Maximizing Image Detail
Getting the most detail from your images is a complicated subject—heck, somebody could write a whole book on it. Wait, somebody already has! I co-authored the revision to Bruce Fraser’s image-sharpening book: Real World Image Sharpening with Adobe Photoshop, Camera Raw, and Lightroom (Peachpit Press). I can’t stuff the 360 or so pages of that book in this section, but I’ll try to give you the most salient points when adjusting the Sharpening and Noise Reduction settings in Lightroom or Camera Raw.
You want to apply sharpening that enhances image detail but doesn’t do harm. Oversharpening is a bigger problem than undersharpening. You can always add additional sharpening down the road, but it’s very difficult to undo the undesirable effects of overly crunchy sharpening. Add the complications of sharpening to noise reduction, and you’re attempting a balancing act between getting the most image detail while reducing visible noise. Trying to arrive at optimal settings for both is difficult and requires experience and knowledge of the tools.
The reason for image sharpening in Lightroom and Camera Raw is to recover and improve the image detail lost when converting the continuous tone light of a scene into pixels captured by the sensor. The act of demosaicing introduces softness due to interpolation. The sensor itself can cause softness because of the use of aliasing filters. Lenses may have defects and can suffer from diffraction at small apertures. As a result, you’ll need to apply sharpening to improve the apparent sharpness or acutance (edge contrast). That edge part is important to understand. You want to sharpen the edges and not the non-edge surfaces in the image. Edge frequency is the determining factor of what Sharpening Radius you want to apply to your image—and arguably one of the more difficult things to determine. I’ll break down sharpening for three common edge type images: high-frequency, low-frequency, and mixed-frequency edges.
High-frequency edge sharpening
To show an example of high-frequency edge sharpening, I chose this image from Bryce Point in Bryce Canyon National Park. It was shot with a Phase One P65+ camera back using a 45mm lens, which has excellent performance. At the final cropped size, the image is 29.6 x 18.8 inches at 300 PPI. Figure 4.69 shows the entire image (only cropped top and bottom a bit). You see the tiny green rectangle? That’s the area of the image I’ll be showing zoomed to 4:1 in Lightroom.
Figure 4.69. The full Bryce Point image as cropped, showing the zoomed area.
I would normally use a screen zoom of 1:1 in Lightroom (100 percent in Camera Raw) while adjusting the sliders, but I’m showing you this portion at 4:1 so detail will show in the halftone reproduction of the book. The aim of the Sharpening and Noise Reduction is to extract as much usable detail in the image while avoiding any oversharpened artifacts. Figure 4.70 shows the image with four sets of Sharpening and Noise Reduction settings.
IMAGE WITH OPTIMIZED SETTINGS AND NOISE REDUCTION LUMINANCE 25
Figure 4.70. Comparing various Detail panel settings at a 4:1 screen zoom.
Obviously, with Sharpening of 0 the image appears very soft at 4:1 zoom, but at the default settings there’s little improvement in the apparent sharpness. Once the Sharpening settings have been optimized, you can clearly see the improvement in the image detail. Yes, there are visible halos—but remember, we’re at 4:1 zoom so it’s overenhanced; viewed at 1:1, those halos are invisible. The issue with the settings in the third optimized image is that increasing the Detail setting to 90 and Amount to 60 has done some undesirable work on the Luminance. The last image has a Noise Reduction Luminance setting of 25 to help mitigate that noise bloom.
So, how did I arrive at those settings? I worked out of order. Yes, generally, you should work from the top down, but I think a case could be made that the first adjustment for Sharpening should be setting the Radius. This image obviously has a lot of small edges, and the Radius should be set to under the 1.0 default. Figure 4.71 shows the default Radius of 1.0 and adjusted to 0.7 while previewing the settings holding the Option key (Mac) or Alt key (Windows).
ADJUSTED RADIUS SETTING OF 0.7
Figure 4.71. Adjusting the Radius setting.
Reducing the Radius setting reduced the width of the halos and tightened up the sharpening of the edges. You might be tempted to simply use the lowest Radius of 0.5 in a case such as this, but that isn’t really optimal—it would undersharpen the edges because the Radius would be too small. There isn’t a lot of difference between 0.5 and 0.7, but the impact will be visible when adjusting the Amount and the Detail sliders. This is the part of fine-tuning that requires experience. Setting the optimal Radius is critical. The next setting to adjust is the Detail slider, shown in Figure 4.72.
The default setting of 25 has very little deconvolution sharpening and a lot of halo suppression. For high-frequency images, I always increase the Detail slider to reduce the suppression and enhance the fine detail. The setting of 90 is pretty high, but this image could withstand the strong setting. You can see the inevitable increase in the sharpening of the noise, which will be addressed with the Noise Reduction Luminance setting (this preview is before the addition of the Noise Reduction). The next step will be to fine-tune the Amount and Masking settings, as shown in Figure 4.73.
ADJUSTING THE MASKING TO 20
Figure 4.73. Adjusting the Amount and Masking settings.
After setting the Radius and Detail sliders, I went back to the top to adjust the Amount. The Amount is a simple “volume” control, and it’s pretty easy to see how far you can go up to the point you’re doing harm by oversharpening. Also, remember that if the overall Amount setting is good for the majority of the image but produces some suboptimal areas of oversharpening, you can use the local minus Sharpening settings in the Graduated Filter or Adjustment Brush to locally reduce the global settings made in the Detail panel. This ability to mitigate oversharpening is an important consideration, because it can change your global sharpening strategy. The Masking was adjusted so that surface areas (non-edge) didn’t receive the full sharpening. Where the mask is white, the full sharpening is applied; where it’s black, the sharpening is substantially reduced. If you use a very high mask setting, the sharpening in the black areas is almost eliminated. Be careful not to use a really high Masking setting when applying strong sharpening to an image with a lot of noise. You can get an unfortunate rippling effect, where no sharpening is applied to the surfaces and high sharpening is applied to the edges.
The Detail panel adjustments would normally be judged at 1:1. Figure 4.74 is a screen zoom at 1:1 (100 percent). However, due to the halftone screen, you won’t see what I saw in Lightroom when I made the adjustments. But it should give you an idea of what things should look like when optimally adjusted.
Figure 4.74. The optimized Detail panel settings at a 1:1 screen zoom.
You might think that would wrap up this section, but no. I mentioned that optimal Detail panel settings might not be optimal for the entire image. In this case, the Sharpening and Noise Reduction were fine for most of the image but not for the sky. To adjust the global settings in the Detail panel, I used a local adjustment brushed in with the Adjustment Brush to modify the settings being applied in the sky. Figure 4.75 shows the masked image area and “before” and “after” detail figures at a 4:1 screen zoom.
4:1 ZOOMED VIEW AFTER
Figure 4.75. The masked area with local settings and a 4:1 zoomed view of the image before and after adjustment.
The differences are subtle but relevant if the maximum image detail and image quality are the goal (which they always are for me). This sort of local modification doesn’t need to be extreme precision with the creation of the mask. Unlike making tone and color adjustments, locally fine-tuning the Sharpening and Noise Reduction settings only requires moderate accuracy of the mask. Since the full-size image is way too large to be reproduced in the book without substantial downsampling (which would make the image detail adjustments invisible), go back and take a look at Figure 4.69 to see the entire image and Figure 4.74 to see the results of the Sharpening and Noise Reduction at 1:1.
Low-frequency edge sharpening
When sharpening an image with low edge frequency, you want to increase the Radius so small areas of texture and noise don’t get overly sharpened. For example, on a person’s face, you want to sharpen the eyes and lips but avoid oversharpening the skin (unless you want to go for that “weathered” look). The image I’m using as an example is a shot of my friend Daniel Ortize’s wife, Roxana Chåzaro, in San Miguel de Allende, Mexico. Daniel is an avid photographer, so I asked his permission to photograph his wife. Figure 4.76 shows the full frame of the image and a 2:1 zoom detail image of Roxana’s eye. The image was captured with a Canon EOS REBEL T1i with an 18–135mm zoom lens at 113mm zoom. The ISO was set to 800 to get a fast enough shutter speed (1/60 second with image stabilization) to avoid camera shake.
Figure 4.76. Low-frequency example image.
The Radius is again the first slider I adjusted. I knew I wanted to increase the Radius to avoid sharpening her skin and the sensor noise visible at ISO 800. I moved from the default 1.0 Radius setting to a setting of 2.0. Figure 4.77 shows the preview of both Radius settings.
PREVIEW OF RADIUS 2.0 SETTING AT 2:1 ZOOM
Figure 4.77. Adjusting the Radius setting.
By increasing the Radius, I increased the sharpening on the strong edges in the eye while mitigating the noise and skin texture. The preview shows a slightly increased sharpening of the noise, but that will be adjusted later when adjusting the edge masking and adding noise reduction. The next adjustment was to adjust the Detail slider (Figure 4.78).
PREVIEW OF ADJUSTED DETAIL SETTING OF 10 AT 2:1 ZOOM
Figure 4.78. Adjusting the Detail slider.
Adjusting the Detail slider is probably going to be difficult to see in the book’s half-tone reproduction, but let me explain what you’re looking at (even if it’s hard to see). At the default Detail setting of 25, there is still a touch of deconvolution sharpening being blended in, which means the high-frequency texture of skin and noise is getting hit too hard. By reducing the Detail slider to 10, much less of the noise is getting sharpened. The next adjustment, Masking, will cut down on the surface sharpening even further. Figure 4.79 shows adjusting the Masking and Amount adjustments.
PREVIEW SHOWING AMOUNT SETTING TO 88
Figure 4.79. Adjusting Masking and Amount Settings.
The Amount of 88 may seem high, but consider that the Radius is 2.0 while Detail is down to 10 and the edge Masking is up to 68—all of which alters the overall impact of the Sharpening. The Sharpening amount needed to be pretty high, and I actually increased the Sharpening around the eyes and lips with a local adjustment. After adjusting the final Sharpening settings, I adjusted the Noise Reduction Luminance setting to +50 to smooth out the noise. Figure 4.80 shows a less-cropped version, still at 2:1 zoom. Yes, it does look a bit crunchy, but remember that you’re seeing the results at two times the actual pixel size; it should look a bit crunchy.
Figure 4.80. Final global Sharpening and Noise Reduction results.
In addition to the global Sharpening and Noise Reduction settings, I also did a little light-duty retouching and local Sharpening and skin adjustments. The spotting was very minimal—Roxana has good skin, but a little negative clarity never hurt anybody! I also wanted to increase the local Sharpening around the eyes and lighten up a bit under the eyes. Figure 4.81 shows four of the local Adjustment Brush masks doing the majority of the touchup.
MASK FOR REDUCING CLARITY –36
MASK FOR INCREASING SHARPNESS +43
MASK FOR EXPOSURE +0.39, CONTRAST –31, SHADOWS +11, AND NOISE +22
MASK FOR CLARITY –15 AND SHARPNESS –15
Figure 4.81. Local Adjustment Brush masks.
Using a local brush with minus Clarity has the effect of reducing the midtone contrast in the texture of the skin, which is really useful without actually blurring (and destroying skin texture). By adding additional Sharpening just around the eyes and lips, I avoided adding any additional global Sharpening. The +43 Sharpness with the Adjustment Brush is added to the global Detail panel settings. Lightening up the skin under the eyes and an additional touch of minus Sharpness and plus Noise with the Adjustment Brush finished things up. Well, okay, there were a couple more tweaks: I added a very gentle adjustment to lighten the whites of the eyes while darkening the round highlight over the pupils. I also lightened the teeth and added a touch of color and saturation to the lips. However, this isn’t a heavy-duty fashion-magazine type of makeover. The goal was to maintain Roxana’s good looks while keeping a natural appearance. Figure 4.82 shows the final adjusted image at a screen zoom of 1:1 (100 percent).
Figure 4.82. Final portrait of Roxana at 1:1 zoom.
Mixed-frequency edge sharpening
Not every image can be lumped into just high-frequency or low-frequency categories. Using the wrong sharpening can adversely impact the final image quality. So, what do you do if you have an image that is predominantly in the low-frequency camp but has important image detail that needs high-frequency sharpening? You have to get smart! You need to take a slightly sneaky way around the issue. Lightroom and Camera Raw can apply only one type of sharpening at a time, so when faced with this problem, I turn to Photoshop and open digital negatives as Smart Objects. Yes, the image must be sent to Photoshop, but when it’s a Smart Object, you retain the capability of editing raw image-processing parameters. Figure 4.83 shows an example of an image with lots of low-frequency areas and some high-frequency areas in the center of the image. The iceberg image was shot at the northern tip of the Antarctic Peninsula near the Weddell Sea using a Canon EOS 1Ds Mark II and a 24–70mm lens.
Figure 4.83. A mixed-frequency image.
The sky and the water are the low-frequency portions of the image, while the iceberg in the center really needs high-frequency sharpening. To apply a mixed-frequency edge sharpening while retaining the ability to edit raw parameters, I had to take a trip to Photoshop.
First, I’ll point out a limitation of Smart Objects: once you take a raw image into Photoshop, you pretty much sever the relationship of the digital negative residing inside of Lightroom and the raw file that’s embedded into a Photoshop file as a Smart Object. Any changes you make to the Smart Object using Camera Raw can’t be easily changed back to your original negative inside of Lightroom. It’s pretty much a one-way street from Lightroom to Photoshop. Yes, you can save the Photoshop file and import it into Lightroom, but then it’s considered a rendered PSD or TIFF file. You can open the edited TIFF or PSD and re-edit the parameters from inside Camera Raw, but those settings aren’t shared with Lightroom. Another slight hiccup regarding editing a Smart Object in Photoshop from Lightroom is that your Lightroom and Camera Raw versions need to be in sync (another good reason to keep Lightroom and Camera Raw up to date).
One way I work around the issue of severed settings from the digital neg and the Photoshop file is to make all the image adjustments in Lightroom and save Snapshots prior to editing the image as a Smart Object. Since Snapshots travel with the file, Snapshots made in Lightroom will show up in Camera Raw as well. Figure 4.84 shows the two Snapshots I made and the sharpening settings I used for this image.
loW-frequency DetaIl panel settIng
Figure 4.84. The Snapshots and the Detail panel settings.
There are a couple things you’ll notice about the different settings: For the high frequency, Detail is all the way up and the Radius is very small. I did this to really bite into the textural detail of the iceberg ice. For the low frequency, The Sharpening Amount slider is all the way up to 150 (that’s rare, but it’s sometimes useful), but the Detail slider is way down to +10 and the Masking slider is up to +50. I’ll admit that I’m actually pushing this a little bit for the purposes of the example, but as you’ll see, the results are very reasonable.
Once the settings are adjusted and the Snapshots are saved, it’s time to open the digital negative as a Camera Raw Smart Object in Photoshop. From the Photo main menu in Lightroom, select Edit In and the Open as Smart Object in Photoshop option. This renders the image into Photoshop and the Smart Object is embedded into the Photoshop file and appears as a special Smart Object layer. Figure 4.85 shows the single Camera Raw Smart Object layer, as well as the result of copying to a new layer and adding a Reveal All layer mask.
THE SMART OBJECT LAYER IN PHOTOSHOP
COPYING THE SMART OBJECT TO A NEW LAYER
Figure 4.85. The digital negative opened in Photoshop as a Smart Object.
In order to copy the Smart Object layer, you need to do it in a special way: in Photoshop under the Layer menu is a flyout menu called Smart Objects. From the flyout menu, select New Smart Object via Copy. This creates a new Smart Object that can have different parametric adjustments from the original Smart Object layer. Simply duplicating the layer won’t work; it needs to be a new Smart Object. Once the new Smart Object layer is in place, double-click the layer icon to open the Smart Object version of Camera Raw shown in Figure 4.86.
Figure 4.86. Opening the Smart Object into Camera Raw.
Since I already saved Snapshots, I didn’t need to fiddle in the Detail panel. I simply selected the low-frequency Sharpening Snapshot and clicked OK. Camera Raw renders the changes back into the Smart Object layer.
To blend the two edge-frequency sharpened layers, you need to create a layer mask. In this case, since the low-frequency image was on the top of the stack, the layer mask needed to reveal the high-frequency layer on the bottom. Figure 4.87 shows the layer mask painted in and the resulting layer. Where the image is red, the low-frequency image will be hidden.
THE LAYERS PANEL SHOWING THE LAYERS
Figure 4.87. The layer mask for hiding the low-frequency sharpening layer.
The payoff for this mixed-edge frequency sharpening is that it allows you to mix and match the optimal Sharpening and Noise Reduction settings while still retaining the raw editing capabilities of Camera Raw. It’s a true blending of the strengths of combining Lightroom, Camera Raw, and Photoshop to optimize your image detail. Figure 4.88 shows the high-frequency layer and the blended layers.
THE MIXED-FREQUENCY SHARPENED IMAGE
Figure 4.88. Comparing the high-frequency sharpening to the mixed-frequency sharpening.