THE USE OF 3D VISUALISATION AND INTERACTION TO

MARKET A NEW NEAR-NET-SHAPE PIN TOOLING

APPLICATION

Kevin Badni

Loughborough University, Department of Design & Technology, Loughborough, U.K.

Keywords: 3D, collision detection, STL parsing, interaction, Director, Pin Tooling.

Abstract: This paper describes the methodologies used to create an interactive 3D multimedia application. The

application was required to fulfil a business need of representing as close to reality as possible, how a new

product technology worked. A number of IT solutions for the application were researched and described

within this paper. The construction of the application is then discussed with the methodology behind a

stereolithography parsing system and the algorithms used to detect collisions being described in detail. The

application has been successfully implemented by the business to build an engaged client basis.

1 INTRODUCTION

Surface Generation is a privately owned UK

company which has a revolutionary Pin Tool, mould

making, process with intention to ‘fundamentally

alter the economics of small volume production of

large products’. (Surface Generation 2007).

Figure 1: Surface Generation’s Pin Tooling. Copyright

Surface Generation 2004-2007.

The Near-net-shape Pin Tooling (NPT)

technology that the company wanted to promote

uses beds of adjustable square pins to create near net

shape surfaces (see Figure 1). These beds are

designed for use as a support structure for composite

tooling systems. Their approach has been designed

to reduce lead time and labour issues which are at

the centre of large component manufacture. Due to

the unique nature of this product Surface Generation

wanted to use information technology as a marketing

device to demonstrate how their unique technology

worked.

The proposed application was to allow the user

to ‘play’ with the technology and so understand by

experience how the technology worked and how the

users decisions could influence the final results. The

hypothesis that interacting with a virtual product

would allow for a customer to gain a better

understanding of the technology involved is backed

up by existing research as described by Schlosser

(2003) ‘The act of directly manipulating a virtual

object should produce clearer mental images than if

this information were acquired passively, regardless

of the user's goal’. This is supported by Rosenblatt

(1965) who describes mental imagery as being a

critical aspect of information processing. The

proposed application would be designed to mirror

the real life technology with the intention to

persuade the customer that the new technology is a

feasible option for them to purchase. To facilitate

this Biocca (1992), states that in theory the

application should “blur the "boundary between

simulation and communication, and virtual and

physical reality". As discussed by Kahneman and

Tversky (1982), the easier and more vividly

individuals can envision a scenario, the higher the

likelihood that they will estimate that the scenario

will actually occur.

325

Badni K. (2008).

THE USE OF 3D VISUALISATION AND INTERACTION TO MARKET A NEW NEAR-NET-SHAPE PIN TOOLING APPLICATION.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - HCI, pages 325-330

DOI: 10.5220/0001678403250330

 SciTePress

This technique of representing the experience by

using vivid mental imagery rather than expecting the

customer to use cognitive elaboration has been

proposed to be a superior method in influencing

intentions (MacInnis and Price 1987). Whilst the

creation of an application that requires little

cognitive elaboration is ideal, it must be

acknowledged that a web based 3D design is always

a compromise between realism and what can be

displayed considering computer software, hardware

and bandwidth. As stated by Davis (1996) complete

imitation is impossible as it is inevitable that some

process of subtraction and modification of

representation is always required. Nevertheless

from a marketing perspective it can be argued that

the more a customer can envisage themselves using

a technology, the more influential it will be in

affecting their expectations to purchase.

2 INFORMATION

TECHNOLOGY OPTIONS

The author was asked to research into possible IT

solutions which could sit on Surface Generation’s

website. The application was to be up and running

within a tight deadline of only three months. The IT

application required specific actions to be

undertaken which generally mirrored the real-life

process used by Surface Generation, these are set out

in Table 1.

Table 1: Application Process.

Actual System Virtual System

Load STL files select STL file

create 3D model

pin bed reset create pin bed

run collision detection run collision detection

position pins position pins

create infill create mould volume

To facilitate these operations a number of

approaches were considered. As the application was

to represent the real-life technology and hardware so

reduce the need for cognitive elaboration, a 3D

virtual model would need to be created. Due to the

application revolving around a virtual 3D model the

first obvious IT choice to be considered was VRML

(Virtual Reality Mark-up Language).

2.1 VRML

VRML is a world standard for web based virtual

reality interfaces. VRML is a scene description

language based on a subset of Open Inventor from

Silicon Graphics that describes the geometry and the

internal behaviour of a 3D scene. Even though

VRML is not a general programming language it

does have some features that can make it a useful

tool for dynamic 3D environments. These features

include the ability to write prototype nodes which

can incorporate scripts to perform various

behaviours within the VRML world. With the

creation of this standard a number of dedicated web

browsers have been produced which are used to

display and control virtual worlds using VRML over

the internet. These browsers are cross platform

compatible and interpret the VRML files displaying

a 3D model within the browser (Badni 2005).

Having investigated the use of VRML as an

application base, a number of negative issues were

raised. VRML is mainly used to visualise 3D

models, it is not normally used to facilitate heavy

computational tasks such as the required collision

detection. VRML requires a dedicated browser to be

used, so the customer would have to download and

install a browser. VRML browser’s rendering

engines would also struggle with the number of

polygons used within the imported STL

(stereolithography) files, making the updated frame

rate rather slow. The frames per second rate (FPS)

is affected by the processor speed, graphics card and

memory available on the customers PC. As this is

unknown, the 3D designer must design the 3D scene

to be as computationally light as possible yet still

maintain a high level of model realism. The 3D

model files would therefore have to contain the

minimum number of polygons (facet shapes that

make up a virtual model) as possible. However as

the application would use complicated STL files

with a large number of polygons (>2500) this would

not always be possible. Having undertaken some

experiments it was clear that having more than 2,500

polygons displayed would reduce the frame per

second rate below the acceptable 10fps rate.

Hartman and Wernecke (1996) recommend that the

designer should aim at achieving a minimum frame

rate of 10 FPS, which is the frame rate required to

simulate visual continuity whilst moving in the

virtual world.

ICEIS 2008 - International Conference on Enterprise Information Systems

326

2.2 Java

The second IT option was to create a bespoke

application using a high level programming

language such as Java which was formalised in

1995. Because the Java language is an interpreted

language, it requires an interpreter (also known as a

Just-in-time compiler) in order to run, so as long as

the customer has ‘The Java Virtual Machine’ on

their PC, the application will theoretically run. The

main advantage that a proposed Java application

had, was due to the nature of Java programming the

computational side of the application could be

relatively easily handled. The major downside of

creating a bespoke application using Java is the

enormous time scale that would be involved. Surface

Generation wanted a working application within

three months which ruled out a bespoke application

being created by just the author.

2.3 Director

The third option, and chosen route was to use

Macromedia’s Director authoring environment.

(now owned by Adobe). Director has an object-

oriented development environment, similar but not

as powerful as Java but much more powerful than

VRML. Director also has a pre built 3D rendering

engine so can handle 3D model files. The other

advantage Director has is the common usage of the

Shockwave plug-in, as over 200 million web users

(58.5% of mature markets, Millward Brown 2007)

having already installed the Shockwave Player.

3 APPLICATION

DEVELOPMENT

3.1 Reading STL Files

The first stage of the application was the reading of

imported STL files. The real life NPT application

uses customer’s STL files to set up the pin positions.

This was to be mirrored in the virtual web

application by selecting from a library of different

STL files. An STL file is a triangular facet based

representation that approximates surface and solid

forms. An STL file as described by Burns (1993)

‘consists of a list of facet data. Each facet is

uniquely identified by a unit normal (a line

perpendicular to the triangle and with a length of

1.0) and by three vertices (corners). The normal and

each vertex are specified by three coordinates each,

so there is a total of twelve numbers stored for each

facet.’

The standard STL format specifies that the

object represented must be located in an all-positive

octant. In other words, all the vertex coordinates of

the facets must have positive numbers. However,

with a few exceptions most STL creation software

allows the facets to be in any arbitrary location in

3D space. Director also works within a 3D space,

however Director’s co-ordinate system is

unfortunately incompatible with an STL file so

ruling out a straightforward direct import. Therefore,

a new STL parsing system had to be written to

interpret the STL models into the Director

environment.

The first part of the parsing system involved

stripping out each individual vertex (3D points)

position from the ASCII STL text file. The idea then

was to recreate the mesh using Directors own rules.

Within Director a vertices value needs to be entered

in to the code that will indicates the number of

entries that will be expected in a linear list of

vectors, each of which describes the location of one

vertex within the 3D model. This vertex list then has

to be cross referenced with individual facets in order

to work out the connections between the vertices

that would create the final triangles. To achieve this

with the different STL models available from the

library a new algorithm had to be created that could

sit within a repeat loop within the application to

determine which faces from the STL file were based

on which vertices. The method undertaken, involved

creating a breakdown list of a number of sample

STL surfaces, (see Figure 2 and Table 2), listing

each vertex which related to each face. A pattern

emerged for both the odd numbered faces and the

even numbered faces, these have been described

below.

Figure 2: Facet STL surface.

THE USE OF 3D VISUALISATION AND INTERACTION TO MARKET A NEW NEAR-NET-SHAPE PIN TOOLING

APPLICATION

327

Table 2: Vertex List.

Face Number Vertex 1 Vertex 2 Vertex 3

1 1 4 2

2 2 4 5

3 2 5 3

4 3 5 6

5 4 7 5

6 5 7 8

7 5 8 6

8 6 8 9

Even Faces = [Vertex1,Vertex1+the

columns+1,Vertex1+1]

Odd faces = [Vertex1+1,Vertex1+the

columns+1,Vertex1+the columns+2]

However a discrepancy was found with vertices

that were on the edge of a surface (i.e. Vertex 6) .

To circumnavigate this issue the arithmetic modulus

operator was used to check if the vertices was on an

edge:

If Vertex1 mod (columns+1) = 0 then

Vertex1=Vertex1+1

End if

This would only come into effect if the STL file

created was not fully closed, otherwise in theory

there would be no edge vertices.

Once the vertex position and connections had

been resolved they were arranged and positioned in

the 3D space in a counter clockwise order when

looking at the triangular face from the outside using

the right-hand rule with the thumb pointing up being

‘front’ (see Figure 3). This rule was applied to make

sure the ‘front’ or back’ side of the triangle was

correctly orientated.

Figure 3: Front face of facet.

Within Director the Camera property has a front

direction vector, any normal that is between 0 and

90 degrees of that vector will be drawn with the

front face, if it is beyond 90 degrees the back face is

drawn. If the back face were drawn the 3D STL

model would appear to have holes in it and would

not work with the collision detection scripts. Once

the faces had been drawn correctly the final step was

to perform a relative scale on the model so that it

would fit within the pin bed footprint. This was due

to the fact that STL files do not contain any

information on scale as STL units are arbitrary.

3.2 Creation of Pin Beds

The second step was to allow the customer to create

a number of different pin beds. There were two

options available within Director to create the beds.

The first was to build the beds in an external 3D

modelling environment and then import them into

Director. This had the advantage that the different

beds could be quickly called upon with the 3D

environment, however it had the downside

increasing the applications size so increasing

download time. With this is mind the other option

was chosen, that of making the pin bed from within

Director itself (see Figure 4). This had the advantage

of adding very little size to the application only lines

of code.

Figure 4: Creation of Pin Beds.

3.3 Collision Detection

Once the 3D model was in place and the pin bed had

been built, the next stage was to move the pins in a

vertical vector to their Near-Net-Shape positions.

To facilitate this, it was necessary to work out which

pins would hit the model and at what vertical

position this would happen. It was originally

ICEIS 2008 - International Conference on Enterprise Information Systems

328

conceived that due to having built a vertex list for

the STL models, a simple comparison of vertex

positions to pin positions could have been used.

However, the STL model could be re-orientated and

repositioned within the Director application so

rendering the list irrelevant. The use of collision

detection was implemented instead. Due to the high

computational requirements of collision detection, it

inevitably incurs a ‘cost’. Costs relating to collision

detection are described by Phelps and Cloutier

(2003) as being a ’certain level of computational

complexity relative to the level of accuracy in

detecting collision. Generally speaking, the more

precise the accuracy of a given methodology, the

more computationally intensive the calculation, and

the larger the drain on the overall program’. To help

reduce this cost a number of algorithms were used

simultaneously within the Director environment.

The first algorithm created a bounding sphere

around the 3D model, to be used to see which pins

were directly below the sphere. This was created by

taking the pixel furthest from the centre of the model

and using the distance vector as a radius. This was a

quick way to see if the sphere was directly over a

pin, if the returned result was negative, then the pin

was ignored, if it was true, the pin was added to a

‘hit list’ of pins that required finer detection work.

The second algorithm undertook the finer

detection work. Taking each of the hit list pins a

form of spatial partitioning was applied to their top

faces. The partitioning split the pin head once

horizontally, and once vertically, dividing the space

into four quadrants of equal size. If the 3D model

was detected within one of the quad regions, that

region was then further subdivided into four more

quadrants providing a double-level detection zone

(see Figure 5). For the larger 150mm pins a third

quad level was introduced to improve accuracy. The

distance from the pin to the 3D model was then held

in an array and overwritten if another part of the 3D

model was closer to that pin, ultimately returning the

lowest collision distance. This algorithm was then

repeated for all lower and upper hit list pins to allow

for the next step of the application to take place.

Figure 5: Quad Partitioning of Pins.

3.4 Pin Movement

Once a lower bed and upper bed collision distance

list had been collated each pin was moved

accordingly to be in contact with the model. The

space left was then in-filled with a moulding

compound, illustrated within the application by a

solid infill block (see Figure 6). The volume of this

infill was then calculated and shown to the user

illustrating the volume of moulding material they

would need in the real world. This fulfilled the

original brief of allowing the customer to affect the

volume calculation outcomes.

Figure 6: Solid Infill (Lighter Colour) Between Upper and

Lower Pin Beds.

THE USE OF 3D VISUALISATION AND INTERACTION TO MARKET A NEW NEAR-NET-SHAPE PIN TOOLING

APPLICATION

329

4 CONCLUSIONS

The creation of an interactive 3D product simulator

was undertaken using the most appropriate

technology available to the author. A number of

options were examined for each stage of the

development, which culminated in a single

application which can be viewed over the web by

any customer who has the Adobe shockwave viewer.

Research has shown that interacting with a

virtual product rather than being exposed to only

passive information is a superior method in

influencing customers intentions. Having a 3D

environment is particularly beneficial to those

customers who lack knowledge of the product or are

unable to vividly imagine the technology as it does

not rely on individuals' self-generated mental

images.

The application is presently being piloted by

Surface Generation who have currently received

positive feedback on the application. This has

enabled them to progress their commercial leads

with these companies, so fulfilling the marketing

objective.

It would be interesting to examine the effect of

having been exposed to virtual interaction with

customers post consumption satisfaction. As the

customer would already be familiar with the

capabilities and options of the technology, would the

real-life technology be easier to use and manipulate?

However it may also result in unrealistic

expectations, so making Surface Generation’s Near-

net-shape Pin tooling system more complicated and

frustrating to use. This would be an interesting area

for further investigation.

REFERENCES

Badni, K.S. (2005), ''Virtual reality design techniques for

web-based historical reconstructions'', Virtual Reality,

9(4), 1st April 2005, p2.

Biocca, Frank (1992), "Communication within Virtual

Reality: Creating a Space for Research," Journal of

Communication, 42 (Autumn), 5-22. p 8

Burns, M (1993). Automated fabrication: Improving

productivity in manufacturing. PTR Prentice Hall.

Davis,S.B. (1996) The DESIGN of Virtual Environments

with particular reference to VRML Report for the

Advisory Group on Computer Graphics, Middlesex

University.p4.p61-65

Hartman, J and Wernecke, J (1996) The VRML2.0

Handbook, building moving worlds on the Web,

Silicon Graphics, Inc. p226

Kahneman, Daniel and Amos Tversky (1982), Judgment

under Uncertainty: Heuristics and Biases, New York:

Cambridge University Press.pp130-154

MacInnis, Deborah J. and Linda L. Price (1987), "The

Role of Imagery in Information Processing: Review

and Extensions," Journal of Consumer Research, 13

(March), pp473-491.

Millward Brown (2007), [Online]. Available:

http://www.adobe.com/products/player_census/metho

dology/ [2007, November 25]

Phelps, A. and Cloutier, A. "Methodologies for Quick

Approximation of 2D Collision Detection Using

Polygon Armatures" by A Phelps and A Cloutier.

Published both at the Directors Online User's Group [

DOUG] and the Macromedia DevNet forum. 2003

Rosenblatt, Louise (1965), Literature as Exploration, New

York: Modern Language Association. Carbondale:

Southern Illinois University Press. pp174-178

Schlosser, Ann E. (2003) “Experiencing Products in the

Virtual World: The Role of Goal and Imagery in

Influencing Attitudes versus Purchase Intentions” The

Journal of consumer research [0093-5301] vol:30 iss:2

p:184

Surface Generation 2007 [Online]. Available

http://www.surfacegeneration.com/iexplorer/history.ht

m [2007, November 25]

ICEIS 2008 - International Conference on Enterprise Information Systems

330