Unlike color images, which can satisfy as JPEGs right out of the camera, digital infrared images almost always benefit from post-capture adjustments that bring out their unusual qualities. After all, infrared is “beyond visible,” so there is no right or wrong approach, unlike color photography.
To produce compelling images, this post-processing should be done with an open mind for artistic possibilities.
Infrared by its nature has no real color or brightness; unlike regular color photography, the human eye offers no guidance as to how infrared should be rendered. On the photo- graphic continuum, visible-light color photography maps most closely to reality, then visible-light black-and-white photography, then infrared and ultra- violet photography, which capture invisible light and have no perceptual reference points in the human visual system. If that premise is too disconcerting, then the best choice might be to stick with conventional visible-light photography.
Part I of this series covered the capturing of digital images in the field— correct exposure, white balance, lens selection, and various other issues. This follow-on article covers what to do with those image files once created. By reading this article, you’ll get a good idea of the variety of applicable techniques (for detailed information, see the diglloyd Guide to Digital Infrared Photography at diglloyd.com).
Workflow with digital infrared images involves the following:
• assessing images for their artistic potential as well as technical quality (sharpness, exposure, etc);
• understanding the technical considerations of color space and gamut, as well as 16-bit versus 8-bit processing;
• processing Raw files for optimal results;
• working with the color channels of an RGB image;
• working with Lab-mode images (the L and/or the a/b color channels);
•converting to grayscale (many approaches);
•advanced techniques: swapping color channels and expanding color, selective color.
The remainder of this article ad- dresses each of these in turn. Regrettably, space limitations do not allow detailed explanations of all of them.
When assessing infrared images, it is best to set aside conventional notions of what constitutes a quality original. An infrared image might at first appear to lack snap, with an overall haze and low contrast—look past this initial impression at the latent image structure. This isn’t always the case, but can be in some circumstances (see Figure 2).
For efficient workflow, be sure to memorize the keyboard shortcuts for automatic adjustments such as Auto Levels in Photoshop, as well as the shortcuts for viewing the individual image channels (Figure 3). The rest of this discussion assumes the use of Adobe Photoshop; other image editors have similar features.
To quickly assess an image, make some instant adjustments to see how an image emerges from the original (later you can use layers to make non-destructive edits to the image file). The steps listed below are techniques for quickly investigating the possibilities.
1. Open the image in your image editor, such as Adobe Photoshop. For Raw files, this means conversion to TIF or JPEG, more on that below (see Processing Raw Files). For the initial evaluation, JPEG is fine, so you might want to set your camera to record Raw+JPEG.
- Immediately use Auto Levels. This neutralizes the color, increases the contrast, and in general makes the image look much more interesting.
- To assess the grayscale rendition, view the R, G, and B color channels individually; this yields a good idea of grayscale potential (as opposed to color rendition). Keyboard shortcuts (cmd-1/2/3) make this very fast.
- For another gander at grayscale potential, convert the image to Lab mode (see Figure 5) and view the L channel; it often has a different tonal distribution than the RGB channels. Try Auto Levels on the a/b color channels of the Lab mode image. The effect will usually be too extreme, but might suggest interesting false- color possibilities.
- Run scripts to generate a large number of variants (such as those included with the diglloyd Guide to Digital Infrared Photography). See Figure 7 for examples.
After the initial assessment, you should have a good idea of whether the image is worth further effort. Often, Auto Levels alone produces a non-optimal but satisfactory result for casual use; further effort can be expended as
desired, including the addition of Photoshop layers for tonal modification, hue/saturation, grayscale conversion, and so on. Do not save the image over the original; if you like what you see, create a master file with adjustment layers, leaving the original image bits alone.
Tip: Create a 16-bit PSD or TIF file for your master file. A recommended practice is to keep the Raw (or JPEG) original in the same directory as this master file, using a naming scheme that correlates them. A file-name suffix on both can also be very helpful for future reference. For example, you might have the two files shown in Figure 4. I like to retain the original Raw file name (_MG_5329) as part of the name so as to maintain a sequence with other images.
Dealing with extremes
Infrared images can sometimes result in extremes of color, depending on the camera, filtration, white balance, and so on. The color can often be outside the range (or gamut) that can be recorded in the most common color space, sRGB. Color that won’t fit is pinned (forced to the same value), resulting in loss of detail. For this reason, avoid the sRGB color space. Instead, use a wide-gamut color space such as Pro Photo RGB, Wide Gamut RGB, or the universal reference Lab.
In Photoshop, verify the color space being used by choosing Edit > Convert to Profile (click Cancel, don’t actually convert). See Figure 5.
The choice of color space won’t matter for many images, but it will matter for some images, so the best habit is to choose a color space that can hold nearly every image. Even if the original image is in gamut, post-processing could send values out of gamut. At the minimum, Adobe RGB 1998 should be used, but Pro Photo RGB or another wide-gamut color space is better (and Lab mode is perfect).
When a wide-gamut color space is used, the range of color can be very large, which means that the steps between values can be quite large if 8- bit mode is used. Therefore, 16-bit mode is strongly advised; 8-bit mode provides only 256 distinct values, which can cause the same sorts of pinning (mushing together) issues as a small- gamut color space does for colors. As with the choice of color space, 16-bit mode won’t matter much for many images, but it will matter for some images; the goal is to not have to worry about such exceptions. In Photoshop, verify 16- versus 8-bit in the Image > Mode menu (see Figure 5).
Processing Raw files into TIF
For all post-processing, a 16-bit TIF file should be the starting point. Set up your Raw converter to produce 16- bit TIF files in a wide-gamut color space such as ProPhoto RGB or Wide Gamut RGB. You can re-save the TIF file as a Photoshop (PSD) file if you prefer. If you aren’t shooting Raw, convert original JPEG files to 16-bit TIF files, retaining the unmodified JPEG original.
Other than white balance, special conversion settings are generally not needed for processing Raw files (see Figure 6).
There are different philosophies with Raw-file conversion. One philosophy is to make as many adjustments as possible in the Raw converter. Another is to defer most adjustments to Photoshop (or similar program). The author prefers the latter approach, because it’s easier to learn Photoshop skills for any and all files produced by an and all cameras than to learn a variety of Raw-file converters and their peculiarities. Also worth noting is that upgrades to Raw converters can change the image-conversion algorithms, which means that you might not be able to produce the same result from the same Raw file after upgrading to a new version of your Raw converter. For this reason, it’s best to avoid Smart Objects and instead stick to a TIF file, making a new conversion only when you choose to do so.
One complication is that not all Raw-file converters are happy about the odd- ball color balance of infrared images; Adobe CameraRaw (for example) does not honor the as-shot white balance from many cameras; this can be really irksome when you’ve selected an in- camera custom white balance that cannot be replicated using CameraRaw. This is one reason I tend to use Canon’s Digital Photo Professional for Canon Raw files, and Nikon Capture NX for Nikon Raw files.
The recommendations for Raw-file processing are simple:
- process using a white balance that produces the effect most similar to what you’d like to start with. This might mean the custom as-shot white balance set in the camera, or it might mean using a white balance tool in the Raw-file converter. Sometimes white-balancing off one portion of the subject matter versus another can produce a duality of color that is more preferable one way or another.
- for optimal results, always convert to 16-bit TIF in a wide-gamut color space.
- apply a minimal level of sharpening during Raw-file conversion; more sharpening can be done later when the image is targeted for printing or Web display.
- be willing to try more than one Raw-file converter; don’t assume your favorite one will produce the best results.
RGB and Lab images
Configure your Raw-file converter to output 16-bit TIF files; they will contain red, green, and blue color channels. Even if your intended result is grayscale (black-and-white), these color channels are very useful. It is not uncommon for
one of the color channels to offer a more interesting tonal scale or contrast than another; this is very helpful in deciding how to proceed. In addition, one color channel might have considerably more digital noise than another so it’s good to have all channels at your disposal. Please refer to the Converting to Gray- scale section.
Unlike RGB images, whose channels co-mingle color and brightness values, Lab mode separates the brightness (luminance) from the color; this is the L channel of the image (see PT May/June 2006 for a detailed article about Lab). The a (magenta/green) and b (blue/ yellow) channels hold the color values. This separation is extremely useful because it allows manipulating color information independently of the brightness (the luminance or L channel). Possibilities include inverting color channels—e.g., flipping color from blue to yellow and/or magenta to green; this is done trivially in Lab mode using the Image > Adjustments > Invert command on the desired channel. The same technique cannot be used in RGB mode because there are three color channels, each containing color and brightness information. However, channel swap- ping can be done in RGB mode (exchanging one channel for another in up to six combinations). See Figure 7 for examples of numerous variants of channel inversion and swapping and grayscale.
Once you’ve opened the 16-bit RGB TIF file in your image editor, you can convert it to Lab mode using Image > Mode > Menu (see Figure 5). Depending on what you want to accomplish with the image, it might be more con- venient to use RGB or Lab mode.
Converting to grayscale
An entire book could be written on converting grayscale images to color. The following is only a partial list of the possibilities; see Figure 8 for screen dialogs of these approaches:
- converting the red or green or blue or L channel into grayscale while discarding the other two channels; do so, view the desired channel, then choose Image > Mode > Gray- scale;
- desaturating using Hue/Saturation;
- using Photoshop’s Channel Mixer to blend the red, green, and blue channels into grayscale, or using a black-and-white adjustment layer;
•using third-party software with alternative conversion choices.
I frequently keep it simple for gray- scale conversion, using the Image > Mode > Grayscale command on the channel that offers the most appealing initial rendition. Which technique is most appropriate depends on your personal preferences and experience, as well as on the image.
In some cases, the L channel offers a low-noise and high-key grayscale rendition that can be very appealing; in other cases one of the red, green, or blue channels is more interesting. Figure 8 shows examples of the L channel used for grayscale conversion.
Advanced techniques include expanding color to accentuate the range of color values: In Lab mode, use a Levels (or Curves) layer to modify the a and b color channels for increased color definition. See Figure 11 for examples of resulting images.
Masking techniques allow color and/or grayscale rendition in certain areas of the image. Figure 12 shows an image in which all color was masked away except for the yellow sunflowers, resulting in a nicely subtle effect. Use a Photoshop layer mask on a Hue/ Saturation layer for such effects.
Unlike visible-light color photography, where accurate rendition of color is frequently a requirement, infrared images are open to the imagination—they can be rendered in monochrome or false- color, and the results are not rejected as wrong because it is usually clear that the image is not a conventional image. Exploring the wide variety of options can help one develop a style that avoids overused effects that announce “infrared image” too obviously.