Daylight Evaluation: The Influence of Vertical Luminance on Visual Preference in Daylit Environment
Authors: Yi-Szu Liao Parsons School of Design, The New School Craig Bernercker Ph.D., FIES, LC, Associate Professor of Lighting Design, Parsons School of Design Rebecca Mintz Parsons School of Design, The New School
Introduction Daylighting is considered as an important architectural design approach not only because it enables reduction of energy consumption but also provides high illuminance, good color rendering to interior space, view connection with outdoors, and many other benefits. However, without careful design, daylight might also cause discomfort glare, veiling effects, and visual fatigue. In order to achieve an ideal constructed lighting environment, there are illuminance-based guidelines to evaluate and design electrically illuminated spaces. Daylight delivers higher illuminance levels to a space; thus, it is not equivalent to use the same metric to evaluate daylight environment. Recent software developments for daylight study, like DIVA, have increased access to luminance-based data, but there is still not a method to effectively relate the data to human impression of daylight environment. This paper is interested in daylight’s impact on human perception and what determines human preference in daylit environment. By studying the relationship between quantifiable data and subjective daylight qualities, this paper aims to provide a way to evaluate spaces for daylighting design.
Background Current horizontal illuminance-based metrics are predominantly used in electric lighting design for measuring humans perceived adequacy of illumination on a task surface. But visual performance studies and visual comfort metrics have shown that luminance is more related to human visual comfort than illuminance [Boyce 1973, Hopkinson 1972, Wienold and Christoffersen 2006]. As daylight typically provides illuminance levels exceeding what is necessary for task performance, the quality of light is most important in daylighting design. Thus, it is valid to use luminance-based metric to discuss daylit environment. A previous daylight preference study conducted an experiment to examine a series of luminance-based visual comfort metrics, trying to evaluate their potential to explain human’s preference in daylit spaces. The result of the study indicates that the metrics that had higher correlation coefficient to human visual comfort response are not based on the entire field of view. The luminance metrics that most correlated with subjective visual preference responses are those for the scene-dependent mask of window area and scene-independent mask of 40∘band in field of view [Van Den Wymelenberg and Inanici 2015]. Both of metrics are vertical-surface-related. Since most contemporary tasks have shifted from physical paper-based horizontal tasks to monitor-based vertical tasks, human visual preference might be mainly affected by the luminance of vertical
surfaces in a space. This study is to explore the correlation between vertical luminance and visual preference. The previous research from Van Den Wymelenberg and Inanici [2015] was done in a small private office. This research is more interested in shared public space, such as a school study room. The methodology established here can be applied to other similar space types. Methodology Research Settings The research setting is a shared study room E301 in 25 E 13th St Parsons School of Design, New York, NY (41° N and 74° W), with a row of southwest facing windows (28° from true South). The room received daylight only from the left side. Electric light was not turned on during the experiment, so there is no electrical lighting influence. The research design includes two experiments: 1) in-person experiment in physical space and 2) computer-generated visualization experiment. Since real space experiment for visual comfort study in daylit environment might be restricted to limited data collection because the results will only come from specific time frames and days. It is hard to collect comprehensive information from broader weather and daylighting conditions. By using computer-generated visualizations, effect of daylight can be better controlled, which allows to choose more extreme conditions and to fully randomize the scenes. In addition to these advantages, the reflectance of the surfaces in the space can also be manipulated, which provides another aspect of variables for researchers to compare the results. Studies have been done by using visualizations in the research as an alternative to using real environment. However, there are also disadvantages of visualization that cannot be neglected, such as perceived inaccuracy from real environments because the 3D model cannot 100% present the real space. Thus, the combination of real environment and visualization studies is needed in order to get the most effective result as possible. The human perception of the room was measured along with the absolute luminance levels in the field of view. Figure 1 demonstrates the room setting as seen from a participant’s point of view. The view remained constant in both in-person and visualization experiments. The visualization was a representation of the physical room.
Section
Plan
Figure 1: Site configuration
In-person experiment The in-person experiments were conducted on March 30th and April 6th in three separate sessions and included a total of 12 participants (4 person for a session) all Parsons graduate school students (8 lighting design major and 4 non-lighting design major). The detail of the time and sky conditions were documented during the experiment [Fig. 2]. Participants spent 30 minutes doing a monitor-based task of their choosing and responding to a questionnaire, discussed in the following section, every 15 minutes to assess their impression of the room under changing daylight conditions. Three assessments were done in one session (at the beginning, 15 minutes in, and at the end), resulting in nine different daylight scenes being evaluated in each of the three sessions [Fig. 3]. Session 1 (03/30/2019)
Session 2 (04/06/2019)
Session 3 (04/06/2019)
0 min
15min
30min
0 min
15min
30min
0 min
15 min
30 min
1:00 pm
1:25 pm
1:43pm
2:34 pm
2:51 pm
3:07 pm
4:07pm
4:23 pm
4:40 pm
P. Cloudy
P. Cloudy
P. Cloudy
Clear Sky
Clear Sky
Clear Sky
Clear Sky
P. Cloudy
Overcast
Figure 2: Time and sky condition records
Figure 3: Images of the scenes took from each in-person experiment session
The independent variable in this research is the variation in luminance distribution of the room surfaces. The variation in interior luminance patterns results from changes in daylighting due to shifting sky conditions over time.
Questionnaire During each assessment, participants were asked to provide their subjective assessments of the space for each scene. The subjective evaluations were conducted using a questionnaire as used in Van Den Wymelenberg and Inanici’s experiment in 2015, with adjustments to suit to the current experiment. The questionnaire items that participants were asked to rate for each scene are the follows:
1. This is a visually comfortable environment for a classroom. (QU1) 2. I am pleased with the visual appearance of the room. (QU2) 3. I like the vertical surface brightness. (QU3) 4. I am satisfied with the amount of light for computer work. (QU4) 5. I am satisfied with the amount of light for paper-based reading work. (QU5) 6. The computer screen is legible and does not have reflections. (QU6) 7. The lighting is distributed well. (QU7) 8. When I look at my computer monitor the scene that I see in front of me seems: (front-scene) (QU8) 9. When I look to my left the scene that I see seems: (left-scene) (QU9) 10. When I look to my right the scene that I see seems: (right-scene) (QU10) 11. I find the ceiling to be: (ceiling) (QU11) For questions one through seven, participants rated the statement using a 5-points Likerttype scale from 1 (strongly disagree) to 5 (strongly agree). For questions eight through eleven, participants rated the visual impressions of brightness using a semantic differential scale from 1 (too dim) to 5 (too bright).
Analysis Method High Dynamic Range (HDR) photography was collected at the same time as subjects were assessing the room. HDR photographs were generated for all nine scenes. HDR photography can be used to collect luminance data, enabling the ability to analyze complex luminance distribution patterns by converting the scenes to quantifiable luminance maps in false color images [Inanici 2006]. The false color images of each inperson experiment scene are presented in Figure 4. The software “Photosphere” was used in the research to analyze the luminance mapping of region interest. In order to study specific area, masks were utilized to isolate the region of interest from the rest of the scene [Inanici 2005]. In this study, 4 masks were applied in order to study each area’s contribution to participants’ responses [Fig. 5].
Figure 4: False color images of each scene
Figure 5: Mask applications for region interest
Results Comparing visual preference responses with perceived brightness of each vertical region (left/window all, front/adjacent wall, and right/inside wall) [Fig. 6], the results suggest that left/window wall is the main factor that influences human visual preference of the entire room. Figure 7 shows the correlation between subjects’ ratings of bright/dim of left wall and of visual preference: the visual preference and perceived brightness are tracking when the left wall mean luminance is perceived as “bright� (from approx. 260 cd/m2 to 415 cd/m2). When mean luminance goes high enough (above 415 cd/m2), then the visual preference ratings drop off. However, when the left/window wall mean luminance is reduced below approximate 250cd/m2, no appreciable difference in the rating of visual preference is observed.
Figure 6: Mean L of all vertical surfaces (Mask_1) versus subjective ratings of visual preference and perceived brightness of window, front, and right wall
Figure 7: Mean L of left/window wall (Mask_3) versus subjective ratings of visual preference and perceived brightness of left/window wall
Perceived front/adjacent wall brightness also attributes to visual preference when subjects perceived the front/adjacent wall as “too bight� (over 200 cd/m2 in this study) [Fig. 8]. Figure 8: Mean L of front/adjacent wall (Mask_2) versus subjective ratings of visual preference and
perceived brightness of front/adjacent wall
For perceived brightness of the right/inside wall related to visual preference, no clear relationship could be observed, indicating that human visual preference is not affected by the right wall and the change in brightness of right/inside wall is not perceived because the left wall predominates the perception of the space. Luminance ratio is not applicable to explain the relationship between visual preference and different daylight qualities because the luminance ratio did not change appreciably from one scene to another, which stays consistently around 1:2 to 1:3 [Fig. 9]. As the luminance of window wall goes up, the luminance of adjacent wall goes up in proportion to it.
Figure 9: Luminance ratio of front wall and window wall versus visual preference ratings
Visualization experiment: The experimental visualization scenes were generated using AGi32’s daylight study renderings. A 3D virtual model of the room used for the in-person experiment was created. The actual reflectances of room surfaces were measured and inputted into the 3D model [Fig. 10]. The method of using illuminance and luminance to determine surface reflectances was used, but actual material specifications were not available to confirm the accuracy of the reflectance data. Additionally, the render may not be perfectly accurate in lighting level. However, the intent was to match the rendered image as close to the real space as possible.
Figure 10: Interior material reflectance values
The choice of the scenes was based on the seasons (Equinox, Summer Solstice, and Winter Solstice), from 9am to 16pm (with one-hour interval), in order to get widest range of daylight conditions. Scenes that required electrical lighting to reach basic task illuminance levels and the scenes having similar daylight conditions were eliminated. Six daylight conditions were chosen to be modeled, with variety of daylight distributions. Variation in reflectance of the pin-up wall (from 0.25 to 0.5) was added to the remaining six conditions, resulting a total of twelve scenes for the visualization experiment. The overview of the twelve scenes is presented in Figure 11. The visualization experiment was conducted through an online survey with a total of 66 participants (29 with design background and 37 with non-design background). In the
Figure 11: Visualizations of daylight conditions used in the experiment
survey, participants were presented with the twelve scenes showing up on the screen one at a time in a random order and asked to evaluate their impression of the space using the questionnaire described below.
Questionnaire: When doing the online survey, participants were asked to provide their subjective assessments of the space for each scene. The subjective evaluations were conducted using
the same questionnaire as used in in-person experiment with the exclusion of questions 5 and 6 (see questionnaire section above), because the judgement was done by looking at images and those questions require an immersive environment.
Analysis Method Instead of using false color images, point-by-point calculations were done in AGi32 for all the twelve scenes by placing 0.5ft x 0.5ft grid of Diffuse Luminance calculation points on each vertical surface within the field of view. Then calculation points were grouped for region interest, as the similar fashion in in-person experiment [Fig. 12].
Figure 21: Groups of calculation points for region interest
Results The results from the visualization study suggest that subjects’ visual preference perception is correlated to the brightness perception of left/window wall and front/adjacent wall [Fig. 13]. Figure 14 shows the perceived brightness of left/window wall and the ratings of visual preference track almost identically and that almost true for front/adjacent wall as well [Fig. 15]. When left/window wall and front/adjacent wall are perceived as “dim” (below approx. 340 cd/m2 on window wall and 190 cd/m2 on adjacent wall), the ratings of visual preference go downward to not preferable. However, the drop off in visual preference at high luminance is not observed as it is in the real space.
Figure 13: Mean L of all vertical surfaces versus subjective ratings of visual preference and perceived brightness of window, front, and right wall
Figure 14: Mean L of left/window wall area versus subjective ratings of visual preference and perceived brightness of left/window wall
Figure 15: Mean L of front/adjacent wall area versus subjective ratings of visual preference and perceived brightness of front/adjacent wall
Despite the brightness of the right wall can be relatively perceived in visualizations, it still does not attribute to human visual preference. Figure 16 shows the ratings of visual preference follow the brightness perception of left/window wall instead of the perceived
brightness of right/inside wall. In addition, comparing two scenes with the same lighting condition but different reflectance (0.25 and 0.5) on the right/inside pin-up wall, no obvious change in the ratings of visual preference responses could be observed [Fig. 17]. The luminance ratios are greater in the visualizations than in the real space (from 1:1.3 to 1:3.1). Figure 18 shows that when the luminance ratio goes lower than 1:1.5, the visual preference is rated toward not preferable. However, the changing of the ratio is simply not great enough, thus, the luminance ratio still doesn’t have an appreciable impact on visual preference.
Figure 16: Mean L of right/inside wall versus subjective ratings of visual preference and perceived brightness of right/inside wall
Figure 17: Comparison of visual preference ratings before and after the modification in pin-up wall reflectance
Figure18: Luminance ratio versus ratings of visual preference
Discussion It is clear from both in-person and visualization study that human visual preference in a daylit space is predominantly influenced by the perceived brightness of window wall in that space. As window wall and adjacent wall are perceived as too bright, visual preference decreases. The inside wall that does not have window does not affect visual preference. This result supports the results of the previous study from Van Den Wymelenberg and Inanici (2005) that the luminance of window area is most related to subjective visual preference responses. However, instead of using scene-dependent window area to explain visual preference, this study shows a clear relationship between luminance of entire window wall (scene-independent) and visual preference in a daylit space, which provides a broader application for more space types. In the in-person experiment, it is not identified that visual preference as uncomfortable when mean luminance of window wall is rated as “dim� (below approx. 260cd/m2). It is reasonable because the study is based on that every scene meets the basic task illuminance levels. Thus, when the mean luminance of window wall drops to a certain degree, the ratings of visual preference goes down slightly then flatten out. The visualization is not capable of creating visual discomfort situation since it cannot represent the high luminance well enough to cause the sensation of visual discomfort. When looking at the scenes on the screen, subjects’ perceived luminance would not nearly as high as it really should be in the real space.
The luminance ratios are greater in the visualizations than in the real space due to different measure methods. In the real space, the luminance levels on the two walls were measure from the same point of view; the window wall was perceived as peripheral view. But in the visualizations, all the calculation points are perpendicular to the measured surfaces; thus, the luminance ratios were the related number of two direct views. However, both experiments could not draw a conclusion about how the luminance ratio affects visual preference. As the sun is the only light source that affects the whole room, the brightness on each wall increases or decreases in proportion, the ratio wouldn’t change drastically in general unless the room is deeper enough which is not applicable in this research. In conclusion, the results from this study suggest that one might need to use the physical space rather than visualizations for studying visual preference in daylit environment, because what we perceive in the visualization does not correlate well with what we experience in the real space. In addition, it is difficult for visualizations to accurately present the actual space. Designers can predict and evaluate daylight qualities through building in-scale physical mock-ups. This study was limited to some degree by the small sample size and lighting conditions in the real space, the absolute luminance numbers in this study can only serve as references. More studies are needed in order to find a clearer relationship between absolute luminance levels and visual preference, which may provide a useful information for future architects and lighting designers when considering daylighting design.
Reference Inanici M. 2005. Per-pixel Lighting Data Analysis. <https://escholarship.org/uc/item/688137zg#main> Accesses 2019 14 Febuary. Inanici M. 2006. Evaluation of High Dynamic Range Photography as a Luminance Data Acquisition System. Lighting Research and Technology. 38(2):123-134. Lee E, Clear R, Ward G, Fernandes L. 2007. Commissioning and verification procedures for the automated roller shade system at the New York Times headquarters, New York, New York. <http://eta-publications.lbl.gov/sites/default/files/nyt-shade-cxprocedures.pdf> Accesses 2019 12 February.
Loe L, Mansfield KP, Rowlands E. 1994. Appearance of lit environment and its relevance in lighting design: experimental study. Lighting Res Technol. 26:119-33. Stokkermans M, Vogels I, De Kort Y, Heynderickx I. 2018. A Comparison of Methodologies to Investigate the Influence of Light on the Atmosphere of a Space. LEUKOS. 14(3):167-191. Van Den Wymelenberg KG, Inanici M. 2010. The Effect of Luminance Distribution Patterns on Occupant Preference in a Daylit Office Environment. LEUKOS. 7(2):103122. Van Den Wymelenberg KG, Inanici M. 2016. Evaluating a New Suite of LuminanceBased Design Metrics for Predicting Human Visual Comfort in Offices with Daylight. LEUKOS. 12(3):113-138.