The Hidden Power of Photoshop Elements 4 PHẦN 5 ppsx

Generally, you will want to choose a setting with a low threshold, which is measured in levels. This means Photoshop Elements will look at the number of levels of difference in the surrounding pixels, and if the number of levels is greater than the threshold, it will apply sharpening based on the settings for Radius and Amount. You should usually keep the Threshold setting between 0 and 5 levels—toward the lower end of this range. In fact, you will often want to use zero tolerance. Low tolerance levels (1 or 2) can keep Photoshop Elements from sharpening what is otherwise image noise. Sharpening noise will only make the image noisier. With that in mind, a good rule of thumb is to raise the threshold more for images with more image noise. This will keep you from enhancing image noise. The only time you will set the threshold higher than 5 is when you want to limit the filter’s effect to high-contrast areas of the image to play up existing contrast and separation of image elements. Radius and Amount might be set quite differently depending on what you are trying to achieve and the ppi and content of the image. In many cases, you might apply the filter twice: once with a low radius for general sharpening and once with a higher radius for broader enhancement of image contrast. The following sections cover these techniques. Affecting Sharpness with the Unsharp Mask Filter Sharpening an image with the Unsharp Mask filter depends on Elements recognizing and enhancing existing edges in an image. In other words, if the image is too soft or blurry to have defined edges, the filter can’t tell where the edge is, so it can’t tell what to enhance. This is why the filter works best to sharpen images that are already characteristically sharp. Although settings can vary depending on the type of image, the size of the image, and the desired result, you will normally maintain the following settings in 300 ppi, 5 × 7" images that have average busyness and contrast: Radius: 0.5 pixels to 3 pixels Amount: 50% to 100% Threshold: 0 levels Note that these are rough guidelines, but they work for a variety of situations. If the con- tent of the image is not busy, lacks focus, and/or is lower contrast, you can tend toward the high end of the ranges; if the image is busier, is relatively sharp, and has high contrast, you would probably tend toward the low end. If the image has more resolution, you would tend toward the higher end of these ranges; less resolution, tend toward the lower. The name Unsharp Mask comes from the traditional darkroom process in which an inverted, blurred (unsharp) duplicate of the original image was sandwiched to mask the exposure dur- ing printing. This helped target and adjust (sharpen) contrast differences. (un)sharpening and boosting contrast ■ 119 4456c05.qxd 3/1/06 3:05 PM Page 119 Figure 5.21 shows an image before and after sharpening. Before sharpening, the image appears slightly soft and perhaps a bit lacking in contrast. A single application of the Unsharp Mask filter in the midpoint for the suggested range increases the contrast and boosts the sharpness (see Figure 5.21). A halo effect occurs when the Unsharp Mask is applied too strongly over areas where flatly dark portions of an image meet flatly light portions, forming a high-contrast edge (see Figure 5.22). Often, a halo effect is more apparent when the applied Radius is short— or not long enough to dissipate the edge of the sharpening effect without being obvious. Not only will the halo tend to blow out (or clip) areas of images, but the image will also distort, and the effect will become unpleasant. You can reduce the halo effect by first undoing the Unsharp treatment and then reap- plying it with either a broader Radius, a lower Amount, or a combination of these or by blending the result. Methods of blending are discussed in later parts of the book. One quick method of blending is to flatten the image (Layer ➔ Flatten Image), duplicate the Background (Layer ➔ Duplicate Layer), apply the Unsharp Mask filter, and reduce the Opacity of the Background copy layer in the Layers palette. Figure 5.22 The original image and an oversharp- ened counterpart: Although some sharpening may be desirable, a halo effect probably will not be. Figure 5.21 Sharpness and con- trast in this image improve with an application of Unsharp Mask. 120 ■ chapter 5: Correcting Image Tone 4456c05.qxd 3/1/06 3:05 PM Page 120 Another option for blending the sharpening effect is to use the Fade tool in the Hidden Power tool set. You can find the Fade tool in the Power_Extras category under Effects on the Styles and Effects palette. Just undo the sharpening (press Command+Z/Ctrl+Z) and double-click the Fade tool. Instructions will appear on-screen. Staying within the guidelines helps you avoid oversharpening and creating halos in high-contrast areas of your images. Better to sharpen a little several times, sharpen a duplicated layer, or try other measures than to sharpen hastily and heavily and damage the image. Raising Local Contrast with Sharpening Adjusting contrast with the Unsharp Mask filter has a much different effect on an image than applying Curves or Levels, because the effect actually compares adjacent pixels rather than adjusting based on a more predictable scheme. The results of the application will be unique as they are dependent on the differences that already exist in the image. Applying Unsharp Mask to increase contrast works well with low-contrast images or images that seem to lack dynamics that increasing dynamic range (with Levels) or contrast (with Curves or Gradient Maps) doesn’t fix. When you are adjusting local contrast with the Unsharp Mask, the Radius might be much higher than suggested for sharpening (50 pixels or more) and the Amount much lower (between 10% and 50%). Again, these are just suggested ranges. The goal of these broad settings is to increase the radius beyond the distance where a halo is noticeable and to force objects in the image to play against one another. Figure 5.23 shows a somewhat low-contrast image of a boat (a) and corrections using the Unsharp Mask filter (b and c). Although it looks okay in color, the image lacks a little pop in black-and-white. By using Unsharp Mask to raise the local contrast, the image ele- ments have more separation from one another. Two applications of Unsharp Mask—one to build local contrast (b) and one to sharpen (c)—make quick work of what would other- wise be an arduous task in masking to separate this boat from its surroundings, creating a result based on existing image differences. The third image (c) includes slight Levels and Curve preset adjustments. Curve presets were used to enhance the contrast that Unsharp Mask brought out, and Levels was used to adjust the final brightness. Some tone was replaced quickly by duplicating the original, moving it to the top of the stack, setting it to darken, and lowering the Opacity to 10%; this filled in areas that sharpening had forced to clip (go to 0% gray). So you see that tech- niques of using Levels, Curves, and Unsharp Mask work together—rather than separately— to produce enhanced results. (un)sharpening and boosting contrast ■ 121 4456c05.qxd 3/1/06 3:05 PM Page 121 Manual Unsharp Masking: Calculations in Action While calculations were briefly introduced in the last chapter, they have many creative and interesting uses, most of which are not immediately obvious. One of the first really useful layer calculations I worked with was manually creating unsharp masking effects. Unsharp masking, as mentioned earlier, was a darkroom process before it was a filter. The photog- rapher doing the developing would sandwich a blurred film negative copy of the image with the original to burn in (increase exposure of) the image shadows. The blur would target the contrasty edges, and the result after the application would be increased shadow detail and a sharper look to the image. This Hidden Power application of unsharp is a little different but builds on the same concept. Imitate a modified version of the darkroom sharpening effect by using the following steps: 1. Open a flattened image to which you’d like to apply an unsharp mask calculation. 2. Duplicate the Background layer. 3. Invert the Background Copy layer (press Command+I/Ctrl+I or choose Filter ➔ Adjustments ➔ Invert), and rename the layer Unsharp Mask. a bc Figure 5.23 The original boat is a little dull (a). After raising the local contrast with the Unsharp Mask (using a broad radius and low per- centage), the boat stands out better from the surround- ings (b). The filter is then applied again (c), but this time to sharpen the image (using a short radius and higher percentage). 122 ■ chapter 5: Correcting Image Tone Copyright © 2005 photosphere.com 4456c05.qxd 3/1/06 3:05 PM Page 122 4. Blur the Unsharp Mask layer using Gaussian Blur. The size of the blur will depend on the resolution of the image and the amount of detail. The more detailed the image, the less blur; the higher the resolution, the greater the blur. Start with 5 pixels for a 3 × 5" image at 300 ppi; use more pixels in the radius for larger images. 5. Change the Layer mode of the Unsharp Mask layer to Overlay. 6. Reduce the opacity of the duplicate layer to 50%; adjust the opacity as desired. The result of these steps is a sophisticated mask based on the content of your image. The Unsharp Mask layer you have created ends up working much like the sandwiched negative in the darkroom process, in pretty much the opposite way than the unsharp mask filter does, reducing image contrast and pulling details from shadows and highlights. Because the effect is nearly the opposite of the Unsharp Mask filter, the two sharpening effects can often be used together to greatly intensify image sharpness. Alternately, you can apply curves or other adjustments to increase contrast to offset the manual unsharpening. The best results for the unsharp masking will often be had by applying the change just to the image tone (or separated Luminosity). If you apply the changes to the color as well, the change will enhance color noise. By targeting tone, color won’t be altered, and you may achieve better results than applying sharpening to tone and color at the same time. A means of doing this in reverse is to add back the color from the original image, which you could do by duplicating the Unsharp Mask layer before making any changes and setting that layer to Color mode. Other adjustments, such as increasing saturation and making a tone adjustment, may be necessary to compensate for alterations that the Unsharp Mask inflicts. The Hidden Power Sharpen tool included with the Hidden Power tools (in Power_ Adjustments) will go through the simple sharpening process described above in steps 1 to 6, allowing you to select the intensity of the blur. The tool goes several steps further, to include the option for Hue/Saturation and Levels corrections, apply the Unsharp Mask filter to tone, and group all of these corrections into a clipping mask. To use this tool, open the image on which you want to use the tool, and double-click the Hidden Power Sharpen tool. You will be able to adjust the intensity of the separate sharpening applica- tions by using layer opacity after the tool runs you through the process. You can view Before and After images of the effect by toggling the view for the Snapshot Composite layer, and you can adjust the whole effect by lowering the Opacity for the Snapshot Composite. This is layer masking and calculations working together to create a powerful tool and repeatable result. (un)sharpening and boosting contrast ■ 123 4456c05.qxd 3/1/06 3:05 PM Page 123 Managing Image Noise Management of noise in your images is another way to adjust tone. Having noise in an image can mean a number of things, from having many objects in an image to something more akin to random digitized information—like you might get when turning on your TV without a cable connection or antenna. The concepts in this section deal mostly with the latter and will help you to reduce or eliminate undesirable patterning, digital image noise, and texture in an image by controlling image noise. At times you may want to edit out, reduce, or even add image noise to achieve particu- lar results in your images. To do this, you will use Blur and Add Noise filters, often in con- junction with other image editing functions, such as layer blending, selection, masking, and perhaps a few Hidden Power tools. The Gaussian Blur filter (Filter ➔ Blur ➔ Gaussian Blur) can blur images or image areas. It does this by averaging the effects of pixels over a radius, which you define by using a slider on the dialog box: the greater the radius, the more intense the blur. As a result of averaging pixels, blurring removes or lessens image noise by lowering the contrast of adja- cent pixels. This averaging softens image edges, smoothing hard lines between areas of contrast, and can mitigate or obliterate image details. Essentially, this is the opposite of sharpening, which enhances existing contrast. The Add Noise filter (Filter ➔ Noise ➔ Add Noise) generates image noise by randomiz- ing color assignments for pixels. There are several choices in the Add Noise dialog box for controlling the filter. The Amount is related to Percentage, and it defines the range of vari- ation possible in creating the noise distribution. As the Amount goes up, the application of the noise is potentially more radical. Very strong applications of noise, like blur, can obliterate detail—in this case by wiping it out with random behavior rather than averag- ing. Noise applications can swiftly become something of a special effect, depending on the ppi of the image. An image with a higher ppi, containing more pixels and resolved detail, will be able to withstand stronger applications of the filter. Generation of noise is affected by the Distribution Type. A Uniform distribution changes the values of individual pixels by selecting a random number within the range defined by the Amount. This number can be the original value plus or minus the amount for each color component. For example, applying an Amount of 25 to a 50% gray image (128 levels) in Grayscale will result in values between 103 and 153 levels of gray for any pixel, each value generated at random. You can see this effect by creating a new image, filling with a 50% gray background, and applying the Add Noise filter. A Gaussian distribution changes the values of individual pixels by selecting a random number based on a Gaussian function. The function creates a tendency to select from the center of the range, but the results can also deviate more strongly from that norm. While the quantitative effects to each pixel can extend beyond what is dictated by Amount, 124 ■ chapter 5: Correcting Image Tone 4456c05.qxd 3/1/06 3:05 PM Page 124 the total effect is the same as uniform distribution—just with greater peaks and valleys in the deviation. Because deviations can be broader, Gaussian noise can appear to be a stronger effect than the Uniform distribution with the same Amount setting, as shown in Figure 5.24. Compare this to the Uniform effect by creating another gray image of the same size, applying the Add Noise filter using the Gaussian setting, and comparing the images side by side. The Monochromatic option applies the filter to only the tonal elements in the image without changing the color. For example, this would keep an RGB image that has been desaturated from generating color noise when the filter is run. So far, neither adding noise nor blurring may sound desirable, because either could be damaging (or at least compromising) to an image. However, used in a controlled fashion, both can enhance an image and make the results look more realistic. For example, an image or image area that is damaged by JPEG compression can be restored, somewhat, by selective blurring. In this case, blurring could potentially dissipate artifacts generated by compression. In a similar way, some types of digital noise can be lessened or removed, as might be effects of film grain, halftone printing, and paper texture. Selective blurring can also help in isolating image subjects by imitating effects of focus, such as depth of field. Original gray Uniform Noise @ 10% Gaussian Noise @ 10% Figure 5.24 A magnification of a 25% gray area shown with Uniform and Gaussian noise distributions of 10% The histograms show the flat appli- cation of noise in the Uniform distribution and the bell-shaped application in Gauss- ian distribution. managing image noise ■ 125 4456c05.qxd 3/1/06 3:05 PM Page 125 On the other side of the coin, most image tones that look natural in an image are not completely flat when you look at them close up. When you attempt to add new objects or elements to an image, such as by painting them in with flat tones or creating an area with blends (for example, to replace a sky), the elements can tend to look too perfect. The result is that the repair will look like a repair: skin tones will look more like a mannequin or caricature; a replaced sky will look like a fill or gradient. Applying noise can mimic a more natural look by randomizing and effectively dithering image information. In addition, you can use Add Noise to create texture or graininess (for example, to mimic film grain). As strange as it may seem, sometimes when blurring won’t solve a problem that you may have thought required noise reduction, applying noise can. Even more often, apply- ing both blur and noise can do the job, adding variation while mediating extremes. Blur and Add Noise filters can be used along with other functions to produce the best results. For example, you might make a selection of a particularly noisy area of an image to isolate it before applying a blur. After using Blur, you may need to use the Add Noise filter to fix the blurred areas so that they don’t seem flat. You also might use a layer mode, mask, or other features to isolate the application so the fix is applied only to the areas where it is intended. Both Add Noise and Blur are best when applied lightly and in combination. Figure 5.25 shows a repair in which noise was used to make an image correction blend better after a blur was applied. The skin on the subject is not bad in the original, but it could appear much more youthful with smaller pores. A selection was made of the subject’s face (mostly using the red channel as a mask—a technique we will look at more later). With the selection loaded, the area of the subject’s skin was copied and pasted to its own layer to isolate it. Once isolated, the copied skin was blurred. The results smoothed the skin, but left it too flat. Noise was added, using Uniform distribution with the Monochromatic check box selected. This returned some of the texture to the skin without leaving it looking too flat and fake. The appearance of the pores was softened first, using the Blur filter to remove noise. Next, the Add Noise filter was used to gently replace some of what was lost in the texture. The result is much smoother skin and a somewhat more youthful smile. Several techniques, including erasing information on the new layer or masking, could have been used to bring back the details that were getting covered. However, Blend Mask, a Hidden Power tool, was applied to the layer to enable the character of wrinkles to blend through based on tone. Again, combinations of tool applications and functions generally work best in achieving goals in an image because no one tool can do it all. You’ll see more of Blend Mask in the following section. We’ll also talk more about noise reduction in Chapter 6 in a section called “Quelling Color Noise.” 126 ■ chapter 5: Correcting Image Tone 4456c05.qxd 3/1/06 3:05 PM Page 126 Masking with Image Tone Masks are very much like selections, in that they can help you isolate image areas and work on them without changing other parts of the image. Masks can be based on tone (as we will look at in this section), color, saturation, manual selection, or other calculated results. Masking is quite flexible. The greatest advan- tage of masking is that it is not permanent; masking allows you to temporarily hide parts of specific layers in the image rather than erasing or otherwise permanently altering the layers. If you want, you can adjust the masking to add and subtract layer content at will by adjusting the layer masks to reveal or hide more of the layer. One frustration with using Photoshop Elements is that it natively uses masks only in conjunction with adjustment layers, and it does not freely allow you to mask any layer in an image. It also doesn’t allow you to work fluidly with alpha channels, which is where Photoshop can store masks and selections. Although you can save a selection in Elements 4, you still can’t attach a mask to any old layer—that is, you can’t unless you have Hidden Power. a masking with image tone ■ 127 c b d Figure 5.25 Original image(a) is shown closeup (b). Blur is added (c), and then noise for texture (d). Copyright © 2005 photosphere.com 4456c05.qxd 3/1/06 3:05 PM Page 127 The Hidden Power tool set can help by offering several tools for working with masks. We’ll explore native possibilities and the Hidden Power enhancements for masking in the next few sections. Native Layer Masking Layer masks are available natively when you are using Adjustment layers (such as Levels or Gradient Maps, found on the Layer ➔ New Adjustment Layer submenu). A layer mask is a mask attached to a specific layer that applies to that layer only. All this means is that the content of that layer will be targeted at the areas that remain unmasked. Masks created with adjust- ment layers reveal all the content of the layer by default. How- ever, you can control layer masking of adjustment layers by making a selection before you create the layer. Instead of trying to explain this technique, it is best shown in an example. 1. Open a new image at 500 pixels × 500 pixels, 72 ppi, RGB, and use a white background. 2. Change the foreground color to red (R: 255, G: 0, B: 0). 3. Fill the Background layer with red. Use the Paint Bucket tool (press K). 4. Create a new Hue/Saturation Adjustment layer (Layer ➔ New Adjustment Layer ➔ Hue/Saturation). 5. Shift the Hue slider to –120 and choose OK. This should turn the image blue. The Layer palette should look like Figure 5.26. 6. Choose the brush tool, and make the foreground color black (press D to set default colors, and then press X to exchange the foreground and background swatches). The foreground color should be black. Change the brush to 100 pixels in diameter using the Options bar. 7. Paint in an M shape (for “mask”) over your image. As you paint, the black will be applied to the layer mask for the adjustment layer, and you will reveal red from the layer below. See the thumbnail change in Figure 5.27. If you want, you can adjust the Hue/Saturation sliders (double-click the Hue/Saturation layer icon to open the Hue/Saturation palette) to see how the image is affected. The black portion of the mask will block the corresponding portion of the layer with which it is associated. While masking is a neat trick, it actually has many interest- ing applications. Open any image, and then double-click the Highlight Mask tool from the Power_Masking category of the Hidden Power tools. This will make a selection of the shadow areas in the image. Now make a Levels adjustment layer 128 ■ chapter 5: Correcting Image Tone Figure 5.26 The layer mask for the adjustment layer is represented in the Layers palette by the thumbnail to the right of the layer thumbnail. In this case there is nothing masked, so the thumbnail is white. Figure 5.27 The mask for the adjustment layer will now have a black M. The image will show a red M surrounded by blue. Black masks out the effect of the layer. 4456c05.qxd 3/1/06 3:05 PM Page 128 [...]... the 0% Location and a white (RGB: 255 , 255 , 255 ) color stop at the 100% Location 12 Add a new color stop for each sample at the Location that corresponds to the B values in the card, and change the color of the stop to the same value Since there are three samples, you will create three stops as shown Location 75% 50 % 25% HSB value 0, 0, 75 0, 0, 50 0, 0, 25 13 Accept the changes by clicking OK on the. .. depending on the size and resolution of the image 4 Point the tip of the Eyedropper at the light gray reference area in the image, and click to sample the color 1 54 ■ chapter 6: Color and Tone Enhancement 5 Double-click the Sample Analysis tool in the Power_ Extras category of Effects This will create a new image from the color sample separate from the RGB and leave a swatch of each of the component... Sample each of the tones in the Sample Analysis image, and note the B value of each Write them down, and average them Close the Sample Analysis image without saving it 7 Repeat steps 4 to 6 for the medium gray and dark gray swatches in the image Your results should be similar to those noted here Color (as B from HSB) 75% 50 % 25% Red 89 51 22 Green 82 59 35 Blue 82 56 30 8 Make an RGB separation of the original... value of 43 makes the red change only 75 percent of the way, leaving some respect for the value you originally measured in the image The following list shows the target values for making the change with 75 percent strength (deviating 25 percent from the original measurement of color in the image): Color Original Value Red 54 Target Value 43 Green 26 36 Blue 36 38 In this case, you would take the same... and resolution of the image Larger images with more resolution can usually stand broader sample areas 4 Point the tip of the Eyedropper at the gray reference area in the image, and click to sample the color 5 Double-click the Sample Analysis tool in the Power_ Extras category of Effects This will create a new image from the color sample separate from the RGB and leave a swatch of each of the component... percentage) In the example, the average value is still 39 ( 54 + 26+36/3) Using the red channel, which measures 54 , your equation would look like the following if you were applying 75 percent of the change: new target value = 39 + ([ 54 – 39] × 0. 25) , or 39 + 4 = 43 ■ 151 Figure 6.10 Gradient Map adjustment layers should be placed between the component tone and its associated color layer  152 ■ chapter 6:... enhance the tonal range See Figure 5. 32 11 Apply the Unsharp Mask filter to the content of the Sharpen layer Be sure you are Figure 5. 32 This Levels adjustment to the mask will make sure that darker portions of the image in the masked layer appear in the image 1 34 ■ chapter 5: Correcting Image Tone Figure 5. 33 This Levels adjustment to the mask will make sure that darker portions of the image in the masked... change the color of the stop to the same value (in the example, HSB: 0, 0, 39) 12 Accept the changes by clicking OK on the Gradient Editor and Gradient Map dialogs 13 Change the name of the Gradient map layer to Gradient Map Template 14 Duplicate the Gradient Map Template, and drag the copy above the Red Component Change the name to Red Map Adjustment 15 Duplicate the Gradient Map Template, and drag the. .. only half of the very long tails in the Green and Blue components and doing an additional adjustment to the midtones in RGB The first part of the adjustment makes the shadows in the image seem pale, but adding a shift in the midtones compensates—pushing more image information toward the shadows Once that’s accomplished, the tone of the image seems more natural and balanced See the result of these Levels... shadows, we can mimic the effect 1 Open the twins_on_beach.psd image on the CD 2 Isolate the highlights in the image using the Shadow Mask tool in the Power_ Masking category of Effects 3 Copy and paste to place the highlights on their own layer, and name the layer Glow This effectively masks the highlights into their own layer 4 Adjust the highlights to be brighter by opening the Levels (press Command+L/Ctrl+L; . Gaussian Blur. The size of the blur will depend on the resolution of the image and the amount of detail. The more detailed the image, the less blur; the higher the resolution, the greater the blur 127 c b d Figure 5. 25 Original image(a) is shown closeup (b). Blur is added (c), and then noise for texture (d). Copyright © 20 05 photosphere.com 44 56 c 05. qxd 3/1/06 3: 05 PM Page 127 The Hidden Power tool. the thumbnail is white. Figure 5. 27 The mask for the adjustment layer will now have a black M. The image will show a red M surrounded by blue. Black masks out the effect of the layer. 44 56 c 05. qxd