Filters are like small software applications within Photoshop that enact sometimes dramatic changes to our images. Some are very simple, almost trivial; some are like full-featured image-editing applications. Most are controlled via a dialog box with a few sliders and usually a preview. Some take over the screen with their own workspace.
There are 70 stand-alone filters plus 2 galleries: the Filter Gallery with its 47 filters, and the Blur Gallery featuring 5 of the ways to blur (out of nearly 20 in total). You can mix each gallery’s filters to create a combination effect, and then add other filters too. It’s dizzying!
Significantly, most filters can (and should) be applied to Smart Objects to become “Smart Filters.” This allows you to experiment more freely, trying combinations of filters with different settings to achieve a desired effect.
Below, I’ll describe many filters, but not all 122! I’ll discuss approaches and strategies and show some examples. I strongly recommend setting aside time to experiment, guided by this text. An additional resource is provide by Adobe itself: their (do a web search for that phrase and you’ll find it immediately). Nothing beats experience, however. Play with filters as often as you can to build a visual vocabulary. The most elaborate is the Camera Raw filter. Its sharpening power is covered in this chapter, but a comprehensive look can be had in the “Editing Raw Files” chapter.
Blurring is an effective way to draw attention to the parts of an image that aren’t blurred. It can also be a way to reduce noise or other artifacts in an image. In design, we may blur an image to have it serve as a backdrop to text or other design elements. With multiple reasons come multiple methods.
When you choose any of the filters in Filter > Blur Gallery, you enter a new workspace that will let you switch from one of those filters to the others, adding one to another or using one to keep an area sharp that another was blurring.
Each uses Pins, points from which an amount of blur emanates. Some have additional interface elements as well. To complete the illusion, you can use that workspace to add noise or motion effects to the blurred areas and bokeh affects to blurred light sources. Finally, you can choose to have an alpha channel made that serves as a map of which areas were blurred and, by its shades of gray, how much.
This example shows a two-pin Field Blur. The lower left pin emits very little blur, but the upper right one emits significantly more. The amount is shown by both the extent of the white in the ring surrounding the pin and by the Blur slider in the panel at right. You can actually drag the white part of the ring to change the amount of blur! Dragging the pin moves the source of blurriness.
To see where (and relatively how much) blur is being applied, you can hold down the M key (for “Mask”). This shows the area of maximum blur with white, and unblurred areas with black. If you check the Save Mask to Channels box, this is what the resultant alpha channel will look like too.
There are other imaging consequences to blurry images that may be desirable. For these, you need the panels at the lower right of the Blur Gallery workspace.
Bokeh is the blooming of light around bright reflections or light sources in a blurred area. Light Bokeh controls the amount of blooming that occurs. Bokeh Color allows those blooms to adopt color from around them. The Light Range is the range of tones that exhibit bokeh (in the example, all tones lighter than 221). You can see the result on the brightly lit bits of those buildings on the right.
Motion Effects are for Path Blur only. This simulates illumination of the blurred subject by strobe (flash), creating sharp versions of the subject within the blurred one.
When parts of a noisy image are blurred, the noise from that area is removed too. However, it can be restored (simulated, actually) with the Noise controls.
Add more pins by clicking where you want one. The cursor will look like this: .
Remove a pin by selecting it and tapping backspace/delete.
Iris Blur creates a sharp central area surrounding the default pin up to four blue circles. From those outward to a faint gray line is the fade area in which sharpness gives way to blur. By dragging the circles, you can adjust the size of the sharp area. Dragging the gray line adjusts the extent of the whole affair. However, modifier keys are powerful here:
- Holding Shift while dragging the outer gray circle forces it to be a circle.
- Option/Alt-dragging the gray circle resizes only the fade area, leaving the inner sharp area as is.
- Option/Alt-dragging one of the blue circles moves only it, creating an asymmetrical sharp iris.
There are also points along the outer gray circle that are special. Dragging the tiny blue circles on it inward or outward allows you to resize the whole interface in one dimension (width or height). Dragging them along the circle rotates the whole iris. Finally, the small square on the gray circle is for changing the iris from elliptical to rectangular.
I use this filter to simulate old or cheap plastic lenses, which have poor edge sharpness.
Tilt-Shift Blur attempts to simulate the focus control of a tilt-shift lens or an old-school camera with a bellows between the lens board and the film plane. With those, one can have a narrow zone of focus with relatively abrupt blurriness in both the foreground and background.
When you choose this blur in the gallery, an initially horizontal zone of sharpness appears around the pin. The fade area, in which focus falls off to the amount of Blur you choose, runs between the solid gray lines and the dashed ones. Beyond the dashes is the blur area. You can drag any of the lines to change those zones. If you carefully drag the blue circles on the solid line, you will rotate the whole effect. Since those are often close to the pin, which is the axis of rotation, it can be a bit tricky. The Distortion slider distorts the area that started at the bottom (before any rotation). Negative values distort parallel to the sharp zone, and positive values distort perpendicular to it. You can see foreground objects getting pushed around one way or the other if the Blur and Distortion values are high enough. If you want that effect on both sides of the sharp area, then check Symmetric Distortion.
Path Blur simulates motion blur. It creates an editable path you use to indicate the direction and magnitude of motion. I often use this to give life to an object I’ve introduced to a scene.
The idea is that the object moved during the exposure. By default, this filter assumes uniform speed and direction. However, you can click on either end point and alter the End Point Speed, as I did above. When you do, a red arrow appears to indicate the direction of motion. The longer the arrow, the faster the motion. I set the starting point to zero, and the point nearest the rocket to nearly 200 pixels of blur to simulate an accelerating object. I pushed the center point, too, to give a slight arc to the motion. I also introduced a little bit of Taper to recover some of the edges that were a little too blurry perpendicular to the motion.
Each end point can have a bit of its own motion, too, if you check Edit Blur Shapes. So you can simulate an object (or the camera making the photo) that wiggled a little at one end of the exposure.
Spin Blur can make wheels or windmills turn. With Spin Blur, we indicate how many degrees of rotation has happened during the exposure. The circular blur area can be made elliptical by dragging the blue circles at its edge, so things that are not directly facing us can be spun too.
Related to the Blur Gallery in that different parts of an image can be blurred differently, Lens Blur also differs in two ways. Since it is more tightly wed to the physics of optics, and is therefore more demanding of our computer hardware, it cannot be applied to a Smart Object. So we have to duplicate the layer to which we want to apply this filter.
The other difference is that, rather than generating a channel to indicate what’s been blurred and by how much, Lens Blur uses an existing alpha channel to indicate depth or distance and then uses that information to simulate a lens’s depth of field.
Recall that alpha channels (like all channels) are grayscale images—see “Saving & Loading Selections” (page 285) for information on how to create these from selections. You can also duplicate one of the color channels to create an alpha channel that looks just like it.
The cover image for this book was made with the help of Adobe Dimension. Each element (the rocket, planet, moon, and background) are different virtual distances from the viewer. In the rendered Photoshop Document that Dimension produces, there’s a group of layers that can be handy for post-render adjustments. One of those layers is a grayscale image called “Depth Information.” In it, white represents the closest element (the rocket), and black the farthest (the backdrop of stars). To use this information in the Lens Blur filter, I needed this image as a channel.
I hid the layers above “Depth Information,” so it was the only one visible. Since it was grayscale, the Red, Green, and Blue channels were all identical to it. So I highlighted one of those, right-clicked it, and chose Duplicate Channel…. I named that duplicate “depth info.”
Apple’s iOS devices can now capture in a format they call HEIC (High Efficiency Image Coding). These may contain a depth channel you can use. For documents without such a useful layer or channel, you will have to construct one. You can select an object to be the subject, then save that selection as a channel.
Access the filter via Filter > Blur > Lens Blur. Within the filter’s interface, direct your attention to the Depth Map section. Choose the alpha channel you created as the Source. Then use the Set Focal Point tool to choose which part of the image should be sharp. The level (shade of gray) at that location in the alpha channel will be used. Anywhere that value is used in the depth map, the image will be sharp. That level is displayed as the Blur Focal Distance.
The Iris section is where the physics simulation kicks in. This filter tries to simulate a real lens’s behavior and properties. The lens properties that affect a blur’s properties are the size, shape, and orientation of the aperture. Typically, the iris of a lens is formed by a number of blades—we choose how many with Shape. The larger the Radius, the more narrow a lens’s depth-of-field, and thus the blurrier things are if they’re not at the focal distance.
The shape of the iris also affects the shape of bokeh, or the blooming of highlights in out-of-focus areas. Bokeh is controlled in the Specular Highlights section.
Finally, Noise can be applied to the blurred areas so their noise level matches that of the areas that remain sharp.
I find this filter works best when sharp subjects are in front of blurred backdrops. Edges of blurred objects in front of sharp ones don’t always blur correctly, as seen below:
Sharpening has fewer methods, especially those that are effective. Do not look to Photoshop to “refocus” the lens that made a blurry photo. No doubt, spy agencies have software for that, but it isn’t Photoshop. However, images that have a bit of softness can be made to appear sharper by enhancing contrast at edges. One filter, Shake Reduction, attempts to be the exception. It analyzes motion in an image and attempts to cancel its effects. In extremely limited cases, it helps.
This filter is one of the two best ways to sharpen an image. All images need a touch of sharpening (at least), but the image in the following examples needs more than that.
I very strongly recommend applying this filter to a Smart Object so it can be revisited or masked if necessary. Once the dialog box opens, I drag one of its corners to make it larger, then I disable Preview. In this way, the original image acts as a large “before” view. A click somewhere in the document window centers the filter’s preview window on that location.
I find the following procedure works very well when approaching this filter.
- After choosing Filter > Sharpen > Smart Sharpen…, start with the Remove menu. Almost always, I choose Lens Blur, as that does active edge and detail detection. Gaussian approximates the imprecise Unsharp Mask filter, but sometimes I simply like the way it looks. Motion Blur requires us to specify the angle of motion that we’re trying to cancel out, but it is rarely effective.
- I set Reduce Noise to 0—for now. If the sharpening I achieve is satisfactory but accentuates noise in the image, I’ll return to this slider and raise it minimally.
- I temporarily set the Amount to its maximum so I can more clearly gauge how much Radius I need. This is the distance from the detected edges that the sharpening effect extends. It should be subliminally small. To see it accurately, zoom both the document window and the preview window to 100% magnification. If you have a high-definition display, you may need to use 200%. In the example, I’m using 300% in the hope that you’ll be able to see the effects of the filter in this book. A typical Radius value is about 1 pixel. This image’s softness was so extreme that I’ve set it to 1.8.
- I then set a more appropriate Amount. For images with large, relatively smooth expanses, you will likely use a lower value (or a higher one with a bit of noise reduction). With highly textural images like this one, you can use higher values, especially when an image is as soft as this one. I find for this image, I leave Amount at the maximum, but with a small amount of Reduce Noise. This is unusual!
- Finally, if either the shadows or highlights exhibit artifacts that the other controls don’t address without compromising the sharpening, you can fade the sharpening from them. For these controls, the Radius is that of an area that should be averaged and judged to be a shadow or highlight area. Tonal Width is how much of the tonal range should be considered a shadow or highlight. Fade Amount is how much to reduce the sharpening in that part of the tonal range.
Many longtime Photoshop users are familiar with, and therefore recommend, this filter. Unlike Smart Sharpen, however, it does not do active edge or detail detection. Instead, it relies on the user to indicate what an edge is by specifying how brightness differs from one side of the edge to the other using a Threshold slider. Unfortunately, a single value may not apply across an entire image.
Besides, the algorithm used by the Unsharp Mask filter resides in Smart Sharpen too. If you’d like to compare algorithms, simply choose Gaussian rather than Lens Blur from the Remove menu in the Smart Sharpen dialog.
Some use Unsharp Mask with low Amount and high Radius settings to add local contrast to an image. I think there are also better ways to do that: Clarity or Texture in the Camera Raw filter, for example.
Curious about the name? It’s derived from a clever and very old film technique that literally masked (hid) the unsharp areas of an image. I used that technique a few times when I had access to lots of film and darkroom time.
Under the right circumstances, blur caused by the motion of the camera/lens can be reduced, sometimes dramatically. It’s best if the lens used was of a longer focal length and the exposure time was too long to freeze any motion of the camera. Here, a 200mm lens for 1/15 second.
The workspace that opens when you choose Filter > Sharpen > Shake Reduction… will automatically perform an analysis of a part of the image to determine the extent and “shape” of the blur in that region. If the lens movement was mostly vertical with a little side-to-side motion, then the Blur Trace that results might be a gentle S-curve like above.
You can move and resize the Blur Estimation Region to analyze a region you think has better edge contrast or better exhibits the motion you want to reverse. You can use the Blur Estimation tool to draw other regions to analyze as well. In fact, this is often necessary because movement can be different in different areas of the image. Each Blur Trace will appear as a white squiggle in a black field. These can be enabled and disabled to experiment with their usefulness. Each region’s trace can be customized with a few settings. I find that a trace’s size, the Blur Trace Bounds, is best set a bit high, at least for the motion blur I’ve encountered. Other values have been either ineffective or contribute to more artifacts. Speaking of which, you will no doubt see ripples around contrast edges in the image while using this filter. These are actually what is helping to cure the shake blur. So if you were to set a trace’s Artifact Suppression to 100%, you will not experience any sharpening. Zero may yield great texture sharpness, but the ripples at edges will be unacceptable. Each region’s trace will need to have this value adjusted multiple times to find the right one. Smoothing removes surface noise if it is bothersome. Lower values are better here too.
When processing a raw image in Adobe Camera Raw or Lightroom, we find what is likely the best is to sharpen in the Detail panel. We can also access this fabulous tool with Photoshop’s Camera Raw filter. In the ACR interface, tapping the Q key presents a before-and-after view (the letter is Y in Lightroom).
Two of the four sliders are very similar to those in Smart Sharpen: Amount and Radius. Just like in Smart Sharpen, Amount controls the amount of contrast enhancement at the edges, and Radius defines how broad those edges are. Remember that sharpening is just careful contrast enhancement at edges in our images.
There are two additional sliders as well: Detail and Masking. Some edges are around extremely tiny elements in an image, like noise. Some are around larger objects. Detail is a way to indicate the scale of those details that are sharpened. A value of 100 means that all details, no matter how small, will be sharpened equally. This can accentuate noise. Interestingly, 0 still allows sharpening, but mostly of the largest edges. Detail shifts sharpening’s emphasis.
The Masking slider is a work of inspired genius. This one control embodies a clever multi-step workflow developed by the late Photoshop master Bruce Fraser that sharpens only strong edges and allows no sharpening at all in some areas. Higher values protect areas like skin tones (so pores don’t get sharpened), but the strong contrast edges of eyelashes will ensure that they are sharpened nicely. To see exactly which areas will be masked, hold down option/Alt while dragging the Masking slider. Areas shown in white will be sharpened, areas in black won’t.
In fact, each of those sliders offers a different view if you hold down the option/Alt key as you drag them. Amount will show you a grayscale version of the image so you can focus on the tonal details (like edge contrast). With Detail, option/Alt gives you a better idea of which edges are being emphasized by sharpening.
If sharpening makes an image’s noise more apparent, there are elaborate controls for Noise Reduction below those for Sharpening in the same Detail panel.
Although the Filter menu is divided into categories, I categorize them a little differently. Many filters can fall into multiple categories. Of course, Photoshop users find unintended uses for filters too. Not for the last time, I encourage you to explore and play!
Add Noise & Grain
Whether I’m trying to simulate film grain (or sensor noise) or add texture to a surface, Filter > Noise > Add Noise… and the Grain filter within the Filter Gallery are useful. Note that only 8-bit/channel images are compatible with the Filter Gallery.
It’s always best to deal with noise in images as close to capture as possible: that’s why my camera is set to capture raw files, and I deal with noise in either Lightroom or Adobe Camera Raw. However, we are sometimes given images that retain noise that is distracting or unpleasant.
Choose Filter > Noise > Reduce Noise…. Note that there are two kinds of noise: luminance (managed with the Strength slider) and color (managed with the Reduce Color Noise slider). Reducing luminance noise risks blurring away actual details, so the dialog offers a slider that forces the filter to find and Preserve Details. If in the course of reducing noise the overall effect still looks soft, use the Sharpen Details slider. You will likely find yourself cycling through the sliders as you compromise the settings of one with the others. If you determine that most of the noise is on one color channel (often, that’s the Blue channel), you can enable Advanced mode. This simply lets you control noise reduction on each channel individually.
Several filters can use channels (either color or alpha channels, or both) as “maps.” In the case of the Lighting Effects filter, the map serves as a way to indicate the texture that the lights shine across. Note that only 8-bit/channel images are compatible with the Lighting Effects filter.
This filter in the Filter > Render menu benefits from a channel used to “map” texture. The usual rule is that dark values on the channel to be used will represent depressions in the surface, while white is used for bumps. A slider in the filter interface can be used to reverse that, if necessary. For the brick wall example above, I chose to make a duplicate of the Blue color channel, as it had good contrast between the bricks and mortar. But since the mortar was white, I used Image > Adjustments > Invert to make a negative of that. I also blurred it slightly with the Gaussian Blur filter set to only a few-pixels Radius.
To work nondestructively, I almost always create a layer above the image. Here’s the tricky part: I fill that layer with 50% gray and set its blend mode to Hard Light (sometimes Overlay). This makes the layer disappear! For extra flexibility, I usually convert that gray layer into a Smart Object too. Later, when I apply Lighting Effects, the light and dark areas it makes will be apparent, superimposed on the layer below.
Interestingly, negative values work with the ambient light and the lights we make too! Setting Ambience to –100 ensures that the only light in the scene is from our lights. Negative Intensity values on the lights causes them to become light absorbers, like Shelob in her tunnels. (Look it up!)
You don’t always have to designate a Texture channel if you just want lights in a photo. But it’s great to have the dimensionality when you need it. With or without a texture, the surface Properties can be set as glossy as you wish with the Gloss slider. The Metallic slider controls the lights’ diffusion (mostly if it’s already glossy).
In this example, I made the default light a spotlight via its Properties. I dragged it up and to the right, and adjusted its hotspot and falloff to aim it down and left. I created a second spotlight and made it red, aiming right from beyond the left edge of the image. To get the brick texture, I chose the alpha channel I made (the duplicate of the Blue channel) in the Texture menu. To actually see it, we have to set the Height slider to something other than 0. Positive values follow the rule of “white is a bump, black is a hole;” negative values reverse this. I set it to a low value (2).
As the name implies, this filter performs a kind of average. With a minute Radius setting (like 1), this filter will look at the color values of the pixels surrounding each pixel in the image. It then assigns the middle values to the pixel under examination. Larger Radius values mean more pixels are being averaged.
The benefit is that tiny blemishes or damage in an image can be eliminated, while still preserving edges. Get to this filter by choosing Filter > Noise Median….
Something from nothing! That’s what these filters provide. Well, most of the filters in Filter > Render can be applied to an empty layer (or Smart Object). There are two exceptions: Flame and Lens Flare. Those need a little preparation.
Clouds & Fibers
Many creative recipes require multiple ingredients. This is why I often use filters in combination to achieve the results I seek. When I need a random starting texture, these two filters can provide that.
Sadly, the Clouds filter doesn’t make nice clouds that you’d use in a sky. And what results from the Fibers filter does not resemble fabric. Both create a random image using the current Foreground and Background colors, so you may wish to set those before applying either filter.
Clouds is completely random. Fibers, however, is guided randomness. It has two properties you can adjust: Variance (higher values yield shorter “fibers”) and Strength (higher values yield a more stretched or well-combed look).
Although there are many controls (menus and sliders) in this filter’s dialog, there is a random element to it too. Algorithms that generate random results are often said to start with a “seed,” which I find pleasantly punny with this filter.
Choose the species (even fanciful Stylized Trees) with either matching Default Leaves or leaves from a different species. Light Direction can be from any angle, from the left to the right of the tree. The other sliders control the density of branches and foliage. The Advanced tab allows you to control the color and contrast of the leaves and branches. When you choose Flat Shading, Photoshop creates a more cartoon-like result. Camera Tilt changes the altitude of the viewer: at 0, the viewer is assumed to be at the tree’s base gazing upward, with higher values elevating the camera and tilting it down. Finally, Leaves Rotation Lock gives a better sense of the leaves’ shapes with a face-on view of them.
This filter needs a path no longer than 3000 pixels to indicate the base or source of the flames. Use one of the Pen tools or other vector shape tools, configuring the tool to draw a Path (rather than Shape) in the Options Bar. Once drawn, create an empty layer to hold the flame, then choose Filter > Render > Flame….
This filter has an extraordinary number of parameters to configure. Whether you desire a single candle flame or a raging inferno, you should be able to get it configured. Note that not all parameters are available for all Flame Types (illustrated above). For example, Angle rotates the flames from their default for two types: that would be vertical with Multiple Flames One Direction and perpendicular to the path you drew for Multiple Flames Path Directed. For Multiple Flames Various Angle, Angle sets the range for random rotations for each flame.
The Length and Width of each flame are easily controlled, as is the Interval, or space between multiples. If you put the flame along a closed path (a “loop”), ensure that Adjust Interval for Loops is checked to have even spacing. The best-looking flames are attained with Quality set to Fine, but expect to wait some time for them to render. All of the illustrated examples use Medium quality. Each path is also shown here for your reference, but the path won’t be part of the resulting flame normally.
The Advanced tab has even more options. The Turbulent and Jag sliders control the undulations of the flames in a coarse or fine way, respectively. To make the bottom of each flame more aligned with each other and your path, use a low number for Flame Bottom Alignment. If you wish to be able to consistently repeat a specific look, disable Randomize Shapes, then use the Arrangement slider. If all the other settings are the same, you’ll get the same results.
The flame’s shape and style can also be chosen, as illustrated below. In all of these examples, the paths were drawn either from the bottom up or from left to right. You may get inverted results otherwise.
This filter simulates the flare that sometimes occurs when making a photograph with a strong light source in the frame. Although you can apply this filter to an image layer directly, or more wisely to a Smart Object, I almost always use a layer filled with black. To revisit my settings later, I make that black-filled layer a Smart Object. This makes it easier to colorize the flare to better match the color of the light source in the image.
I also use the Info panel to note the light source’s position; that is, its distance (in pixels) from the image’s left and top edges.
The majority of the filters that I put in this category attempt to make one medium (photography, usually) appear to be made in another.
I’ve heard heated conversations about this filter and whether or not it can make a photo look like an oil painting. On a couple of occasions, I’ve used it to create a usable interpretation of a photo that was decently composed but either lacked sharpness or had some other flaw. Perhaps I’m not as critical of this filter as some folks because I’m not a painter!
The dialog that appears when you choose Filter > Stylize > Oil Paint offers several parameters to adjust to get the look you desire.
The greatest impact is had with first two, Stylization and Cleanliness, which control stroke smoothness and length. High Stylization and low Cleanliness yield a more Van Gogh–like look, while reversing those sliders yields a more subtle result. Scale controls the paint’s thickness and Bristle Detail controls the depth of the ruts left by the bristles in each stroke. Control how light interacts with the paint by adjusting the Angle and Shine sliders.
A halftone is the dot pattern you see when a printed image is magnified. Of course, this is typically a necessity of making a printed piece, but many like it as an aesthetic.
There are at least three ways to create a halftone effect in Photoshop. There are two filters, discussed below, which simulate true halftones. The process that generates a true halftone, but only with black dots, involves converting an image to Grayscale mode, and then to a mode called Bitmap, choosing a halftone dot pattern and resolution along the way. The filters are easier and less of a commitment but are just effects, not true halftones. But for creative ends, that is just fine!
This filter is accessed by choosing Filter > Pixelate > Color Halftone…. It generates a fairly good simulation of CMYK dots. In fact, I use its result to explain the four-color process to my students. In the filter dialog, you choose the dot size (Max. Radius) and the Screen Angle for each ink color. The defaults represent typical values used by printers, but they can be changed, of course. Unfortunately, there is no preview. You have to commit it to see the result. That’s why it’s always good to use a Smart Object.
When the image is grayscale, we need only worry about the settings for the black dots. In either color or grayscale, the dots are opaque and vary in size: larger for dark areas and smaller for light ones.
Filter Gallery Halftone
This filter creates a monochromatic halftone effect from color or grayscale images (but only in 8-bit/channel mode) using the current Foreground color. It can be accessed from Filter > Filter Gallery…. Once there, you’ll find Halftone in the Sketch section.
Its dots are more uniform in size, unlike a true halftone, but it does create light dots on black for dark areas, and dark dots on white for light ones, like a halftone should. It also makes rather muddy dots, unless you increase its Contrast setting. Finally, it can produce circles or lines instead of dots, if you prefer.
Other Filter Gallery Examples
Note that Filter Gallery filters do not work in 16- or 32-bit per channel modes. But if you create a Smart Object of one or more layers while in those modes, you can then convert the containing document to 8-bits/channel, retaining the greater bit depth inside the S.O.
But the name “Gallery” should be a small hint of the good news: the many filters in the Filter Gallery can be combined in a way similar to applying multiple filters to a Smart Object. Go to Filter > Filter Gallery… and plan to experiment with the nearly 50 effects you find there.
For quicker access to these filters in the Filter menu, enable Show all Filter Gallery groups and names in Preferences > Plugins.
Many of the filters in the Sketch and Artistic sections create strokes like those created by a brush, pen, etc. Their length, width, intensity, and resulting contrast are often the parameters you can adjust. Some, like the Graphic Pen filter, will use the current Foreground color.
I use Chrome and Plastic Wrap to create indistinct reflections. It helps to blur the image first.
Some filters use grayscale images as “maps” to control distortion or shading. A Glass filter texture map shows how the glass’s thickness changes with gradation. Areas with no transitions are flat glass and don’t distort. The center of the map used here acts a bit like a lens.
Time for some bending and twisting! When you visit the Filter > Distort menu, try to keep in mind that not all the names should be taken at face value. For example, the Pinch filter can bloat things if you use a negative Amount. Some of these filters are quite simple, others require preparation, and some are as deep as small software applications.
There are also several filters in the Filter Gallery that distort. Examples of Glass and Ocean Ripple are shown in the previous section. Texturizer is tempting, but significantly better results can be had with a combination of Lighting Effects (page 313) and Displace (page 329), though it’s admittedly more work.
Pinch vs Spherize
Neither of these filters previews in the document window, but they do have their own small preview panes. Spherize bloats and Pinch pinches, with positive Amount values, and the opposite is true with negative values. I would describe the distortion produced by Pinch as looser, and that produced by Spherize as better delineated at its elliptical edge. Both distort elliptical areas (circular if the image is square).
When applying these filters to a selected area, the results differ if the layer is a Smart Object or just a pixel layer. If it’s an S.O., the entire image is distorted, but a filter mask is created to hide the parts outside the original selection. If the layer is an ordinary pixel layer, then the distortion will take place only within the selection, blending in with the undistorted pixels at the edge—but the selection needs to be elliptical.
The Liquify filter (Filter > Liquify) was another program that Adobe bought and re-engineered. With it, we can sculpt an image, pushing pixels to and fro. It’s powerful and fun. You have certainly seen its use in fashion photography to augment or diminish subjects’ attributes.
This filter has face detection, noting the presence of faces and letting you target each one so you can add a smile or frown or adjust other parts of the face’s structure with sliders or the Face tool. The other tools are for more manual (or extreme) manipulation. Each uses a “brush” that can be resized with sliders, or more quickly with the bracket keys ([ or ]). Most notable is the Forward Warp tool, used to sculpt an image.
I use the Twirl Clockwise tool to curl hair or flip paper corners. Holding option/Alt twirls counter clockwise.
Pucker and Bloat shrink or enlarge what’s within the brush perimeter.
The Mask tools are used to protect areas of the image (Freeze Mask tool) or allow editing (Thaw Mask tool). To correct distortion that is too abrupt, use the Smooth tool. To restore parts of the image either completely or partially to their original state, use the Reconstruct tool. Light dabs are best to gently nudge the image back to its pristine state. The Reconstruct… button provides a complete or partial undo of the whole adventure. It allows you to input how much of the total distortion remains. Dragging the slider triggers a preview so you needn’t guess what value is right.
Finally, the Push Left tool creates a rather messy and uncontrollable distortion. Use it with discretion.
Adaptive Wide Angle
When dealing with an image captured with a wide-angle lens, especially a fish-eye lens, lines that were straight in the world may be curved in the image. This is also true when creating a panoramic image with File > Automate > Photomerge…. The Adaptive Wide Angle filter allows you to choose lines in the image that you would like to straighten, likely at the cost of more distortion elsewhere.
Although this filter purports to help with correcting converging perspective lines common in photos of buildings, the Camera Raw filter has a better solution: see “Transform Tool & Upright Perspective Correction” (page 362).
With panoramic images, however, this filter does a lovely job of straightening horizon lines that are curved if not composed in the image’s center, for example. Simply drag the Constraint tool from one end of the horizon to the other.
Wave & Ripple
These two filters, both of which are found by going to Filter > Distort, create undulations. The result of the Ripple filter is more uniform and geometric. The Wave filter provides from 1 to 999 wave generators, creating undulations of random wavelength and amplitude within ranges you specify. Imagine several people of various sizes bobbing in a pool, and the resulting wave action.
This guided randomness yields a more “natural” distortion. Sadly, neither filter has more than a small preview in its dialog box.
The Wave filter offers two other wave types besides the smooth Sine wave. Triangle yields a sawtooth distortion, and Square creates an offset block pattern.
With work, this filter can be an effective way to have an image appear to follow the convolutions of a dimensional subject. For example, I’ve used it to warp type around the curving pages of an open book. The difficult part is creating the displacement map that this filter requires.
This map needs to be a separate document whose channels’ tones (shades of gray) control the horizontal and/or vertical displacement of the image to which the filter is applied.
If the displacement map is grayscale (has one channel), then its shades can control both horizontal and vertical displacement; otherwise, the first two channels deal with each respectively. A displacement map filled with 50% gray causes no displacement at all. Lighter shades move corresponding image pixels left and/or up; darker shades move the pixels right and/or down. If the map pixels are white, 127 levels above middle gray, then the image is shifted 127 pixels. If that’s too much (or too little), the filter’s dialog offers fields to enter a percentage by which to adjust the movement both horizontally and vertically. Since there is no preview, a lot of trial and error is required, so allot time for your experiments with this filter! Ensure the map document is a PSD with the same pixel dimensions as the one you’re distorting.
I most often use the Displace filter to enhance the realism of other effects. In the example above, I used the Lighting Effects filter to create a repoussé effect on a photo of sheet metal. See “Lighting Effects” (page 313) for details on it. To minimize complication, I intended to use the same image for both the Lighting Effects filter’s texture channel and the Displace filter’s displacement map. I made a middle-gray color fill layer with a white type layer above it. I simply duplicated one of the color channels by right-clicking it, creating an alpha channel. This I blurred slightly, resulting in what these filters would interpret as a flat area rising up to type-shaped bumps.
Unfortunately, the Displace filter requires that its map be a separate document. So I right-clicked the alpha channel, chose Duplicate Channel…, and specified New as the document where the dupe should go. This generates a new document, which I saved in a convenient place.
Now, all was ready—finally! I made the sheet metal image a Smart Object since I knew the first attempts would not be quite right. I applied the Lighting Effects filter, creating a spotlight using the alpha channel as the Texture Channel. Then I turned to the Displace filter.
After a few attempts, I decided that the image pixels should be displaced up and right about 25 pixels at most, or 20% of what the Displace filter would do if Horizontal Scale and Vertical Scale were set to 100. So I used a value of 20 in both those fields in the filter’s dialog box. However, the white parts of the displacement map image would move image pixels to the left, so I entered –20 into the Horizontal Scale field.
I was not concerned about whether the Displacement Map was stretched or tiled, as it was the same size as the image. Also, since the map was gray at the edges, I knew there would be no displacement there, and thus no Undefined Areas would result.
In your favorite web browser, search for the phrase “cylinder mirror anamorphosis,” and you’ll see the premise behind this filter. My friend Blake Garner applies this filter, performs other distortions (the Wind filter, I believe), and then applies the Polar Coordinates filter again to convert back to rectangular coordinates. Fun!
If you should come upon a historic image distorted in this way, you can use this filter to convert it to rectangular coordinates, since it can convert in either direction.
The filters in Filter > Other tend to be for very special purposes. Here are the most commonly used of these hard-to-categorize filters.
Most practically, this filter is used to aid the creation of patterns. With it, we literally move, or “offset,” the edges of an image. For example, if I were using an image that’s 3000 pixels wide and 2000 pixels high, I might move the top edge down about 500 pixels and the left edge rightward about 700 pixels (I like to avoid moving the edges to the center).
Rather than leave a void at the top or left, the default behavior is to wrap the bottom of the image around to the top, and the right side around the left. When an image is made into a pattern, this is its behavior: it tiles. To ensure that obvious seams don’t appear when the pattern is made, I retouch the offset image to make its former top and bottom edges transition more seamlessly into each other. The same with the left and right. I then repeat the Offset filter to see if my retouching introduced any new seams, which I remove. When repeated application of the Offset filter reveals no more odd transitions, I can use Edit > Define Pattern….
Useful almost exclusively for masks (or other occasional high-contrast grayscale images), the Maximum and Minimum filters make white areas larger or smaller, respectively, by the number of pixels you specify. If there are sharp turns (corners), you can choose Squareness from the Preserve menu to keep them corners. If you’d prefer to round them off, choose Roundness.
The Gaussian Blur filter is sometimes referred to as a low-pass filter because it keeps only low-frequency content. Fine details (high-frequency content) are lost when blurred. The High Pass filter is the opposite. Only fine details remain, and soft textures and blurriness are removed. The result is mostly middle gray with fine edges throughout.
When a blend mode that hides middle gray (Hard Light, Overlay, etc.) is applied to the result, only the edge details remain. When blended with the original image, the result is a sharper image. Some Photoshop old-timers claim this method has virtues. However, there are many new ways to sharpen that are superior and more direct. That said, there are a few multi-filter processes in which I’ve needed this filter to provide the stronger edge contrast I needed.