CAMERA



Autor: Dennis Jackstien


Lighting technology: Basics and characteristics (3 / 10) - Colour rendering

In this article we deal with the extensive topic of colour reproduction.

The complex of "colour rendering / colour quality" has gained enormously in importance, especially due to the spread of LED lights in film and TV. Cheap, simple LED products often do not achieve good colour rendering, which then leads to falsified colour impressions in the picture. However, fluorescent tubes and other lamp technologies are also affected.

What does colour rendering mean?

Colour rendering describes how naturally and unadulterated a light source can render colours. The decisive factor is the spectral composition of the light source.

The following example explains the subject very clearly:

We have 2 luminaires with a colour temperature of exactly 3,200 Kelvin and identical colour coordinates. On the left a typical halogen spotlight, on the right a theatre spotlight with red, green and blue LEDs (RGB). Visually, the light of both spotlights looks exactly the same, but it is spectrally completely different.


Figure 1 - Halogen & RGB LED Spectrum [Image: Dennis Jackstien]


If the same motif is now illuminated with both spotlights, completely different colour impressions result.


Figure 2 - Color falsifications of RGB LEDs (right) [Image: Dennis Jackstien]

Put simply: A light source that emits hardly any yellow-orange spectral components cannot correctly reproduce yellow-orange colours (such as peppers in the motif). What is not emitted cannot be reflected back (correctly).
It is therefore important that all the luminaires we use have a well-filled, uniform spectrum.

How can colour rendering be measured and evaluated?

The best known method is the Color Rendering Index (CRI). In this case, 8 test colors are illuminated with a reference light source (e.g. halogen or daylight D65) and then with the test headlight. The color differences between the two tests are then calculated for all 8 test colors.


Figure 3 - The 8 test colors of the CRI

If there are no color differences at all, the CRI maximum value of 100 is reached. The stronger the color differences, the smaller the CRI value will be and it can in principle fall below 0. Of course, all these calculations are now carried out using software solutions that only require the light spectrum of the test light source as input. For high-quality applications, such as film and TV lighting, a CRI of 90 is usually set as the minimum requirement. High-quality LED spotlights generally achieve this value today.

Problems of the CRI method

The CRI procedure was developed in the 1960s and is particularly unsuitable for today's film and TV light ratings.
1) Only 8 largely unsaturated test colors are too few to evaluate LEDs sufficiently. From time to time, additional test colours, such as R9 = saturated red, are also indicated. However, the standard CRI only knows these 8 test colors.
2) The colorimetry on which the CRI calculations are based is based on the spectral perception of the human eye. Camera sensors usually perceive colors differently, which is why headlamps with a high CRI can still display color shifts in the camera image.

Alternative color rendering calculations

TM-30-15

The TM-30-15 is a new variant of the CRI. It uses 99 test colors instead of only 8 and calculates the color range (Rg) as well as the color rendering (Rf). For the latter, light sources that make colours look particularly saturated are evaluated positively. For the film and TV sector, however, Rf is the much more relevant value. The TM-30-15 offers extensive evaluations, e.g. a precise graphic representation of color shifts.


Figure 4 - Color Shift Information for TM-30-15


TLCI-2012

The Television Lighting Consistency Index (TLCI) is the only method that uses a camera sensor as the calculation basis. The evaluation is not based on the human eye, but on the sensor specification of the EBU-HD standard camera (ITU Rec.709). The method was developed by the BBC. There are 24 test colours and exact details of individual colour deviations, e.g. for post-processing by a colourist. The TLCI is recommended for applications with TV cameras (3 chip). However, it does not work for film cameras and 1-chip sensors in general.

Spectral Similarity Index (SSI)

The SSI developed by AMPAS goes a completely different way. There are no test colours and it is based neither on the human eye nor on a camera sensor. The SSI only compares the spectral values of a test light source with those of the reference. Since camera sensors (as well as humans) perceive colors very differently and therefore CRI, TM-30 or TLCI can at best offer an approximation, this direct evaluation of the SSI is a consistent approach. However, no peculiarities of the spectral receiver (eye or sensor) are taken into account, so that in case of doubt the results are inferior to other methods that were selected to match the receiver (eye, sensor 3-chip camera, etc.). 

Conclusion

All methods should only be consulted for a first orientation. Only a real test with the respective camera is really meaningful and resilient.

Related Articles: