08-KP4-11 Hagerhall-Hassan.fm Page 292 Wednesday, …20av%20landskapsbilder... ·...

292 KART OG PLAN 4–2011

Visning av landskapsbilder på en stor kurvet skjerm– Visningsbetingelsenes effekt på evalueringeneCaroline M Hägerhäll and Ramzi Hassan

Vitenskapelig bedømt (refereed) artikkel

Caroline M. Hägerhäll og Ramzi Hassan: Displaying landscape photos on a large curved screen – The impact of viewingcondition on evaluations.

KART OG PLAN, Vol. 71, pp. 292–301, P.O.B. 5003, NO-1432 Ås, ISSN 0047-3278

Photographic representations are commonly used as substitutes for on-site experience of landscapes in research on hu-man evaluations of landscape. However, there is an ongoing discussion about the validity of such representations. In thisstudy we explore how screen size affects evaluations. We also present an approach in which 2D static photographs areused in combination with a digital 3D environment, combining the advantages of a photorealistic representation of veg-etation and a digital model. Forty-two respondents, divided into two equal groups, evaluated forests with three differentdegrees of biodiversity viewed either on a 17 inch computer screen or on a 7m x 3m curved display. Groups evaluated theforests using the Basic Emotional Qualities scale, where subjects are asked to judge how they would feel after havingspent a period of time in the environment presented. Although the overall landscape evaluation was not statistically sig-nificant under the different screen conditions, interesting differences in evaluation of the different forests were foundthat suggest that the evaluations were affected by viewing conditions.

Key words: landscape experience, emotion, biodiversity, forest, curved display, landscape modelling techniques, virtualreality theater

Caroline M Hägerhäll: Professor, Agr.Dr., Department of Work Science, Business Economics and Environmental Psy-chology, Swedish University of Agricultural Sciences (SLU). P.O Box 88. SE-230 53, Alnarp, Sweden. E-mail: [email protected]

Ramzi Hassan: Associate Professor, Dr. Scient., Department of Landscape Architecture and Spatial Planning, Norwe-gian University of Life Sciences (UMB) P.O.BOX 5029, 1432 Aas – Norway. E-Mail: [email protected]

IntroductionField experiments are costly and time con-suming and limit the number of environ-ments that can be tested. Photographic rep-resentations are therefore commonly used asa substitute for on-site experience of land-scapes in research on human evaluations oflandscapes. Furthermore, photo based stud-ies provide experimental control and allowthe study of future landscapes using simula-tion techniques. The validity of photo basedstudies is currently under discussion.

Studies have been carried out on varioustypes of representations, such as normalphotographs, panoramic photographs andvideos. The results concerning validity areinconsistent and point to a complex relation-ship between stimuli and evaluations (Dan-iel T. C. & Meitner M.M. 2001; Hull R.B. &Stewart W.P.1992; Kellomaki S. & Savolain-

en R. 1984; Shuttleworth S. 1980; StampsA.E. 1990; Oh K. 1994). One important dif-ference between an on-site experience andviewing a photographic representation is inthe sense of presence, that is, the differencebetween being in or looking at the landscape.Apart from on-site factors that could affectother senses than vision, the extent of thevisual environment differs greatly if, for ex-ample, a person is looking at a small sizephoto or is surrounded by the environmentin the field. (IJsselsteijn W. & Riva G. 2003;Lombard M. & Ditton T. 1997; Flach J.M. &Holden J.G. 1998; Slater M. 2009).

The majority of landscape preference andlandscape experience tests are performed us-ing a standard computer screen or projector.The aim of this study is to investigate possi-bilities for combining the control of a labora-tory setting and photographic representa-

08-KP4-11_Hagerhall-Hassan.fm Page 292 Wednesday, November 23, 2011 9:09 AM

Visning av landskapsbilder på en stor kurvet skjerm

KART OG PLAN 4–2011 293

tions with a larger degree of presenceachieved by showing static photos on a largecurved display and using Virtual Realitytechnology. The hypothesis is that the expe-rience of environments in this setting woulddiffer from the experience when viewing thesame environments on a small normal com-puter screen. We hypothesize that responseswill be stronger and that there will be a larg-er degree of presence.

Commercially available 3D software isavailable which could create 3D representa-tions of vegetation and ecosystems. However,3D digital models of a forest, for example,would not be as realistic as photographs of areal forest. Creating digital 3D models is alsotime consuming. Photographs, on the otherhand, represent real cases and could be pro-duced in a relatively shorter time. In thisstudy we introduce a method using 3D im-mersion using 2D photo based material. Thestrong 2D illustrative characteristic of a pho-tograph is combined with a 3D environmentin order to enhance the feeling of presence inthe environments. This creates a new experi-ence that differs from the experience of a dig-ital 3D model. In this study we investigatehow Virtual Reality technology can enhancethe use of 2D photo presentation.

MethodParticipantsForty-two respondents participated in thestudy, equally divided between the two setups:small computer screen and large curved dis-play. All were students at the Norwegian Uni-versity of Life Sciences. Most of the partici-pants were from the Department of LandscapeArchitecture and Spatial Planning but a fewcame from The Department of Ecology andNatural Resource Management. The samplefor the small computer screen consisted of 5males from 22–29 years old (mean 25.6), and 16females from 19–25 years old (mean 21.8).Thesample for the large curved display consisted of9 males 22–34 years old (mean 26.4) and 12 fe-males 22–32 years old (mean 25.25).

EnvironmentsThe environments in this study are forestswith differing degrees of biodiversity, catego-

rized as low, middle or high diversity basedon the visible degree of layering of vegetationand the ground cover vegetation. Sites wereselected in collaboration with local authori-ties and a landscape ecologist at the Depart-ment of Landscape Architecture and SpatialPlanning. Local landscapes were used in or-der to allow the subjects to return to the sitesfor future studies, and to provide sites thatwere familiar to the test subjects.

StimuliThe study was conducted in mid-October andover a short time span, since the visual envi-ronment changes rapidly at that time of theyear. Sampling was done under similarweather conditions to minimize effects ofweather on the responses. Numerous pic-tures were taken at each locations to providea good variety from which to choose, and fourimages were finally selected for use in thetest. One image was selected from each of thethree diversities, and the fourth was a low di-versity forest with more apparent human in-fluence, Figure 1-4.

The technical preparation of the stimuli isof particular importance and can be de-scribed as follows. All pictures were takenwith a Canon EOS 30D digital single-lens re-flex camera mounted with a wide angle lens,Canon EFS 10–22 mm. A tripod was used toavoid blurred photos and to ensure that allpictures were taken from the exact same po-sition at each viewpoint. The pictures weretaken in Canon RAW file format (CR2) formaximum resolution. The large curved dis-play (curving 160 degrees with 7m width x3m height) is part of a virtual reality theaterused for visualizing and interacting withthree-dimensional models. The projectiondisplay has a resolution of 3840x1805 pixels.To adapt the format to match these dimen-sions we needed to take several pictures, sideby side, at every location, which we thenmerged together digitally. By manually lock-ing the shutter speed, aperture, focus pointand focus length we achieved the same light-ing conditions and viewing area in concur-rent pictures for each location. Using desig-nated software, we stitched the photos to-gether to create the correct size and view.Several programs were used to process the

KP4-2011.book 93 Thursday, November 17, 2011 3:02 PM

Bedømt (refereed) artikkel Caroline M Hägerhäll and Ramzi Hassan


pictures. First, the RAW-format file was con-verted to high quality jpeg file format withCanon Digital Photo Professional. Then westitched two or three pictures together withCanon PhotoStitch. Finally, the pictureswere altered and adjusted with Adobe Pho-toshop and Google Picasa.

One problem that arose during the prepa-ration process was the distortion of perspec-tive that occurs when using a wide anglelens. We were photographing in relatively

cramped areas, in forests and amongst ob-structive landforms, and we needed to use awide angle lens to capture the whole scenefrom a short distance, without losing theground or sky from view. The wide angle lenstends to distort the perspective as it ap-proaches the 180 degree view, making it dif-ficult to merge them without mismatching.We overcame this problem by adjusting theresults manually in Photoshop to fix the mis-matching.

Figure 01: Forest with low diversity. The ground vegetation has few species, and there is al-most no vegetation in the middle layer. The main vegetation is trees of one species.

Figure 02: Forest with middle diversity. The ground cover still has low diversity but the mid-dle layer is richer and several tree species are present.

KP4-2011.book Page 294 Thursday, November 17, 2011 3:02 PM



Preparing images for the large curved displayAs stated earlier, the photos were to beshown in two settings: a traditional compu-ter screen and a large curved screen. Thesmall screen setting in the computer lab pre-sented no problems. It was easy to show 2D

picture on a 17 inch screen with display res-olution:1280X1024 pixels.

Concerning the test using the large curvedscreen, special procedures needed to be de-veloped. The main issue was how to projectthe photos which were prepared previouslyonto the curved screen with some kind of 3Dimmersion.

Figure 03: Forest with high biodiversity. All layers contain many species.

Figure 04: Forest of low diversity, but containing stands with both planted (on the right) andnatural (on the left) structure.




The curved display is located in a virtual re-ality laboratory (Figure 6). This VR-Lab con-sist of a 3D visualization system of threeChristie Digital Mirage S+2K front projec-tors with 1400x1050 native resolution and3000 ANSI lumens brightness per projector.A warping and blending module for projec-tion onto cylindrical surface is integrated in-side each projector and is optimized for 3Dprojection. The projectors project a stereo-scopic image to a screen curving 160 degrees(7m width x 3m height) around an audienceof maximum 24 people. The 3D visualization

systems are powered by a cluster of threeworkstations (3.8 GHZ with 4GB RAM), eachof which has Nvidia FX 4500G PCI graphiccard with 512 MB graphic memory.. Eon Re-ality Concave system us used as the main op-erating system to run the VR-Lab. The dis-play technology used for testing is based onactive stereo technology combined with LCDshutter glasses, see Figure 7. Active stereodisplay technology fits in here since it repre-sents a bright and homogeneous picture infull pixel resolution. Due to the high separa-tion of the individual stereoscopic pictures,

��

��

��

��

��

� ��

Figure 05: Process of preparing photos, for both tests.

Figure 06: 3D illustration showing the setup of the VR-Lab with the curved screen.

08-KP4-11_Hagerhall-Hassan.fm Page 296 Wednesday, November 23, 2011 9:13 AM



perception of depth is not impaired by doublecontours. Furthermore, no compromises arerequired between the stereoscopic mode andresolution.

We developed a solution based on the prin-ciple of creating a 3D model of the field ofview for each forest scenario. The field ofview is basically the representation of theprojected view on the curved display. Afterstudying the properties of the field of viewon the curved screen, we made a 3D modelfrom the data using 3D modelling software(Google SketchUp). We then took the forestimage and draped it (paint) over the 3Dmodel screen. The whole process involvedis shown in Figures 8, 9, 10 and 11. Theprocess was carried out for each forest sce-nario.

Figure 07: Concept sketch of Active stereosystem (Belleman R.G. et al 2005)

��

Figure 08: Studying the field of view by projecting a grid over the curved screen.

Figure 10: Adjusting the image and the centre view, then draping the image over the 3D mod-el of the field of view, which represents the model of the curved screen.

Figure 09: Creating a 3D representation of the field of view in the 3D modelling softwareSketchUp.




The central viewpoint was placed at eyeheight (1,8 meters above the bottom of themodel) and 3 meters out from the middle ofthe concave curve. We could now place themodel into the VR-Lab as a 3D model. Whenshowing the pictures in this way, we couldsimulate a walking movement in the model.This was done by moving the control joystickforwards and looking sideways. When mov-ing forward we were actually walking to-wards the model, and looking left or rightturned the virtual person. Using this tech-nique we could move around in the picture ina fluent way, giving the feeling of movingaround in the landscape. To enhance the lev-el of immersion we used the 3D active stere-oscopic projection in combination with 3DLCD shutter glasses.

Test formOur emotional state can influence how weperceive an environment. In this study weused the Basic Emotional Qualities index.This instrument is based on the Human–En-vironment Interaction model (HEI) devel-oped by Küller (Küller 1991, 2004; Johans-son and Laike 2007). The HEI model takes aneuropsychological perspective on emotionsand describes a basic emotional process thatinvolve the four components activation(e.g. alert/sleepy), orientation (e.g. interest-ed/bored), evaluation (e.g. happy/sad) andcontrol (e.g. confident/hesitant).The specific

emotions we experience are combinations ofdifferent levels of these components.

The Basic Emotional Qualities index formwas first translated from the original Swed-ish version to Norwegian before the test. Therespondents are instructed to look at eachpicture and asked to rate how they wouldfeel if they had spent an hour in the depictedenvironment. Their response is based on 12adjectives (for example rested, safe, interest-ed, angry, efficient, sleepy, happy) represent-ing different emotions and four levels foreach adjective. The respondents are asked tomark the level they think corresponds to howthey would feel. For example: very tired,quite tired, quite rested and much rested.

Test procedureBefore both tests, the participants were giv-en a short presentation of the goals of thestudy and the test procedure. The respond-ents were told it was a test comparing theviewing of images on different screens, butnot what the environments depicted or thatthe environments differed in biodiversity. Inthe VR–lab the images were shown to thewhole group of participants at the same time,as shown in Figure 12. Every forest wasshown for approximately 5 minutes, startingwith some panning and zooming by the sys-tem administrator. The image was shown foras long as necessary, to allow everyone tocomplete the test form.

Figure 11: Projecting the 3D model (with the forest image scenario) on the curved screen at theVR-Lab.




The procedure for the test in the computerlab differed in some ways from the VR-lab test.It was carried out in three different computerlabs at the same time. One folder for the testwas placed on the server available on all thecomputers. This folder contained four morefolders, one for each of the forests. Every par-ticipant had his or her own screen and compu-ter and could decide for themselves how muchtime they needed to complete the form.

The order of presentation was the same inboth tests; high diversity, low diversity, mid-dle diversity, and the alternative low diversi-ty environment (Figure 4). The order was en-sured by the systems administrator in theVR-lab test and by the numbering of foldersin the computer lab test.

AnalysesAn average emotional index, based on all theratings on the adjectives, was computed foreach test participant for every forest. An aver-age emotional index, based on all the ratingson the adjectives, was computed for each testparticipant for every forest. The index was

then used in the further statistical analysesin SPSS and STATISTICA. To analyze the ef-fects of the screens and the forest diversityconditions, a Repeated Measures ANOVA wasused, with diversity as the within-subjectsfactor and the screen as the between-subjectsfactor1. Significance level was set to 0.05.

ResultsThe results of the ANOVA showed thatscreen type did not have a significant effecton the participant’s response on the emotion-al index. The level of diversity, in contrast,did have a significant effect on the emotionalresponse.2 No significant interaction be-tween screen type and diversity was found.However, the response to different forestswas affected in different ways by the screencondition. As shown in Figure 13, the emo-tional response to the high diversity forestwas higher in the large screen condition thanin the small screen condition. The responseto the middle diversity forest was the samein both screen conditions. The low diversityforests had a higher response index in the

Figure 12: Photo from the test sessions at the VR-lab with the curved screen.

1. An important strength of a within-subjects design is that it controls for individual differences. We expected such differences to be bigger in relation to experience of biodiversity than in relation to the screen conditions.

2. (F3,120 = 25.84; p = 0.000)




small screen condition. This is interestingsince it means that responses to the high andlow biodiversity conditions are affected in op-posite ways by different screens. The re-

sponse to the middle diversity forest is notaffected by viewing condition. Interestingly,that emotional index was highest for the lowdiversity forests.

DiscussionOur results, surprisingly, show no statistical-ly significant difference between the twopresentation methods. The larger screen wasexpected to add more presence, but this didnot have a significant effect on emotions.There are indications, however, that the ef-fect of presentation method differs in forestsof different diversities. This is very interest-ing and calls for further research.

Screen size affects the amount of detailthe respondents can see, and both positiveand negative aspects of the environment canbe enhanced by increasing the size of thescreen. The high diversity forest has a lot ofdetail, for instance from foliage. The emo-tional response to this forest type was higheron the large screen than the small screen,which might be due to blurring of these de-tails on a small screen. The small screen doesnot give the viewer as clear an impression ofthis type of forest as the large screen. Theamount of detail in a low diversity forest is

lower. As a result, photos of a low diversityforest could be less exciting on the largerscreen, as everything is clearly visible.

Another factor influencing the resultscould be the low position of the sun on the ho-rizon due to the autumn season. In the moreopen, low diversity forests, sunlight throughthe trees is visibly striking and might haveeffected the evaluation of these images, mak-ing them more interesting. This effect couldhave confounded the effect of vegetation di-versity. Even though we tried to make thephotos as alike as possible, the sunlight ismost visible in the lower diversity forests.

This could explain why the emotional in-dex is higher in low diversity forests underboth screen conditions than in any of the oth-er forests types. It should also be noted thatsound conditions in the two different lab set-tings could have had an effect on the results.Conditions in the VR lab setting were lesscomfortable due to noise from the large pro-jectors.

�

Figure 13: Mean emotional index score for the four forest images in the two screen conditions.Forest type, HDv- high diversity, MDv- middle diversity, LDGv- low diversity, LDBv low di-versity alternative.




A main aim of this study was to explore theusefulness of VR technology and a facilitywith curved screen, for presenting environ-ments to respondents in photo based land-scape experience studies. Although we did notfind an effect of screen size on response to for-est diversity, the development of the modeland procedures for showing 2D static imageson the large curves screen of the VR facilitywas successful and has great value for futurestudies. Using the VR-lab environments forlandscape experience studies opens many op-portunities to simulate field experience, in-cluding work with senses other than vision.Adding such effects was outside the scope ofthis study, which focused on effects of size ofthe projected image. In future studies varia-bles such as sound, wind and maybe evensmell could be included. Further work withvisual aspects of the image, for instance theperception of depth, is also called for. A great-er sense of depth is expected to add to the feel-ing of presence; this could be tested usingemerging techniques that can produce stereo-scopic(3D) effects in standard 2D photos.

The focus in this study was on screen sizeand projecting 2D static images, but the ap-proach also allowed for some movement (for-ward and backward and side to side) in themodel. One advantage of virtual reality envi-ronments is that real-time movement can beadded. The type of motion is more interac-tive, and hence more natural, than what canbe achieved, for example, in a film of a walkin an environment.

AcknowledgementsWe thank Morten A Kirkemo and AmundHareland who worked on this project as astudent assignment.

ReferencesBelleman, R.G., Stolk, B. and De Vries, R., 2005,

Immersive Virtual Reality on commodity hard-ware. Unpublished paper. Universiteit vanAmsterdam. http://staff.science.uva.nl/~rob-bel/publications/asci2001.pdf

Daniel, T.C. and Meirtner, M. M., 2001, Represen-tational validity of landscape visualizations:the effects of graphical realism on perceived

scenic beauty of forest vistas. Journal of Envi-ronmental Psychology, 21, 61–72.

Flach, J.M. and Holden, J.G.,1998, The reality ofexperience: Gibson’s Way. Presence, 7, 90–95.

Hull, R. B. and Stewart, W. P., 1992, Validity ofphoto-based scenic beauty judgements. Journalof Environmental Psychology, 12,101–114.

IJsselsteijn, W. and Riva, G., 2003, Being There:The experience of presence in mediated envi-ronments. In Riva G, Davide F and IJsselsteijnW. A. (eds). Being there: Concepts, Effects andMeasurements of User Presence in SyntheticEnvironments. Amsterdam: Ios Press, pp 3–16

Johansson, M. and Laike, T., 2007, Intention toRespond to Local Wind Turbines: The Role ofAttitudes and Visual Perception. Wind Energy,10, 435–451.

Kellomaki, S. and Savolainen, R., 1984, The scenicvalue of the forest landscape as assessed in thefield and the laboratory. Landscape Planning,11, 97–107.

Küller, R. ,1991, Environmental assessments froma neuropsychological perspective. In Gärling T.and Evans G.W. (eds.). Environment, Cognitionand Action: An integrated approach. NewYork:Oxford University Press. 1991, pp.111–147.

Küller R., 2004, Emotional model of human-envi-ronment interaction. In Evaluation in Progress-Strategies for Environmental Research andImplementation. Abstracts of the 18th IAPSConference [CD-ROM]. Martens B, Keul AG(eds.). Vienna University of Technology: Vienna

Lombard, M. and Ditton, T., 1997, At the heart of itall: the concept of presence. Journal of Compu-ter Mediated Communication, 3 (2),0.

Oh, K. ,1994, A perceptual evaluation of computer-based landscape simulations. Landscape &Urban Planning, 28, 201–216.

Shuttleworth, S., 1980, The use of photographs asan environmental presentation medium inlandscape studies. Journal of EnvironmentalManagement, 11, 61–76.

Slater, M., 2009. Place illusion and plausibility canlead to realistic behavior in immersive virtualenvironments. The Royal Society, 364,3549–3557.

Stamps, A. E., 1990, Use of photographs to simu-late environments: A meta-analysis. Perceptualand Motor Skills, 71, 907–913.


08-KP4-11 Hagerhall-Hassan.fm Page 292 Wednesday, …20av%20landskapsbilder... ·...

Documents

Transcript of 08-KP4-11 Hagerhall-Hassan.fm Page 292 Wednesday, …20av%20landskapsbilder... ·...