9.4. Matching a Knockout to Its Background Hopefully, you have already collected the images you're going to include over the background image in your composite. Of course, if it's a collage, matching the background is probably pretty easy. In all likelihood, you won't even see a background. If you do, just make sure it's a color that flatters the images you're pasting on top it. Also, make sure it's not lit or patterned in such a way that it detracts from the images, rather than enhancing them. Since that sort of thing is always very subjective, it's hard to give you more specific advice than that. On the other hand, if you're making a montage, there are lots of "gotchas" to pay attention to if you expect your end result to really look as though the photo actually happened, rather than being cut and pasted together. 9.4.1. Matching Exposure and Color Balance The first thing you want to do is open the images at the same time and match exposure (brightness and contrast) and color balance. Here are the steps for this stage of the workflow. It may seem strange, but you'll want to do all the above-mentioned adjustments within the image. So just start by dropping each image and its layers onto your background image. Go to the Layers palette and place it right alongside your composite image. Double-click the Hand tool to make the image fill as much of the screen as possible so that you can easily distinguish one image from the other. Name each layer for the subject of its knockout. This ID process is important because you will be doing a lot of work with layers and layer groups in this compositing workflow. So you want to be able to easily find the objects you're looking for. Also, you're going to be attaching a number of other layers to each of these images and you want to group all those layers so they can be easily identified and changed as the client requests. Make your color balance and exposure adjustments. To keep complexity to a minimum and to save space, try doing all of this with Levels adjustment layer. Be sure you attach each of these adjustment layers to their respective images with a Clipping Mask (right-click on the layer and choose Make Clipping Mask). You'll find the recommended Levels routine in the "Levels for Shadow, Highlight, and Overall Brightness" section in Chapter 6. At the end of that routine, don't go back to the color layers to adjust color balance until you've made the initial exposure adjustments for all the images. Adjust the Levels for all the channels, the RGB channel, and overall brightness for all the images, then go back to the Levels adjustment layer for each image. Double-click the Adjustment icon and the Levels dialog will appear for each image. From the Channel Menu, choose the channel whose color will contribute the most to correcting the color balance to most closely match what you want as the overall color balance for your composite image. In other words, you want the color balance of your knockouts to match that of the background image. (If you're making a collage, you'll want the color balance of the images to match one another. However, the easiest way to do that is to use one image as the image to which you want to match all others.)  | Don't use the Match Color command for this job. More often than not, you'll be very unhappy with the results. |
|
Repeat the step above for each of the other images. Match exposure for all the images. You've already pretty much done that with the Levels adjustment layer by adjusting the Midtone slider in the RGB channel. However, you may need to have control over contrast in some areas of the brightness range, so add a Curves layer to those images that need it. Be sure to use a Clipping Mask on that layer to attach it to only the specific image you're adjusting. Make the Curves adjustments as you've been instructed in the "Curves for Adjusting Contrast in Specific Brightness Areas" section in Chapter 6. Repeat the step above for each of the other images.
Figure 9-17 shows how the layer structure looks at this stage of the workflownot taking into account any other layers you might have created in making each of these images before they were included in the composite. Figure 9-17. Because it's a collage, it's easy to exaggerate the effects and to show what layer effects do for a collage. In a montage, the subject matter, scale, and edge blending would be the only differences. 9.4.2. Lighting and Casting Shadows One of the hardest things to keep consistent between imported objects and the background photo is the hardness/softness of shadow edges and the predominant directions of primary and secondary lighting. If you're lucky enough to have the time to specifically shoot the objects you'll be importing or to specify assigning them to others, make sure the photographer knows the time of day, year, and location that the background photo was made. The photographer or his assignee should also take along a copy of the background photo when shooting objects to be added. That way, you can at least position the subject in regard to the lighting, put up a scrim if the light needs softening, or move the subject into the shade if the original was taken on a cloudy day. Back to reality. All too often, it's just impossible to control all those elements when the picture was shot. So what you have to do is to imitate lighting and perspective in Photoshop. Following are some steps that will be helpful in most instances: Shoot the incoming photos in flat lighting as much as possible. That is, in the shade or on a cloudy day. If you have the time and the manpower, it's also a good idea to use a soft reflector so that there are no deep shadows. Take the photo from the same distance and angle at which it will be seen in the photo. Create a new layer above the imported object, fill it with 50 percent gray (Edit Fill 50 percent gray), then change the Blend Mode of the gray layer to Overlay, then add highlights by painting in white on the gray layer. Add shadows by painting in black on the gray layer. Experiment with the softness of the brush and the Opacity of the paint to create a believable intensity of highlights and shadows. - DistortLens Correction. The Lens Correction dialog performs many tricks to correct lens distortion. All you want to do at this point is to fool with the Vertical Perspective and Horizontal Perspective sliders and the Angle dial (to rotate the image as needed). The sliders are rather self-explanatory, but you should know that the further you drag the cursor away from the angle ring after click on the approximate angle, the easier it will be to change the angle in precise increments.
9.4.3. Fixing Focus There will be times (but definitely not always) when you want to change the focus in the background image in a montage to approximate a more shallow depth of field. Doing so tends to emphasize the subject and is especially useful if you're placing a head shot against a new background. Figure 9-18 illustrates that point. Figure 9-18. A montage using depth of field simulated by the Lens Blur filter. Notice that objects closer to the model are in sharper focus than those far away. On the left, the image as it would have appeared without the Lens Blur filter. If the subject is grounded (appears to be standing or resting on a flat surface) or appears to be near another object that will be in sharp focus, you will definitely want to use the Lens Blur filter (FilterBlurLens Blur). The Lens Blur filter allows you to simulate depth of field by keeping the area nearest the subject in sharp focus while blurring the objects that are far away from the subject considerably more. Figure 9-18 shows the result of using the lens blur filter on a composited object. Here's how to use the Lens Blur filter to blur the background: Make your knockout and place it in the background image. Light the knockout to match the lighting in the background image, as instructed in the previous section. Make an alpha mask that protects the areas you want to keep sharp and then "graduates" to the area(s) that you want to blur. You can skip this step if nothing in the picture is on the same plane as the subject. For instance, in a head shot you rarely see any part of the model touching the objects in the background. Make a new layer at the top of the layer stack. Make a selection of the areas that you want to keep in focus. Then feather the selection (SelectionFeather) so that it graduates from black to clear over as much distance as its going to take to keep the nearer objects relatively sharp compare to the farthest objects. The area in which the farthest objects reside should be pure black. In this image, I also used black and gray paint to paint over areas that weren't in the primary plane of focus, such as the leaves overhead and the park bench. The park bench is medium gray to ensure that it wouldn't be as out of focus as more distant objects. When you've finished painting the mask, turn off all the other layers and select the entire image (Cmd/Ctrl-A). Open the Channels palette and click the New Channel icon. A new channel, titled Alpha 1 will appear. Name this channel Lens Blur. While it's still selected, press Cmd/Ctrl-V to paste the mask into the channel. You can see the resultant mask as it looked over the background in Figure 9-19. Figure 9-19. This focus mask was painted onto a new, transparent layer after the knockout image had been placed. Notice that in Figure 9-19, I also painted a little spot of black at the bottom of the image. I'd have done the same at the top of the image, but the need to mask the leaves would have hidden the spot. If there's nothing but transparency at the top and bottom of the layer, the mask will be centered in the alpha channel when you paste it in. Once it's pasted in, simply use the Eraser tool to remove the black spots. Make sure the background image's layer is selected. This is where you use the Lens Blur filter. Choose FilterBlurLens Blur. The Lens Blur dialog appears, as seen in Figure 9-20. Created a new, empty layer beneath the Knockout layer. Made a rectangular selection and feathered it by 70 pixels so that it would gradiate from white to gray to white. I filled the rectangle with light gray then dragged the selection to the next position (making sure the New Selection Icon was selected in the Options bar) with the Rectangular Selection tool and filled it again until I had what looked like a striped background or curtain (see Figure 9-21). Figure 9-21. You can see how the selection rectangle feathers and how the Transform marquee takes it into account. A variation on this technique could also be done with Vanishing Point. Once all the stripes were made, I made a new Rectangular selection that covered the bottom portion of the image. I then feathered that selection by 100 pixels so that the stripes would "bend" when I transformed the rectangle to make a perspective transformation. To make this transformation, you will need some space all around the image. Double-click the Hand tool to fill the screen with the image window. Choose the Zoom tool, press Opt/Alt and click in the center of the image. Each time you click, the image will become progressively smaller until you have all the room you need. You can also drag the corner of the window so you have more room from side to side. To make the perspective transformation, all I needed to do was press Cmd/Ctrl-T, then press Cmd/Ctrl and drag each of the corner rectangles directly out to the side and equidistant from the original position. I wanted to "light" the background so that Kashi actually looked like she was standing there and so that the part of the background that was on the floor was slightly darker. I created a new layer above the background layer, made another feathered rectangular selection, filled it with a light gray, and then used the Multiply Blend Mode so that the layer below would show through. NOTE An excellent way to make non-seamless walls and floors is to use the Patch tool. Just make a selection around the edges that should be seamless and then move it onto the floor. This technique works best when the patterns in the floor and wall aren't too pronounced or too different from one another. I needed to "anchor" Kashi with a shadow. I made a new layer, made a freehand selection under her feet, feathered it by 120 pixels, and filled it with a 35 percent gray. I then changed that layer to Multiply mode. I also used Free Transform to stretch the right side of the shadow so that it was cast further to the right than to the left. Finally, I used the Burn tool to really darken the areas immediately under her foot and the coat.
9.4.4.2. Lens Correction filter This filter is discussed a bit more in the "Lens correction" section in Chapter 8. It's perfect for correcting any tilts you might have forgotten to correct in Camera Raw, for getting rid of vignettes and lens distortion, and for correcting the overall perspective. I particularly like the way it lets you correct perspective by rotating the image both left and right and top to bottom, which allows you to visually fine-tune the results. Another use is for creating backdrops. In Figure 9-22, I created a wall on one layer and the floor on another and used the Lens Correction filter to position the wall and floor in separate operations. Figure 9-22. The back wall and the floor were from separate photographs and perspective-corrected with the lens distortion filter. The floor then needed further perspective correction using Free Transform. 9.4.4.3. Vanishing Point filter Vanishing Point is one of the coolest new features in Photoshop CS2. It is especially useful for architectural photography and cleaning up flat surfaces. It is also great at letting you redecorate by adding new doors, windows, or paintingsall in perfect perspective. Figure 9-23 shows the before and after of one such situation. Pay attention to the workflow, this feature is anything but intuitive. Figure 9-24 shows you the images I started with and the image I ended with. Obviously, I was just fooling around to make a point. I'll leave it up to you to do the serious stuff. Figure 9-23. On the right, you see this alleyway as it has been "bricked" in Vanishing Point. Figure 9-24. The image in the Vanishing Point interface. Vanishing Point works by having you set a perspective grid for all the flat surfaces in your image into which you're going to place perspective. If you want to do that according to the workflow laws set by yours truly, first duplicate and flatten the background image. If you don't do that, you'll find this process unrecoverably destructive. Minimize the destruction by putting the Vanishing Point stuff on its own layer. So as soon as you've duplicated and flattened the image (and closed the original image so you don't accidentally overwrite it), open the Layers palette and click the New Layer icon. A new blank layer will appear. If you're going to "perspectivize" an image from an outside source, open it, duplicate it, then flatten and re-size the duplicate so that it's close to being the size in the final image. Also, if that object is a flat object (that's the only kind that will work) that was photographed from an angle, use Lens Correction to correct its perspective. Once you've done all that, press Cmd/Ctrl-A to select all, Cmd/Ctrl-C to copy it to the clipboard. If your target image will use outside images in this effect, you'll have to open them in a separate window, select the portion of the image you want to use, and copy it to the clipboard (Cmd/Ctrl-C). Be sure you don't do anything to replace that image in the clipboard before you're ready to use it. Just in case you do, keep that image open so that you can repeat pasting it to the clipboard if you need to.
Whew! Alrighty! We're finally actually ready to use Vanishing Point. Make sure the new layer is the one selected, not the background layer. What you see should be the background image, the Layers palette and the image you're going to use to surface a portion of the background image. Open the image you're going to use on the surface of the background image. Press Cmd/Ctrl-C to copy it to the clipboard. Choose FilterVanishing Point. Youll see the image in the Vanishing Point interface, shown in Figure 9-24, except the perspective grid won't be in place. Choose the Create Plane tool in the toolbox. It looks like a perspective grid. You're going to draw a perspective grid that looks like the one in Figure 9-25. Click to set the first point, drag to set the next corner, then the next. In this image, I used the four corners of the window to keep the perspective correct. If you make a mistake, do not press Esc! It'll just drop everything you've done up to that point and return you to the original image. Figure 9-25. The bricked wall. If you've laid out the original plane properly, it'll be blue. If it's not geometrically correct, it'll be red. Use the Edit Plane tool (arrow) to move the corner points until the grid is suddenly blue. Stretch the grid up the gray wall on the right. Just grab the center right handle and pull it up the wall. When you get to the edge of the image, change the direction of the grid to lie flat along the white part of the image that contains the incoming picture(s). Then, go back and stretch the bottom of the grid straight out until it reaches the bottom of the image. Press Cmd/Ctrl-V once the Grid is in place. The image you're using to resurface your image will appear. Be sure to place it in an area of the image that you're not resurfacing. The image will be selected as soon as you paste it. Press Opt/Alt and drag the contents of the selection onto the perspective grid. As soon as you do, the image will transform itself to the correct perspective, regardless of which plane you drag it to. Be sure to keep the Marquis tool selected. This time, make sure you've chosen On from the Heal menu. Place the cursor inside the selection and press Opt/Alt again and immediately drag your surface image to the next location. Keep repeating this step until your entire grid is filled. In this image, I've created a brick wall and driveway. You can see the result in Figure 9-25. Click the OK button and you will return to Photoshop. Now choose SelectLoad Selection9.4.5. Matching Size When you composite one photograph into another, the image(s) added to the background image should always be at the same resolution per inch as the background imageor higher. You may be able to get away with images that are slightly too small, but otherwise the image is going to lack detail and contain too much noise. Figure 9-26 shows you a background image with an incoming knockout superimposed on it in a new layer on the left, after resizing but before sharpening and noise matching. On the right, you see how the image looks after it's been resized and the noise level reduced. Figure 9-26. The truck on the right looks like it's been here all along, thanks to noise matching, color balancing, sharpening, and shadow casting. Because you often have to downsize composited objects to give them the right scale in regards to their location in the background, you'll probably lose much or all of the noise (the equivalent of grain in film) that was originally in the image. Always do this when the object is on a separate layer from the background. It's also a very good idea to duplicate the original layer and then hide it beneath the background layer. Then, if you're asked to add more or less noise, you can delete the layer to which you added the noise, pull duplicate the original layer, and make a change. No need to search around to find the original, re-do the color matching, etc. After you've transformed the image down to the size where it fits realistically into the composition, you will generally need to do two things: match image noise and re-sharpen. Both should be done with tools that let you see both the knockout and the background superimposed over one another and should have a preview option so that you can match sharpness and noise visually. NOTE If you have multiple images that need to be added in and resized, you may be able to save disk space by pasting one or more of the originals onto a transparent layer. Then, should you need to remanipulate one of the originals, you can simply select that one subject and lift it to a new layer with our workflow's most useful shortcut keys: Cmd/Ctrl-J. This is the routine to re-size, sharpen, and match noise for the image in Figure 9-27: Figure 9-27. The Transform command used to cast the shadow as the light from the front of the tunnel would cast it. Open the background and knockout images at the same time. Cut and copy the knockout image's layer, then Paste it into the background image where it becomes a new layer. Free Transform (Cmd/Ctrl-T) the knockout to fit the size, angle, and perspective you want it to have. Ideally, you should have to reduce its size or keep it the same size, rather than enlarge it. Create the shadow: duplicate the layer and then drag it below the original. This is going to be the Shadow layer. Lock the layer's transparency by clicking the Transparency Lock icon near the top of the palette. Choose EditFillBlack. You now have a black silhouette of your knockout. Make a gradient selection for the entire image. This is going to allow you to make a shadow that becomes more diffuse as it recedes from the light source. To do this, go to the Channels palette and click the New Channel Icon. A black channel named Alpha 1 will appear. Rename it Truck Shadow. Now choose the Gradient fill tool, a rectangular fill, and a black to white gradient. Fill the channel with the black to white gradient, spreading the gradient portion as far as possible. Return to the Shadow Layer and turn off all the other layers so that you can clearly see what you're doing. Choose Select Load Selection. In the Load Selection dialog, choose Truck Shadow (or whatever youve called the layer and channel you're turning into a shadow). Drag the resultant marquee so that its lower line (which happens to be the center of the gradient) is slightly below the bottom of the shadow silhouette. Use the Gaussian Blur filter to soften the edges of the shadow, while the gradient mask makes the shadow more diffuse as it is projected further from the light source. For that reason, you want to diffuse the whole thing to an extreme so that it can graduate to that extreme. You may have to reapply the Gaussian blur filter several times to blur it enough for a realistic effect. In Figure 9-28, you see not only the marquee line mentioned in the step above, but the Gaussian Blur dialog and the preview of the result. Figure 9-28. The Gaussian Blur filter has been used in conjunction with a feathered selection to make the shadow diffuse over distance. Match color balance, noise, and sharpen the downsized image to match the sharpness of its surroundings. Choose a Color Balance adjustment layer. When its dialog opens, don't do anything, just click OK. Then, while the layer is still selected, right-click and select Create Clipping Mask. Match the noise and sharpening. Double-click the Zoom icon so you can see detail clearly. Now match the noise. Choose FilterNoiseAdd Noise. The Add Noise dialog, shown in Figure 9-29 appears. Drag the Amount slider to the left. You will probably need a very small number, so when you get the grain structure that comes closest to matching the Background image, type in the exact percentage you think is most likely to work. When youve got it, click OK. Figure 9-29. The Add Noise dialog. Add the sharpening to the same layer orif you're concerned that you might have to sharpen to a different degree latercopy the noise matching layer first and put it on top of the stack. Then choose FilterSharpenSmart Sharpen. The Smart Sharpen dialog will open, as shown in Figure 9-30. Choose the settings you see in the figure. You need to be careful not to oversharpen because you will have to do final sharpening for the printer you are using and that could cause dramatic oversharpening. Figure 9-30. The Smart Sharpen dialog box.
|
