H 51997 No. 105 Mar/Apr 2017 www.pld-m.

com
No. 105 Mar/Apr 2017
CHINESE | GERMAN | ENGLISH | TURKISH

Magazine for
professional lighting design

MAIN TOPIC
Facade lighting

LIGHTING DESIGN
Klubhaus St.Pauli
in Hamburg/DE

Museum in Basel/CH

C3A in Córdoba/ES

CityLights in Paris/FR

Motel One in London/UK

Palazzo di Vigonovo
in Vigonovo/IT

Carnegie Hall
in New York City/US

study
Facade lighting on
residential buildings

research
Colour preference
under white LEDs

Magazine for the PLD Alliance, official magazine of the Professional Lighting Design Convention
62

Can we standardise the human eye?

An assessment of colour preference under white LEDs by individuals with different iris colours.
Text: Dr. Karolina M. Zielinska-Dabkowska, PhD, Veronika Labancová, Dr. Amardeep M. Dugar, PhD

Does the iris colour of individuals have an impact on their preference of colour under white light emitting diodes (LEDs)?
To address this question, a study was conducted to examine how people with different iris colours respond to the colour and
brightness of three scene settings under illumination from LEDs.

1. Introduction The eye colour behaves in the colour vision has undergone years of from LEDs. This study explores the
Various psychophysical factors same way as skin colour – the darker intense study and modelling, very role of iris colour in the preference
affect human colour discrimination the skin or eyes, the more light is little evidence has been found on of colour under light emitted from
such as the crystalline lens, the blocked (8). Therefore, people with the role of iris colour in the prefe- LEDs by attempting to answer these
retina, the visual pathway, or the dark eyes can better handle high rence of colour under light emitted questions.
visual cortex. Researchers (1–4) glare situations, such as strong day-
argue that iris colour might also have light at noon but can barely see in
an influence on human colour the dark even after adaptation due
discrimination. The iris itself comprises to a relatively high concentration of
of two tissue layers: the innermost melanin, which absorbs both short
layer comprises pigmented cells and long wavelengths of light (9).
known as the iris pigment epithelium; On the other hand, people with blue
the outermost layer, also referred to as eyes might have a higher risk of
the anterior iridial stroma, comprises macular degeneration and other
mainly loosely arranged connective diseases affected by light due to
tissue, fibroblasts and melanocytes more light rays passing through the
(5). The main factors that differentiate lens (10–12). However, this variation
iris colour are the density and in iris colours occurs only in white
cellular composition of the stroma Caucasians or their descendants,
(6). The melanocyte cells and not in any other human race
aggregated in the anterior border with brown eyes (4,7,13).
layer of the stroma store melanin in a But does the iris colour influence
special organelle within their human colour preference under light
cytoplasm, which is referred to as from white LEDs? Do people with
the melanosome; white light is brown eyes prefer colours under
absorbed and reflected as a light from white LEDs differently
spectrum of wavelengths giving rise than people with blue or green-hazel
to the three common iris colours eyes? And do the basic properties of
namely blue, green-hazel, and LEDs such as the Colour Rendering
brown: these are the result of low, Index (CRI) and luminous flux have
moderate and high melanin levels, any relationship to iris colour and Figure 1: The distance from the observer to the objects and the location of the
and the numbers of melanin colour preference? If so, what is that light sources was based on the viewing angle of the observer and the average
particles respectively (7). relationship? While human stereo height of the observer.
 RESEARCH Colour preference 63
under white LEDs

Figure 2: Experimental set-up in Wismar. Figure 3: Experimental set-up in London.

2. Methods others. As an additional precaution, were used as the basis for the three
Participants were presented with an black cylinders were placed on the scenes as depicted in Figure 2.
experimental set-up and expected to luminaires to prevent light spill and In London, the experiment was
complete a questionnaire. Similar to avoid participants looking directly conducted inside a rented room
experiments were conducted in two into the luminaires. The viewing measuring 8.5 metres (width) x 8.5
different locations: Wismar/Germany angle was adjusted, based on the metres (length) x 4 metres (height).
and London/UK. height of an average person (1.65 to The luminaires were suspended
The entire experiment in each 1.70 metres). All lighting scenes from a ceiling-mounted lighting
location lasted for approximately ten were static and switched on at the track at a height of three metres.
hours. The experiment itself consisted same time. All luminaires performed The ceiling and walls were painted
of three scenes generated by three at 100 per cent output. The illumi- white and the floor was covered
downlights with commercially nance levels of the realized scenes with beige laminate wood panels.
available LED modules, which varied were not considered as part of the The room set-up was very similar to
in CRI (80/95) and luminous flux experiment, as the focus was more Wismar, the only difference being
(1000lm/2000lm). A Correlated on light source characteristics. that objects for all three lighting
Colour Temperature (CCT) of 3000K In Wismar, the experiment was scenes were placed on one long
was maintained for all three scenes. conducted in the Lighting Labor- table covered with a black tablecloth
atory at Hochschule Wismar: instead of on separate stands, as
2.1. Methods – Physical Set-up University of Applied Sciences, depicted in Figure 3. Figure 4: A line at a distance of 1.5
It was important to define an easily Technology, Business and Design. As a general precaution, parti- metres from the displayed objects was
repeatable physical set-up of three The laboratory in which the cipants were permitted to view the marked on the floor. Participants were
scenes as shown in Figure 1. experiment was carried out consists displayed objects only from a speci- asked to position themselves behind
The visual composition of each of a room measuring 9.2 metres fied distance. To ensure this, a line at this line during the experiment to
scene was to be kept as compact as (width) x 7.2 metres (length) x 9 a distance of 1.5 metres was drawn ensure exact viewing conditions for all
possible for easy comparability while metres (height). The ceiling, walls on the floor and participants were participants.
viewing so as to help participants to and floor are made of untreated requested to view the three scenes
evaluate scenes under the same concrete. The movable steel grid from behind this line, as depicted in
conditions. The three displays and ceiling with integrated luminaires Figure 4. The distance of this line
their corresponding three luminaires was lowered to a height of three from the scenes was based on the
were positioned at a distance of metres. Three display stands, each viewing angle of the observer and
1.20 metres from one another. measuring 40 x 40 centimetres and the average height of the observers.
It was assumed that the lighting of at a typical ergonomic height for The LED manufacturer Xicato donated
one scene would not impact the stands and counters of 1.20 metres, the three requested luminaire
64

Figure 5: Colour measurement summary of Xicato LED modules: XSM 8030 – 2000 C, XSM 9530 – samples with specific LED parameters,
1000 B, XSM 8030 – 1000 B with 350 nm to 750 nm, measurements taken using a GL Optic mini- as depicted in Figures 5 and 6, and
spectrometer, Type MSM1/ DE. This information, although originally provided by Xicato, was checked Table 1. The involvement of Xicato
and validated using the GL Spectrosoft software program. as a partner company was based on
their reputation as a worldwide
leader in producing highquality LED
modules with excellent colour
rendering properties. The selected
light sources vary in two basic
parameters: colour rendering index
[CRI] and luminous flux [lm].
However, each of the downlights
had the same beam angle, thereby
providing a similar beam spread for
the three scenes. Three identical
compositions of different cubes
were designed to enable participant
evaluation of colour and brightness
preference. The CRI method
comprising a mixture of basic and
additional colour samples, as
depicted in Figure 7, was used
considering its wide acceptance as
an approach for evaluating colour
rendition of white light sources (14).
Nine large cubes (5 x 5 x 5 centi-
metres) of saturated colours (three
each of TCS09 Strong red, TCS010
Strong yellow and TCS012 Strong
blue) and eighteen small cubes (3 x
Figure 6: Comparative spectral power distribution (SPD) curves of Xicato modules: XSM 8030 – 2000 3 x 3 centimetres) of pastel shades
C, XSM 9530 – 1000 B, XSM 8030 – 1000 B with halogen, fluorescent and ceramic metal halide lamps (three each of TCS02 Dark greyish
as well as reference illuminant D65 from the Commission Internationale de L’Eclairage (CIE). yellow, TCS03 Strong yellow green,
TCS05 Light bluish green, TCS07
Light violet, TCS011 Strong green
and White) were arranged in the
Scene type Scene 1 Scene 2 Scene 3
composition depicted in Figure 8.
Light source XSM 8030 – 2000 C XSM 9530 – 1000 B XSM 8030 – 1000 B TCS09, TSC10, TCS012 were used
characteristics (XSM 80 Series) (XSM Artist Series) (XSM 80 Series) primarily because they are the most
Luminous flux (lm) 2000 1000 1000 commonly used colours in day-to-
day activities and participants
CRI [Ra] 80 [81.3] 95 [97.6] 80 [80.4]
can easily relate to these colours.
CCT (K) 3000 3000 3000 TCS011 was used because it is an

Table 1: The different characteristics of the Xicato light sources used in the three scene settings.
 RESEARCH Colour preference 65
under white LEDs

Light Dark Strong Moderate Light Light

greyish greyish yellow yellowish bluish Light Light reddish
red yellow green green green blue violet purple

Light Moderate
Strong Strong Strong Strong yellowish olive
red yellow green blue pink green

Figure 7: CIE–recommended basic and additional colour samples for testing CRI.

indicator for correct rendering of the cal). Construct validity refers to the Demographics Wismar London
wavelength range between 520 nm extent to which a test captures a
Age group
and 560 nm, which is crucial for specific theoretical construct or trait.
haemoglobin reflectance and Of the 133 people who agreed to 18 to 29 19 36
therefore skin colour of humans. participate in the experiments 30 to 39 11 30
TCS013 was not used, despite the conducted in the two locations, 40 to 49 13 9
fact that it best represents the 55 were from Wismar and 78 were
Caucasian skin colour, based on the from London. The participants from 50 to 59 12 3
assumption that participants from Wismar included students and Gender
other ethnic origins might not be staff from the university, while the Male 28 46
able to relate to this colour as their parti-cipants from London included
Female 27 32
natural skin colour. A differential members of the general public as
selection in textures, with the large well as staff and guests from a Eye colour
cubes having a semi-glossy finish restaurant located in the adjacent Blue 24 27
and the small cubes having a matt building. Table 2 lists a brief demo-
Green-hazel 10 24
finish, was primarily done to provide graphic analysis of the 133 partici-
visual hierarchy and to generate pants from Wismar and London. Brown 21 27
interest in the composition. The majority [> 34%] of participants
involved in both experiment set-ups Table 2: Overview of participant demographics from Wismar and London.
2.2. Methods – were between the ages of 18 and
Sample Population 29.
Given that the intention was to ob-
tain feedback from a wide range of 2.3. Methods – Questionnaire
people with different iris colours, a A questionnaire was designed which
generic invitation applicable to a first asked participants to study the
general population was prepared. three scenes visually, and select
However, the participants were their most and least preferred sce-
sampled from two different geo- nes in terms of colour quality. It then
graphical and cultural locations in asked participants to select their
order to counter the validity of the most and least preferred scenes in
study. While there are various terms of brightness quality. These
different kinds of validity, the three questions were compiled on the
important ones include: internal, basis that participants might have a
external and construct validity. differential preference for the three
Internal validity refers to whether scenes based on their individual iris
the effects observed in a study are colours. Although literature already
due to the manipulation of the inde- reveals that people with different iris
pendent variable and not some colours respond differently to
other factor. External validity refers different brightness conditions, the Figure 8: The displayed objects consisted of nine large cubes in saturated colours
to the extent to which the results of test for brightness preference was and eighteen small cubes in pastel colours, based on CIE recommendations.
a study can be generalised to other also included as a control test to
settings (situations), other people evaluate the overall validity of the
(population) and over time (histori- experiment.
66

3. Results and Analysis All hypothesis tests ultimately use separately as a pre-cautionary interesting to note that Scene 1 with
The Chi-Square Test Calculator was a p-value to weigh the strength of measure, it yielded similar results. the 80 CRI – 2000lm LED module re-
used to calculate the p-values and the evidence (what the data reveals ceived almost the same number of
analyse the statistical significance of about the respective population). 3.1. Results and Analysis – votes as Scene 2 with the 95 CRI –
the quantitative data obtained on The p-value is a number between Colour Preference 1000lm LED module. This implies
colour and brightness preferences. 0 and 1 and interpreted in the Table 3 lists the descriptive and that CRI might not be the most
When a hypothesis test is performed, following way: a small p-value inferential statistics obtained for the accurate metric for measuring
the p-value or calculated probability (typically ≤ 0.05) indicates strong participants’ preference for colour people’s preference for the colour
helps in determining the statistical evidence against the null hypo- quality. Participants rated the rendering properties of light
significance of the results. Hypothe- thesis, so it is rejected; a large appearance of Scene 2 with the 95 sources.
sis tests are used to test the validity p-value (> 0.05) indicates weak CRI – 1000lm LED module as their
of a claim that is made about a popu- evidence against the null hypothesis, most preferred scene. Participants 3.2. Results and Analysis –
lation. A no-change claim is called so it is accepted. rated Scene 3 with the 80 CRI – Brightness Preference
the Null hypothesis, which in this As the ratios of the number of 1000lm LED module as their least Table 4 lists the descriptive and
case states that there is no relation- participants with different eye preferred scene. However, no inferential statistics obtained for the
ship between iris colour and colour colours varied considerably between conclusive evidence was found to participants’ preference for bright-
preference under the light emitted Wismar and London, it was decided relate participant preferences for ness quality. Participants rated the
by LEDs. The Alternative hypothesis to perform these tests on the entire colour quality based on their iris appearance of Scene 1 with the 80
is believed if the null hypothesis is set of 133 participants, instead of colours as p > 0.05 for both tests. CRI – 2000lm LED module as their
concluded to be untrue, which in treating them as two separate The analysis of colour preferences most preferred scene. Participants
this case states there is a relationship samples. While the same test of the showed that participants prefer rated Scene 3 with the 80 CRI –
between iris colour and colour individual samples from Wismar and coloured objects illuminated using 1000lm LED module as their least
preference. London was also conducted higher CRI light sources. It is preferred scene. However, no

Scenes Scene 1 Scene 2 Scene 3 p χ2 (df = 2) Scenes Scene 1 Scene 2 Scene 3 p χ2 (df = 2)
Observed cell totals, Observed cell totals,
(expected cell totals) and (expected cell totals) and
[Chi-square statistics [Chi-square statistics
for each cell] for each cell]
Most Eye Brown 24 21 7 0.700 2.195 Most Eye Brown 26 15 10 0.286 5.017
pre- colour (21.50) (22.68) (7.82) pre- colour (22.24) (20.32) (8.44)
ferred [0.29] [0.12] [0.09] ferred [0.64] [1.39] [0.29]
scene Green- 12 scene Green- 11 16 3
11 6
hazel (11.99) (12.65) (4.36) hazel (13.08) (11.95) (4.96)
[0.00] [0.21] [0.62] [0.33] [1.37] [0.78]
Blue 19 26 7 Blue 21 22 9
(21.50) (22.68) (7.82) (22.68) (20.72) (8.60)
[0.29] [0.49] [0.09] [0.12] [0.08] [0.02]
Totals 55 58 20 Totals 58 53 22

Least Eye Brown 15 9 24 0.197 6.025 Least Eye Brown 15 8 28 0.631 2.578
pre- colour (14.08) (8.30) (25.62) pre- colour (13.04) (6.90) (31.06)
ferred [0.06] [0.06] [0.10] ferred [0.30] [0.17] [0.30]
scene Green- 13 scene Green- 8 2 20
2 18
hazel (9.68) (5.71) (17.62) hazel (7.67) (4.06) (18.27)
[1.14] [2.41] [0.01] [0.01] [1.05] [0.16]
Blue 11 12 29 Blue 11 8 33
(15.25) (8.99) (27.76) (13.29) (7.04) (31.67)
[1.18] [1.01] [0.06] [0.40] [0.13] [0.06]
Totals 39 23 71 Totals 34 18 81

Table 3: Chi-square statistics for the most and least preferred scenes in terms Table 4: Chi-square statistics for the most and least preferred scenes in terms
of colour. of brightness.
 RESEARCH Colour preference 67
under white LEDs

conclusive evidence was found to and lower CRI – higher luminous Literaturverzeichnis
relate participant preferences for flux, produced almost the same
brightness based on their iris colours results leading to the possibility that 1. Taylor HR, B M, S W, NM B, SB B, FS R. Visible light and
as p > 0.05 for both tests. The analy- these two properties of light sources risk of age-related macular degeneration. Trans Am Opht
sis for brightness pre- are more closely related than initially halmol Soc. 1990;88:163–73.
ference shows that participants obvious. Future testing could lead 2. Franssen L, Coppens JE, Berg TJTP van den. Grading of
prefer coloured objects illuminated to more tangible results, such as Iris Colour with an Extended Photographic Reference Set.
using a higher luminous flux. It is energy savings through the use of J Optom. 2008;1(1):36–40.
interesting to note that Scene 2 with higher CRI – lower luminous flux 3. Zlatkova M, Beirne RO, Hinds NA. Colour discrimination
the 95 CRI – 1000lm LED module re- light sources in lit environments in individuals with light and dark irides: an evaluation of
ceived almost the same amount of where perceived brightness is of the effects of intraocular straylight and retinal illumination.
votes as Scene 1 with the 80 CRI – primary concern. J Opt Soc Am A. 2014;31(4):A268–73.
2000lm LED module. This implies 4. Dain SJ, Cassimaty VT, Psarakis DT. Differences in FM100-
that a high CRI can be considered as Acknowledgements Hue test performance related to iris colour may be due to
a suitable alternative to high The authors thank Roger Sexton and pupil size as well as presumed amounts of macular
luminous flux for improving the Patrick van der Meulen from pigmentation. Clin Experiment Ophthalmol. 2004;87
perceived brightness of objects. Xicato for providing the necessary (4–5):322–5.
data, equipment and infrastructure 5. Park DJJ, Karesh J. Topographic Anatomy of the Eye: An
4. Conclusions for conducting the tests. Overview. In: Tasman W, editor. Duane’s Ophthalmology
Three broad conclusions can be The authors also thank Lisa Woods [Internet]. Philadelphia, USA: Lippincott Williams &
drawn from this study. Firstly, there for help with the statistical data Wilkins Publishers, Inc.; 2006. Available from:
seems to be no statistically analysis. http://www.oculist.net/downaton502/prof/ebook/
significant relationship between duanes/pages/v7/v7c001.html
colour preference from LED light 6. Imesch PD, Wallow IHL, Albert DM. The colour of the
sources and iris colour. However, human eye: A review of morphologic correlates and of
given the fact that the test for some conditions that affect iridial pigmentation. Surv
brightness preference also yielded Ophthalmol. 1997;41(2):S117–S123.
similar results could imply that 7. Sturm RA, Frudakis TN. Eye colour portals into
future tests should consider pigmentation genes and ancestry. TRENDS Genet.
additional parameters such as light 2004;20(8):327–32.
sources with more natural and 8. van den Berg TJ, IJspeert JK, de Waard PW. Dependence
vibrant colour rendering properties, of intraocular straylight on pigmentation and light
comparisons with daylight, and transmission through the ocular wall. Vision Res.
larger sample sizes to yield more sta- 1991;31:1361–1367.
tistically significant results. 9. Fox DL. Biochromy, natural colouration of living things.
Secondly, CRI does not appear to Berkeley, USA: University of California Press;
be an accurate metric for measuring 1979. 244 p.
people’s preference for the colour 10. Weiter JJ, Delori FC, Wing GL, Fitch KA. Relationship of
rendering properties of light senile macular degeneration to ocular pigmentation. Am J
sources. The two scenes illuminated Ophthalmol. 1985;99:185–187.
using a higher CRI – lower luminous 11. Nicolas CM, Robman LD, Tikellis G, Dimitrov PN, Dowrick
flux, and lower CRI – higher A, Guymer RH, et al. ris colour, ethnic origin and
luminous flux, produced almost the progression of age-related macular degeneration. Clin
same results, leading to the Experiment Ophthalmol. 2003;31(6):465–469.
possibility that a different metric 12. Frank RN, Puklin JE, Stock C, Canter LA. No Title.
might be required for measuring Trans Am Ophthalmol Soc. 2000;98:109–15.
peoples’ preference of the colour 13. Eiberg H, Mohr J. Assignment of genes coding for brown
rendering properties of light eye colour (BEY2) and brown hair colour (HCL3) on
sources. chromosome 15q. Eur J Hum Genet. 1996;4(4):237–241.
Thirdly, using light sources with 14. CIE. Method of measuring and specifying colour
a higher CRI can influence the rendering properties of light sources. Vienna/AUSTRIA:
perceived brightness of objects. CIE 013.3-1995; 1995.
The two scenes illuminated using
a higher CRI – lower luminous flux,