This article is provided courtesy of Peachpit Press. All text and images © Dan Moughamian, unless otherwise noted.

This article details useful techniques for creating surreal composite images in Photoshop CS4. We all have our own definition of what it means for something to be "surreal," but the idea is to bring elements together seamlessly that you traditionally don't associate with one another.

For this example, I utilized three stock images from iStockphoto: the amazing Iguaçu Falls in South America, a woman taking in her surroundings, and a statue of Buddha (Figure 1).

Figure 1. I'll show you how to blend these three stock images into one finished landscape.

Aside from the context provided by the subjects themselves, the lighting and the coloring of those subjects are also important. I made sure that the shot of the woman and the statue were captured in something approximating cloudy, diffuse light (similar to the shot of the falls), and that the colors of the subjects would blend well with the background scene. For example, the off-white coloring of the Buddha status could blend well with the water in the falls or the clouds; the woman's dark clothing could blend well with the rocks or grassy areas in the waterfall shot.

Artistically, the goal of this composite was to bridge the plausible and implausible in a single scene, while providing some message to the viewer. Here the message is left to the imagination. It could be environmental, spiritual, or a mix of the two. It's up to you as the artist to decide what you want your scenes to mean.

I used the falls as the starting point. This provided areas in the foreground, middle-ground, and background for placing my other images into the scene, effectively widening the scope of creative options. A photograph of a woman with her back to the camera gives us an opportunity to add a human element, similar to what we might imagine in the real world at an eco-tourist location like this. Granted, she's not necessarily dressed for the occasion, but sometimes details like this can bring the viewer in to really examine things; everything doesn't have to be perfectly logical. Finally, the statue of Buddha provides the surreal element to this scene.

NOTE: Often, especially in the world of marketing and advertising, stock images are a common source of material for compositing projects. If you don't have your own set of images to use, exploring sites like iStockphoto.com can be a big help, and usually it's quite easy to set up an account. Just be aware of each image's licensing limitations when choosing shots for your composite artwork.

Gathering All the Shots into One Document
The first step is to open the shot of the falls and save it as a PSD file. Next, I had to bring in the shots of the Buddha statue and the woman. We want to maintain as much editing flexibility as possible when compositing, so use File > Place to bring all of your images into the main scene (see Figure 2). This will create a Smart Object Layer for each placed image. Note that Smart Object Layers have distinct icons and layer masks in the Layers panel. These masks will be important later in the process.

Figure 2. Using the waterfalls as a backdrop, the other elements are added to the document as Smart Objects using File > Place. Click on the image to see a larger version.

Once you've gathered your images together in one document, the next problem is usually the simplest to address: Placed images often require scaling to make the visual illusion possible. Anything that looks obviously too small or large will detract from the quality of the final image. For this concept, the woman needed to be smaller so that she could be placed somewhere in the foreground or middle-ground of the landscape shot (among the rocks and grass); the statue of Buddha also needed to be smaller, although it ends up being a very large object relative to the scene as a whole.

To scale a Smart Object Layer, select the layer and choose Edit > Free Transform (Ctrl-T on the PC, Command-T on the Mac). Next, hold down the Shift key to constrain the image's proportions, click a corner handle, and drag it inward to reduce the image (as shown in Figure 3), or drag outward to enlarge it. Click the Enter or Return key when you have a good approximation of the size you want to use.

Figure 3. Use the Free Transform command to scale Smart Object Layers properly. Click on the image to see a larger version.

NOTE: You should be able to scale your Smart Object Layer up to its native pixel dimensions before you notice any softening. Reducing the size of an image layer usually doesn't have a pronounced impact on quality, though obviously the details get smaller as you go. Don't worry if you're not absolutely sure how large you want each layer to be -- just get close. You're likely to tweak the scaling during the project.

Beginning the Detail Work
The next part of the process is placing the Smart Object Layers in the appropriate area of the scene so you can begin your detail work. Since you won't yet have masks in place to hide the extraneous subject background (in this example, the white area around the Buddha statue and the nature scene surrounding the woman), you can reduce their opacity by 40–50% to allow some of the background to show through as we're roughing out our scene (see Figure 4). Once you've got the final masks created for your layers, you'll make pixel-precise placements, so again just get close when starting out.

NOTE: Masks will be covered in detail on the next page of this article.

Figure 4. While you will eventually mask away the parts of your subjects you don't want to see, a quicker way to get started is to reduce their opacity temporarily to find the proper image placements. Click on the image to see a larger version.

Another placement choice involves depth. Sometimes that means deciding whether you want to create the appearance that one of your subjects is "between" two other elements in the background scene. The alternative is to place one or more of your subjects directly "onto" a surface in your scene that's not hidden by other parts of the scene (such as the rocks along the river plateaus in this example). This placement only requires that you mask your subject so that it appears to be naturally at rest on the surface in question.

To place a subject "between" areas of your scene, you have two options:

* Zoom in closely and precisely mask away the areas of the subject that must be hidden for any given placement.

* Mask out the background of your subject (once), and then sandwich that layer between two copies of the background image. From there, you can mask away the top background layer to reveal your subject. To demonstrate this trick, I "floated" the statue of Buddha between the vegetation in the middle of the scene and the falls in the background (see Figure 5).

Figure 5. Using two copies of your background layer to "sandwich" a subject allows you to mask away parts of the scene quickly, revealing your subject. Click on the image to see a larger version.

I generally prefer the "sandwich" technique, as it allows me to make a precise mask of my placed subject one time and then move it from place to place, masking and unmasking things more quickly than I could otherwise. So, for this example, I first masked the exact shape of the Buddha statue, placed the statue in the right spot, and then masked away the top background layer to reveal the statue "ascending" toward the top of the vegetation.

It's important to get the brush hardness correct in these situations. If you mask with too soft a brush, you can end up with an unnatural "glow" or haloing effect where the placed subject is revealed, instead of slightly crisp edge details as you would expect to see in the real world.

The Next Step
On page 2, I'll discuss masking in more detail. On page 3, I'll explain how to make improvements to the tone, colors, and focus of your subjects, so that they appear to blend more accurately with the parts of the scene they occupy.

Go to page 2.

Return to page 1.

On page 1, we gathered three stock images into one document, scaling various elements of the composite to fit together visually. Now we'll consider how we can mask away the extraneous parts of our source images, so that the remainder appears to be part of the original scene.

For the waterfall example, I started with the woman standing alongside the river and park bench. I needed to add her to the waterfall scene in a realistic way, so my three primary tasks were placing the woman into the scene, scaling her to fit the placement (both of which were discussed on page 1), and then masking away any extraneous areas from the original photo that would detract from the composite scene.

Having already chosen a placement and scale (foreground shoreline), I had a couple of masking choices. I could either integrate some of the grass and leafy material under the woman's feet, or I could try to hide everything in the scene except for her. I went with the latter choice so that I would have the option of placing her directly onto the rocks or shoreline, without having to mask and unmask the areas around her feet as I repositioned her.

To get started with my masks, I like to use the Quick Selection tool to make an approximate outline (see Figure 1). This method is fast and provides a good starting point.

Figure 1. There are many ways to create a layer mask for your subjects. For this subject, the Quick Select tool was used to generate an outline of the model's body.

For this image, I tapped W to activate the Quick Selection tool, and then sized the brush cursor down so that the cursor fit within the smallest part of the subject's body -- her head.

TIP: To size the brush cursor down, press the left bracket ([). Press the right bracket (]) to increase the brush cursor's size.

Next, I placed the stylus on the tablet and slowly dragged the cursor around the periphery of the subject's body. In many cases, if you have a strong tonal contrast, as I did here, the entire subject is selected quickly with this technique.

TIP: It's common for extra bits of picture to be included when using Quick Selection. To remove them from the selection, pick up your stylus, hold down Alt/Option, and then make a separate brush stroke over the edge of the extraneous regions.

Once you have a decent selection outline of your subject, it's easy to turn that selection into a layer mask. Click the target layer; then click the Add Layer Mask button at the bottom of the Layers panel (see Figure 2). This action automatically masks away the rest of the layer so that only the outlined region remains visible.

Figure 2. Creating layer masks from selections is a good way to get a quick start on the process, saving many brush strokes.

Perfecting the shape of a layer mask requires two separate steps. First, I generally use the Brush tool to paint around the edges of the subject with a combination of white (reveal masked area), black (hide visible area), and gray (partially reveal or hide a masked area), so that a precise and relatively smooth edge is created. As shown in Figure 3, I zoomed in to 300% and brushed along the edges of the subject's figure with a small white brush at about 70% hardness. The reason for this brushwork is that the final Quick Selection was hiding the outermost edges of the woman's clothing. Later in this procedure we'll uniformly shrink the selection (leaving only the woman visible) by using the Mask Edge function.

Figure 3. Use the Brush tool to perfect the shape of the layer mask that hides the extraneous material around the subject.

Brush hardness can be an important factor when tweaking a layer mask. As a general rule, it's rare that you'll use a brush with a 100% Hardness setting to create mask edges. Next time you're outside, pay attention to the edges of things in the distance. Cars, people, vegetation -- do any of them appear to have perfectly defined edges (like an illustration)? Or is there a tiny bit of haze and softness in many cases? This slight degree of softness is what you're trying to mimic when masking your subject.

One new development for masking workflows in Photoshop CS4 is the Rotate tool. This tool allows you to rotate your image preview quickly, so that you don't have to brush along awkward diagonal edges. This feature makes it easy to brush around complex shapes with your stylus and tablet (see Figure 4).

Figure 4. The Rotate tool simplifies brushing around complex shapes when perfecting a mask.

To use the Rotate tool, hold down the R key, place the stylus on the tablet, and move right or left to rotate the preview. Once you've made your brush strokes and you want to move to a different orientation, just hold down R again and continue rotating, or tap Escape to remove the temporary rotation effect. If you prefer, you can use the Rotate Angle field in the Options bar to specify a numeric value for the rotation amount.

The final step in creating the masks for the woman and the Buddha figure in the example was to make a final correction by using the Refine Mask dialog. Make sure that the layer mask is selected in the Layers panel; then click the Mask Edge button on the Masks panel. This action opens the Refine Mask dialog, give you precise control over the entire boundary of the mask shape (see Figure 5).

Figure 5. The Refine Mask dialog (accessed from the oddly named Mask Edge button in the Masks panel) is a great way to make uniform adjustments to the shape of the entire mask contour.

My goal was to "shrink" the mask a little bit so that it stopped right at the edge of the woman's hair and clothes. I accomplished this using a combination of the Radius, Contrast, and Contract/Expand sliders, along with the "mask on black" and "mask on white" options. Generally I keep the Smooth and Feather settings at low values when I'm trying to maintain a degree of realism. Sometimes you may find that most of your mask looks perfect but one area still has extra image data showing through, or perhaps a bit too much taken away. For these cases, go back and touch up the mask with the Brush tool; then try again.

Figure 6 shows how the woman looks as she relates to the rest of the scene after the final masking tweaks were all applied.

Figure 6. Creating precise layer masks around the images you've placed into your scene is crucial to creating the beginning of a believable look.

As I continued working with this image, I masked away the white areas behind the Buddha (seen in the other image) using similar techniques, except that in this case I selected all the pure white pixels by using the Magic Wand tool and a low Tolerance value. Sometimes you only need the most basic tools to get a good head start on your layer masks!

The Next Step
On page 3, I will describe the details of fine-tuning the tones and colors of our placed images, so that they appear to fit within the lighting conditions and general surroundings of the waterfall.

Go to page 3.

Return to page 2.

Return to page 1.

Fool the viewer's eyes into believing the impossible. In the final part of this article on the creating surreal landscapes, you'll learn how to pull together all the pieces of your composite image. We'll work with the colors, tones, and textures in each image so that they blend together properly.

Keep in mind as you work that you may need to go back and slightly tweak masks, placements, or scaling repeatedly as you make more changes to the image. Compositing is in iterative process.

Let's take a look at color. As I noted at the beginning of this article, my goal when searching for images to use for the composite was to make sure that all my selections had lighting that was generally similar to that of the overcast sky above the waterfalls. The original photo of the woman in the park certainly had the same type of moody lighting, and to a lesser degree the photo of the Buddha did as well. Both have relatively even, flat lighting. Another part of the light equation is the color that's imparted onto the subjects of each shot. The more overcast the shot, the more blue things seem to look. We'll also need to pay attention to the color of the cloth on the Buddha statue.

To tweak colors and tones, I like to use Photoshop's adjustment layers because they're non-destructive, and you can change their settings later if you decide that you don't like the result. They also have the advantage that they can target specific layers and regions of an image, by utilizing their built-in layer masks.

For the Buddha statue in this example, I used the Photo Filter adjustment (Layer > New Adjustment Layer > Photo Filter) and chose the #80 Cooling Filter. From there, I moved the control slider to the right until the orange cloth looked like it fit into the scene. To make sure that the adjustment affected only the Buddha statue and not the other layers, I clicked the Clip to Layer button on the Adjustments panel (circled in Figure 1).

Figure 1. To isolate an adjustment to one layer, place the adjustment over the target layer in the Layers panel, and click the Clip to Layer button on the Adjustments panel.

However, once the blue was added, isolating the change to the layer with the statue wasn't enough. The stone elements of the statue became overly blue in color, so I needed to isolate the adjustment to just the orange cloth. To do this, I used the adjustment's built-in layer mask and masked the stone areas with a black brush (see Figure 2).

Figure 2. Once the masked Photo Filter adjustment was applied to the cloth, the image looked more natural as a part of the overcast scene surrounding the falls.

Next, to give the statue a slightly more weathered look, as we might expect to find here, I added a Vibrance adjustment layer (Layer > New Adjustment Layer > Vibrance) and then reduced the Vibrance setting substantially. This technique allows you to make substantive changes to the overall saturation of the layer without clipping the colors to gray or white. I used the same Clip to Layer button as before, to ensure that the Vibrance settings affected only the statue.

After handling the orange cloth on the Buddha statue, I used the same technique to fade the color of the woman's clothing a little bit. I didn't make any adjustments for blue this time because it was unnecessary, due to the overcast light under which the model was originally photographed. In fact, she was so blue that I had to warm her up just a bit, so I added another Photo Filter adjustment layer and made a slight boost with the #81 Warming filter.

Another important step with these two images is to correct their tonality to more closely match the lighting of the falls. The Buddha statue is brighter than it should be for this scene and the woman a tad darker than she should be. To make the correction, I used a Curves adjustment layer for both images (Layer > New Adjustment Layer > Curves), clipping each to its respective target layer. For Photoshop CS4, adjusting Curves is a more intuitive process. Simply click the little hand-arrow widget in the Adjustments panel (see Figure 3), move the cursor over the area where you want to modify the tones, and then click-and-drag upward to brighten or downward to darken.

Figure 3. Using the new on-document Curves adjustment, you can isolate corrections to specific areas, such as brightening the woman's jeans.

For tonality of the larger scene, one final set of corrections was needed: using the Burn tool (shortcut O or Shift-O, depending on which tool is active) to darken bits of white water around the woman as well as the vegetation in front of the Buddha statue. In both cases, I used Exposure values under 20% and left the Protect Tones option active.

Now the image was close to taking its final form. One extra touch was to make sure that I used a bit of Lens Blur on the Buddha statue, so its level of sharpness mimicked the level of focus present in the vegetation. To do this, I highlighted the statue's Smart Objects layer, chose Filter > Blur > Gaussian Blur, and applied a blurring value of .3 -- just enough that the stone textures were less obvious at a glance. On closer examination, the textures on the woman's clothing exhibited roughly the same level of detail as the textures on the rock on which she's "standing," so I opted not to modify her clothing or coloring any further. These types of decisions can be subjective, and are a matter of personal preference in many cases.

The final tweaks included slightly modifying the woman's scale and position on the rocks. While it's left to the viewer to determine exactly how large are those rocks and clumps of grass along the river, the model seemed a tad too small, and I wanted to make it appear as though she's stepping up onto the precipice. From there, I just cropped in along the right edge of the frame to remove some extraneous details to the right of the falls, and I was done. These minor changes shored up the composition (if you'll pardon the bad pun) and when zoomed out, allow the viewer to take in everything at a glance (see Figure 4). Notice also how the rock formations at the lower right of the image draw your eye back to the Buddha statue.

Figure 4. Final image after making multiple adjustments to color, tone, position, and scale. Notice the more weathered look of the Buddha statue and the flatly lit appearance of the woman and the water around her.

Summary
It's often the little details that can make the difference between a good composite and a great one. As you work through these types of editing tasks, experiment with different adjustments and lighting styles. The goal here was to make things look realistic, even though we know intuitively that such a scene doesn't exist in the real world. The whole point is to find a style or purpose that speaks to you, and go for it!

© 2009 Pearson Education, Inc. All rights reserved.