)
2. You set 10 for Amount, 1 for Radius and 0 for Threshold on both USM, and you advise the Action users to fune tune the USM parameters to suit images. As the pre-set Amount values are very low, I can hardly notice if the image is sharpened or not. Therfore I think I need to increase the Amount values to a certain extent, but I am concerned about the following:
(1) Supposing I apply the 100(maybe too high) for Amount to each USM, then my image might be severely degraded by double application of USM,
(2) I understand that Light Halos is rather critical than Dark Halos, so I’d better increase Amount only on Darken-blend-mode layer while leaving the USM parameters on the other layer as set in the given Action.
In fact, I am ashamed to raise such a low level question as this. My Photoshop knowledge is not deep although I have been studying Photoshop and digital darkroon stuff very hard for several months. I hope this doesn’t cause any inconvenience to you.
Don
A. Cemal Ekin
Hello Don,
Layer masking in Photoshop is based on a simple principle. Black hides, white reveals, to the degree of “blackness”. You can paint with black on the parts of the mask where the layer’s impact should be hidden. Generally, but not always, it is better to use a soft brush with a low opacity set on the brush toolbar which is visible when the brush tool is selected. First click on the mask, a thin frame will appear around it to show that it is targeted. Then, using the brush, directly on the image paint on those areas where the effect should be hidden. If you use a brush opacity of, say 30% it will take repeated brush strokes to build up enough black density to start showing the hiding effect. Remember that while you keep the mouse button down you are painting with the “same” brush stroke. When yo let go of the mouse button, that brush stroke ends and when you start painting again, it will be a new brush stroke. If you want to see what and where you are painting, Alt-click on the layer mask. It will show you what areas are painted. Click on the layer icon itself to show your image. To hide the effect of the mask, Shift-click on the layer mask, a red X will appear showing you that the mask is hidden. Shift-click on the mask again to make it active again, the red X will disappear.
Layers are the most potent tool in Photoshop, and also can be the most confusing. Keep at it, it will start making sense.
Cemal
Don Park
Hello Cemal,
Due to time difference(GMT +9 here), I’ve just seen your kind reply which was sent only 22 minutes after my posting for Help. I am deeply touched to see this quick and self-explanatory response to my elementary level of querries.
Regrading how to create a LIGHTEN layer mask to eliminate White Halos which may be produced by Unsharp Mask, I am sure that I can handle it.
May I ask you of one more favor to advise me about the My Concern #2 described in my original post?
Thanks in advance for your help for me to understand and make best use of Split Sharpening concept in my digital photogrphy life.
Don
A. Cemal Ekin
Don, that question escaped my attention. Here is how I suggest you proceed. Think of images with lots of detail, say a landscape with trees, branches, leaves, etc. Let’s call this kind of image “high frequency” detail image. Now think of a portrait of a soft skinned woman or a baby. even a round smooth ceramic urn. Let’s call this kind if image “low frequency”. The radius value of 1 is a good starting point from this point on. Remember a rule of thumb “low frequency images require higher radius, high frequency images require lower radius values. For that imaginary landscape you may use a radius somewhere between say 0.7 to 1. For the soft detailed portrait, you may use a radius value of 1-1.5. These are just hypothetical numbers but should give you a general idea. For the amount, you need to push it until it is too much then pull it back. Lower radius sharpening may accept higher amounts but when you push higher radius sharpening to large amounts you may see stronger artifacts. Also remember the output purpose, screen sharpening should look just right on the screen at intended display size. However, print sharpening should look a little over the top with small white halos visible. This will take some experimenting to get a feel for it. Sharpen, print, evaluate, adjust sharpening, print …
Take a look at an earlier post I have which presents the slides of a workshop I ran:
https://www.keptlight.com/sharpening-workshop/
Cemal
Don Park
Please allow me to add a couple more questions here.
After reading your article Part 1, I downloaded the Ekin Sharpen Action. I did the same things reading Part 2 and 3. After reading through all three articles, I loaded the last Action set linked to the Part 3 article.
1. I guess I only need to play the last Action titled “Sharpen: Split USM SO”. Am I right?
2. Just for clarity, what does “SO” satnd for?
3. I applied Split USM SO action, and changed Amount value from pre-set 10 to initially 100 and decreased to 70 AND Radius value from 1.0 to 0.5 on both USM layers. With relatively low values, I see my image still over-sharpened. FYI, my image is high resolution JPG(840×560 px, 584KB). Is this result we can normally anticipate from double application of Unsharp Mask filters even at lower values setting?
4. I tried to create Lighten layer mask on the Smart Filters already exists inside Smart Object layer by brushing the every edge on the 300% view image with black. The reason I chose 300% view was that I couldn’t see any halos at 100% and 200%, I can see something around the edges at 300%, but I am not sure whether those are enlarged jagged shape of pixels or so-called white halos.
I truly hope that I am not causing headaches to you. However, I would appreciate highly if you understand that I really want to have this wonderful tools as one of the most important asset in my digital workflow.
Don
A. Cemal Ekin
Hi Don,
SO stands for Smart Object, and the filter applied to it becomes a smart filter and its values can be changed.
You may use any one of the actions, but not all or more than once without taking some precautions
Your image is actually a very small, low resolution image and will not take too much to push it over the top. I would try to use a low radius and low amount on that image.
When you view a small image like that at 300% zoom, you will see the pixelation which may be exaggerated by sharpening. As you paint on the filter mask with black you should see the sharpening effect gradually disappear. To get a feel for it, over sharpen the image so you can clearly see the halos, then mask the edges with black. Just to see quick effect, use the brush at 100% opacity.
Cemal
Don Park
Hello Cemal,
As per your guide, I tried again with a duplicate of 26 MB PSD file. On purpose, I put 1.5 in Radius and 200 in Amount on each USM layer. Even in 100% view, yes, I clearly see the halos around edges. With 100% opacity soft brush of black color, I painted all the edges on the image enlarged at 200%. Yes, I confirmed the layer mask eliminate the halos. This test was to see the halos and the layer mask effect to control the halos. Because I used 100% opacity brush, the sharpening effect disappeared completely from edges, other than edges I would have no problem with opacity controllability. The question remaining is to find out where would be the right opacity % to each image in work. Up to this point, am I right?
Next I am encoutering a fundamental issue. Sharpening is a kind of illusion. By lightening the light and darkening the dark around edges, we want to increase the contrast which cause illusion of sharpness. Now I need to compare the two different sharpening approach, i.e. (1) Split Sharpening, and (2) Edge Masks Sharpening.
Split Sharpening allows me to shapen an entire image including fine details and meaningful textures residing at both edges and non-edges. In addition, I can control White Halos taking advantage of Ekin split Sharpening methodology. However, I have to lower the sharpening already made to the edges to control halos. As of now, I don’t know what would be the optimum opacity for brush masking.
Edge Sharpening allows me to sharpen only edges on the image(of course there could be some % of sharpening applied to non-edges depending on the black/mid/white slider adjustment on Levels adjustment layer). If a selective sharpening is needed, then I could add an another step for this during my digital workflow. Edge sharpening will also create White Halos like Split Sharpening. What if I apply the “Blend If” to the output of Edge Sharpening, i.e. to move whte slider from 255 to 205/240 and black slider from 0 to 15/50 of This Layer. By cutting the range of sharpening to the highlight below tonal value of 240 with gradual transition down to 205, I might eliminate(or at least alleviate) white halos. If this idea turns to be vialble, then I don’t have to agonize on tricky parameter settings in USM twice and layer masking, etc. My life would be a lot easier in that case.
I’d like to hear your opinion. Again, please be aware that I am in the porcess of learing Photoshop, so my thoughts are not well organized and must be ridiculous to some extent.
Don
A. Cemal Ekin
Hello again Don,
Think about a building corner, a clean edge against a plain blue sky. In all likelihood that edge will look as sharp as it needs to be even without any sharpening applied. Our eyes will perceive it as “sharp”. So, from those clearly sharp edges you can remove most, if not all sharpening artifacts. The degree to which you mask these edges is controlled by the brush opacity and how dense a mask you create.
In terms of edge sharpening vs split sharpening, there is no reason you cannot combine the two. Instead of painting on the mask manually with brush strokes, you can create an edge mask and apply that to the split sharpened smart object. Even after that, you can fine tune the mask with a brush if you wanted to. Edge sharpening essentially is straight sharpening with a mask in place. If you know how to make the edge mask, you can create it and save it as an alpha channel. After applying the best split sharpening, you can use that alpha channel to mask the split sharpened smart object.
Experiment!