Digital infrared photography is far more convenient and practical than film-based infrared shooting, but it remains technically challenging. By understanding exposure and white balance, hot-spots, backfocus, and sharpness, you’ll avoid frustrating trial and error and quickly begin to enjoy excellent results. I’ve been shooting digital infrared for half a decade now, and those years of experience have taught me a great deal. In this article, I share my key discoveries that make shooting digital infrared images a productive and satisfying effort. A follow-up article will delve into post-processing techniques.
Infrared is the spectral band beyond deep red, starting at around 700 nanometers (visible light runs from 400-700nm) A previous PT article (May/June 2005) covered converting digital cameras for infrared; the conversion removes the infrared-blocking glass over the digital sensor and then substitutes one that passes infrared (and blocks visible light), either over the sensor itself, or by using a filter over the lens. A few cameras (such as the Nikon D2H and Leica M8) incorporate anemic infrared-blocking sensor-cover glass, and can be used in their unmodified state using filtration over the lens; the B+W 092 filter is recommended for such use. This article discusses shooting infrared images once you have an infrared-capable camera (and may help persuade you to convert a camera, if you haven’t already).
Setting expectations for infrared
1. Plan on being thrilled with some images and disappointed with others. And remember that compelling form and structure masked by harsh visible light and/or the distraction of color can be revealed in infrared. Experiment!
2. Plan on dedicating a camera for infrared use (camera modification). Using screw-on filters on an unmodified camera does work, but with inferior results at much slower shutter speeds. You will be discouraged from freely exploring infrared in varied shooting situations, due to the tedium of compos- ing and exposing as compared with free- form handheld shooting.
3. Plan on your share of “soft” (unsharp) images—accurate focus requires more effort than with visible light. Backfocus is a frequent problem. Infrared images can be as sharp as color images, but accurate focus is critical—autofocus is error-prone with most lenses.
4. Plan on paying more attention to correct exposure—cameras are not designed to meter infrared. Experience and patience are required to learn optimal exposures.
5. Plan on shooting in Raw if your camera offers it. Infrared images can often benefit from substantial adjustments (especially white balance); these are best converted in 16-bit mode from a Raw file.
6. Plan on using an image editor (e.g., Photoshop) to optimize your infrared images. Almost every infrared image can benefit from some adjustment.
7. Plan on lens purchase(s)—one or more of your favorite lenses might not perform optimally in infrared, might have unacceptable hot spots, and so on. You might have to use specific lenses to achieve top-quality results in infrared.
The usual candidates for infrared include blue sky and/or water set against foliage of some kind for a dramatic effect (Figure 2). Less commonly seen are portraits, artificial materials, and animals (Figure 1), all of which offer great potential for infrared. The high reflectivity of plant matter results from its cellular structure, so while leaves are highly reflective of infrared, so are such things as dead leaves and grass or wood. Anodized metal alloys can be very interesting in infrared (camera lenses themselves make good subjects). Covered-over graffiti can often be revealed by infrared; this makes infrared useful for detecting changes and repairs to documents, rugs, and so on. Infrared penetrates skin (Figure 3), erasing wrinkles and blemishes, but revealing veins in youthful skin and even showing whether hair color is “original equipment.”
Exposure and white balance
Accurate exposure makes a huge difference to preserving detail and minimizing sensor noise, especially in the blue channel (which is often dark) and the red channel (which runs a constant risk of being blown out). Yet obtaining an optimal exposure can be tricky, because the camera meter continues to meter visible light. Even if infrared filters are used on the lens (which makes it rather difficult to compose), the built-in metering doesn’t understand infrared exposure, especially with the variable response of the red/green/blue sensor dyes. Exposure compensation does work if the lighting is consistent, but can also produce wildly inconsistent results with visibly minor changes in lighting; your best route to reliable results is manual exposure coupled with liberal use of the in-camera histogram.
Using the camera’s histogram meaningfully requires a custom white balance because most digital cameras take white balance into consideration for the image histogram. When white balance is set to Daylight (or Cloudy or Shade), the image shown on the LCD will be extremely red, and the histogram and/or blinking highlights will mislead as to correct exposure (daylight white balance for infrared skews the red channel strongly to the right). Always set a custom white balance using grass or other foliage. However, the exact subject matter is not critical; choose a subject that yields a mostly monochromatic result (light blue or yellow is typical) with visible differentiation for plants against sky or water. Experiment until you find a white balance that produces results that are visually interesting and (ideally) some evidence of “false color” emerges (typically blue/yellow). A custom white balance is important when shooting Raw and absolutely critical when shooting JPEG files—it prevents any individual color channel (usually red) from being blown out (Figures 4 and 5).
Camera response to infrared varies based on the type of sensor and the filtration used. In general, a histogram with a daylight white balance shows a red channel shifted well to the right, while green and blue channels are shifted more to the left (some cameras feature a blue channel shifted strongly to the left—e.g., underexposed). Figures 4 and 5 show RGB histograms using a custom white balance and daylight white balance; Figure 5 also shows an in-camera histogram from the Nikon D70, which offers luminance (brightness) only, a less-than-ideal situation with infrared photography, but not usually a problem so long as a custom white balance has been set.
I strongly recommend bracketing, because even Raw files can deliver surprises—sometimes an “overexposed” image can deliver superior results, especially if the green and/or blue channels are extracted by themselves for a preferred tonal rendition. Bracketing as much as two stops over and under the “correct” exposure can be a very useful practice and can also be used for HDR (high dynamic range) images.
Raw versus JPEG
Using the Raw format should always be preferred over JPEG; white balance adjustments alone require a lot of “stretching” of the image data, at least with cameras using filters in the 700nm range (deeper into infrared, the red/green/blue sensor dyes all go transparent and respond about equally). In addition, the green and blue channels are often quite underexposed compared to the red channel (even using a neutralizing white balance) and so if they are to be used, a Raw file with 12 or 14 bits of precision adds considerable leeway over an 8-bit “pre-baked” JPEG file. A Raw infrared image exposed badly can often be salvaged from one of the color channels; a badly exposed JPEG image goes into the trash.
Another reason to shoot Raw is that image contrast can be extremely low. When image contrast is extremely low, “stretching” it into a wider tonal range fares very badly if the data are only 8 bits. With a Raw file, at least 10 bits are available (4× the precision), and some cameras offer 12-bit or even 14-bit Raw files, for 16× or 64× the precision of 8 bits. The Raw file should be processed into a 16-bit TIF file where manipulation in 16-bit mode produces higher-quality results.
A hot spot is a bright central area that becomes brighter and more defined as the lens is stopped down, typically associated with a significant color variance from center to corners (bluish at the center and yellowish in the corners). There is no single cause for a hot spot, but lens coatings exert a very strong influence. One or more lens elements also can cause reflections off the digital sensor. (For direct comparisons on how important lens coatings are to infrared performance, please see the “Zeiss ZF Prototype Lenses for Infrared” article at diglloyd.com.)
Hot spots are rare with visible-light images because modern lens coatings are specifically optimized for visible-light wave- lengths (400–650nm). Unfortunately, this optimization sacrifices performance in the ultraviolet and infrared range, particularly off-axis, where infrared tends to be attenuated. (Certain specialty lenses such as the Coastal Optics 60mm ƒ/4 APO macro use elaborate lens coatings that perform well in ultraviolet, visible light, and infrared; see diglloyd.com for details.)
A faint hot spot can be dealt with by careful processing and is frequently masked by the subject matter. Moderate and strong hot spots are more difficult to deal with. They can destroy image contrast in the central portion of the frame and can even “blow out” detail so that it cannot be recovered. Even if a hot spot can be dealt with in post-processing, this is tedious and time-consuming work that is best avoided.
Although some lenses have moderate or strong hot-spot issues, they can be used successfully so long as they are not stopped down beyond ƒ/5.6–ƒ/8 and careful attention is paid to both subject matter and shading the lens from extraneous light. Although a hot spot degrades image contrast, a mild hot spot can actually be used to advantage to help lighten a dark central area—but such “bonus shots” aren’t all that common, and the hot spot invariably reduces image contrast and thus perceived sharpness.
Test a lens for a hot spot by stopping down to ƒ/16 and shooting a uniform surface such as the sky, or place a sheet of translucent material such as vellum over the front of the lens. Focus the lens at infinity. See Figure 6 for an example of how hot spots vary by focal length and aperture. Most lenses do not exhibit such severe problems, but many do. In general, lenses with many elements are prone to hot spots, but surprisingly many “prime” lenses, such as 50mm lenses, also have strong hot spots. Hot spots can appear to come and go because of the combination of subject matter, aperture, and ambient light. A strong light source in the frame can produce an extreme problem with some lenses, so be sure to use a lens shade or shade the lens from any extraneous light, especially direct sunlight.
Flare is invariably worse in infrared than in visible-light photography. While hot spots are a distinct phenomenon from lens flare (Figure 7), the same conditions that cause lens flare can exacerbate hot spots.
Because hot spots worsen upon stopping down (often beginning at ƒ/8), a problem lens may be perfectly usable at ƒ/5.6 or ƒ/8. All modern SLR lenses are multi-coated, and multi-coating is formulated (perhaps as a side-effect) to atten- uate infrared (see the SchneiderOptics.com filter-transmission graphs for B+W filters for single-coated versus multi-coated filters). Zeiss recently announced prototypes of their “ZF” line with single coatings; the author tested them against their multi-coated peers, and they perform extremely well in infrared (see “Zeiss ZF Prototype Lenses for Infrared” at diglloyd.com for details).
Lenses exist that show no signs of a hot spot, even under adverse lighting at ƒ/16; these are the lenses one naturally gravitates to after having once been burned by a hot spot in a favorite image. Examples of such lenses include the Nikon 85mm ƒ/2.8 PC-Micro-Nikkor and the Nikon 28–70 ƒ/2.8 AF-S (Nikon lenses can be used on Canon EOS with an adapter). The Canon EF 135mm ƒ/2L is also superb in infrared (but beware of backfocus). After testing more than 60 Nikon and Canon lenses, I concluded that Canon lenses are more susceptible to hot spots than Nikon lenses. Each lens manufacturer has its own “secret sauce” for lens coatings, which is applied consistently across the lens line, and Canon’s lens coatings seem less “friendly” to infrared than Nikon’s in most cases.
Backfocus means that the actual image is focused behind the desired plane of focus (i.e., is farther away); the desired plane of focus is therefore blurred. With most lenses, you need to compensate for infrared by focusing at a closer distance than in visible light (unless the camera has been adjusted; see Figure 8 for an example of backfocus). Unless a lens is extremely well corrected for color aberrations (true apochromatic or superachromat), infrared light focuses differently than visible light—and even apochromatic lenses might not fully correct for infrared focus. Stopping down does mitigate backfocus error, but sometimes even ƒ/11 cannot fully compensate, and the entire zone of sharp focus remains shifted from its intended location. Uncorrected chromatic aberrations also extend into the infrared band (700–1100nm for digital cameras), and the individual red/green/blue color channels can display focus shifts relative to each other, making “correct focus” ambiguous.
A point-and-shoot infrared camera focuses with the same sensor that makes the image, so focus is accurate. By comparison, a digital SLR uses a separate optical path for autofocus and/or viewfinder (“eyeball”) focus, and even if adjusted for infrared focus using one lens, other lenses and even other focal lengths of the same lens (zoom) might not be accurate. The various companies that modify cam- eras for infrared can adjust the modified camera to com- pensate for the infrared focus shift, but this adjustment might or might not work well for many lenses or even dif- ferent focal lengths of the same zoom lens. For example, the author’s Nikon 200–400mm zoom focuses perfectly at 400mm, but is strongly backfocused at 200mm. On the other hand, Nikon’s 18–200 VR focuses perfectly at all focal lengths on the author’s Nikon D70-IR. However, high technology is coming to the rescue with the advent of “live view” digital SLRs, which allow focusing from the sensor itself.
Contrast and sharpness
A common misconception suggests that infrared is inherently less sharp than visible-light photography. While this can be true for some lenses, many lenses offer performance in infrared that is every bit as good as that of visible-light photography (and sometimes even better when a narrow spectral range is used). Most of the time it is inaccurate focus (backfocus) that yields unsharp images, not any inherent optical shortcoming.
Another common misconception is that infrared is inherently high contrast, but many infrared images show extremely low contrast. High-contrast images typically result from daylight (sunlight) shooting and low-contrast images typically result under cloudy or shaded conditions. The subject matter itself can produce extremely low-contrast results, such as when foliage all renders as a uniform monochrome tone.
False color and monochrome
The term “false color” refers to the strange-looking color that results when digital images are processed by the unwitting camera or Raw-file converter that was designed for a visible- light image. Typical results are blue/yellow or blue/red, but color channels can be inverted or swapped (see Figure 8 ) or stretched for other effects. (Part II of this article will discuss processing images for false color.)
False-color occurs because infrared light penetrates the color dyes of a digital camera’s sensor to varying degrees, depending on wavelength. Deep red and short-wave infrared (~700nm) affects the red photosites the most and the blue photosites the least; this brightness disparity results in the false-color effect when interpreted by a Raw-file converter or camera. By 800–900nm, the digital sensor color dyes become transparent to infrared and so the resulting images become completely monochrome—all color channels see more or less the same brightness—e.g., the same shade of gray. For pleas- ing false-color effects, use filtration in the 715nm range; filters in the 800+ range largely eliminate such effects. A camera converted using a 715nm filter can still work using additional filtration for “deep” infrared. Both Canon and Nikon DSLRs offer pleasing false-color results using filtration in the 700nm range; Fuji cameras tend to offer less differentiation.
The quality level and satisfaction you gain from shooting infrared digitally will depend on how well the technical challenges are understood and managed, particularly exposure and focusing. Don’t settle for “pretty good” when you can obtain outstanding results.