Visual Variables in Physical Environments and Virtual Environments: an Experiment

IntroductionWhere 3D ‘virtual-worlds’ are intended to communicate the architects’, designers' or planners’ ideas, they are widely assumed to give the end-user an accurate preview of the anticipated experience of the actual building. There are however indications that suggest that differences between the professionals' intentions and the end-users' perceptions arise through the representational medium employed, leading to dissatisfactory performance evaluations (Mullins et al., 2003). Particularly in matters where participation of the public in the planning and design process is given greater importance, there is a need to develop methods to minimize these variations of perception (Holmgren and Svensson, 2001). This study is part of an attempt to broaden the understanding of the use of digital technologies in architects' design-representation and to develop means to enhance their use in public participatory projects.Aim - General Research Questions AskedWith regard to physical buildings, to what extent do architectural intentions and user expectations coincide through the means of various representative media; specifically CAVE and PANORAMA environments.Does prior knowledge of a physical building or environment have any effect on the perception of simulations of virtual environments and screen media?HypothesesThere is a correlation between perceiving subjects’ spatial ability and their understanding of simulated environments. There are measurable differences in the perceptions of laypeople and professionals in the information gained from representations of existing and yet-to-built environments. More accurate perceptions in environments of spatial simulation will be found where there is prior experience and tacit knowledge of the given physical environment. More accurate perceptions of screen representations will be found where there is prior experience of the spatially simulated environment.Methodology - VR and Physical ModelsA CAVE model and interactive VRML model of an existing (physical) building are used for the experiment. The physical building is itself used in the study and provides benchmarking for quantifiable values of visual variables of objects within the building. Specifically, the foyer volume was used in all three environments.Visual VariablesTest objects in the physical building were introduced shapes of varying size, colour and texture (see figure 1). These were attached to walls, columns, etc. at differing distances in relation to the subject’s standpoint. With a departure point in Granum and Musaeus’ set of static object properties, test objects included variables for position, pose, size, shape, colour and texture (Granum and Musaeus, 2002). Observation distance was measured in relation to the size of object, where the dimension of 100cm of the largest object = 1 Standard Distance Unit (SDU). Following Gram and Musaeus, a range of important characteristics like inter-object distance and observer-to-object-distance are thus related in a meaningful way to the size property of the objects. Objects had 3 differing sizes to monitor the effects of observation distance on shape recognition.The objects positioned in the physical building were represented in the virtual environments by scaled simulations in the same (virtual) positions.Table 1 summarises the variables used.SubjectsThe participants in the experiment were comprised of four groups in roughly equal numbers: Group P1: professionals, with no previous spatial experience of the building.Group P2: professionals, with previous spatial experience of the building.Group L1: lay public with no previous spatial experience of the building. Group L2: lay public with previous spatial experience of the building.The sample comprised 70 subjects.ProcedureSubjects were individually tested in random order. Procedures A and B were alternated successively. Subjects were first shown a 1:20 scaled drawing of the test objects. The drawing was available for re-showing throughout the remaining procedure. Presentation in each of the 3 environments comprised 3 standpoints with 3 objects visible from each standpoint. Standpoints were precisely the same in each environment. Subjects were asked to identify the visible objects in relation to the scaled drawing, with each answer (27 in total) recorded on the questionnaire immediately.Procedure A: Half of the subjects (the groups P1 and L1) were shown the 9 objects in each of the test environments, starting in the PANORAMA, followed by the CAVE and finally in the physical building. Procedure B: procedure A was reversed for the remaining half of the subjects; that is, the groups P2 and L2 were shown the 9 objects in each of the test environments, starting in the physical building, followed by the CAVE and finally in the PANORAMA.ConditionsSpatial perception in a 3D virtual environment unfamiliar to the subject. This condition was reproduced in procedure A for groups P1 and L1.Spatial perception in an existing environment familiar to the subject. This condition was reproduced in procedure B for groups P2 and L2.InstrumentsQuestionnaires were developed to enable the identification of accuracy in relation to the specific perceptual references used in the investigation, by subjects with training in spatial ability (professionals) and those without (laypeople). Questions were posed as multiple choices, for example:” From standpoint ‘A’, do you judge the visible square shape to be: S1, or S2, or S3? (Choose only one)”.External validity can be measured by the ‘training effects’ of people engaging with virtual reality and transferring their knowledge to the real world (or vice versa).ResultsSubjects’ answers were grouped into four and tested for accuracy against actual recorded shape positions, shapes and sizes.Data AnalysisAn analysis of the data collected as described above is presented in the paper. The aim of analysis is to identify patterns and noteworthy, statistically significant discrepancies which provide reasonable evidence to support the hypotheses stated above. DiscussionThe findings lead to a discussion which point to the implications for the use of virtual environments and which suggests means of inhibiting differences that arise between the professionals' intentions and the end-users' perceptions of the represented intentions. The paper attempts to broaden the understanding of the use of digital technologies in design-representation and to develop means to enhance their use in public participatory projects.