As white as a...colour calibration target
The Macbeth ColorChecker(R) is often observed in digitised images adjacent to the subject being imaged. It is a colour calibration target used widely by photographers to achieve consistent colour within a studio environment. Good colour management allows the photographer to have continuity to achieve the same result with any camera. The rectangular cardboard target consists of a grid of 24 squares of colour samples, each with a measurable spectral reflectance. Reflectance refers to the fraction of incident light reflected at an interface. The spectral reflectance of these patches does not change under different lighting conditions in the visible spectrum (this is not the case in the ultra-violet and infra-red – see footnote* below), so are reliable to track colour changes in this range.
Figure 1: Calibration target shown over f.86v of Cotton Nero A.x. during imaging of Sir Gawain & the Green Knight. The target is set to be on the same focal plane as the folio. Targets are often cropped out of final processed images.
The idea of a colour chart came about in a 1976 paper in the Journal of Applied Photographic Engineering by C. S McCamy, H. Marcus and J.G. Davidson entitled A Color-Rendition Chart. The abstract states “A color chart has been developed to facilitate quantitative or visual evaluations of color reproduction processes employed in photography, television, and printing.” Their paper has been cited over 350 times to date. The original chart consisted of a 4 x 6 array of patches, each 5 cm square.
Figure 2: The original colour chart consisted of square patches of side 5 cm. The same size chart is still available and used today.
There are still 24 patches on modern colour calibration targets but smaller versions are now available with patches measuring 1 cm wide. The X-Rite ColorChecker(R) Classic target used in our lab is shown below with a scale, focusing target, and reference number that we attached.
Figure 3: The ColorChecker(R) Classic target has 24 colours in a 6 x 4 grid. The colours are painted in matte on smooth paper and surrounded by a black cardboard border.
The colours are roughly divided into four kinds. The top row is composed of colours which approximate natural objects such as human skin (dark and light), blue sky, the green colour of a leaf, and a blue chicory flower. The second row is made of miscellaneous colours encompassing a good range of test colours. The third row is comprised of the primary (blue, green, red) and secondary (yellow, magenta, cyan) colours, and the fourth row represents a uniform gray lightness scale ranging from brilliant white to black.
Figure 4: Colours in the calibration target. These colours are precisely measured and can be described in terms of the Munsell color system (a colour space describing colours in terms of their hue, lightness and chroma).
Larger colour calibration targets do exist such as the ColorChecker(R) Digital SG which boasts a gamut of 140 colours.
Figure 5: ColourChecker(R) Digital SG boasts the widest colour gamut available. Its design is based on the original ColorChecker(R) target but is enhanced for digital photography. Image copyright X-Rite, from X-Rite website.
For consistent colour, photographers can take a shot of the calibration target with the camera set to capture raw files. Shooting raw is the only way the camera chip can capture all of the information available in the scene. The image is opened in image processing software such as Photoshop, and a script is run on the image which opens it multiple times with different settings. Results are measured and a status is generated with values which can be used to fine-tune the camera’s colour calibration and get processed colour to match the original scene (or alternatively to distort the colour for special effects!).
While colour calibration targets are on the whole produced in the same way using the same materials, on average, every colour target is ever-so-slightly different. The colour difference may be very small and only measureable using other scientific methods. Colour difference is a metric of interest in colour science - the standard metric being Delta E (ΔE). This definition allows colour difference to be quantified in a way which is more reliable than just using adjectives, a practise which is detrimental to anyone whose work is colour critical!
Our multispectral imaging system captures images in Lab colour space, where L is lightness and a and b are colour-opponent dimensions. Lab colour space approximates human vision and is device independent. It includes all perceivable colours with RGB and CMYK spaces (see our previous post What the CMYK? Colour spaces and printing) sitting within its larger gamut, so file sizes are generally much larger. Values for L, a, and b can be tracked once the image has been white balanced using the white colour patch on our calibration target as a reference. However, Lab files don’t open in all software packages so quite often it is necessary to transform images into other spaces such as RGB, but the original Lab file is always stored.
Colour Science is a fascinating and growing area of research. For fun you can try out this Color IQ test from the X-Rite website to learn more about how you see colour, and to find out where you can get your own targets.
Christina Duffy (@DuffyChristina)
*While the Macbeth ColorChecker(R) provides 24 colours with consistent spectral reflectance under typical lighting conditions in the visible spectrum, it does not behave similarly in the ultra violet or infrared parts of the Electromagnetic Spectrum. Another material such as Spectralon is required for imaging outside of the visible range. The property which defines a diffusely reflecting surface (i.e. an ideal “matte”) is called Lambertian reflectance and Spectralon exhibits highly Lambertian behaviour with a spectral reflectance of >99% from 400-1500nm and >95% from 250-2500 nm. Spectralon is a fluoropolymer - others include PVF, PVDF and PTFE (Teflon). Spectralon has the highest diffuse reflectance of any known material over IR (infra-red), VIS (visible) and NIR (near-infrared) regions of the spectrum, and is therefore very expensive, but necessary to track colour difference during multispectral imaging.
McCamy, C.S.; Marcus, H.; Davidson, J.G A, 1976, A Color-Rendition Chart, Journal of Applied Photographic Engineering Volume 2, Number 3, pp 95-99
X-Rite website, or follow X-Rite on Twittter