1

Using Geo-Located Augmented Reality for Community

Evaluation

Mohammad Ashraf Khan and Andy Dong

University of Sydney, Australia

Conventional practices of two-dimensional representation of three-dimensional objects remain an impediment to end-user engagement in

participatory urban design. An alternative is to harness geo-located

augmented reality (GAR) technology to embed life-sized virtual three-

dimensional images at the actual site of proposed interventions. This format offers closest to real-life visualizations for end-users, enabling them to firstly

comprehend and then express feelings concerning future proposals. This

paper presents an iPhone web-app that capitalizes on the Layar browser’s GAR interface to tip the economies of scale in favor of intimately attached

users of public space, rather than often remotely detached clients and their

commissioned designers. Walk-around virtual images of public space designs can be viewed and commented via iPhones by the public. It further

allows users to display their own ideas as alternatives, thus in effect serving

as an instrument for advancement of CAAD-enabled participative

environmental design in general and the idea of reclamation of authorship of public space in particular. This paper briefly describes the development of a

prototype, including its preliminary evaluation, and then highlights a study

to determine the 3D rendering performance parameters of GAR technology, as the core component of the idea. The paper concludes with a discussion of

future implications.

1.0 Introduction

This paper builds upon the work of Hii, Zhou, Karlekar, Schneider, and Lu

[1] by adding the alternative of using outdoor geo-located augmented reality

(GAR) technology available on smart phones upon their research on palm-

2

sized alternative specialized apparatuses for outdoor viewing of life-sized

geo-located 3D virtual models of architectural or urban design proposals.

Whereas the cited work is grounded in the search for achieving improved dimensional consistency across architectural or urban design spatial

decisions and their representations, as an advanced approach to

environmental design, in the present paper this research trajectory is viewed specifically for its significance as a potential option to address the

information accessibility gap on the side of end-users when they are invited

to participate or collaborate in spatial design decision-making processes.

This gap is the result of the conventional practice to obligate untrained users to comprehend and comment on two-dimensional representations of

three-dimensional or four-dimensional spatial design decisions. Constraints

that result from coding and decoding back and forth across 2D and 3D formats are already recognized to be an internal systemic gap in the design

discipline itself [2], [3]. It is therefore all the more problematic for any party

placed external to this process, including foremost end-users. They are expected to decode, comprehend, and contribute informed feedback on the

basis of data contained in representational formats that may have taken years

of training for design professionals to themselves gain proficiency. Lay

members of the public have to intercommunicate across a missing dimension of spatial cognition within the span of a single exposure. Their engagement

is therefore at best constricted and at worst skewed, indicating an urgent

need for improvements, as already discussed in published studies [4], [5], as well as in a number of reviews of on-going international work to address this

gap [6], [7], [8].

In moving towards alternatives, one of the directions that can be explored

further is the usage of outdoor augmented reality technology. Work completed thus far, including the one mentioned at the outset, has tended to

rely on custom-configured portable equipment [9], [10], [11], [12], [13]. A

possibility that remains to be investigated is the usage of mobile phones that are now offering comparable in-built capabilities on Android and iOs4.0

platforms via the Layar browser [14].

This is the trajectory of the present paper. It describes the development and evaluation of a project undertaken specifically for the purpose of

overcoming the above mentioned information accessibility gap by

harnessing GAR capability available specifically on the iPhone/iOS4.0

interface. It pursues the research question of whether dimensional consistency between spatial design decisions and their representation is at all

doable in real-time or not. This research introduces a potential solution, in

the form of a dedicated iPhone web-app that enables prospective users to view full-scale walk-around virtual images of proposed public space designs

at their actual prospective locations. Its present functionality allows users to

even display their own ideas as propositions, albeit within certain limitations of textural detail, thus in effect serving as an instrument for advancement of

3

CAAD-enabled participative environmental design in general and the idea of

reclamation of authorship of public space in particular.

This idea relies heavily however, on the extent to which current technological capability of the iPhone/Layar combination supports real-time

visualization of geo-located virtual 3D objects super-imposed on live video

screen views of their placement. This paper is correspondingly devoted mainly to this key component. It covers salient features of development and

evaluation of the web-app briefly and then focuses in detail on a study

carried out to determine 3D rendering performance parameters of current

mobile handset-based GAR technology. This is followed by a listing of reflective inferences, concluding with a discussion of future implications.

Research Context

This research builds on existing body of work dedicated to development of

CAAD-based technological alternatives for bringing end-users into the spatial design process, as already reviewed in detail by a number of authors,

including Hanzl [6], and with reference to cultural considerations in

particular, by Foni, Papagiannakis, and Magnenat-Thalmann [15]. Those that relate specifically to the use of GAR technology to facilitate participative

design of public spaces are more recent and relatively few [1], [9], [10], [11],

[12], [13] while a larger number of previous works deal with the broader

spectrum of underlying issues, bound at both ends by two mutually convergent directions of research: development of IT-based alternatives for

improvement of public consultation processes; and, technological

innovations for achieving dimensional consistency in the representation of spatial design through realistic life-sized 3D in-situ outdoor imagery.

At the core of this published discourse is a concern to aid end-users to

understand visual information more intuitively, with the ensuing body of

work divided into two main prevailing viewpoints. One group of authors who have conducted an empirical study on the problem of cognitive

processing of 2D versus 3D images and related perception of scales, report

that 2D representations supported by rich supplementary information are a more effective medium than 3D visualizations [16], while others report

benefits of one option over the other to be associated with the nature of the

task in question [17], [18]. This later view supports the option of using 3D visualizations as best suited for representation of 3D data in the context of

spatial design decision-making processes. Most recent arguments on this

topic include the view that ‘complex problems can be recognized faster with

the aid of an interactive adjustment of suitable viewpoints’ [19], which points in favor of real-time responsive 3D simulations, while another

argument is that stronger user engagement is associated with the extent of

realism captured by a given medium [20]. It is on the basis of similar reasons

4

that a comprehensive review by Hanzl [6], on research and practice related

to the usage of IT tools in participatory spatial design concludes that 3D

augmented reality systems hold the most potential as the future direction of this field [21].

Majority of AR-based works however, including usage of GIS-based 3D

graphics or virtual models, are constrained by the requirement of PC and large screen equipment [22], or even in the case of hand-held devices,

specialised equipment and set-ups. This leaves a potential for the use of

more ubiquitously available devices, such as GAR-capable mobile phones,

while in quantitative terms on-line discussion forums offer a maximum catchment of end-user feedback [2]. A combination of these two

technologies therefore holds promise for the most conducive interface for

public engagement in architectural or urban design processes. End-users otherwise stand largely excluded in the prevalent model of practice, of

design decisions being taken at times by commercially-driven developers or

government bureaucrats and their chosen consultants, who may have marginal or nil contact with the context in question [23].

An Alternative

Drawing on the above thinking, the basic concept underlying this project

was to use the iPhone iOS4.0 platform to harness GAR technology offered

by the Layar browser [14], to create a more lucrative interface for spontaneous inclusion of users in public space design processes. A specially

formatted website dedicated for use on iPhones, termed as a web-app in the

iPhone user and developer community, was produced as a prototype. Highlights of the development process are given below, followed afterwards

by a concise documentation of its testing, with the aim to showcase the

innovative value of the underlying idea and its merit as a direction of future

research.

GAR Technology

Technological setup of the project was developed based on an approach that

can be best described to be grounded in Dewey’s concept of pragmatism [24]. Augmented reality was identified as a better suited option based firstly

on future projections of market value, as described by Cheng [25], in turn

reporting on Juniper Research: that usage of AR services is ‘expected to reach USD 732 million by 2014’ [25]. The iPhone platform was selected,

because it offered an HTML-based prototype to first be developed and tested

(as completed within the scope of the present project), as a precursor to a

full-fledged iOS4.0 app.

5

Design Development

A user-centred design (UCD) approach was pursued [26], in combination

with the interaction design (IXD) iterative approach [27], comprising: prospective user interviews; a contextual inquiry and review of related

literature, to determine basic issues involved; use of a low-fidelity prototype,

to determine tentative organisational and operational decisions; and, a usability test with a hi-fi prototype, to address unforeseen operational bugs

in the design.

A key suggestion that arose from interviews was to supplement the

envisaged 3D images with availability at the site of down-sized samples of actual materials proposed to be used in the implementation of the given

urban design object. This suggestion was partially instrumental in

formulating experiments on the rendering capability of the Layar browser described later. Existing community consultation activity was identified as

the context for the prototype, and was taken into account via published

reviews on this topic [28], available video recordings of typical community consultations, and also by examining archived/live online discussion forums

that simulate or offer an alternative to such sessions.

Prototype Implementation

Data obtained from above user need identification was translated first into an organisational schematic design, comprising infrastructural or functional

decisions, and an operational design scheme, comprising superstructural

decisions such as operational sequencing and GUI aesthetics.. Both of these

two main types of decisions were taken rapidly, in line with the limited scope of the present prototype phase, with a view to further improvement in

successive development or prototyping stages that may be taken up

separately as a future project. The prototype used for the present study consisted of a three component

setup: at the centre was the Layar browser, with the HTML website serving

as the first point of contact, and a PHP-database (DB) set-up as the storage for URLs of 3D objects (files). Information provided by the Layar browser

developer website was used to configure a dedicated layer to function in

accordance with the aim of the webapp, while the apple developer website

was used to finely adjust the main website with HTML meta-tags to display on the iPhone screen as a consistent web-app. A screenshot of the web-app

GAR display is included here (see Figure 1, on next page).

Evaluation

The scenario selected for an evaluation study was usage of the product by postgraduate students in a university setting. Five tasks were assigned to the

6

Fig. 1 – Screenshot showing GAR

functionality via the Layar browser (16 cubes superimposed on live video)

voluntary participants: to locate a pre-assigned 3D life-size object and bring

into view; and, to recognise and successfully deploy link to the ‘feedback’

webpage. Results revealed areas that needed improvement but were overall positive: participants were impressed by the novelty of the product; and, they

took more time to understand the navigation system than planned. These

indicated the necessity to further streamline the sequence and transition

between each function of the prototype. Highlights of this study, conducted

according to the think-aloud method [29], are given here as the basis for subsequent 3D performance testing.

Aims

This evaluation had two aims: to collect data concerning anticipated operation of the web-app, with a view to identification of needs for further

improvement; and, to measure learnability of the user interface (UI).

Participants

Five university students were recruited on a voluntary basis to participate in the study. One of the participants was from a background other than

environmental design, in order to check the appeal of this web-app to other

types of professionals.

Method

Participants were requested to perform five tasks: exploration (view pre-

defined object); interaction (submit feedback for pre-defined object); design-1 (upload real-time GPS coordinates); design-2 (upload a 3D file name);

7

design-3 (retrieve feedback left by others). Data was recorded in the form of

a video of the tasks completed by each participant. Participant responses

captured in these videos were noted down according to a predetermined coding system, based on the open coding concept of the grounded theory

approach [30] (see Figure 2).

Analysis

Most significant of all the observations was regarding the GPS functionality of iOS4.0. Although issues of inaccuracies and delays were known from

beforehand, this study helped identify a certain pattern which could reveal

insights for a possible work-around to these issues. This pattern became evident while participants made multiple attempts to view geo-located 3-D

images. In the first sweep of the assigned geo-position of the image the

browser displayed a distance in the range of hundreds of meters, while in the second sweep it displayed a distance of few tens of meters. The first set of

distances happened to coincide with the location of one of the nearest towers

for 3G signals, while the latter happened to be closer to the actual distance to

the geo-located image. Large amount of the time spent during the operation was in reading the text instructions at various stages. Participants felt

burdened by this and the visceral content of the web-app also appeared to

have suffered on account of it. A number of positive comments were made by the participants (‘awesome’, ‘cool’, and ‘great’), but appeared to be

bogged down by the number of steps involved as well, especially due to non-

automated data entry of GPS coordinates.

PROGRESS

Comprehension : participant is trying to gather information to arrive at a decision

to interact

Selection : participant is considering which option to choose among available possibilities of interaction.

Change : participant has completed the task or has decided otherwise and wishes to return to the previous menu.

Wait : time taken for upload of desired feature.

ERROR

Retrial : unexpected outcome from a particular selection, need to back-track and try again.

Search : participant is searching for a way to overcome an unexpected outcome.

Failure : operation has come to a complete halt, participant has given up

on attempts to overcome an error.

Intervention : evaluator has to intervene and help participant to take the next action(s).

Fig. 2 - Evaluation Coding

8

Reflection

A Java script patch that directly feeds GPS coordinates into a given input

form had already been identified but was saved for later implementation. It is now realized that it would have been useful to have incorporated that patch

within the prototype that was used for the current testing exercise. On the

whole this evaluation confirmed the envisaged potential of the basic idea of the prototype but at the same time emphasises the need for smoother

operation.

3D Performance Testing

Besides the above user evaluation a pilot study for independent investigative

performance testing [31], of the Layar browser was also conducted. Aim of this pilot study was to determine the 3D performance parameters (3DPP) of

the Layar browser. Usage of the word performance here underlines the fact

that the 3D display of this browser is made possible through the coordination and convergence of a number of technologies and related operative factors,

including GPS information exchange, digital compass readings in sync with

any change of direction, accelerometer sensitivity, rapid two-way internet communication, and the live video screen display of the device used.

Parameters here mean the variables that influence the performance of the

Layar browser, in terms of time, quality, accuracy, and most of all,

reliability. Topics covered include: official developer guidelines (as published by the Layar executive team); format compatibility; gross file

sizes; number of polygons/triangles; texture effects; height settings; accuracy

of geo-locative functionality; animation options; and, upload or response times. Recommendations based on the inferences drawn from this test are

given in the form of a flow diagram that illustrates a possible methodology

for extracting optimal 3D performance from the Layar browser.

Layar Guidelines – Complexity

A tutorial for creation of 3D content has been provided by Layar, in which a

number considerations have been listed, out of which one of the most

significant is the constraint of maximum surfaces – polygons converted to triangles – of any single object visible simultaneously to be not more than

10,000. This is double the recommended limit of 5000 given for the previous

version. Files in .obj/.mtl format have to be converted via the online Layar converter into .l3d format. Modelling must be mesh-based (not NURB-

based), with unit for vertex coordinates set/scaled to 1m, centred on the grid,

for display on screen-size 320x480. Normals must be facing outwards, and

small parts are recommended to be extruded out of larger parts (to prevent creation of unnecessary/non-viewable surfaces). Faces that are not meant to

9

be seen can be avoided/deleted altogether. Details are suggested to be added

via texture images, with vertex points connected preferably to squares (to

avoid NGONS, or faces with more than five vertices). Floating vertices or overlapping faces should be avoided and models should always be created in

two versions: a simple/faster version, meant to be displayed when the image

occupies less than 128 pixels on the screen; and, a full version that is expected to upload at a slower rate. A 2D icon has also been recommended,

to be displayed when the model pixels are less than the icon size. To explore

the possibilities within these constraints was the main objective of this test

exercise.

Layar Guidelines – Texture

Another major constraint is that multiple textures (such as for bump mapping

or normal mapping), are not supported as yet, only setting up of a diffuse color to show material. Texture images (optimized .jpeg[non-transparent] or

.png files) are required to have their width and height be in (or re-sized to)

powers of 2 (that is, 16x16, 32x32, 64x64, 256x128, and so on [no need to be 1024x1024, because the display is only 320x480]). Texture images are

advised to be kept small (combined into a single UV-mapped/canvas texture

file in the same directory as the wavefront file[.obj/.mtl]), because larger

textures require more bandwidth and are mentioned in these published guidelines to have a performance impact on the client (testing revealed this

impact to be in terms of slower upload times and flickering displays). Work-

arounds like baked ambient occlusion maps can be used to fake lighting effects and also the UV/UWV-unwrap function can be used to render with

images. Transparency in textures is supported. However, blending is not

supported (therefore in the case of partial transparency, alpha value < 0.1 (or

= 0), is not rendered at all, and objects behind the face will show. This allows cut-outs to be made with textures. While for 1.0 > alpha > 0.1, the

image is blended with the camera view but not with the image behind.

Techniques not supported are: multisubs; standard materials selected from the software library (must be standalone image files); and, software-enabled

shading.

Layar Guidelines – Animation and Other Options

Further details provided in the guidelines, include colour (transparency, or

alpha values, are not supported as yet), MIME type (not defined), built-in

orientation settings, terminology, availability of basic editing in the Layar

converter (smooth/hard edges, flip faces, materials optimization, preview [with statistical info], rotate and scale), programmable ‘transform’ variables

(angle, rel and scale). Animation is mentioned in the published

documentation as a texture manipulation option. Recommendations include: maximum image size 256x256; maximum total file size 1.0MB; and,

maximum number of frames up to 25 (can be more, but not exceeding 10 if

10

more than one objects are to be viewed simultaneously). All these guidelines

were taken into account and used to develop the test models, and thereby to

test the potential possibilities within each parameter, beginning with format compatibility.

Compatible Formats

The objective of this test was to determine options available leading to the final stage of conversion of .obj/.mtl files into .l3d files by Layar’s online

converter interface. The expected outcome was a list of 3D modelling

software that can be used to produce virtual objects in addition to those

already suggested in the tutorial for devlopers provided by Layar (Blender[open source, but less intuitive], Google SketchupPro [easy to use,

but less detailing capacity or control over normals], 3DMax [professional

tool but expensive]). The main finding was the possibility to use 3D photogrammetric software, such as 3dsom.com (option to create .obj files),

and online modelling/viewing tools, such as 3DVIA.com (export in .dae then

convert to .obj via 3DMax/Sketchup). The latter has the added advantage of preview on iPhone (via 3DVIA app).

Gross File Sizes

Though Layar has suggested a maximum number of polygons for 3D

models, it is possible to reduce file sizes while maintaining detail via polygon-reducing software, such as Polygon Cruncher

(http://www.download3k.com/Install-Polygon-Cruncher.html). Optimum file

size has been reported by users/researchers as 0.4MB. Though both these two variables, namely number of triangles (after converting any polygons)

and file sizes, appear to be directly related, detail is dependent on the

additional variable of texture or rendering. Therefore perception of detail

ultimately varies from object to object depending on the particular combination of number of polygons and textures optimally suitable for the

purpose of virtual simulation of each given object. To arrive at a satisfactory

combination in the case of highly complex objects or those that involve crucial detail could entail a number of iterations of polygon reduction and

texture richness reduction, before arriving at a blend that can be reliably

displayed by the browser. File sizes used in the testing exercise ranged from 0.00075MB to 1.99MB. Though upload time appeared to vary in direct

proportion to file size, this was not directly related to viewability of the

display. For example, in the case of test nos. 9 and 10, involving objects with

the highest file sizes, the objects did load after long intervals but only if detected to be located beyond a certain minimum distance from the device.

Specifically, the object in test no.10 came into view when the fixed location

setting was increased from 7.0m to 21.0m. In contrast to this, in the case of test no. 19, even at this distance an individual piece of texture, which

11

happened to be larger than 2048x1024px, was only visible in a flickering

state (approx. frequency 10-20 flashes per sec.).

Number of Triangles vs. Detail

It was observed that the level of detail available within the Layar browser’s

recommended limitations of file size and number of polygons could suffice

most urban or public spatial design purposes. It would nevertheless be useful to further explore the possibilities within this parameter by further tests using

models of buildings in which a high level of detail may be critical for

realization of the original design intent, for example, a historic building with

a relatively simple overall shape but a high level of intricate ornamentation. Such a test could reveal insights about the extent to which detail can be

included while modelling proposed objects.

Texture Options

Numerous texture options were included in the testing, ranging from

single/flat colouration to intricate but regular patterns, and complex/artistic

compositions. There appeared to be no significant relationship between number of textures and the overall file size, as evident from a comparison of

test nos. 9 and 10 against test no. 19. Although the file size of the object in

the latter is less than the objects in the former tests, the total number of

textures is more. Upload time can be seen to remain unaffected by the approximately 30% increase in texture items. Quality of the display was

nevertheless affected, with the largest individual sized texture visible only as

a flicker (as mentioned above).

Height Setting

All the objects were placed at eye-level height. That is, their base was set at

approximately 1.7 meters below the line of sight (value of 0.0 on the Z-axis),

in order to simulate the view from the standing position of an average adult. There is as yet no functionality in the Layar browser for any kind of altitude

change. It remains fixed to a predefined value set at the time of the object

definition in the corresponding DB table. Further testing could be carried out with various fixed values to check any impacts on upload times and quality

of the display, including manual variation of altitude change values.

Animation

Within the time scope of this study only one attempt could be made for an

animated texture test (test no. 21). Textures used for this test were kept

deliberately larger (1024x2048) than the recommended size, and the object

was also large (approx. 15m square base and 24m height). It registered in the browser at a distance of approx. 80m during viewing but failed to render and

made the browser crash after approx. 10 seconds.

12

Geo-Location Accuracy

As the browser relies on GPS functionality of the host device, a high level of

inaccuracy can be experienced, depending at times on access to satellite signals, or the positions of nearest service-provider towers (in triangulation

mode). At the designing or modelling stage this can be taken into account by

obtaining prior information about the GPS functionality with respect to the specific location where the 3D simulation is actually intended to be carried

out. When using fixed location settings, the distance to the objects displayed

in the browser appeared to be accurate to within 2-3 meters. However, in

location detection mode the displayed distance was found to be often fluctuating and unreliable (0-500meters).

Upload/Response Time

This was measured as the time it takes for the given object to register on the screen as present, in the shape of a blob, till its full shape, size and texture to

become visible. There was no evidence for incremental rendering, in all

cases the object’s true form, when downloaded, would appear at once. For simple objects with file sizes less than 0.5MB the upload time was within 5

sec.. With larger file sizes this increased to 19 seconds, and in one case it

was 27 sec. (test no. 9). Although care was taken to choose a test location

that allowed for full ISP signal reception, variation in quality of ISP signal was still noted to affect the results in some specific cases, as shown in a

selected portion of the Test Log table (see Table 1, on next page).

Miscellaneous/Unresolved Observations

One key factor which remained only partially explored is the GPS

mechanism of the device, specifically its switching between IP-triangulation

method and satellite information. Understanding this parameter could

possibly hold the key to predicting the inaccuracies and drifts in the 3D image display on the device screen.

Recommendations

A core recommendation that emerges from this testing exercise is that a direct relation between file size and upload time exists. However, file size is

not directly related to the potential for visual detail and quality of the object.

It is therefore possible to achieve considerable detail and visual richness by striking a balance between numbers of triangles and the sizes and complexity

of textures. This means a number of modelling iterations are required for

extracting optimum 3D performance from the Layar browser. It is also

useful to take into account the intended purpose of the object/simulation in question. For example, if the purpose is to simulate a proposed large

building at its actual anticipated location then it may not be necessary or

beneficial to view this object at a close range. In that case it would not be

13

TEST LOG [no. of triangles is the value displayed by the Layar converter (cnvrtr)] ['reduced' is 0.1 or less of original, icon is a 32x32pixels .png file]

model visual variations

volume (metres) [z,x,y]

tri-angles

nor-mals

file size [MB]

texture upload time (seconds)

accuracy (metres)

screenshots [cnvrtr/iPhone]

notes (observations)

SIMPLE OBJECT WITH RICH TEXTURE - AND TESSALATIONS

7 cube

steel

simple/

small

/texture

2.7x3.0x2.4 12 6 0.002650 1x steel

3.00 3.0

Note: file size increases with large file size of images used for creating texture, to test whether multiple use of the same image affects file size or number of images does.

8 cube16 with8

textures

simple/

multiple/

textures

1.0x5.0x7.6 192 6 0.698867 8 x

HRimages

8.00 3.0

9 cube16 with16

textures

simple

/multiple

/textures

1.0x5.0x7.6 192 6 1.983034 16 x

HRimages

50.00++ Error

(fxd loc - 7m) 27.00 (full signal/21m)

3.0

10 cube32 with16

textures

simple/

multiple/

textures

2.2x5.0x7.6 384 6 1.992378 16 x

HRimages

Error (fxd loc - 7m) 19.00 (fxd loc - 28m)

3.0

no significant change in file size - faces are doubled but textures are constant. At 21m the image flashes on the screen and then goes into re-load mode.

Table 1 – Selected view of the Test Log table

14

Decision-Making Flow Diagram

for Optimum 3D Performance

of Geo-Located Augmented Reality (G-AR) Interface

(iPhone / Layar Browser)

PURPOSE

is the G-AR simulation being set-up for public consultation? social networking?

for urban design decisions? public art decisions?

EXPECTED

UPLOAD TIME

[<0.5MB] = 4-5 sec.

[>1.0MB] = 8-10 sec.

[>2.0MB = 20-25 sec.

SCALE

small / large

FILE SIZE decide an optimum file size

suitable for the

purpose/scale of the project

MINIMUM

VIEWABLE DISTANCE

[<0.5MB] = any

[>1.5MB] = only when object fits in

LAYAR DESIGN MANUAL

http://layar.pbworks.com/f/3D+in+Layar_The+manual_v1.0.pdf

includes step-by-step design process guidelines

for texture/detail manipulation

TRIANGLE COUNT use optimization software

to reduce count

[polygoncruncher / ]

TEXTURE detail / quality / size

CHECK intended file size achieved or not

LAYAR 3D CONVERTER fine adjustments and/or animation

L3D file

(ready for upload)

FIRST STAGE consider all aspects and decide

FILE SIZE

SECOND STAGE consider Layar Guide

and decide MODEL SCHEME

3D MODELING SOFTWARE GSU-Pro/3D-Max/A-CAD/Blender

[export format: .obj]

THIRD STAGE construction and checking

iterations

Fig. 3 – Flow diagram showing recommendations of 3DPP

testing as a decision-making tool for users of GAR technology

15

necessary to add intricate detail to the rendering. If the facade detail happens

to be imperative, then a small portion of the object could be assigned the

required detail instead of the whole object. Conversely, in the case of a small item of street furniture or public art is to be simulated then it may be possible

to achieve a high level of detail with the low number of triangles required to

express its overall volume. On the basis of these recommendations a summarised process flow diagram has been formulated (see Figure 3, on

previous page), to serve as a tool for users of GAR technology.

Reflection

This study has opened the significance of fully understanding the 3DPP of the Layar browser. This is necessary to gain any sort of predictability for its

usage in non-gaming contexts, such as professional urban design or public

art applications. This test has been limited to a preliminary exploration, indicating the need for a more systematic, involved and rigorous exercise.

This would entail the use of a wider range of models bearing incrementally

varied properties, in order to more precisely determine the corresponding expected performance of the browser, in terms of: upload times; set of

conditions required for optimum GPS accuracy; end-impact of simulations

(to what extent the purpose of the simulation is achieved or not);

minimum/maximum distance at which the objects can be viewed; and, animation possibilities. This study serves as evidence for the need to conduct

comprehensive investigation. An increased range of objects and an increased

range of test scenarios could possibly yield sufficient information to allow the Layar browser to be used predictably and reliably in para-professional or

industrial applications. Such an investigation may require careful planning

and extensive time-labour resources.

Conclusion

Overall this study has demonstrated, albeit within certain limitations, the significance of GAR technology available on smart phones as a potential for

addressing the information accessibility gap that currently constrains end-

users to engage more meaningfully in participatory design processes. Results from the evaluation, though indicating a critical need for several

improvements in the prototype design, correspond to the initial research

question by providing evidence for the possibility of using this technology to achieve dimensional consistency between spatial design decisions and their

representations as an alternative to overcome the stated gap in participatory

design practice. Among the limitations of this study that could be improved

upon in future work in this direction, the most significant is the accuracy of geo-positioning offered at present by the Layar browser. Although

alleviation of this constraint can be expected with foreseeable improvements

16

in GPS services worldwide, it is nevertheless a serious inhibition for the

concept described in this paper at present. In terms of future implications,

the 3DPP study offers a ready reference for successive usage of GAR technology for representation of spatial design information. In general this

project opens the avenue for further development of the basic concept by

transporting it now into the realm of iOS4.0 development. This prospect opens the further possibility of building into this concept a feature for on-

the-fly designing of 3D objects as well, as already partially available in the

iPhone app called GD3D offered by Google. It would be fortuitous if the

time required for this further development could coincide with the availability of better GPS services globally. Furthermore, if decreasing

trends in prices of smart phones also continues then this concept could truly

serve as a basis for a significant shift in CAAD-enabled end-user involvement in spatial design processes.

Acknowledgements:

This project has benefitted during development stages from input of a

number of individuals, including foremost: Dr. Martin Tomitch, Dr. Andrew

Vande Moere, Dr. Xiangyu Wang, Kazjon Grace, Rui Wang, David Bartollo, Michael Sarroff and Julie Whickers.

References

1. Hii, J. C., Zhou, Z. Y., Karlekar, J., Schneider, M., Lu, W. Q. (2009) Outdoor

Mobile Augmented Reality For Past and Future On-Site Architectural

Visualizations. T. Tidafi, T. Dorta, Joining Languages Cultures and Visions (pp.

557-571), Montreal: University De Montreal.

2. Barton, J., Plume, J., Parolin, B. (2005) Public Participation in a Spatial

Decision Support System for Public Housing. Computers, Environment and Urban Systems, 630-652.

3. DSL-NUS . (2009) Design Space Lab, Department of Architecture. Retrieved

January 5, 2011, from National University of Singapore:

http://www.arch.nus.edu.sg/dsl/index.html.

4. Carson, L., White, S., Hendriks, C., Palmer, J. (2002) Community Consultation

in Environmental Policy Making. The Drawing Board: An Australian Review of

Public Affairs (School of Economics and Political Science), University of

Sydney, 1-13.

5. Carson, L., Hartz-Karp, J. (2005) Adapting And Combining Deliberative

Designs. Retrieved August 2010, from 21stcenturydialogue.com:

http://www.21stcenturydialogue.com/resources/GastilChapter.pdf 6. Hanzl, M. (2007) Information Technology as a Tool for Public Participation in

Urban Planning: A Review of Experiments and Potentials. Design Studies, (28)

289-307.

17

7. Coates, B., Kavanagh, D., Judd, B., Unsworth, L. (2008) Telling the Story:

Community Engagement in Bonnyrigg. Sydney: Housing NSW (Bonnyrigg

Living Communities Project).

8. Bates-Brkljac, N., Counsell, J. (2007) Issues in Use of Computer Visualisation

of Large-Scale Urban Developments as Planning Support Tools. Real Corp 007

Proceedings (pp. 59-69). Vienna: Real Corp.

9. Vlahakis, V., Karigiannis, J., Tsotros, M., Gounaris, M., Almeida, L., Stricker,

D., Gleue, Christou, I. T., Carlucci, R., Ioannidis, N. (2001). Archeoguide: First

Results of an Augmented Reality, Mobile Computing System in Cultural Heritage Sites. Proceedings of the 2001 Conference on Virtual Reality,

Archeology, and Cultural Heritage (pp. 131-140). Athens: ACM, New York.

10. Papagiannakis, G., Schertenleib, S., O’Kennedy, B. (2005). Mixing Virtual and

Real Scenes in the Site of Ancient Pompeii. Computer Animation and Virtual

Worlds, (16) 11-24.

11. Tonn, C., Petzold, F., Bimber, O., Grundhofer, A., Donath, D. (2008) Spatial

Augmented Reality for Architecture – Designing and Planning With and Within

Existing Buildings. International Journal of Architectural Computing, 41-58.

12. Santos, P., Acri, D., Gierlinger, T., Schmedt, H., Stork, A. (2010) Supporting

Outdoor Mixed Reality Applications for Architecture and Cultural Heritage.

Proceedings of the 2010 Spring Simulation Multiconference (p. Article No. 190) Orlando: ACM, New York.

13. Torpus, J. (2010) Lifeclipper3. Retrieved January 8, 2011, from

http://www.lifeclipper3.torpus.com/

14. Layar. (2010) See the World. Retrieved May 2010, from Layar:

http://www.layar.com/

15. Foni, A. E., Papagiannakis, G., Magnenat-Thalmann, N. (2010) A Taxonomy of

Visualization Strategies for Cultural Heritage Applications. ACM Journal on

Computing and Cultural Heritage, Vol. 3, No. 1, Article 1.

16. Smallman, H. S., St. John, M., Oonk, H. M., Cowen, M. B. (2001) Information

Availability in 2D and 3D Displays. Applied Perception, 51-58.

17. Nowell, L., Schulman, R., Hix, D. (2002) Graphical Encoding for Information Visualization: an Empirical Study. IEEE Symposium on Information

Visualization, 2002 (pp. 43-50) Boston: IEEE .

18. Marchak, F. M., Cleveland, W. S., Rogowitz, B. E., Wickens, C. D. (1993) The

Psychology of Visualization. In Proceedings of the 4th conference on

Visualization 1993 (pp. 351-354) Washington DC: IEEE Computer Society.

19. Pantförder, D., Vogel-Heuser, B. (2009) Benefit and Evaluation of Interactive

3D Process Data Visualization in Operator Training of Plant Manufacturing

Industry. Proceedings of the 2009 IEEE International Conference on Systems,

Man and Cybernetics (pp. 824-829) San Antonio: IEEE Press.

20. Neto, P. L. (2006) Public Perception in Contemporary Portugal: The Digital

Representation of Space. Journal of Urban Design, 11(3) 347 — 366.

21. Estrin, D. (2010) Participatory Sensing: Applications and Architecture [Internet Predictions]. Internet Computing, IEEE , 14(1)12-42.

22. Kingston, R., Carver, S., A, E., Turton, I. (2000) Web-based public

participation geographical information systems: an aid to local environmental

decision-making. Computers, Environment and Urban Systems, 109-125.

23. Minton, A. (2009) Ground Control – Synopsis. Retrieved January 19, 2011,

from Anna Minton: http://www.annaminton.com/synopsis.htm

18

24. Kadlec, A. (2007) Dewey's Critical Pragmatism. Lexington Books.

25. Cheng, K. (2010) Augmented Reality: Is It Real? Should We Care? Retrieved

April 2010, from kev/null: http://kevnull.com/2010/02/augmented-reality-is-it-

real-should-we-care.html

26. Mansfield, J. (2009) Design Principles. Retrieved July 2010, from James

Mansfield: http://www.jamesmansfield.id.au/design-principles/

27. Verplank, B. (2005) IxDSketchBook. Retrieved from Bill Verplank:

http://www.billverplank.com/IxDSketchBook.pdf

28. Cuthill, M. (2001) Developing local government policy and processes for community consultation and participation. Urban Policy and Research, 19: 2,

183 — 202.

29. Moritz, F., Meinel, C. (2010) Mobile Web Usability Evaluation - Combining

the Modified Think Aloud Method with the Testing of Emotional, Cognitive

and Conative Aspects of the Usage of a Web Application. ACIS 9th

International Conference on Computer and Information Science, 2010, (pp.

International Conference on Computer and Information Science, 2010).

30. Strauss, A. L., Corbin, J. M. (1998) Basics of qualitative research: techniques

and procedures for developing grounded theory. SAGE.

31. Barber, S. (2010) Investigation vs. Validation Retrieved October 2010, from

PerfTestPlus, Inc.: http://www.perftestplus.com/resources/014PeakPerf.pdf