Project Examples

Image

To truly take advantage of hidden powers, they must be exercised in context. Part III shows some brief examples of bringing techniques together and choosing various approaches.

Creating the Workbench Files

This section is for the intrepid digital explorers who love to tinker with bits and ones and zeroes. Although I have provided instructions for a couple of basic experiment files, you should always be ready to create your own specific experiments to test various properties and characteristics of whatever tool, feature, or technique you want to better understand. Often, understanding comes through building a better vocabulary, and in the Photoshop world that vocabulary is more likely to be visual and conceptual than a new set of words.

What makes a good experiment file is the ability to let you control the variables involved, and to reduce the distractions and attachment we have to our own work. Using files that reduce the environment to colors, brightness, shapes, and so forth helps us leave behind any attachment we have to working on our photos. For example, if you are just trying out blending modes with a photograph, you are likely to judge an effect based on whether you like the results in the moment. If you come across something that is unappealing, you are probably going to just move on and leave that blending mode behind and not bother trying it again in the future.

Such is the fate of Hard Mix and Dissolve, two blending modes typically ignored or used for singular effects. In the following pages, I hope you find a little of the fascination I feel when experimenting—and that leads to a discovery or two!

Gradient Experiments File

Remember that my intent with this book is not to show you the fastest, coolest way to edit photos, but to expose the tools and features in a way that helps you build a stronger foundation for your own work. Building and using these files is a way for you to push beyond what I’ve shown in the previous pages. If you have favorite techniques, plug-ins, or other tools I haven’t described, try them out with these files or build something new.

Image

Each individual experiment file should be built in its own document. I prefer to use a document of 2000 px by 2000 px in 16-bit RGB, with a resolution of 300 px/inch, and either a white or transparent background.

Basic Color Bars

The first file is something I’ve used for years to see how blending modes behave over basic colors and luminosity values. Ensure that your working space is set up for your normal workflow, including such settings as gamma, color profiles, and so on. If you haven’t changed any of this before, it’s likely you don’t have to do anything more, but you can learn about the basics on the following site: helpx.adobe.com. Search for “Color settings Photoshop” and browse the results.

Image

In your new experiment document, select the Gradient tool and then open the Gradients panel (Window > Gradients). Create a new preset group by clicking the Create New Group button at the bottom of the panel (it looks like a folder) and name the group Utility Grads.

Tip

The New button does not start you off with a clean, basic gradient. Only click the New button after you have set up the gradient you want to save and you have entered a name in the Name field.

Press D to load the default foreground and background colors (black and white, respectively), then from the Gradients panel open the Basics folder and select the Foreground To Background gradient—this will now of course be black to white. If you have a layer active in the Layers panel, clicking a gradient preset will either add a new Gradient Adjustment layer if the layer has content, or convert a blank layer to a Gradient Adjustment. Name the layer B-W 0% Smooth.

Double-click the icon on the Gradient Map layer to open the Gradient Fill dialog box, and set the Angle to 0º. Then click the Gradient swatch to open the Gradient Editor dialog box. Foreground To Background should already be loaded in the Name field.

In the editor area below the presets, find the Smoothness setting and enter 0 (zero). Smoothness interpolates luminosity values to look more uniform to your eye using a Gaussian distribution; setting the value to zero gives a linear transition between color stops. This will be important when using the Posterize adjustment later on.

Give your modified gradient a new name such as B-W 0% Smooth, and then click New to save it (see Tip). Drag the new preset to the Utility Grads group. As you create new utility gradients (such as the black-white-black gradient we created in the “Gradient Map Luminosity Selection” section of the “Selections & Masking” chapter in Part II), place them here for easy use and organization. Click OK in the Gradient Editor and Gradient Fill dialog boxes to close them.

Note

I have changed the default cyan-colored guides to magenta for better visibility. You can set the guide color in Preferences (Photoshop > Preferences in macOS, Edit > Preferences in Windows) in the Guides, Grid and Slices category.

Back in your document, create a blank layer to work on and add a new guide layout (View > New Guide Layout) with one column and eight rows, ensuring that Gutter and all other numerical settings are blank or 0.

Image

Switch to the Rectangular Marquee tool, then create a selection from the top horizontal row. If you have not previously done so, ensure snapping is on (View > Snap). Click the foreground color swatch in the Tools panel and choose RGB red (RGB: 255, 0, 0). Fill the selection area on your blank layer with the red foreground color by tapping Option+Delete/Alt+Backspace.

Image
Image

Skip a row, then repeat the process for green (RGB: 0, 255, 0) on row 3 and blue (RGB: 0, 0, 255) on row 5.

For row 7, you need to create a color spectrum. In the Gradient Editor, make a new gradient preset with the settings shown in the following table. In addition, set Brightness and Saturation to 100% and Smoothness to 0%. You can start from any preset you like so long as it uses 100% Opacity; it’s probably easiest to choose a simple preset so you can add each stop individually.

LOCATION

COLOR

0

Red

(255, 0, 0)

16

Yellow

(255, 255, 0)

32

Green

(0, 255, 0)

48

Cyan

(0, 255, 255)

65

Blue

(0, 0, 255)

82

Magenta

(255, 0, 255)

100

Red

(255, 0, 0)

This new gradient gives you a spectrum that will repeat if you use it with the Radial gradient, and gives a full account of the primary and secondary RGB colors. Save your new preset (I called mine Spectrum - RyGcBmR), and use the Gradient tool to drag from left to right across row 7, the next-to-last row on your blank layer.

Name the layer with your color bars Blend, and then duplicate it naming the duplicate Reference. Use the Move tool to drag the Reference layer downward (hold Shift to constrain the movement) until it snaps to fill the even-numbered rows in Blend. Click the Lock All button with the Reference layer active to prevent it from being changed. You’ll use this to compare the results of experimenting on the Blend layer so differences are easier to see.

Your layer stack should now look like this.

Image

Turn off the visibility of the Blend and Reference layers for now, and add a blank layer between the B-W 0% Smooth layer (the gradient) and Blend layer. Switch to the Gradient tool and load the B-W 0% Smooth preset. Hold the M key to temporarily load the Marquee tool, then drag to select row 7. Release M, and drag from the bottom edge of the row 7 to the top edge so the selection gets a uniform black-to-white gradient. Repeat for the row 8. Here’s what the newly created gradient layer should look like.

Image

The reason for this is because the last two rows are spectra from left to right, so a horizontal gradient won’t evenly interact with every color when we start changing blending modes. Having the black-to-white gradient vertical here ensures that each of the colors gets to blend with every level of gray.

To use this file as an experimental workbench, make the Reference and Blend layers visible again and change the blending mode of the latter. Watch how the colors behave based on how they blend with the gray layers below. You can also clip adjustment layers to the Blend layer which gives you control over the brightness, saturation, and hue of the Blend layer colors depending on which adjustment you choose. Because the adjustment layer itself does not apply a blending mode to the Blend layer, you’ll still have to choose a blending mode and/or change the opacity and fill in order to see a blended effect.

Image

Although this particular set up doesn’t seem too exciting, it’s kind of like a “Hello, World!” programming project meant only to demonstrate that something works, leaving the how to future efforts. It allows you to see how the fundamental RGB colors behave when blended with different values of gray. One fun variation is to place some kind of color control layer between the Blend layer and the black-and-white gradient base layer. Try a Solid Color Fill layer set to Color blending, a Color Lookup adjustment layer, or even a Gradient Map layer.

More complex and useful project files can be built from these principles. Before moving on, though, take a look through the various blending modes, including the Special 8 that respond differently to the Fill slider (the Special 8 blending modes respond differently to the Fill slider than the Opacity slider. Read about them in the “Blending Modes” chapter of Part IV, “References”). As you see changes (or no change), try to describe to yourself why the result makes sense. For example, the previous image is the result of the Lighter Color blending mode.

Remembering that Photoshop assigns a weighting value to luminosity that approximates human visual perception, then it should be clear that there is some point along a black-to-white gradient where the gray value is brighter than the color. The point where the solid horizontal color bars stop is exactly where the gray value and the luminosity of the color are identical. The spectrum along the bottom shows the same effect for every RGB color at full brightness and saturation.

Image

You can test the idea using a Curves adjustment layer clipped to the Blend layer.

Changing the RGB curve changes the relative brightness. Check out the result of a typical contrast S curve, such as the Increase Contrast (RGB) preset in the Curves Properties panel. What changed and what is the same?

Image

Did you notice that the “pure” primary and secondary colors did not change at all? The intermediate, mixed colors did change. This result tells you that in order to adjust contrast in an image, some colors have to change. If, instead, you return the curve to its default setting and then choose the Blue channel and drag its right-side control down to the bottom of the graph, you’ll see that the blue bar does shrink until it’s gone, and the blue region of the spectrum turns black.

You can also reset the Curve and then explore how self-blends work by using blending modes on the Curves layer while leaving the Blend layer set to Lighter Color. Use Multiply for the Curves layer.

Image

Notice the depression in the intermediate areas? Again, the color bars haven’t moved. Once you’ve built this experiment file, see if you can approximate some of the blending mode results with various combinations of adjustment layers.

You can also try adding images or other gradients between the various layers. Use an Invert adjustment layer, also clipped to the Blend layer. Just play and see what you can discover, but always think about the results you see and ask yourself “why?”

Want to explore the difference between Fill and Opacity? Set the Blend layer to Hard Mix and turn off visibility of the adjustment layers. I have some thoughts on why these behaviors are different in the “Blending Modes” chapter in Part IV, “Reference”.

Image

Let’s move on to using the experiment file to create a preset. Set all blending modes back to Normal, all opacity and blending values to 100%, ensure no selections are active, and turn off any adjustment layers. Add a Black & White adjustment layer above the Blend layer and set its blending mode to Luminosity. This is the default black-and-white conversion that Photoshop thinks you want to start with. What’s wrong?

The default conversion results in red and blue getting a little lighter, and green getting darker. In other words, this is not a direct, neutral conversion of luminosity values. Using this setup, start by adjusting the Reds, Greens, and Blues sliders so those colors match the Reference layer. Then go back and adjust the secondary sliders. If you need more precision than dragging the controls provides, hold Option/Alt and then drag on the slider name, which slows down the movement of the control.

Image

Once you have achieved a true neutral setting, save the preset for easy recall later (click the panel options menu in the upper-right corner of the Properties panel and choose Save Black & White Preset). I use this preset as a starting point when applying a B&W adjustment layer for detailed contrast work, as in the B&W Control Freak technique from the “Color & Value” chapter of Part II, “Techniques.”

Image

How about an experiment that tells us something about how the Opacity setting works with color?

Turn off the Blend layer, and unlock the Reference layer. At the very top of the layer stack place a Posterize adjustment layer, leaving the default value setting of 4.

The solid color bars have not changed, but look at the gradients. The black-white gradient now has the expected four solid bands, but the spectrum is more complex. What’s going on?

Image

Posterize splits each channel into the specific number of levels, evenly distributed between black and white. That means each channel gets four levels in this case, which you can verify by looking at each channel individually. The reason there are not exactly 12 bars in the spectrum is because of overlap. In particular, pay attention to the Red channel being split because we created the spectrum such that RGB red occurs at both edges of the document. However, each of the channels has the same gray values, and the same amount of gray (you can measure the width and brightness of each gray bar on each channel if you like).

Image

Back in the RGB Composite channel, slowly lower the Opacity setting of the Reference layer while watching the spectrum. At about 75%, there should be a distinct change in pattern, and the edges of the solid bars may change slightly. By 70%, the solid bars obviously are being affected, and additional banding shows up in the spectrum.

Remember that the Posterize adjustment is affecting the result of the blend, not just the content of the Reference layer. The angles that show up in the spectrum are due to the way the compositing equation handles the alpha component, and represent the dividing point between the luminosity values that Posterize has used for each channel. If that word salad doesn’t mean much to you, don’t worry about it. Just pay attention to the regular pattern, which means the opacity value is smooth across the entire range of colors.

Image

RGB Triangle

Of course, we are not limited to this rather simple design. Because there are three color channels, we can also build a triangle spectrum that shows a more logical mix of colors and allows us to explore interactions using patterns. This one is a little trickier and requires some precision while creating it.

To create the equilateral triangle, start with another square document (this time with a transparent background), and create a new guide layout of six columns and five rows, with zero gutter. Ensure that View > Snap is selected. With the Rectangular Marquee tool, drag out a selection as shown in this diagram.

With the selection still active, choose Select > Transform Selection. When the transform options appear, select Toggle Reference Point and drag the reference point from the center to the lower left of the selection. Moving the reference allows the selection to rotate from the corner.

Image

In the options bar, enter −60 for the rotation angle. With the transform still active, chose View > Rulers and drag a guide down from the top of your document to the corner indicated here. Leaving the transform active allows the guide to snap to the corner. Press Return/Enter a couple of times, then Command+D (macOS)/Ctrl+D (Windows) to deselect.

Image

Use the Polygonal Lasso tool to connect the corners you just used, then fill with black and deselect. We’re going to fill on the channels directly, and Photoshop will not add pixels to a blank channel. The black fill is there to “catch” the gradient we’ll apply shortly.

Image

We have one more trick, which is finding the midpoints of the triangle’s sides. Again, drag a rectangular marquee of any width from the top of the triangle to the bottom, then drag a guide down until it snaps to the middle of the selected region—that is the half-way point of the height of the triangle, and one of the guides we previously created should intersect with it precisely on the edge of the triangle. Deselect the marquee.

Image

And now you know a stupid Photoshop trick: how to create an equilateral triangle. That’s an accomplishment!

Okay, remember how I said to create a preset of the 0% smooth black-white gradient? Select the Gradient tool and load that preset up. Open the Channels panel, and select the Red channel. Drag from the apex of the triangle to the center point of its base. It should be white on the top, black on the bottom.

Image

Repeat for the other two channels: Drag from the lower-right vertex to the left side’s center point for Green, and from the lower-left vertex to the right side’s center point for Blue. You may hide the guides now if you like. Click back on the RGB composite channel, then rename the triangle layer to Base. Your final triangle will show saturated channel colors in its angles and close to 35% gray in the exact center.

Add a Curves adjustment layer above that and set the Curves blending mode to Hard Mix.

Image

Cool, huh?

Now tweak the curve and watch the pattern change (try the Solarize preset we created in the “Helper Layers” chapter of Part II, “Techniques”). Remember that setting a blending mode on an adjustment layer is effectively a self-blend of everything below the adjustment (or what the adjustment layer is clipped to).

Variations you can try with this setup:

  • Duplicate the base triangle and use Cmd/Ctrl+I to invert the colors.

  • Duplicate then rotate the triangle to see how different colors interact with various blending modes as well as Opacity and Fill settings.

  • Add adjustment layers between the triangles or clipped to the top one.

  • Use a Posterize adjustment in various places (keep the levels settings fairly low).

  • Use this to explore color interactions, generate palettes of color chips, and so on.

Hue Wheel

Another favorite of mine that probably looks more familiar is a hue circle. This takes advantage of the spectrum gradient we created. In another new document sized 2000 px by 2000 px, create a guide layout of two columns and two rows, so that you get an intersecting guide right in the middle of your canvas. Switch to the Elliptical Marquee tool, ensure snapping is on, then hold Option/Alt+Shift while you drag out from the center guide intersection so that most of the square canvas is taken up with the selection.

Image

Select the Gradient tool, load the Spectrum - RyGcBmR preset we created earlier, and select Angle Gradient. On a blank layer, drag from the center to either the top or right of the selection area. The direction in which you drag places red along that axis. In HSB color space, RGB Red is considered 0°, while green is 120° and blue is 240°. For this version, I chose red to be at the top for consistency with the RGB triangle.

Name your new layer Hue Wheel.

With the selection still active, add a blank layer above your hue wheel, then use the B-W 0% Smooth preset with the radial fill gradient (select the Reverse option) and drag from the center to the edge of the selection. This should give you a gradient from white in the center to black at the circumference. Name this layer Luminosity.

Create another blank layer, then edit the gradient to replace the white color stop with fully saturated red (Hue: 0, Saturation: 100, Brightness: 100). Drag again from the center to the circumference, creating a gradient with red in the center and black at the edge. Name this layer Saturation. Here’s the layer stack so far.

Turn off the Saturation layer, and set the Luminosity layer blending mode to Luminosity.

Image

With your fresh, new experiment file ready to go, group the Luminosity and Saturation layers, then start adding adjustment layers and blending mode changes as before. This particular version allows for incremental rotations so you can align virtually any combination of colors. Adding saturation and luminosity layers enables you to see new patterns based on those characteristics in addition to opacity, fill, blending, and so forth.

Honestly, this one is also just a lot of fun to play with. The layer names are just to keep track of the original settings, but please do try all the different blending modes! Duplicate the Hue Wheel layer and rotate it, add some Posterize and other adjustment layers, create flat copies and apply filters—go nuts! Part of the value of these exercises is just to have fun and see what happens. Sometimes the results will be quite boring, but other times they may spark a new idea or give you a creative boost.

Solving Problems

I tinker with these and many more experiments all the time, building little files to investigate certain questions. The methodology here is similar to any scientific endeavor in that I take time to think about what I want to explore, and ask myself questions about how best to reduce the variables. The experiments themselves are not meant to be pretty, but they are meant to let me see things in an unambiguous way. Sometimes that means making multiple experiments. Although the projects I’ve talked about here focus on the RGB display spectrum, I also dive into virtually everything I can find to play with in Photoshop, including filters, plug-ins, tools, and any other element that is exposed in the user interface.

Sometimes these experiments help me reverse engineer certain techniques, such as simulating classical realism style painting. Other times they show me the limits of some function or technique, which both sets a boundary and poses new questions: What if I want to go further than the technique allows? The general process is to think about the challenge ahead and then to break it into smaller tasks. Often times, the value of an experiment—including a failed one—is how it helps shape your thinking about a problem, or expands your vocabulary regarding other situations.

We can obviously go far beyond tinkering with color and brightness, and use various filters. A classic example is using Difference blending mode with a slightly altered version of some layer content to grab the differences as a selection, as we did with in the “Hue/Saturation Selection” section of the “Selections & Masking” chapter in Part II. Instead of altering hue, what if we slightly blur a layer? The result is that high-contrast areas are now low-contrast, so we end up with a display of high-contrast edges that can be selected and used as a mask.

Image

That mask is a great starting point for selective sharpening.

Moving deeper into filters, let’s say you want to describe the distortion effects from using the Path Blur feature in Blur Gallery. Applying a path blur to a photo is a great place to start, but how do you know if you like it? How does the blur change as the paths are changed? What would you use to show the changes, and better yet, how would you teach yourself better control over those changes?

First, we know we want something that doesn’t resemble our own art work so we can evaluate the effect and controls on their own without being influenced by whether we like the result. We also know we want to reduce the amount of variables to only the essential. The variables we typically have to work with are color and brightness, so what minimizes both? A layer filled with 50% gray meets that description, right?

Image

But performing a task that moves pixels around is boring and pointless if you can’t see those pixels. We can add some visible pixels with the Noise filter (Filter > Noise > Add Noise). A modest amount will do, so let’s fill the gray layer with 10% monochromatic and Gaussian distribution.

Next, we can apply some blur (Filter > Blur Gallery > Path Blur).

Now we have a baseline for seeing the exact effect, and we can build from there. Changing the path lets us see the scope of the effect and how certain controls behave. You now have another entry into your vocabulary—you understand something more about the Path Blur feature. You just discovered a new hidden power, but that power is yours. It does not belong to Photoshop.

Image

The experiment is useless without context, though. How do you fold this new power into your work? Would it be good for building textures? Can you control it well enough to imply complex motion?

Image

Now What?

We’re going to apply general problem-solving skills in the next few projects, building on the material provided in Part II, “Techniques.” The purpose of going through experiments at this point is to provide you with a way to address challenges you will face in your editing and retouching, by giving you a way to reset your mind. As you identify the tasks at hand, you are likely to discover that the instructions for a given technique don’t exactly solve the issue. You cannot simply type in values for certain dialog boxes or exactly follow provided steps when you’re dealing with images that are vastly different than the examples provided.

Working on experiment files when you’re not under a deadline or don’t have an important project in front of you allows your familiarity with the features, techniques, and tools to grow. You’ll recognize patterns and behaviors more quickly. You will start to see new possibilities simply because your mind is making connections. Experiments give us more points of reference, and more points mean more pathways to travel.

And sometimes we just need a break. Building experiments to solve problems is great, but we do need more. Build experiments to play, to kill time; the things you discover will leave little seeds in your imagination, and when the time is right those seeds will begin to sprout.

In the projects that follow, I will not be providing details on every single step because the point is to share how I think about approaching the images. All of the projects rely on techniques previously described in the book, but they also include additional tools and methods I have not talked about earlier. The reason for this is to show that the material provided is not meant to stand alone and solve all of your imaging problems; they were always meant to expand your Photoshop vocabulary, to empower you as the artist. That means it is on you to mix this new knowledge with whatever else you can find, what you already know, and what you will learn when you put this book down.

Image

Basic Portrait Retouching

The next three sections are all about seeing various techniques in action. We’ll start with a basic portrait retouch that introduces frequency separation, then move to something a little more complicated with stylization and compositing. After that, we’ll look at Frequency Separation as a technique and how we can augment it with blending modes and adjustment layers.

Image

Let’s revisit this portrait of a dancer friend of mine. You tried a basic application of dodge and burn back in the “Dodge & Burn Methods” section of the “Dodge & Burn” chapter in Part II. It’s worth going into a little more detail here.

Before we do anything else, however, let’s figure out the things we want to accomplish with this portrait. I find that either writing down a small list or making annotations on an image helps me keep focused and organized. In this case, I’m going to use both annotations and a list that breaks down the various tasks.

Image

When first approaching an image, I may not have a final vision in mind, but I can identify some elements to address. Here’s a list of what I can immediately think of for the dancer portrait:

  • Darken the background.

  • Clean up the stray hairs.

  • Smooth out the brightness variations on her skin.

  • Even out the skin color.

  • Add some highlights to her eyes.

  • Get rid of the odd shadows and textures on her neck.

Image

Just putting a few things down should suggest some techniques right away, such as dodge and burn to deal with the eye highlights and skin variations, for example.

There is no required order for these tasks, but conventional wisdom says to start from big features to small details. For this project, however, I’m going to jump around a little because I want to show you the various techniques in play rather than giving you a workflow to follow.

The simple thing to take care of is the background, and I used Content Aware Fill on a new layer to take care of the gap left by the handheld backdrop. To darken the rest, I used a solid black color layer set to about 75% opacity. Both of these layers were grouped together, then I used Select Subject to create a loose mask of the model, and applied that to the group. To give her a little depth, I added to the group a Gradient Fill layer set to Linear Dodge (Add) at 20% Fill. Don’t worry about the gradient colors used; I just tried a few combinations until I found something I liked. All I wanted was a tiny bit of subtle texture behind her.

Image

Looking next at the skin and hair, I added a single pass of Frequency Separation (which you’ll explore in its own project later in this chapter). Frequency Separation is a method of isolating color information from detail information, allowing you to work on these elements individually. On the high-frequency layer, the most usual approach is to use the Clone Stamp or Healing Brush to remove small blemishes. Take a look at the high-frequency layer before and after (I turned off the background changes to make things more visible).

Image

Notice the small details, such as some hair strands and a few skin details, have been removed. I tend to use the Spot Healing Brush; in this case, I used a brush size that is just a little larger than the features I need to address (say, a skin blemish or speck of makeup). But I also like to use the Luminosity blending mode when possible. For this kind of scenario, tiny features stand out because of contrast that includes differences in brightness. Replacing or smoothing the brightness changes makes some features much less noticeable without obscuring or removing the feature itself.

The regular Healing brush is also very effective, but does require the extra step of choosing a sample region. However, this gives more control and I prefer to use it for strongly directional texture such as hair. Looking at the stray hairs, most are lighter than their surroundings, which suggests using the Darken blending mode. This preserves more of the base texture while removing the brighter regions. I turned off the adjusted background to make the changes to flyaway hairs more visible.

This blending mode approach is also well suited to dealing with the flyaway hairs that invade the background.

Image

Rembrandt Portrait

Back in Part II within the “Dodge & Burn Methods” section of the “Dodge & Burn” chapter, we looked at a few different approaches to adjusting contrast in an image manually. In this project, we’re going to choose various techniques to achieve different goals. For this project, I wanted to let the final result be influenced by Dutch painting master, Rembrandt van Rijn. His work is typified by rich, warm colors, a spartan background, and a sense of realism in portraying his subjects. The mechanics of his look were frequently accomplished by limiting the color range, using a single light source, and implying detail rather than painting it directly.

Image

My starting image is a portrait I shot specifically for this effect. Because I can visualize the end result reasonably well, I am left with choosing techniques to support that goal. Here’s a brief list of tasks to accomplish:

  • Reduce the overall brightness.

  • Apply a strong, warm overall tone.

  • Draw focus to the center of the face.

  • Clean up various details and skin blemishes.

  • Replace the background.

  • Apply a texture.

Image

Because I would be replacing the background, I wanted to start with a mask. My initial attempt at using the Select Subject option of the Quick Select tool was really good, but missed some of the hat and made the edges a little rough. Using the Gradient Map Luminosity trick from the “Gradient Map Luminosity Selection” section of the “Selections & Masking” chapter in Part II allowed the tool to make a much more complete selection. The secret is to ensure the adjustment layer is active so the AI has something to look at.

Note

As this book went to press, Adobe updated the Select Subject tools with Sensei AI, and they are amazing and magical! That notwithstanding, the techniques here are sound and useful because even magic sometimes fails.

Image

With the mask applied, I just needed to add a black layer for the background until I was ready to composite the model with my final scene.

Image

From here, I wanted to apply some rough shading around the entire model. The fast way to do this is with a new layer set to Overlay (ensuring the brush is in Normal blending mode). I was not worried too much about introducing color artifacts, so I had some liberty to really darken things quite a bit. If I had wanted to take more care in preserving color and avoiding harsh transitions, I would have used the two Curves adjustment layer method discussed in “Dodge & Burn Methods”. My soft round brush was set to around 8% flow and is about 1/4 the size of the face, which still required some care to prevent building up too much dark density. In just a few broad strokes I had the overall shape I wanted.

Image

For the next pass, I wanted to shape the face and clean up some of the skin variations. This required another Overlay layer, but also a helper layer to show only Luminosity (“Helper Layers” chapter in Part II, “Techniques”). The Luminosity layer gets toggled on and off during the process, so make sure you are always painting on the correct layer—remember to use good layer naming conventions!

Image

For final contrast details, I used a Soft Light layer and just went around making little highlights and shadows, in particular along fabric seams and around the eyes. Soft Light gives just a little different effect than Overlay, providing a touch more contrast to work with.

Image

This is about as far as I wanted to go with the model alone, so after a little spot healing to clean up some dust and stray pixels it was time to group the correction layers together and copy the mask from the subject to the group. I planned to copy a flattened layer to my background composite, so I wanted to ensure that stray corrections would not be caught up. The subject was completely isolated on the merged layer, but all the changes were in place. I was ready to easily paste this layer into my new scene.

Image

I prepared the basic scene, which is a composite of two images with a single, admittedly sloppy mask to blend them. The subject covered most of the elements that I didn’t want in the final work, such as the plates and coffee pots that don’t really seem to fit in a portrait from the late 1600s.

Image

Since the plan is to darken the background quite a bit, there’s no need to fuss over details here. I added a dark tobacco color via a Solid Color fill layer set to Normal and 70% opacity to set the color tone, then I added a Levels adjustment to make everything darker.

Image

With some context for lighting and overall feel in place, I refined the model. I added some color first by creating a Solid Color fill adjustment layer and clipping it to the cut-out model. Sampling a color from the edge of the fire gave me a great tone to start with.

Vivid light will be a great choice for blending this kind of atmospheric color because it combines Color Dodge with Color Burn. The initial result is going to be odd and harsh, as you can see.

Image

There are two key controls to deal with this. First, adjusting the color itself gives a better preview of the final result. Open the Color Picker and first adjust the Saturation (S) and Brightness (B) controls. This is very much a subjective adjustment, because predicting an exact color to use is going to be nearly impossible. Selecting a color from the edge of the fire is only meant to be a starting point.

Image

Color Dodge and Color Burn are among the Special 8 blending modes that respond differently to the Fill slider than they do the Opacity slider. Vivid Light is a combination of the two, so it is also part of the Special 8. Lowering the Fill value is the second major control that helps the model look more like he fits into the environment. Here, Fill is reduced to about 50%.

Image

The model needs a little glow from the ambient light, and that calls for Color Dodge. Each shoulder gets a little splash of color, also sampled from the fire, but because the lighting intensity is different on each side, I chose to use separate layers for each side. I also created a mask from the cutout model’s outline, and then inverted it to constrain the color to only the model. Each shoulder gets a slightly different Fill and Opacity value, but uses the same color.

To finish up the composition, a scratched copper texture set to Overlay blending provided some extra grit, but it has its own color that pushed everything too dark. I wanted the texture, but not the color. I removed the hue easily by clipping a Hue/Saturation adjustment layer to the texture layer, and then dragging the saturation slider all the way to the left.

Image
Image

Frequency Separation

Frequency Separation is an extremely popular and powerful technique, having grown out of a need to isolate different characteristics for more targeted editing. While it’s been around for years in professional (and let’s face it, geek) circles, it is mostly seen as a tool for working on facial skin in portraits. The idea is pretty straightforward to describe: By splitting high and low frequencies in an image, it is easier to work on each independently without affecting the other. This enables faster, more precise editing, and in some ways is more flexible than working on the base image by itself.

Image

However, most people are left wondering what frequencies are in the first place, and once that is cleared up, how do you choose the range of frequencies you need to work on?

Image

I want to give you some background on the name of the technique, and then use that working definition as a way to expand into a somewhat novel approach. Frequency is a term that relates the idea of an event or occurrence happening repeatedly through time. A ticking clock, the sun rising and setting, and even heartbeats are all examples of frequency when you measure how often they happen over the course of a year, a day, or a second. Frequencies also measure very tiny things like sound waves and light, and we observe the effects as pitch and color.

When we talk about frequency in the visual domain, we are talking about patterns and scale of things we can see. This applies to objects as well as textures and small features in photographs. As noted above, skin has a so-called high frequency we can examine because of pores, small blemishes, and even fine hair; so, high is about texture and detail, while low is about color and shading. What makes it high is not the size of the feature in the real world, but the relative changes in color and brightness that allow us to see details in an image. Because digital images are made up of pixels, the number of pixels involved in a transition from bright to dark is a basic way of describing frequency.

Image

High frequency implies the change from bright to dark occurs over a very small number of pixels. So a low frequency feature is one that occurs over lots of pixels. An example of high frequency would be the transition from white to black being a hard edge or boundary. Low frequency is a gradient transition. With a combination of filtering and blending modes, Photoshop lets us treat these frequencies differently.

This is incredibly useful because it often happens that frequencies overlap. Again, the pores (high frequency) in a model’s skin (low frequency) pretty much always go together. Note that highlights and skin blemishes can be either high or low frequency depending on size and edges, so when considering how to set up your tools, think about the size of the boundary or transition.

Image

Ok, enough background! Let’s get to the techniques and how to make them work for you.

Let’s start with the most popular example of skin retouching. Take a moment to think about the size of the features you want to correct. In this example, the pores are only a few pixels across, so keep that in mind as you choose your settings. Once we have the elements in place, I’ll explain the technique a bit more.

Also, it’s worth pointing out that there are many variations on this technique. I’m going to start by showing you a common, basic version with the addition of my choice to use tool blending modes while working with some of the tools.

Start by duplicating your background image two times. Name the top copy High and the other copy Low. We are going to first create a version of the image that has no high frequency detail.

Image

Turn off the High copy, select Low, and choose Filter > Blur > Gaussian Blur. The goal is to apply just exactly enough blur to remove the details you want to edit. Typically, the blur radius you select should be 0.5 to 3 times the size of the feature, so a pore that takes up 4 pixels would require a blur radius of anywhere from 2 to 12. When the feature is low contrast, you can rely on smaller radius values, but increase a little if the features have more contrast. There are some caveats to be aware of, which I’ll discuss a little later on.

Next, select the High layer, make it visible, and then change the blending mode to Linear Light.

With the High layer still active, choose Image > Apply Image and choose Low from the Layer menu. Use the following recipe:

  • Blending Mode: Subtract

  • Invert: Deselected

  • Scale: 2

  • Offset: 128

  • Preserve Transparency: Deselected

  • Mask: Deselected

In plain language, Apply Image takes the current image (an unaltered copy of the original) and removes the blurred version, including hue, saturation, and brightness. Photoshop then divides the result of the initial blend by the Scale factor, and then adds the Offset value (in units of grayscale brightness) to that output. In simpler terms, Photoshop subtracts the Low layer from the High layer, divides that result by two, and then adds back 50% gray. Dividing by two reduces the color contributions, but also darkens the image significantly. Adding 50% gray back increases the brightness without adding color.

What’s left is something that looks similar to using the High Pass filter, which is commonly used for sharpening.

The result is that you have two separate layers that combine to give you your original image back. The High layer contains only detail information. If you switch its blending mode back to Normal, you’ll see that you’ve isolated the details and lost both color and brightness values. I like to add the blur pixel value to the High frequency layer name in case I need to use another frequency separation pass.

Image

Next, duplicate both High and Low frequency layers, and clip each to its original. Set the clipped layer blending modes to Normal, then lock the original High and Low layers to prevent editing. Group all four layers and name the group Frequency Separation.

These clipped layers allow you to work non-destructively, giving you the option to reduce opacity or mask some of the corrections later on.

Now the question is whether you should work first on the high- or low-frequency layers. Honestly, I go back and forth depending on the image. For this example, I started with the Low layer to try and distinguish between natural skin variations and blemish colors. My goal in such cases is not to make perfect skin, but to make the model look natural and realistic. Starting by evening out the color helps me decide which skin features to retain, which to minimize, and which to remove.

Image

For a long time, my preferred approach was to select areas with the Lasso tool, then run Gaussian Blur on the selection, using a low value but repeating it several times. More recently, I’ve switched to using the Mixer Brush to blend colors because it feels more intuitive, and prevents some odd boundaries you get from the edge of the lasso selection. There is a caveat, however: It becomes extremely easy to mix everything to the point that you destroy the character and produce a plastic sheen or other bizarre texture. The rule of thumb is to ease into the blending! Go slow, use your power with restraint.

For blending, I use a round, soft Mixer Brush with the following settings:

  • Load Brush: Deselected

  • Clean Brush After Stroke: Selected

  • Wet: 20%

  • Load: 20%

  • Mix: 10%

  • Flow: 10%

  • Sample All Layers: Deselected

Image

Remember to work on only the copies of your High and Low layers!

The brush size should be a little on the large side at first. For portraits, I like to use the model’s eyeball as a rough gauge. My first pass around the image is done very lightly to address harsh transitions and edges, then working smaller to discolorations, and finally blemishes. I find working large to small sometimes reduces the effect of blemishes, and when working towards a natural result you may not have to do much more than smooth over the color variations.

The discoloration from blemishes is pretty much gone by this point, leaving small defects in the skin to deal with.

Image

Now you can work on the High copy layer with any number of tools. An easy one to start with is the Spot Healing Brush with a 100% hard edge, sized to just a little larger than the feature you want to heal. That means you will be constantly adjusting the size. Just remember that the “feature” may not be an entire blemish; the brush should only have to deal with small details. Work as small as reasonable.

I also prefer to start with the Spot Healing Brush set to Luminosity blending. In many situations this won’t be very noticeable because the layer is already mostly gray, but it does help prevent stray color pixels from showing up if you use a large radius for the blurring step. However, there are two other blending modes that are really helpful for fine details. When there is a dark feature surrounded by lighter colored details, try Lighten blending mode on the Healing tools.

Image

Notice that the darker patches are now gone, but the lighter hairs are still present on the skin. Similarly, using Darken when a light feature is surrounded by darker areas works well to preserve the local texture.

You can also use the regular Healing Brush, Patch tool, or Clone Stamp. Each has its own pros and cons, and the Clone Stamp has all the normal tool blending modes available. There is no real right and wrong here, so try each of them out to see when you might use each.

A word of caution, though: Ensure you are always working on the correct layer, and check the tool options so that tool is not sampling from all layers. If something looks horribly wrong, check the Sampling and Blending Mode options as well as your working layer!

Variations

There are lots of variations available with this technique. Using the Mixer Brush on the low-frequency layer tends to change all the elements of color, including moving brightness around. This can cause unwanted changes in the shape of facial features. You can correct this with subsequent use of dodge and burn, or you can avoid some of it by separating out the color and luminosity layers. If you want to try this out, you can isolate Color and Luminosity layers as described in the “Helper Layers” chapter in Part II, “Techniques.” You will want to treat the Luminosity layer with dodge and burn techniques rather than the Mixer Brush or blurring tools. Just be warned that this approach takes a lot more attention to detail and patience, because you are working on three different layers.

Another option is to try different kinds of blur. A popular method that seeks to preserve edges uses the Median Blur filter. This is really great for dealing with architectural and textile photos where you need to move large blocks of color around within the bounded areas.

There is something to consider when choosing your blur size, too. Different types of textures or detail may not be completely isolated from the color layer, so you should get in the habit of checking out the results of your blurred color layer. If some of the detail is left behind because the edge is larger than the radius you set for the blur, you risk damaging that texture by blending or giving it additional blur. On the other end, if you use too large of a blur radius, you don’t have much wiggle room when trying to correct smaller defects. And at some point the blur is so large that it will not line up properly with the detail layer.

That’s why I suggest you get comfortable with the idea of using multiple frequency separation passes for different textures and amounts of detail. This will be more common when dealing with images where you have closeups of both skin and fabric, or multiple textiles that need to be retouched.

Image
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
18.189.171.52