Infrared cinematography opens up a whole new spectrum of light not visible to the unaided eye. This has the potential to give otherwise ordinary scenes a surreal and dream-like appearance. In this article, we explore several of the unique applications and technical hurdles.
The colors of visible light are created by electromagnetic radiation over a range of frequencies. In order of decreasing wavelength, these sensations are typically described as red, orange, yellow, green, blue, indigo and violet:
However, visible light is only a small subset of what reflects off everyday objects. Just beyond violet lies ultra-violet (UV), which is perceivable by some birds and insects, and is typically associated with sunburn-causing light. At the other end of the spectrum, infrared (IR) lies just beyond red, and is ordinarily associated with night vision and thermal imaging.
Digital cameras are capable of recording the full spectrum from 300 to 1200 nm, but doing so would not produce realistic images. Standard digital sensors therefore utilize a special filter to block most light outside the visible range:
Note: Above plot is intended to be qualitative and representative of the average digital camera.
However, many hobbyists still create images with the remaining IR and UV light by using lens filters that block visible light. This has the advantage of not needing to modify the camera and void a warranty, but also requires prohibitively long exposures, so the approach is limited to still photographs and time lapse sequences. True infrared motion capture requires specially-modifying the camera sensor to unlock its full spectral potential.
IR FILTERS & APPEARANCE
Without infrared filtration, cameras record images that are effectively superimposed versions from both visible and near infrared (NIR) light. Light transmitted to the sensor therefore increases substantially, and since infrared gets recorded by all three color channels, color saturation typically decreases accordingly:
For a pronounced infrared look, many typically use an "IR pass" or ND filter to reduce the amount of visible light contributing to the image. Common filter options include the Hoya R72, B+W 092 and Wratten 89b, among others. These generally improve contrast similar to using a polarizing filter. However without white balancing, images appear reddish straight from the camera:
Although there is not necessarily a "correct" white balance as there often is with visible light, a common convention is to base this off the foliage, in part because trees and leaves are typically the lightest objects in an infrared image. Doing so improves contrast dramatically, and turns landscape scenes into surreal snowscapes with tinted skies.
Even after white balancing, pre-visualizing infrared imagery is far more difficult than with standard cinematography. Just as one cannot create general rules for which materials will appear blue, there are unfortunately no hard and fast rules about which objects will be most reflective to infrared light. Cinematographers have to develop an intuition for infrared as they would with any other color.
In general, foliage, clouds and skin reflect more NIR and appear lighter, whereas clear blue skies and most tree trunks appear darker. Smooth water surfaces also become darker, but this is primarily because they are reflecting a darker sky. The overall effect helps manage contrast in otherwise harsh midday light, similar to how using a red filter enhances contrast with monochrome imagery. With studio work, the skin on models will appear smoother and ghostly white, but veins might also appear darker and more pronounced if they're close to the surface. With clothing, red pigments will often become brighter than white pigments were with visible light.
CONTROLLING FALSE COLORATION
Since infrared light is beyond our visual capabilities, any corresponding coloration is technically false color. However, this does not necessarily mean that the color isn't physically meaningful. Red color filters are slightly more transparent to shorter wavelength NIR (700 to 800 nm) than are green and blue color filters.
Note: Above plots are intended to be qualitative and representative of the average digital camera. Actual color sensitivities may vary substantially, but the general trends illustrated above typically still hold true. See the tutorial on Bayer sensors for a background on color filters with digital sensors.
This is why images may have a magenta color cast straight from the camera, prior to any white balancing. After white balancing, images are therefore effectively bichromatic, and any color represents an object's relative reflectivity to short and long wavelength NIR. The image can then be post-processed using a variety of hues, depending on the desired effect:
However, the degree and type of coloration depends on where the IR filter's cut-off occurs. Standard IR pass filters typically exclude all light below about 700 to 720 nm, for example, whereas aggressive IR filters may exclude all light below 800 nm or higher. Since infrared at 800 to 1200 nm typically penetrates red, green and blue visible light filters equally, aggressive IR filters produce imagery that is effectively monochromatic. Common aggressive IR filters include the Hoya RM90, Wratten 87 or 87c, and the B+W 093.
TIPS FOR OPTIMAL RESULTS
Infrared cinematography also sets new standards for equipment and technique. Key considerations include:
Infrared cinematography opens up many new creative possibilities, and has the potential to immerse viewers in a visual experience their eyes were never designed to see. One may even find themselves looking at everyday objects anew, to try and assess how those will be rendered outside the visible spectrum. However, as with any new technique, infrared also comes with added technical considerations. Special attention needs to be paid to new filter types, the spectrum emitted by lighting, lens performance and post production coloration. Once all this knowledge has been integrated though, the potential applications are powerful and diverse. These include improved night and low-light sensitivity, surreal or dreamlike music videos, astronomical time-lapse sequences and haze-penetrating aerial footage, among other possibilities.