MOVING CLOUDS ON A VIRTUAL SKY AFFECT WELL-BEING AND
SUBJECTIVE TIREDNESS POSITIVELY
Oliver Stefani1, Matthias Bues1, Achim Pross1, Sandra Mebben1, Phil Westner1
Heinrich Dudel2, Horst Rudolph2
1 Fraunhofer IAO, Stuttgart, Germany, 2 Trilux, Arnsberg, Germany
To investigate the hypothesis that a simulation of natural daylight such as a sky with moving
clouds would be desirable in offices and if this reproduction increases well-being and reduces
subjective tiredness, we developed a luminous ceiling (Virtual Sky) and examined 30 test
persons for three times during three weeks. In this paper we provide a technical description of
Virtual Sky and the test conditions. We compared static light to two different dynamic lighting
conditions. Our findings indicate that the acceptance of dynamic lighting depends on the
performed tasks. Furthermore, for test candidates without access to natural daylight,
subjective tiredness after one working day was significantly lower under a cloud animation
compared to static light. Well-being decreased significantly less under a cloud animation
compared to static light.
Keywords: Dynamic Lighting, cloud animation, LED based luminous ceiling, subjective
tiredness, well-being, spatial resoluted light
1 Introduction and Objective
For more than 150,000 years humans have predominantly lived and worked outside. The
human visual system is optimized for natural light and the entrainment of the circadian rhythm
of several biological functions is controlled by natural daylight. Despite of this evolutionary
conditioning imprint we have been living mostly inside buildings with limited daylight or
artificial light since the industrial revolution some 150 years ago. Our 30 test participants only
spent on average 15 minutes a day outside. An average American even spends just five
minutes a day under an unobstructed sky. The lack of natural daylight is known to cause
winter depression (SAD, Seasonal Affective Disorder) and sleep disturbances (Terman, 2005
and Wirz-Justice, 2009). Various former research has shown that light is the most important
“Zeitgeber” of the body’s inner clock (Brainard, 1995). Due to architectural and geographical
restrictions, completely natural light in buildings, which would be the ideal, can rarely be
ensured. Missing natural light makes artificial lighting indispensable, especially in winter and
after dusk. We developed and evaluated a Virtual Sky (Fig. 1) to create dynamic light (cloud
animation) that increases well-being by reproducing some of the most important
characteristics of natural light.
Figure 1 – Virtual Sky at Fraunhofer IAO
2 Related Work
Former studies examining dynamic light in office environments have shown that light changes
have positive effects on well-being of people: motivation increases and the acceptance of
dynamic light is very high (Dehoff, 2002).
Effects on well-being and motivation by changing light were investigated by Fleischer
(Fleischer, 2001). Lighting consisted of luminaires that slowly changed between direct and
indirect light. The ratio between direct and indirect lighting was changed according to the time
of the day or to weather conditions. It was shown that pleasure rises with higher illuminance
and a large indirect component. With an increase of the direct component and an increase of
illuminance arousal also rises. However, Fleischer suggests that such a lighting condition
should not be applied throughout the day since it can eventually cause bad feelings.
Dynamic lighting in schools can also change learning behaviour (Schulte-Marktwort, 2010). It
was shown that a cold colour temperature increases concentration and a warm colour
temperature provides a relaxing environment. The colour temperature (and illuminance) was
changed from 5600K to 3000K after 25 minutes. A previous study conducted by us showed a
preference of dynamic over static lighting by the majority of participants (Stefani, 2010), which
led us to conceive the more in-depth study described here.
3.1 Technical description of the Virtual Sky
To reproduce the outdoor lighting conditions, particularly its dynamics, we developed the
Virtual Sky which is a luminous ceiling consisting of a matrix of 34560 LEDs (RGBW) behind a
diffusor foil (Barrisol). It has a size of 4.5m x 7.5m and can be controlled like a display with a
resolution of 12 by 20 pixels, each LED tile representing one pixel (Fig. 2). The LED tiles are
equipped with AB-P144-292x292-E-WD+RGB LEDs (Daylight White: 5500K; R: 620-630 nm,
Green: 520-535 nm, B: 465-475 nm) from LEiDs GmbH & Co. KG.
Figure 2 – LED-tiles of the Virtual Sky
Virtual Sky provides colour temperatures from 1550 K to 27000 K, a high brightness (3000 lux
on the floor) and a good colour rendering index (up to 95 at 5600 K). This means that
scenarios like a blue sky with a yellow sun, a cloudy sky and sunrise/sunset can be simulated.
Figure 1 shows a setting with a strong contrast between the blue sky and the white clouds.
3.2 Definition of the dynamic light parameters
For our study, we developed sequences of subtle moving clouds (significantly lower contrast
compared to the scenario shown in figure 1). Although the cloud pattern was constantly
changing, it was not easily identifiable at first sight. Light changes on the desk however were
clearly visible. Figure 3 shows the spectrum and variance that has been applied in the study.
The grey lines show the variance during the cloud animation. Spectral metrology was carried
out using a CS-1000 spectroradiometer from Konica Minolta. The diffuse light from the Virtual
Sky was supplemented with direct static warm white (3000 K) light (TRILUX Neximo).
Table 1 shows the lighting parameters for the three conditions the participants were exposed
to during the study (S: Static, LD: Low Dynamic, HD: High Dynamic = cloud animation). Each
measurement represents values on the desk including the direct light from the TRILUX
luminaire. Measurements were carried out using a LEiDs LED+LIGHT check, Gigahertz-Optik
BTS256 and Gigahertz-Optik HCT-99. Figure 4 shows the illuminance change over time for
the HD condition.
Figure 3 – Spectrum and variance applied in the study
Figure 4 – Illuminance change on the desk during cloud animation (HD)
Table 1 – Light parameters
Maximum and minimum values also refer to five different measurement points on the desk
(measurements taken in the four corners and in the middle of the desk).
30 test persons were examined in three sessions. We selected 14 healthy women and 16
healthy men between 22 - 33 years of age (Ø 24.8). The health of the test persons was
evaluated by entry questionnaires. Exclusion criteria were smoking, medication or drug
consumption, shift work within the last three months, and transmeridian flights in excess of
three time zones up to three months prior to study, pregnancy, allergy, donation of blood
within the last four weeks, epilepsy, migraine, neurological or psychiatric suffering. Each
participant could choose the office tasks to perform, but had to perform the same tasks during
each session. Participants were asked to maintain similar nutrition behaviour, a regular sleep-
wake schedule as well as the same way of travelling to the lab. These parameters, as well as
participants’ time under natural daylight, outside weather conditions, etc. were recorded and
analysed. Participants were not allowed to have long discussions with each other. The
ambient room temperature varied between 19°C and 22°C. During the entire study protocol,
which comprised a total of three weeks in April 2011, each test person was examined in a
session from 9:00 am until 5:00 pm, each on the same weekday with a different predefined
lighting scenario per session (the order was mixed):
Session 1: Static light (S) with a fixed cloud pattern.
Session 2: Low dynamic light (LD) with changes every 90 minutes from cold white (6900K)
to neutral white (4600K) and back to cold white.
Session 3: High dynamic light (HD) with continuous light changes, similar to moving
clouds (cloud animation).
Questionnaires about the current condition had to be filled out every 90 minutes, feelings
were recorded on a scale from 1 (= not at all) to 5 (= very much). The Karolinska Sleepiness
Scale (KSS) (Gillberg, 1994) was used to evaluate the subjective sleepiness.
Figure 5 – Test room
4.1 External influencing factors
First of all we investigated the tiredness of the participants in relation to various external
influencing factors. Our data show no correlations between the time spent under natural
daylight (before the test) and tiredness in the morning (r = -0.01) but we found a correlation
between sleep duration and tiredness in the morning (r = -0.33).
Half of the participants were placed close to the windows with lowered shade while the other
half sat close to the wall, behind a baffle, resulting in significantly less influence from natural
daylight. The setup of the study environment is shown in figure 5.
4.2 Tiredness and well-being in the evening
At 5 pm participants were slightly more tired under S than under HD (Ø difference 0.37, p =
.07). There was no difference between S and LD (Ø difference 0.16, p = .31) (using the KSS
scale from 1 = extremely alert to 10 = extremely sleepy, falls asleep all the time).
If we only consider the 15 participants sitting at the wall side, hence having less influence
from natural daylight from the window, we find significantly less tiredness under HD than
under S (Ø difference 0.93, p = .03). Between LD and S there is only little difference (Ø
difference 0.27, p = .22) (Fig 6).
Figure 7 shows the effects that occurred after seven hours of treatment at 5pm (on a scale
from 1 – 5) and the results of a t-test (one sided, paired Student’s t-test) between S and HD.
We would like to point out that the feeling “excellent” has a highly significant difference of
0.54 (p = .003) between S (Ø 2.75) and HD (Ø 3.29).
Figure 8 shows that participants felt more nervous (Ø 1.69) and annoyed (Ø 1.62) with LD
than with S (Ø 1.31) with p = .03 and p = .05 (on a scale from 0 to 5). There is no significant
difference between S and HD for nervous and annoyed (Ø difference < 0.12, p > .33).
Figure 6 – Tiredness of participants sitting next to the wall at 5 pm
Figure 7 – Feelings after seven hours treatment at 5 pm
Figure 7 – Evaluation of negative feelings after seven hours treatment at 5 pm
4.3 Change of tiredness and well-being during the day
We investigated the sense of freshness (0 = not at all; 5 = very fresh) of all participants but
could not find significant differences. We discovered however that the 15 participants at the
wall side were a little fresher in the afternoon with HD than with S (Fig. 8). We found a
correlation between the time of the day and the freshness for S (in terms of decreasing
freshness under static light) but not for LD and HD. There is a tendency (p = .091) that with
LD and HD freshness oscillates more than with S.
Figure 8 – Freshness of the 15 participants sitting next to the wall decreases during the day
(with a low just before lunch)
We summarized positive feelings (easygoing, nonchalant, relaxed, comfortable, splendid,
laid-back, placid, cheerful, calm, joyful) and compared the mean values at 9 am with the
mean values at 5 pm. Statistical analyses yielded significantly reduced positive feelings while
being exposed to S compared to HD (Ø difference 0.21, p = .001) (Fig. 9).
Figure 9 – Decreasing positive feelings from 9 am to 5 pm
4.4 Preferences and open questions
After each session we asked people how they liked the lighting scenario on a scale between 0
(not at all) and 100 (very much) and could not find a significant difference. However,
separating the groups according to the type of work they did during the test brings up some
interesting effects. Participants who did concentrated work slightly preferred S over LD and
HD (Ø difference > 10, p < .13) (Fig.10). Participants who did creative work significantly
preferred HD over S (Ø difference 25, p = .001) (Fig. 11).
We would like to point out that participants did not appreciate the fast transition from cold
white to warm white under LD. Although the light changed only every 90 minutes it changed
from 6900K to 4600K within 5 seconds. We assume this very fast transition to be the reason
for the aversion of this scenario.
Participants were asked what they liked and disliked most about each of the lighting
scenarios. Open answers for the cloud animation (HD) were:
What did you like most about the cloud animation (HD)?
I liked the light changes in general (10x)
The change of light kept me awake; was energizing (7x)
I liked the bright light phases (6x)
I liked the cold white light phases (5x)
It feels like being outside (3x)
I liked the warm white light phases (2x)
What did you dislike most about the cloud animation (HD)?
The change of light distracted me from work; it was difficult to concentrate (5x)
Dark phases were too dark (5x)
I had eyestrain (3x)
Sometimes the light was too bright (2x)
Figure 10 – For concentrated work S is preferred over LD
Figure 11 – For creative work HD is preferred over S
Our LED based luminous ceiling displaying moving clouds (HD) in an office both improved
well-being and reduced tiredness in the evening. Positive feelings decreased significantly less
under moving clouds compared to static light (S). Nevertheless during slow dynamics (LD)
participants disliked the harsh transition (within 5 seconds) from working light (6900K for 90
min) to break light (4600K during break-time).
Participants who did creative work preferred the moving clouds over static light whereas
participants who did concentrated work preferred static light. Since our experiment was
conducted in a real office environment, measured effects of dynamic light are relatively small.
We expect stronger effects in laboratory settings. Currently we are developing various
dynamic cloud patterns in order to further investigate optimal settings for different working
tasks and contexts. Maybe dynamic lighting is like "music for the eyes": Sometimes one
prefers relaxing music, sometimes inspiring music and sometimes just silence.
This work was sponsored by TRILUX GmbH & Co. KG.
Brainard G. C., Bernecker C. A., 1995 The effects of light on human physiology and
behaviour, New Delhi: CIE Conference
Dehoff P. 2002 The impact of changing light on the well-being of people at work, Proceedings
of the 5th International Conference on Energy Efficient Lighting, Nice, ISBN 91-631-2523-4
Session 14 Dynamic lighting.
Fleischer S. 2001 Die psychologische Wirkung veränderlicher Kunstlichtsituationen auf den
Menschen, ETH Thesis Nr. 14033
Fleischer S. 2001 Effect of brightness distribution and light colours on office staff, Lux Europa
Gillberg M., Kecklund G. and Åkerstedt T. 1994 Relations between performance and
subjective ratings of sleepiness during a night awake. Sleep 17, 236-241
Schulte-Marktwort M 2010 Laborstudie zur Replizierung der Wirksamkeit von dynamischen
Licht im Schulunterricht, Universitätsklinikum Hamburg-Eppendorf, Klinik und Poliklinik für
Kinder- und Jugendpsychosomatik, Studie II: Ergebnisbericht
Stefani, O.; Bues, M.; Pross, A.; Hille, K.; Spath, D. 2010 Lichtdynamik am
Wissensarbeitsplatz der Zukunft. In: Neue Arbeits- und Lebenswelten gestalten Dortmund,
Terman M., Terman J.S. 2005 Light therapy, in Principles and Practice of Sleep Medicine, 4th
ed. Edited by Kryger MH, Roth T, Dement WC. Philadelphia, Elsevier, 1424–1442
Wirz-Justice A., Benedetti .F & Terman M. 2009 Chronotherapeutics for Depression. A Manual
for Clinicians. Basel, S.Karger AG.