12.5 Separation of Subject

As has been noted numerous times in this text, the ability to separate the subject from its background before applying digital enhancements to an image is probably the most powerful darkroom technique available to today’s wildlife photographers.  In Photoshop, this can be accomplished very easily in many cases via the use of the Quick Selection tool.  In section 10.6 we outlined a number of techniques for precisely selecting the bird using the powerful tools provided by Photoshop.  For some bird photos, selecting the subject can still be quite difficult, however, particularly if there are branches or leaves partially occluding the bird; legs and beaks and tails often complicate matters a bit as well, since they’re usually thinner and of a different color than the rest of the bird.  Nevertheless, even in difficult situations it’s often worthwhile to put in the effort to separate the bird from its background, so that the subject can be processed separately from the rest of the scene.

Fig. 12.5.1: Separating the subject from the background is one
of the first tasks to be performed in my own workflow.  This allows
me to adjust the brightness, sharpness, and saturation of the
bird differently from the background.  This can be useful not
only for correcting visual defects in the subject, but also for
making the bird stand out more (or perhaps less).

     As we described in section 10.6, simply selecting the bird via one of Photoshop’s standard selection tools may not be sufficient for the types of subtle processing that you’d like to apply.  In particular, it’s often a good idea to feather the selection via the Select > Modify > Feather menu option, so that the selection has a soft edge.  When the selection is given a soft edge in this way, any processing that is applied to the selection will gradually fall off in intensity around the borders of the selection; this can help to reduce unnatural boundary effects, especially for filters involving brightness or blurring.  Prior to feathering you’ll often want to contract the selection via the Select > Modify > Contract menu option.  The figure below illustrates the related task of expanding the selection.

Fig. 12.5.2: Depending on what kind of processing you’re about
to apply to the bird, you may want to expand or contract the selection
by a few pixels, and to apply some feathering, to supress any edge
artifacts that may occur as a result of further processing.

    Once you’ve got the bird selected, the next thing you should do is save the selection to a named channel, via the Select > Save Selection menu option.  Saved selections are saved in the PSD file when you save your work, so they’ll be available during subsequent post-processing sessions on the same image.  You’ll typically want to first save a non-feathered version of the selection, since you can always re-load the selection from memory and feather it as needed, whereas un-feathering a saved, feathered selection is rather more difficult.  You can, of course, save multiple versions of a selection (such as a feathered and a non-feather version, or versions feathered at different radii) by simply creating different channel names when saving the selection.

Fig. 12.5.3: You should always save any nontrivial selection
into a new channel, so you can recall it later.  Once you’ve
saved the selection, you can then either apply processing
directly to the selected area, or first copy the selected area
to a new layer. 

     After saving the selection of the bird, you can then either apply processing filters directly to the selection, or duplicate the selection into its own, separate layer (by pressing Cmd-J / Ctrl-J).  There are many advantages to creating a separate layer for the bird; just about the only disadvantage is the increased memory space required to do so, though on most computer systems this won’t be a practical limitation (though your file sizes will also increase, and this may be more of a limitation for you if your hard drive is almost full).
    Once you’ve separated the bird and its background into different layers, you’re then free to process those layers differently by simply clicking on a given layer before invoking any image processing filters or painting tools.  Once you’ve got a part of the image separated into its own layer, you can then select parts of that layer (with the standard selection tools) and copy those parts to layers of their own, if desired.  For example, once you’ve got the bird on its own layer, you can then separate out the face, the chest, the wings, and the tail all onto their own, separate layers, and then process those layers differently.  One reason for doing so is to be able to correct local lighting deficiencies that affect specific parts of the bird.  For example, when using flash to illuminate warblers, I often find that the highly reflective flight feathers often result in feather glare that I then need to fix in postprocess.  I do this by selecting just the part of the bird exhibiting the glare and then applying the Highlights tool to reduce the glare.  Conversely, some parts of the bird typically end up less well lit than others, and for these parts I like to apply the Shadows or Levels tool to dispel the shadows and create the impression of more even lighting.  Though these operations don’t always require that parts of the bird be selected before applying the adjustment, doing so often provides for greater control and reduces unintended side effects on other parts of the image.

Fig. 12.5.4: Once you’ve separated the bird and the background
into different layers, you can apply different processing to them,
allowing you to adjust how strongly the bird stands out.  The use
of layers requires more memory, but affords enormous flexibility.

     Keep in mind that while we talk about separation of the subject, what we really mean in many cases is separation of the foreground from the background, where the foreground will typically include not only the bird but also its perching substrate.  The perching substrate will, of course, typically be in focus just as well as the bird, so attempts to de-emphasize that substrate (by processing it the same as the background) may result in an unnatural effect that is apparent to the viewer.  However, because the perching substrate will almost always have a different texture and color than the bird itself, it’s typically worthwhile to separate the substrate from the bird, so that you’ll generatelly want to parse the image into three parts: the bird, the in-focus portion of the perching substrate, and the out-of-focus background. 

Fig. 12.5.5: An example of differential processing of foreground
and background.  Left: the original image.  Right: after rendering
clouds into the background.  The sky was selected via Color Range
and then duplicated twice; one copy was brightened and the other
darkened.  Render > Clouds was then applied to a layer mask on
the top layer, allowing the brighter lower layer to show through
as clouds.  The whole process took about two minutes.

    Note that for many purposes (especially when preparing images to be posted on the internet in low-resolution), you won’t necessarily need to separate the subject from the background with exacting precision; we already alluded to this in some sense when discussing the issue of feathering.  Since birds are three dimensional and the imaging plane is two-dimensional, the borders of the bird will often look acceptable if they’re less distinct than the main part of the bird (especially when the depth-of-field is razor thin).  For this reason, it’s often possible to get away with a selection boundary that only roughly follows the precise outline of the actual bird.  For some applications, you’ll want that selection boundary to fall completely within the bird, while for others you’ll want the boundary to completely contain the bird. 
    As noted earlier, it’s often desirable to create multiple versions of the bird’s selection, to use for different purposes.  These different versions can often be obtained from a single initial selection, via some combination of Contract, Expand, and Feather operations applied to the original selection.  You can also use the Select > Inverse operation to obtain a selection of everything except the bird.  Note also that feathering can be performed by going into Quick Mask mode (by pressing the Q key) and using a soft-edged brush with reduced opacity to paint a gradient around the boundary of the selected region; this technique can be more difficult to use effectively, however. 
    If you’re going to be creating a new layer from the selection, it’s sometimes best to err on the side of including too much of the subject in the new layer
i.e., using a hard boundary rather than feathering it.  Once you’ve got the subject on its own layer and have applied some adjustments to it (e.g., color, brightness, sharpness, etc.), you can then use a layer mask (see chapter 13) or the Eraser tool to soften the edges of the layer (using a soft edge and reduced opacity in both cases).  Blending the modified subject back into the scene in this way requires more work, but has the advantage that you get instant visual feedback and can adjust your handling of the tool accordingly.  By contrast, the use of Select > Feather would normally be applied prior to any qualitative adjustments to the subject, so you won’t know until later whether you’ve chosen an appropriate feathering radius.  An added advantage of using a layer mask to do the blending is that it can always be adjusted again later if you apply additional processing to either the subject or the background (obviously, either could affect the blending of the bird into the scene).