We are given certain numbers to grade a light’s “ color quality”, but what is the grading criteria? What are these numbers measuring? Since quality is such a subjective opinion, measurements should instead be made quantitatively. It is important to understand what these numbers are actually measuring in an attempt to create a standard for “quality”.
A collection of Metamerisms. Whether the color space is given the wrong white point from which to calculate (wrong CCT), or the fixture is outputting a narrower band of the visible spectrum, the perceived color of objects is extremely affected. Subtler differences can still present problems in post-production, or limit the creative palette. Conversely, using a metamarism to your advantage can be a creative advantage in achieving a desired look.
The most commonly used measurement is or . CCT maps out the locations of the of “full spectrum” light sources. This is to say, to a human vision system, the point that has no visible shift towards any hue. Because the human vision system is psychologically based, our brains are constantly reinterpreting color space to focus around this point. Therefore, it is dynamic. In an effort to translate this dynamic point into non-dynamic color systems, Deane B. Judd transformed the of onto the as a curve.
A Plankian radiation temperature tool that helps visualize the SPD of a blackbody radiator at any temperature. Adjust the temperature with the slider on the right. Use the magnifying glasses to change the scale and get a better view at how the distribution would look on a relative value based SPD.
Spectral Power Distribution – commonly referred to as “spectrum” demonstrates the amount of light being output from a source (y-axis) at different wavelengths (x-axis). Above, we see the Spectral Power Distribution of a 3280Kº tungsten source that appears as white light because it outputs all wavelengths of visible light. Because there are more longer wavelengths at greater intensity than shorter wavelengths, the light appears as a more red, or Orange White. When set to 3200K, a camera will balance the spectrum evenly and treat this orange-white as the true white point – something our eyes do naturally. If set to 5600K, the camera will see the deficiencies in shorter wavelengths and the light will appear very orange.
This concept is still used today. This curve, known as the , maps the SPD of an illuminant onto the 1931 CIE color space. When the fall along the blackbody locus, or “Planckian curve,” it is assigned a “correlated color temperature,” a value that correlates the assumed spectral power distribution (SPD) to be the same as that of a blackbody radiator at the same temperature. If the source does not register its x, y coordinates on the blackbody locus in the chromaticity diagram, it is said to have either a green (registers above the curve) or magenta (registers below the curve) shift. This shift is represented within the x, y system as a of coordinates and is referred to as .
These measurements are all incredibly useful and have been in use for decades. Film stocks were built around them and gels were manufactured according to how to shift these numbers to a desired look. Because incandescent light sources ARE blackbody radiators and were the most widely used lights in film, this system was simple and adjustments were straightforward. Even HMIs could work easily enough within this system because of how full their spectrum was. Although they had possibilities of , most of the time, their SPDs were close enough to the average SPD of “daylight” at mid-morning and mid-afternoon that they registered the same on daylight film stocks, which were designed around the daylight SPD. However, as new “discontinuous” light sources became more widely used, the effect they could have on how a film stock, and eventually camera sensor, could interpret and render colors in a predictable manner was devastating. Illuminants such as sodium vapor, mercury vapor, commercial fluorescents, and eventually LEDs or Solid-State lighting, did not follow the Blackbody locus, or even the later developed Daylight locus. They created white light by manipulating color systems into the interpretation of being white – usually by mixing 2 or 3 more discrete wavelength collections in an equal enough manner to mathematically disguise themselves as a Blackbody source, falling somewhere on one of the now two loci. The downside to this is that how they actually reflected colors off of colored objects did not correlate to a blackbody radiator. Because the SPD was missing crucial wavelengths, so was that object’s reflected light SPD, and it now appeared as a different collection of wavelengths that sometimes landed in a very different location on the x,y chart. It was clear that more ways of judging light sources, besides CCT, G/M shift, x,y coordinates and ⊿uv were needed. With the variety of sources now in use, it has become important to give a number to a light’s overall ability to render reflected colors accurately within a given color system, also known as a light’s fidelity. But which system to use?
Since most of this previous work had all been centered around scientists studying the human vision system, this became the basis for the first system in measuring color rendering – CRI (Color Rendering Index). With the advent of the digital age, we have launched into a new era of lighting and color spaces. Now, CRI has proven to be essentially useless for reasons discussed later. In its place, there are several other indexes that seek to provide a measurement to color fidelity, and even the reflected color gamut of a source. These indexes offer measurements of fidelity in the context of either:
The Spectral Response of the Human Eye – Each curve represents each cone’s responsiveness to different wavelengths along the spectrum. The red cones respond mostly to longer wavelengths, the green cones to medium wavelengths, and the blue cones to shorter wavelengths. Note: Violet shows up on the short end of the visible spectrum because our red cones have slight sensitivity to shorter wavelengths in addition to their high sensitivity to longer wavelengths; yet, magenta represents no wavelengths of visible light.
The physical, biological and psychological process of creating the experience of Color.
Two Well-known color spaces: The Munsell System, Used by HSL/HSI and the RGB system
The 1931 CIE x, y Chromaticity diagram with plotted Blackbody Locus.
While a camera system’s interpretation is the most useful, because all cameras interpret color differently, a different measurement for every system is needed. Currently we only have a measurement for broadcast cameras, therefore, we must also consider the other contexts when determining how useful a source is at accomplishing a given look — whether that is naturalism, or something more targeted.
There is no way to guarantee that a specific camera will respond to a given fixture the way the cinematographer desires unless camera tests with that fixture are performed, so it is crucial that all values and measurements are considered within the context of usage. If a source scores poorly in several measurements, but very well on SSI of a practical source that has been decided will play heavily into the mood of a scene, that fixture would be a much better choice to use to mimic that setting than another fixture which has higher measurements elsewhere, but not in SSI. If the production design involves a good deal of glossy, saturated set decoration, TM-30 Rg should be paid attention to more. Shooting a documentary on 3CCD broadcast cameras? TLCI is most likely the most important measurement for you. Lighting an underwater dream sequence with theatrical blues and greens could mean choosing fixtures with less deficiencies in the cyan wavelengths — even if their white light fidelity isn’t high. Every different scenario offers different solutions and what is important is that measurements are not used as a grading system to judge how “good” a light is overall, but instead, they are used as indications of which tool is the right one for the job at hand.