Zbornik - SDR

“Colour rendering index of LED light sources” R-4
5 Updating the CIE test method
During the past decades several attempts were made to update the test method for colour
rendering determination.
Investigations were carried out to test the usefulness of the CIE Test Samples 12,13,,14,15 and the
chromatic adaptation formulas 16,17 , testing the method for practical sources as well 18 . CIE tackled the
question several times, technical committees were established, and after five to ten years closed
down, as they could not find a solution that every party would have agreed upon. The last such
committee, CIE TC 1-33, was established in 1991 and closed down in 1999. It was unable to
recommend a new colour rendering index formula, but published its closing remarks 19 , and in this
publication formulated some ideas that could provide guidance for future research.
Visual experiments were performed already before the new proposal (see e.g. 20 ) and further
experiments were started 21,22 partly to get an alternative description of colour rendering 23 . These
experiments have shown that there are certainly better ways to describe the colour rendering
properties of light sources, especially of white LED sources, where the white colour is produced by
mixing the light of some red-, green- and blue LEDs. The CIE Division 1, responsible for light,
colour and vision decided not to change the current colour rendering index calculation method, but to
develop a new descriptor, such as “colour appearance rendering” or “colour quality index”. A
Technical Committee was established in 2006 to investigate this question.
6 Supplementary methods to describe colour quality of light sources
The wish to supplement colour rendering with further quality descriptors is not new. Judd coined
the term flattery index already in 1967 24 . The flattery index was intended to describe whether a light
source renders colours in a more pleasant (flattery) way then an other source. Jerome discussed the
differences between flattery and rendition in detail 25 . Later the word preference was used instead of
flattery 26 . Thornton’s calculation showed that colour rendering and colour preference indices do not
have their optimum value at the same spectral distribution 28 . Some experiments tried to combine the
colour preference and colour rendition aspect in such a way that the maximum of colour rendition
remained if the test source had the same SPD as the reference illuminant, but the worsening of the
index was slower if the colour difference between the sample illuminated by the test source
compared to the illumination by the reference illuminant deviated in the direction of higher chroma,
or e.g. in case of complexion towards redder hues 27 . Other ideas went into the direction to develop a
colour discrimination index, as there are a number of tasks where the discrimination between small
colour differences is important 28,29 . All these can be supported by simulation experiments 30 . Also
Davis and Ohno published on improved colour quality metrics 31 .
The comfort experience in an interior setting is also influenced by the colour quality of the
lighting. Bellchambers investigated visual clarity 32 and found correlation between visual clarity,
illumination and colour rendering. Other investigations tried to correlate the different aspects of
lighting quality as well (see e.g. 33 ).
An interesting new approach is based on the hue shift of many colours that shows which hues are
highly distorted compared to a reference and which are rendered correctly 34,35 .
7 Colour quality simulation
Our recent studies go in a similar direction by starting from the supposition that if a designer has
carefully chosen the colours of an environment to be pleasant under one light source, i.e. the observer
“Lighting engineering 2006” stran 31