Identifying and Controlling Color Reproduction Variables
Photography and lighting are both processes with many variables which affect outcome. The challenge for a beginner is learning how the variables relate to each other and affect the final result. The starting point for process control is to first identify all the variables, and then identify which have the most impact on the desired final result. A basic tenet of process control is: You can't control what you can't measure. so another important part of process control is determining the best way to monitor the results. In this tutorial I'll provide an overview of the approach I use for monitoring and controlling color for those new to photography.
Color Variables in Digital Photography
Two variables affect the way color is recored in a digital camera:
- Color Temperature of the Light
- In photography the color of temperature of light is expressed in degrees Kelvin. The color temperature scale is based on the concept of how the appearance of an object changes from a dull red glow to bluish-white as it is heated. Tungsten and Halogen lights typically have color temperatures in the range of 2800-3200 K. Sunlight varies greatly depending on time of day and atmospheric conditions, but on a clear day at noon the color temperature is in the range of 5500 - 6000K. But under overcast conditions the color the color temperature will increase to around 6,500 and will be even higher in open shade which is illuminated by sunlight reflected off of the blue sky. The human visual system adapts to changes in ambient color temp perceptively based on known reference values. Our eyes scan a scene, find the lightest and darkest tones and those assumed to be neutral and then the brain adjusts perception to make them seem neutral. It sounds a bit like a riddle, but the reason we perceive a white shirt or sheet of paper as white under any light is because we know (or assume) the object is white.
- White Balance Setting of the Camera
- Color photography is based on the fact red, green and blue lights of equal intensity, when overlapped together will create white and varying proportions of the three colors will produce all the other colors we perceive. A digital camera sensor is has millions of individual sites which absorb photons and convert them it voltages which are then converted in to binary numerical values. A color image is created by covering the sensor sites with red, green, and blue filters, which combined will create the entire range of colors we see on the camera LCD playback and our computer screens. Unlike our eyes the white balance setting of a camera can't adapt based on expectations. The red, green and blue sensor sites simply record however much of each narrow band of wavelengths the light sources generate. Some sources like the noon day sun and flash produce a relatively even balance of red, green and blue energy. Artificial sources come in a wide variety of continuous and discontinuous color spectrums.
A digital camera has white balance pre-sets for a variety of lighting conditions. On the Canon cameras I used they are:
The camera also has an "Auto White Balance" (AWB) setting which attempts to mimic the way our eyes adapt to different ambient color temperatures. It works by analyzing the photo after it is taken, finding the brightest area below clipping, assuming it is white, adjusting the RGB values in the file recorded by the sensor to make them equal there, then adjusting the entire photo by the same amount. But because it bases the adjustment on the assumption the brightest area in the photo is neutral it does not perform well when the brightest area isn't white or when there is more than one color temperature of light illuminating the scene.
- 3200K - Tungsten
- 4000K - White Fluorescent
- 5200K - Daylight
- 6000K - Cloudy,Twi-light, Sunset
- 6000K - Flash
- 7000K - Open Shade
Custom White Balance
The Achilles Heel of AWB is the variability of the area measured. What is needed for process control of color is a known neutral benchmark the camera can rely on to set the RGB values equally. There are a wide variety of gray cards available for use as a color control benchmark. I use a Kodak 18% Gray Card from the Publication R-29 Kit. Photographing the card, then using that image to set Custom WB via the camera menu ensures that the RAW images I capture will be displayed in the editor with neutral color balance. Setting the Custom WB doesn't affect the capture of the RAW file, it simply appends the information about the color balance settings to the file header so the editor knows how to display it. After setting Custom WB I shoot the card again, which allows me to check the card with the eyedropper tool to verify the RGB values on the card are in fact neutral.
Setting color balance to technically neutral with Custom WB off a gray card provides a consistent starting benchmark for evaluation of color. Including a shot of the the card in test shots after setting Custom WB allows me to use what the camera captured to benchmark what is seen on the monitor.
Mixed Lighting - A Color Management Dilemma
Mixed lighting where two different color sources are lighting a scene are a dilemma when setting white balance. Averaging two different sources such as tungsten and flash results in nothing looking normal perceptually in the resulting photo. Perception is based on expectations and we expect faces in photos to look neutral, not halfway between blue and orange. The worst scenario is when the two sources hit a subject from different directions, such as tungsten lights in the ceiling making the foreheads of subjects orange while the WB set for flash renders the front of the faces normally.
A similar mixed source can occur outdoors when shooting under a canopy of trees and adding flash. The foliage acts like a filter which gives the sunlight a green cast. The greenish cast is difficult to spot by eye because the eye adapts to it. Even in the final photos an untrained eye will misinterpret the color cast as being flat lighting because green + pinkish skin results in flat, grayish skin tones. In ambient light alone the problem of the green bias is easily solved by placing a gray card near the face and doing custom WB from it. The Custom WB will shift all the color towards magenta to make it neutral. The problem occurs when flash is either added to the uncorrected greenish light, or the + magenta corrected ambient.
The Solution? Avoid Mixed Lighting
The best solution to the mixed lighting dilemma is to avoid it whenever possible. When using flash indoors in a tungsten environment the flash can be gelled with a 1/2 CTO orange filter with camera WB set via Custom WB to the gelled flash. When the indoor ambient lights are fluorescent gelling the flash is more problematical because fluorescent lights come in a wide range of color temperatures. I deal with fluorescent ambient with a bifurcated strategy of using only the ambient light for the wide establishing shots of the room setting, and overpowering the ambient with flash for close-up shots of the people. When using flash the effect of the ambient light on the exposure can be minimize by shortening the shutter time. Because the flash duration is so short it isn't affected by shutter speed and shooting at 1/250th at f/8 will allow the flash to dominate, minimizing but not completely eliminating the color bias from the room lighting.
There is no good work-around solution for a mixed lighting situation outdoors such as under the trees so my advice is simply not to shoot under trees with flash. Instead utilize the natural lighting and use white reflectors for fill as needed, setting Custom WB off a gray card held near the face.
If mixed lighting situations can't be avoided set Custom WB off a gray card held near the face and as always take another shot after setting Custom WB as a reference when editing.
Color Balance Decision Tree When Shooting
If shooting RAW, which I recommend, process management is more focused on keeping the color balance consistent shot-to-shot under a given set of lighting conditions vs getting it perfect in the camera as when shooting JPGs. In RAW even if the color isn't perfect out of the camera all the files can be batched corrected by adjusting one, then copying the settings to the others in the batch. I also strongly recommend the use of a gray card at least 8" x 10", large enough to cover the center circle of the viewfinder, the zone used for setting Custom WB on a Canon camera. It is also helpful to have smaller neutral gray targets which can be placed in the scene as "click to correct" editing references.
When encountering a new shooting situation the first step is to have situational awareness of the ambient light direction, intensity, and color temperature. I make color management decisions outdoors relative to a baseline of daylight, which is ideally balanced in RGB. When in doubt about the color temperature of the ambient light I set the camera WB to daylight, fire off a test shot, then look at the playback. Indoors under tungsten or fluorescent lighting I will set the WB to the corresponding pre-set to evaluate it. If the color looks OK using the pre-set that will be the extent of my color management when shooting, but I will include a gray card target in shots when possible as a guide for evaluating and adjusting color when editing.
If using a camera which displays the histograms of each color separately shooting a gray card with the camera set to Daylight outdoors and the closest pre-set indoors will inform you whether the ambient light is neutral (all channels line up on the histogram)or color biases relative to the daylight baseline (channel histograms out of alignment). It isn't critical to know the color temperature of the ambient, only a general impression of whether it looks warmer / cooler, or green/magenta vs. what you perceive by eye. Remember your eyes will adapt while the camera, set to Daylight WB, will give you a more objective view relative to that baseline. If the color histograms in the camera playback show a significant deviation from neutral the same shot can be used to set Custom WB. After setting Custom WB shoot the card again and the channel histograms from the card should be lined up.
Using flash almost always involves creating a mixed lighting situation to some degree. Non-diffused flash is typically a bit warmer (yellow) in color than ambient in open shade and slightly cooler (blue) than direct sunlight. Used indoor flash is significantly different than either tungsten and most fluorescents. If the flash is the source lighting the foreground it will become the dominant source perceptually in close-up photos with people in the foreground so the best strategy for making the color seem as "normal" as possible is getting the color on the faces neutral by setting WB based on the flash color temp. However in wider shots where people are not the main center of interest turning the flash off and shooting ambient only, or bouncing the flash up off the ceiling with WB set to the ambient light will produce more natural looking results than a different color, more brightly illuminated foreground. When a seamless meshing of flash and ambient is desired set the camera WB to match the ambient then add a color gel over the flash to match the room lighting.
Editing - Where Color Management Gets Derailed
Blind Faith in Monitor Calibration Tools
Having a calibrated monitor is important, but one shouldn't place blind faith in the calibration tools. Let's say we buy a $200 calibration device, use it, but use it incorrectly resulting in not setting the calibration correctly. Perceptions are based on expectations, and expecting our freshly "calibrated" monitor and your baseline, we open a file shot outdoors at 2PM under clear skys with Daylight WB but looks like it is off color. Trusting what we see on the monitor we edit the file so it looks normal. Then we print it and the results don't come close to matching the monitor or how we remembered the original scene. The problem in that case isn't the camera or the printer it is trusting our eyes based on the assumption the calibration process worked correctly. Our eyes are weakest link because they adapt, even when looking at an image on a monitor.
The simple expedient of including a gray card target in a test shot can prevent that scenario. The way a camera is engineered to work a file containing a image of a neutral gray card and a familiar face captured with Custom WB set of the same card should have equal RGB values on the card if measured with the eye dropper tool in the editor. That file, verified with the equally RGB eyedropper readings then become our "Standard Baseline" for color. If the RGB values on the card are equal it should also look neutral on the monitor with the face next to the card looking "normal" as seen by eye. If for some reason the image on the monitor doesn't look normal it tells us, based on the camera baseline, that the monitor isn't properly calibrated.
If the test shot with the face and gray card, shot with Custom WB off the card, is sent to the printer, letting the printer manage the color, the resulting print may not match the saturation of the colors of the original scene or the monitor, but it should provide a realistic perceptual rendering of the original scene because that is the way the color management aware printing process in the middle works.
Since our camera is what is generating the color which is being managed in the other stages of the workflow getting it neutral with Custom WB off a known neutral gray card will make all the subsequent steps of the reproduction process, editing and printing, simpler and more consistent shot-to-shot and session to session. Setting Custom WB is not a substitute for calibrating the monitor, but it provided an objective "by the numbers" process control cross-check to validate the monitor calibration. If we shoot a file with Custom WB set with a gray card and it doesn't look neutral when opened then the monitor calibration should be suspect.
More detailed tutorials on photographic color management using a camera / gray card baseline can be found in the Camera and Post Processing Technique sections of the Table of Contents