CRI testing of film lighting fixtures and Daylight CCT Measurements
Today's study looks at the measurement of CRI accuracy in lighting fixtures, how they differ, how to be diligent when choosing fixtures, and the differences in Daylight CCT measurements.
The CIE (International Commission on Illumination) standardizes CRI (Color Rendering Index) testing among manufacturers of illuminants. These standardized tests look for a given fixture or source of illumination's color reproduction accuracy across two different metrics:
1 - Spectral Power Distribution (SPD)
Spectral Power Distribution tests the illuminant's energy output across the visible wavelength range. The spectral power distribution thus gives us the light's correlated color temperature (CCT). This CCT measurement then sets a reference illuminant for the white point of the fixture (where the light sits at a neutral white point). Below 5,000 K the reference is a blackbody radiator, and at 5,000 K and above it is a D-range illuminant (Daylight Illuminants). A 5,500 K daylight source would be compared against a D-range of D55.
2 - Spectral Reflectivity
Spectral Reflectance is the fraction of light that a surface will reflect at each wavelength. In CRI testing, this is measured for each test color sample with a spectrophotometer, and the CIE publishes these values as a standard. The wavelengths measured with SPD are direct from the source, and can be measured with a spectrometer, or in CRI tests a very fancy machine. The light reflected off each sample is the illuminant's SPD multiplied by the surface's reflectivity, and then weighed by the human eye's color matching functions (the CIE standard observer). The measured wavelengths' stimuli are calculated as XYZ values, then subsequently converted into a CIE color space (CIE 1964) where the color shift is measured. In the case of a digital imaging sensor (a camera), the stimuli are instead weighed by the camera's red, green, and blue sensitivity curves, and then encoded into a video standard (BT.1886, Rec. 709, Log).
From these two metrics we can begin to measure the CRI value of a fixture with the following:
The SPD
The CCT
The Spectral Reflectivity
The Munsell Book of Color
The Munsell Book of Color samples became the CRI testing standard in 1965. This method scores the color shift of 8 light, unsaturated colors:
a light grayish red
a dark grayish yellow
a strong yellow-green
a moderate yellowish green
a light bluish green
a light blue
a light violet
a light reddish purple
The mean of these colors' shifts is calculated to a CRI value. Generally a fixture with CRI values between 85 and 100 is considered to provide good color rendition. In 1974, six more CRI test colors (R9–R14) were added to the standard. These six additional colors are generally more saturated, but they are scored separately and are not part of the main CRI average. There was large debate if this was the most accurate way to test for CRI values in the wake of HMI and LED fixtures. Thus, in 2012, the EBU introduced the TLCI (Television Lighting Consistency Index), which tests fixtures against the RGB Bayer sensors of digital cinema cameras rather than the human eye.
strong red
strong yellow
strong green
strong blue
a light yellowish pink
a moderate olive green
Although, with the addition of six additional colors to CRI testing, there is a much broader range of CIE color samples that could be tested by manufacturers of film and television lighting that are not required within the CRI standards. This means that there is likely a difference between manufacturers who are testing their fixtures with different ranges of CIE colors. Manufacturers can test for a full range of color shift based on a chosen CIE spectrum.
A 95 CRI on an Aputure light might not be the same as a 96 CRI of a Neewer light or as accurate as a Nanlux. There's an attempt at some more consistency, but it's very important to check how a manufacturer has tested for CRI on their fixture before investing or bringing it onto a production where color rendition is of the utmost importance.
Daylight sources
Compared to a blackbody radiator at about 5,800 K, the Sun matches up on a wavelength curve with the blackbody radiator at about 5,777 K. While we think of daylight sources or a daylight-balanced camera at 5,600 K, it's likely closer to 5,800 K, which would explain why I generally like to shoot my daylight sources 200 K warmer at a 5,800 K white balance in camera. One of the things that might make it appear like 5,600 K is just atmospheric scattering (Rayleigh and aerosol scattering, and ozone, water and oxygen absorption), and obviously light reflected off of Earth's surface could change this.
Direct sunlight is generally warmer on average in more northerly latitudes, because the sun sits lower in the sky and passes through more atmosphere. If you were shooting closer to the equator, you would be looking at 5,800 K for a daylight balance white balance or a source. In northern latitudes, let's say in Alaska, this might be anywhere from 5,600 to 5,400 K.
TLDR:
CRI scores how accurately a light renders color compared to a reference source, based on 8 standard color samples. Manufacturers test and report CRI differently, so a 95 from one brand may not match a 95 from another. Check how a fixture was tested before you trust the number. Daylight is also warmer than the 5,600K standard suggests: the sun is closer to 5,800K, and direct sun gets warmer at higher latitudes, around 5,400–5,600K in places like Alaska