Shape Modeling with Sketched Feature Lines in ... - helen perkunder

Diplomarbeit

Shape Modeling with Sketched Feature Lines in Immersive 3D Environments Helen Perkunder

Gutachter: Prof. Dr. Marc Alexa Prof. Dr. Gesche Joost Betreuer:

Dr. Johann Habakuk Israel (Fraunhofer-Institut IPK, Berlin)

Datum:

22.2.2010

Technische Universität Berlin Fakultät IV - Elektrotechnik und Informatik Institut für Technisch-Naturwissenschaftliche Anwendungen Fachgebiet Computer Graphics

Eidesstattliche Erklärung

Die selbständige und eigenhändige Anfertigung versichere ich an Eides Statt.

Berlin, 22.2.2010

Helen Perkunder

Zusammenfassung

Immersive 3D Umgebungen werden in der Virtuellen Produkt Entwicklung eingesetzt, um Designer im Design-Prozess zu unterstützen. Verschiedene Ansätze konzentrieren sich dabei auf immersive Skizziertechniken, da das Skizzieren eine wichtige Technik in den frühen Phasen des Produkt-Designs ist. Diese Diplomarbeit betrachtet die Frage, ob eine automatische Formerzeugung aus skizzierten Eingabelinien in immersiven 3D Umgebungen den Skizzierprozess während dieser frühen Designstadien unterstützt. Zu diesem Zweck wurden die Modellerzeugungs- und -deformations-Algorithmen der skizzenbasierten Modellieranwendung FiberMesh [NISA07] in eine immersive 3D Umgebung gebracht und die Interaktionstechniken angepasst. Eine vergleichende Benutzeruntersuchung mit zwölf Designstudenten und professionellen Designern wurde durchgeführt. Linienbasiertes Skizzieren in einer 3D Umgebung und skizzenbasiertes Modellieren, letzteres sowohl in einer 3D als auch in einer 2D Umgebung, wurden verglichen. Die Analyse der Studie ergab wenige Unterschiede zwischen den Bedingungen, aber es konnten zwei Ergebnisse gefunden werden: Die Usability für eine kreative Skizzieraufgabe wurde für das linienbasierte Skizzieren in der immersiven 3D Umgebung als höher empfunden als für das entwickelte skizzenbasierte Modellieren in derselben Umgebung. Die Formmodellierung in der immersiven 3D Umgebung wurde als stimulierender und attraktiver als unter 2D Bedingungen empfunden. Die Diplomarbeit wurde am Fraunhofer-Institut für Produktionsanlagen und Konstruktionstechnik (IPK), Berlin geschrieben.

Abstract

Immersive 3D environments are used in virtual product creation to support designers during the design process. Various approaches exist that focus on immersive sketching techniques, as sketching is a prominent technique in the early phases of product design. This thesis addresses the question whether automatic shape creation from sketched input strokes in an immersive 3D environment supports the sketching process in these early design stages. For this purpose, model creation and deformation algorithms of the desktop sketch-based modeling application FiberMesh [NISA07] were transferred to an immersive 3D environment and interaction techniques were adapted. A comparative user study was conducted among twelve design students and professional designers. Comparison was made between line-based sketching in a 3D environment and sketch-based modeling, both in a 3D and 2D environment. The analysis of the study yielded few differences between the conditions, but two findings were made: The usability for a creative sketching task was perceived as higher for line-based sketching in an immersive 3D environment than for the provided sketch-based modeling in the same environment. The shape modeling in the immersive 3D environment was perceived as more stimulating and attractive than under the 2D condition. The thesis was written at Fraunhofer-Institute for Production Systems and Design Technology (IPK), Berlin.

Contents

Zusammenfassung

i

Abstract

ii

1 Introduction

1

2 Background and Related Work 2.1 Purpose of Sketches . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Design as Design Problem Solving . . . . . . . . . . . 2.1.2 Sketching as Externalization and Self-Communication 2.1.3 Design Process . . . . . . . . . . . . . . . . . . . . . 2.2 3D Sketching - Preceding Expert Discussions and User Study 2.3 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

4 4 4 5 6 7 9

. . . . . . . . .

16 16 17 17 18 20 20 23 25 26

3 Approach and Application 3.1 Intention . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Initial System . . . . . . . . . . . . . . . . . . . . . . 3.2.1 SketchApp . . . . . . . . . . . . . . . . . . . . 3.2.2 Tangible User Interface Pen . . . . . . . . . . 3.3 Initial Application FiberMesh . . . . . . . . . . . . . 3.3.1 FiberMesh User Interface . . . . . . . . . . . . 3.3.2 Mesh Creation and Deformation in FiberMesh 3.3.3 OpenMesh . . . . . . . . . . . . . . . . . . . . 3.4 Resulting ImmersiveFiberMesh Application . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

Contents

3.5 4 User 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9

iv

3.4.1 User instructions . . . . . . . . . . 3.4.2 Usage scenario . . . . . . . . . . . 3.4.3 Virtual 3D Object Creation . . . . 3.4.4 Removal of Intersections and Loops 3.4.5 Extension to Multiple Instances . . 3.4.6 Color Palette . . . . . . . . . . . . 3.4.7 Deformation . . . . . . . . . . . . . Discussion regarding Sketching Properties Study Hypotheses . . . . . . . . Evaluated conditions . . . Method of collecting data Participants . . . . . . . . Tasks . . . . . . . . . . . . Procedure of the study . . Evaluation . . . . . . . . . Results . . . . . . . . . . . Discussion of results . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

. . . . . . . . .

. . . . . . . .

28 28 30 32 33 33 33 35

. . . . . . . . .

39 40 40 42 46 47 48 49 50 54

5 Discussion and Outlook

57

Bibliography

59

A User Study

67

CHAPTER 1 Introduction

Cave paintings are thought to be an invention of the Cro-Magnons that lived around 32.000 years ago. They developed the ability to draw pictures on the walls of rock caves. In our time, a new ability has been developed: Modern CAVE1 painting is drawing into air. Human beings are three-dimensional creatures that perceive and interact in a three-dimensional world. ”Every thing has a shape. We can see, touch, even hear shape. It is the fundamental concept for interaction with the world we live in” [Ale02]. From early childhood on, humans draw to represent objects or ideas and to invent new ones. Drawings that can be created and perceived as three-dimensional images may offer a new way to reproduce and invent reality. Current research in the field of virtual product development is concerned with sketching. Sketching is an important method in product development and engineering design, especially during the early design phases ([R¨02], [HSS98]). In this context, sketches serve as an externalization of mental concepts [Tve02]. Moreover, various authors describe sketching as a reflective process of self-communication, in which the designer draws a sketch, reflects on the drawn image and generates new ideas while working with the sketch ([Tve02], [HSS98], [Bux07]).

1

CAVE: Cave Automatic Virtual Environment

1 Introduction

2

This thesis addresses the question whether automatic shape creation from sketched input strokes in an immersive 3D environment supports the sketching process in the early design phases. For this purpose the model creation and deformation algorithms of the desktop sketch-based modeling tool FiberMesh [NISA07] were combined with the input technique and visual perception of an immersive 3D environment. Sketch-based modeling is a field of research that specifically addresses the problem of 3D shape creation from sketched input. The developed application combines significant aspects of the initial system [IWM+ 09] and FiberMesh: The immersive system provides spatial perception and spatial interaction and FiberMesh provides an easy and fast way of creating virtual 3D objects. The thesis is based on recent studies on sketching in 3D environments. Israel et al. [IWM+ 09] conducted a focus group expert discussion on the potentials and limitations of 3D sketching in immersive virtual environments that revealed expectations of experts towards 3D sketching media. A follow-up comparative user study among designers on line-based sketching in immersive 3D environments showed that designers see an advantage for the sketching process in such environments, especially with regard to the system’s ability to foster inspiration and spatial thinking [IWM+ 09] (see chapter 2.2). To investigate whether the applied approach of sketch-based modeling in an immersive 3D environment can provide further support to the sketching process in early design-phases, a comparative user study among twelve design students and professionals was conducted. The thesis is sectioned as follows: Chapter 2 describes the purpose of sketches for designers and two recent

1 Introduction

3

studies on 3D sketching. Additionally, this chapter introduces related work to this thesis. Chapter 3 describes the initial immersive system and the application FiberMesh and introduces the integrated approach ImmersiveFiberMesh. Chapter 4 describes the conducted user study among design students and professionals. Chapter 5 concludes the thesis with a discussion of the presented work and an outlook.

CHAPTER 2 Background and Related Work

This chapter describes the purpose of sketches with regard to the design process and introduces preceding user studies on 3D sketching. Following that, further work related to this thesis is introduced.

2.1 Purpose of Sketches This section describes the purpose sketches serve for designers in relation to the design process. Designers use sketches to reduce the mental workload during the solution of a design problem and sketches serve as a means of self-communication. Sketches are an important design tool during the early phases of product design.

2.1.1 Design as Design Problem Solving A problem can be described using three components: The initial state, the desired target state and the set of operators to gain the target state from the initial state [R¨02]. Problems can be well or ill defined. Considering well defined problems, all three components are exactly specified. The solution of these problems consists in applying the operations correctly and in correct order. On the other hand, in ill-defined problems the three components are indistinct and vague [R¨02].

2.1 Purpose of Sketches

5

Designing work is ”design problem solving”. The designer thinks about and develops a mental model of a non-existing future object, which additionally should have useful new features [HSS98]. Design problems are often ill defined, complex problems [R¨02]. A correct and complete mental representation is needed as a search space in order to solve a problem. Ideally a designer should be able to select the optimal feature pattern of the solution from the Cartesian product of all possible features. But human working memory is limited and is therefore a bottleneck in the process of creative design [HSS98]. Human working memory has a key role in the mental processing of information. In research this part of the memory is either considered as the activated part of the long-term memory or as an independent cognitive structure for medium-term memorizing and processing of information. It can be seen as a functional unit that both processes relevant information and keeps it activated. These two mechanisms require the same resources, resulting in a trade-off relation: A high memory workload reduces the quality of the processing of the information. Conversely, while processing difficult information the activation of this information can only be maintained to a certain degree [Sac01]. Designers use sketches to relieve memory workload.

2.1.2 Sketching as Externalization and Self-Communication Sketches can be seen as an external storage and sketching as an externalization of ideas and mental images. Sketches ”are a kind of external representation serving as a cognitive tool to augment memory and information processing by relieving the mind of some of those burdens.” [Tve02] Tversky points out that written language can have the same function, but sketches convey visuospatial ideas directly [Tve02]. Additionally to the benefit of externalization of mental concepts, sketches are used as a means of self-communication. Designers develop new ideas

2.1 Purpose of Sketches

6

while sketching. They externalize an idea and then use the sketch as an inspiration for further ideas ([Tve02], [HSS98], [Bux07]). ”Sketchers make sketches with certain ideas and goals in mind, but fortuitously, may see new objects and configurations in their sketches.” [Tve02] This leads to an iteration of externalization and reflection on the sketch. Designers act and perceive in turns, taking the sketch as a means of self-communication. Figure 2.1 illustrates this process.

Figure 2.1: Sketching as a process of externalization and reflection resulting in self-communication [IWM+ 09]

2.1.3 Design Process A design process passes different states. Pahl et al. classify the design problem-solving process into four main phases [PBFG07]: 1. identification and clarification of the problem, 2. development of a frame conception of solutions, 3. design of the favored solution, and 4. working out the details (translation from [HSS98]). At the beginning of the design-process designers have an indistinct idea of the functionality and appearance of a product. Sketches are used to concretize and to complete the initial vague idea [Doe98]. As distinct from modeling with CAD tools, sketching is more in accordance with creative thinking and acting

2.2 3D Sketching - Preceding Expert Discussions and User Study

7

[Pet01]. Consequently, sketches are an important design tool during the early phases of product design that are primarily the identification and clarification of the problem and the development of a frame conception of solutions. Those early phases are believed to be most important for innovativeness and costs of product development [HSS98].

2.2 3D Sketching - Preceding Expert Discussions and User Study 3D sketches that can be drawn immediately into space differ considerably from 2D sketches, engineering and CAD drawings concerning creation and information content [IZ07]. This thesis derives from studies on 3D sketching conducted at Fraunhofer-Institute IPK, Berlin. To investigate potentials and limitations of 3D sketching, Israel et al. led two focus group interviews1 and conducted a comparative user study among designers [IWM+ 09]. The focus group interviews among 14 design experts revealed expectations of the professionals concerning immersive 3D sketching. Four key advantages were expected in comparison to traditional sketching methods: • Spatiality: The possibility to work in space directly. • One-to-one proportionality: The possibility to draw in one-to-one scale. • Associations: The possibility to integrate existing models into the working process. • Formability: The possibility to manually deform virtual sketches.

1

A focus group is a ”carefully planned discussion designed to obtain perceptions on a defined area of interest in a permissive, nonthreatening environment.” [Kru94]. Participants are selected because they share certain characteristics related to the topic at hand [Kru94].

2.2 3D Sketching - Preceding Expert Discussions and User Study

8

The majority of the experts expected a positive impact on their work. 3D sketching also had a positive emotional connotation. The participants anticipated that sketching on paper will not be replaced but will coexist with 3D sketching [IWM+ 09]. On the basis of these findings Israel et al. conducted a comparative study among 24 students and professionals of furniture and interior design. Sketching in 3D and 2D was compared in the same technical environment. The study intended to find and investigate advantages of immersive 3D space and its additional degrees of freedom for the sketching process [IWM+ 09]. The setting provided line-based sketching in a five-wall Virtual Reality CAVE with magnetic tracking and interaction with a pen. Participants were asked to sketch furniture in 2D and in 3D in the CAVE. The 2D condition was performed on an imaginary wall in the center of the CAVE [IWM+ 09]. All dimensions of the questionnaire used in this study, AttrakDiff21 [HBK03] (user-perceived pragmatic quality, the hedonic qualities identification and stimulation and the overall attraction), showed significant preferences of the users in favor of 3D sketching. Externalization, self-communication and speed of realization of ideas were rated significantly higher for the 3D sketching condition. A qualitative content analysis of user statements showed that the most prominent benefits of 3D sketching was seen in the sketching technique itself, underlining especially the system’s ability to foster inspiration and spatial recognition and thinking. The main drawbacks of the 3D immersive system were seen in the sensorimotor control functions (e.g. in drawing straight lines and finding connection points) and the lack of haptic feedback. The drawing skills improved over time during the experiment, which can be seen as a clue that training would soften these difficulties [IWM+ 09].

1

For a detailed description of the questionnaire AttrakDiff2 see section 4.3.

2.3 Related Work

9

The results of these studies suggest that designers see subjective advantages in a 3D sketching technique and accept it as a new sketching medium.

2.3 Related Work The work presented in this thesis combines automatic virtual object generation from input strokes with the potential of a 3D immersive environment. Related work comes from the field of sketch-based modeling and drawing in virtual reality environments. This section introduces some relevant work in these fields. Sketch-based modeling is a method to transfer drawing input into 3D models. There is a diversity of approaches in this current field of research. Motivation comes from the interest to support early design phases in which sketching is a major tool. The focus of many researchers is set on intuitive and simplified modeling techniques that avoid extensive parameter input [CA09]. An early approach of sketch-based modeling is SKETCH [ZHH96]. The approach could be called ’Gesture Created Primitives’ [CA09]. Applications of that kind interpret drawing input as gestures or symbolic instructions to determine a user-intended primitive 3D object. SKETCH uses a gestural interface based on simplified line drawings of primitives. After the user sketches the salient features of a variety of 3D primitives, SKETCH places the corresponding geometry in the scene [ZHH96]. Most of the scenes are rectilinear. Figure 2.2 shows two scenes created with SKETCH. Another example of this category is Chateau [IH07]. In Igarashi and Hughes’s ’suggestive interface’, the user gives hints to the system about a desired operation for the construction of a 3D drawing. This is done by highlighting graphical components in the scene. A set of suggestion engines analyze the input, infer possible operations and suggest corresponding results as thumbnails [IH07]. Figure 2.3 shows a screenshot of the prototype Chateau.

2.3 Related Work

10

Figure 2.2: Two scenes created with SKETCH [ZHH96]

Figure 2.3: A screenshot of Chateau: hints of the user (red lines) and suggested operations (thumbnails at bottom) [IH07]

These approaches are based on primitive forms, a class of objects that are unlike the irregular or fluent forms of sketches. The interactive process of giving hints to the system answered by suggested results may also not be ideal to support the fluency of sketching. Instead of self-communicating by means of the sketch, the user communicates with the program. The process of sketching is constantly interrupted in a pattern of call and response, where the user chooses from a set of suggestions. This procedure may hinder a flow of creation. Cherlin et al. [CSSJ05] developed a desktop application that simulates 2D drawing techniques. A common technique for 2D sketches is to first draw the constructive curves (i.e. the silhouette) of an object followed by spiral strokes that connect the constructive curves to create a volume. Another method

2.3 Related Work

11

is to draw single strokes onto the surface to depict sharp contours. Cherlin et al. call these methods ’spiral method’ and ’scribble method’ respectively [CSSJ05]. To simulate the spiral method, they generate a rotational blending surface from two constructive curves. Their approach is to combine the surface of revolution (i.e. a surface generated by rotating a two-dimensional curve about an axis) and the ruled surface (a surface is ruled if through every point on it there is a straight line that lies on the surface, e.g. a cylinder or a cone) in order to find the parametric description of a rotational blending surface. To achieve this, they move and scale a circle along the midpoints of the two constructive curves. Figure 2.4 illustrates this method.

Figure 2.4: Rotational blending. Left: Constructive curves (green), center curve (blue) and scaled circle. Right: completed surface [CSSJ05]

To apply the scribble method, the user draws a cross-sectional stroke (i.e. the relief) onto the constructive curves of the object. For the rotational blending, the shape of the relief is moved along and scaled within the constructive curves. Shapes can be deformed using stroke-based deformation [CSSJ05]. Figure 2.5 illustrates the scribble method.

2.3 Related Work

12

Figure 2.5: Scribble method. From left to right: two constructive strokes (black), a cross sectional stroke (red), result in front and side view [CSSJ05]

This method allows for a wide range of objects. The user controls the form and level of detail of the model by additional strokes. This method would be promising to use in an immersive 3D environment. Keefe et al. regard drawing in virtual reality environments from an artistic point of view. CavePainting [KFM+ 01] is a system for artistic painting in a CAVE. Various types of 3D brushstrokes are featured. These strokes are layered and arranged in space to compose a 3D scene, in the same way that brushstrokes are added to a canvas (figure 2.6).

Figure 2.6: A Painting with CavePainting: Wedding Day - Daniel Keefe [KFM+ 01]

2.3 Related Work

13

CavePainting emphasizes artistic use in providing different artistic input devices and sophisticated color and stroke choosers. The application focuses on a diversity of input strokes but does not feature object-creation. FreeDrawer [WS01] is an application for the Responsive Workbench that generates spline-based free-form surfaces. It is based on a curve network that serves as the outlines of an exoskeleton that describe the shape of an object. Curves are drawn directly into space with a pen and new curves are woven automatically into the existing curve network. The user can then add surfaces in each closed loop within the curve network by pointing with the pen to a segment of the respective loop (figure 2.7) [WS01].

Figure 2.7: FreeDrawer: Curves and added surfaces [WS01]

This is an approach of sketch-based modeling in a virtual 3D environment. The design procedure is divided into two phases: creating the exoskeleton and then filling the surfaces. This can be seen as a technique that corresponds to drawing with pen and paper: first sketching the outline of a shape and then coloring in the surfaces. A range of systems use an input stroke to create rotund objects. This approach can be called ’Blobby Inflation’ [CA09] or ’Blob editing systems’ [CSSJ05]. A prominent example of this category is Teddy [IMT99]. Teddy is run as a desktop application. Figure 2.8 shows the user interface of Teddy.

2.3 Related Work

14

Figure 2.8: Teddy user interface, run on a graphical tablet [IMT99]

For mesh construction, a 2D input stroke is resampled to a smooth polyline. The polyline is closed to a polygon and triangulated using Constrained Delaunay Triangulation. In order to inflate the mesh, a spine is found and each vertex of the spine is elevated in proportion to the average of edge-lengths to connected boundary-vertices (figure 2.9 a, b). The connecting edges are converted to quarter ovals (figure 2.9 c) and neighboring elevated edges are then sewn together (figure 2.9 d).

Figure 2.9: Teddy: Mesh construction [IMT99]

The generated mesh is copied to the other side. Models are topologically equivalent to a sphere. Extrusion, cut, smoothing and bending are featured. Teddy was designed for rapid construction of approximate models. The resulting models have a ”hand-crafted feel” [IMT99]. FiberMesh [NISA07] follows the same basic idea but uses different algorithms.

2.3 Related Work

15

In FiberMesh the mesh is defined by the initial and added input-strokes which stay on the mesh and serve as deformation handles (a detailed description follows in chapter 3.3 ’Initial Application FiberMesh’). Blobby inflation is straightforward and does what can be associated with the name ’sketch-based-modeling’, ”the ability to draw an object and have it literally pop into 3-dimensions.” [CA09] In this thesis this approach was used for object creation. Its straightforwardness is a quality that was anticipated to be supportive to the design process in immersive 3D environments.

CHAPTER 3 Approach and Application

The software developed for this thesis integrates algorithms of the sketchbased modeling tool FiberMesh [NISA07] into an existing immersive sketching system [IWM+ 09]. After outlining the intention of the approach, this chapter describes FiberMesh and the initial system and introduces the integrated application. Finally, the properties of the system with respect to the sketching process are discussed.

3.1 Intention The purpose of this thesis was to combine sketch-based modeling with the visual perception and interaction provided by an immersive 3D environment, in order to investigate whether this approach can support early design phases. In the field of virtual product creation, immersive 3D environments are used in order to provide tools for engineering design that support the designer during the whole design process. Special attention is given to the early design phases as they are considered to have great impact on quality and costs of product development [HSS98]. During these early stages of product design, sketches are a major tool for designers to generate and concretize ideas (cf. chapter 2.1). Sketch-based modeling is a current field of research that addresses the attempt to complement traditional sketching techniques with immediate model creation [CA09].

3.2 Initial System

17

The combination aims at an enhancement of self-communication during the sketching process, as the act of sketching can be regarded as an iteration of externalization and reflection, resulting in self-communication (cf. chapter 2.1.2). Sketching in immersive 3D environments alters both, the externalization (i.e. the act of drawing) and the reflection (i.e. the visual perception of the sketch), in a profound way. The approach of this thesis to further enhance externalization was to provide automatic sketch-based object creation. When creating a 3D model from the input of a simple stroke, the system generates information, which speeds up the process of creating 3D objects. This increase in speed may support the sketching process, as quickness can be seen as an important property of sketching (cf. [Bux07], chapter 3.5).

3.2 Initial System The application developed for this thesis is based on an existing immersive sketching system. This section introduces the initial system that consists of the application SketchApp [IWM+ 09] and the input devices of the system. SketchApp was also used as a condition of the comparative user study conducted for this thesis. The technical equipment consists of a cubiform five-wall Virtual Reality CAVE with rear projection and an edge length of 2.5 meters. The system uses active stereo LCD shutter glasses (CrystalEyes) and magnetic tracking (Ascension MotionStar).

3.2.1 SketchApp The SketchApp application provides drawing of lines in an immersive 3D environment. It was developed at Fraunhofer-Institute IPK in order to investigate whether line-based sketching in a 3D environment supports the early design process.

3.2 Initial System

18

The input device of SketchApp is a pressure-sensitive pen. Strokes are generated using strings of quads, altering side-lengths according to the pressure on the pen. Data from the tracking system is smoothed. The color of the stroke can be altered by keyboard input and loading of external models is supported. A physical slider is used to undo input operations. The slider is located on a table outside the CAVE. SketchApp makes use of OpenSG as a scenegraph system [Ope09]. OpenSG is a development for virtual reality applications and is based on OpenGL.

3.2.2 Tangible User Interface Pen The input device of the system is a pen that belongs to the category of Tangible User Interfaces (TUIs). Ishi et al. describe TUIs as follows: ”The key idea of TUIs is to give physical forms to digital information. The physical forms serve as both representations and controls for their digital counterparts. TUI makes digital information directly manipulatable with our hands, and perceptible through our peripheral senses by physically embodying it.” [Ish08] TUIs can be regarded as an essential part of the system, providing a physical access to the task at hand. To borrow the embodiment of the TUI from everyday life can contribute to making the use of the system intuitive by helping users to understand the usage ’with their hands’. With regard to this, the system makes use of the pencil and paper metaphor. Metaphors based on learned behavior can help to transfer the use of a device to a subconscious level, which is a prerequisite for efficient and intuitive interaction [HI07]. Hurtienne and Israel give a definition of intuitive use: ”A technical system is intuitively usable if the user’s subconscious application of knowledge leads to effective interaction.” [HI07] A task can be divided into two subtasks: The task at hand which the user wants to accomplish (e.g. to create a sketch), and the interaction task the user has to perform to achieve the desired result (e.g. to take a pen and draw). The use of different systems to perform a task at hand can result in

3.2 Initial System

19

markedly different interaction tasks. Ideally, all mental resources should be available to perform the task at hand. If the interaction task is performed on a subconscious level, mental resources can be used for the present task [IHP+ 09]. The pencil and paper metaphor helps to activate subconscious knowledge because it is based on drawing as an ability learned by most humans from early childhood on. It is an evident choice for a sketching application. Pache and Lindemann even regard the use of this metaphor as a requirement for immersive 3D sketching tools [PL03]. Although the ability of sketching on paper has to be transformed to 3D space, the metaphor could offer a key to the usage of the system. Drascic and Milgram describe a perceptional problem, the accommodationvergence conflict: even though the stereoscopic images are perceived in the middle of the CAVE, the eyes focus on the walls of the CAVE where the images that create the stereoscopic impression are displayed [DM96]. When drawing with a physical pen, the eyes try to focus at the same time on the physical pen and the extracted stroke that is displayed on the walls. To avoid this perceptional problem, an additional virtual model of the pen is displayed. The model of the pen and the drawing can now be focused as a whole. The pen employs a forcesensor, enabling the user to draw wider lines when pressing harder. Figure 3.1 shows the pen.

Figure 3.1: Tangible user interface pen

3.3 Initial Application FiberMesh

20

3.3 Initial Application FiberMesh The application for this thesis uses algorithms of FiberMesh [NISA07] for shape creation and deformation. This section describes the user interface of FiberMesh and gives a short introduction to the mesh creation and deformation algorithms. FiberMesh was also used as a condition of the comparative user study conducted for this thesis.

3.3.1 FiberMesh User Interface FiberMesh [NISA07] is a system for freeform modeling, based on silhouette sketching. It creates and interactively changes a 3D model from 2D input strokes. The user’s strokes remain visible on the model and serve as handles to change the geometry of the model [NISA07]. The mesh is internally represented as a triangle mesh of OpenMesh [Kob09]. OpenMesh uses a halfedge data structure. FiberMesh features the creation of a single model that can be manipulated. Figure 3.2 illustrates the main operations of FiberMesh.


21

Figure 3.2: Operations of FiberMesh (from top to bottom): creation, deformation, extrusion, cut and tunnel [NISA07] (altered)

The FiberMesh interface provides five main operations: Creation of an object Initially a silhouette is drawn onto the canvas. When the user releases the mouse-button, a rotund 3D model is calculated and displayed. Deformation of an object To deform the model the user pulls at a curve. The size of the region of interest to be deformed is proportional to the amount of pulling and propagates to connected curves. Extrusion of an object To create an extrusion, the user first draws a closed stroke onto the model


22

and then rotates the model to finally draw the side view of the intended extrusion. Cutting of an object To cut, the user draws a stroke across the model and indicates by mouse click on the model which part should be deleted. The implicit assumption is that the cut-plane has no gradient in the z-axis of the canvas. Thus, to cut the object in the intended way the model has to be rotated. Tunnel an object To tunnel, the user draws two closed strokes on opposite sides of the model. Additionally, the user is enabled to add, smooth and erase strokes and to type-change a stroke between smooth (constraining the surface to be smooth across it) and sharp (placing positional constraints with only C 0 continuity) [NISA07]. Figure 3.3 shows a result of FiberMesh.

Figure 3.3: FiberMesh: Creation from a professional 2D animation artist, modeled in 20 minutes [NISA07]


23

3.3.2 Mesh Creation and Deformation in FiberMesh The mesh-creation algorithm of FiberMesh basically functions as follows: To create a mesh, the initial input stroke is smoothed and divided into sequences of equal length. From this stroke a regular triangle grid is calculated. All vertices in the grid except the ones on the stroke are duplicated and the two grids are elevated by a small ε into opposing directions, ’glued’ together by the boundary vertices. The elevation is applied for the mesh-optimization algorithms to inflate the mesh in the last step of the mesh-creation procedure. It ensures that the results are nontrivial and leads to a faster convergence of the algorithm. The general approach towards interactive surface-based modeling is that the user gives constraints for the shape of an object and the system calculates the result in minimizing a function that represents characteristics of the object to be preserved, e.g. the fairness of the mesh. The ”basic paradigm, finding a suitable representation/discretization for a curve/surface and deforming it such that it minimizes a certain energy functional subject to user-provided constraints, is the key ingredient to all surface-based modeling methods” [Nea07]. Meshes and curves that are used in geometry models are piecewise linear structures. The current mathematical field of discrete differential geometry (DDG) deals with these structures and tries to assign fundamental properties of continuous geometric objects (such as curves and surfaces) to discrete objects (such as piecewise linear curves and polygonal meshes) [GS08]. In piecewise linear curves the first order and second order discrete differentials are L0 = vi − vi−1 L1 = vi −

1 X vj |Ni | j∈Ni


24

where vi are the vertices and Ni are the adjacent vertices to vi . The number of adjacent vertices |Ni | in a closed curve is 2. First order differentials are the edge vectors in the piecewise linear curve. The second order differential is a vector known as uniformly weighted, discrete Laplacian vector. This vector can be seen as a measure for detail and can be interpreted as a curvature-like value [Nea07]. If the piecewise linear curves are irregularly sampled, the uniformly weighted Laplacian vector does not point in normal direction and a weighted term has to be used. For piecewise linear surfaces a corresponding weighted Laplacian vector exists. FiberMesh uses these vectors to represent curvature of curves and surfaces. Figure 3.4 shows the first and second order differentials of a piecewise linear curve.

Figure 3.4: Discrete first- and second-order differentials of a piecewise linear curve [Nea07]

To deform the mesh, the user pulls at a vertex of a control curve. The system then calculates the region of interest to be affected by the deformation. This region is deformed such that the new Laplacian vectors are as close as possible to the original Laplacians [Nea07]. Discrete Laplacians are not invariant with regard to global rotations of the mesh, but for small rotations linearized rotation matrices can be used [SCOL+ 04]. Therefore, the gross rotation is calculated iteratively by concatenation of small delta rotations [Nea07]. Figure 3.5 shows the rotated local coordinate frames after curve deformation.


25

Figure 3.5: Rotated local coordinate frames (red) after curve deformation [Nea07]

The deformation algorithm consists of two main steps: curve deformation and surface optimization. During pulling, these operations are solved sequentially to gain an interactive update of the mesh [Nea07]: 1. The user deforms a curve by pulling at a deformation handle. 2. The curve deformation is calculated using discrete Laplacians. 3. The new curve positions are used as input for surface optimization, which takes the curve positions as positional constraints.

3.3.3 OpenMesh FiberMesh uses the triangle mesh of OpenMesh [Kob09] as internal representation. OpenMesh provides arbitrary polygonal meshes and meshes with triangles only, the latter having the advantage of featuring faster algorithms. OpenMesh is based on a halfedge data structure splitting each edge into two directed halfedges. Halfedge data structures have explicit representations of vertices, faces and halfedges. Connectivity information is stored in the halfedges. OpenMesh provides classes of circulators that encapsulate the iteration over faces, vertices and halfedges. Each halfedge of the data structure stores: 1. a handle to its end-vertex 2. a handle to the face it bounds

3.4 Resulting ImmersiveFiberMesh Application

26

3. a handle to the next halfedge that bounds the same face 4. a handle to the opposite halfedge Figure 3.6 illustrates the halfedge data structure of OpenMesh.

Figure 3.6: Halfedge data structure [Kob09] (altered)

Halfedge data structures are very efficient. Each new entry in the data structure (i.e. vertex, halfedge or face) requires constant memory space. Each query can be done in constant time and is independent of the size of the model.

3.4 Resulting ImmersiveFiberMesh Application The new application ImmersiveFiberMesh combines significant aspects of the applications SketchApp and FiberMesh: • SketchApp and the 3D environment provide spatial perception and spatial interaction, and • FiberMesh provides an easy and fast way of creating and manipulating virtual 3D objects. After giving a broad overview of the approach, this chapter illustrates the functionality of ImmersiveFiberMesh. For this purpose, instructions for new users of the system are listed and a usage scenario is described. Finally, aspects of the implementation are discussed.


27

The functionality of FiberMesh was combined with the spatial visualization and interaction in an immersive 3D environment in an attempt to provide a fast and immersive way to create and judge objects. The approach handles interaction without parameter input at the expense of level of detail. The user’s influence on the first creation of the object is limited to the specification of the silhouette. The rotund shape and the depth of the resulting object is given by the system. To change the depth and shape of an object, deformation can be used. Still, the system in its present state is limited to the category of rotund objects. FiberMesh [NISA07] was designed as a desktop application. Consequently the concept is based on 2D input strokes. This approach was not changed in this thesis and FiberMesh was used as a software module. With regard to this the resulting application has the function of a software interface. It recalculates the input from the 3D input devices and transfers the result to the FiberMesh algorithm. The response of FiberMesh is then transfered to OpenSG to be displayed in the CAVE. Figure 3.7 illustrates the interplay of the components of ImmersiveFiberMesh.

Figure 3.7: Components of ImmersiveFiberMesh

The essential functionality of mesh creation and mesh deformation were made available in the immersive 3D environment. For this purpose, the essential classes of FiberMesh were integrated into ImmersiveFiberMesh. Mesh


28

creation was extended to multiple instances. Loops and intersections were removed from the input to improve interaction.

3.4.1 User instructions To introduce new users to the system, few instructions are needed. A typical introduction to the system contains the following instructions: 1. Usage of the pen The pen should be held in the way of a fountain pen to use fine motor skills. In order to draw a line the pen has to be pressed. Drawing can be done in the whole space of the CAVE. 2. Creation of blobs To create a blob (i.e. a rotund virtual 3D object), its silhouette is drawn roughly in a plane. Releasing the pen results in the display of a blob that matches the input stroke. 3. Deformation The gripper is held like the gripper of a handyman. Gross motor skills are used. Pulling at an input stroke deforms a blob. Strokes can be added by drawing an open stroke near a blob. The stroke has to overlap the initial input stroke of the blob. The stroke is then wrapped around the whole blob and can also be used for deformation. 4. Delete A slider can be used to delete blobs. First the slider works in both directions, making blobs invisible and visible respectively. After drawing anew, invisible blobs are deleted and removed from the history.

3.4.2 Usage scenario The user interacts with the system in first drawing an input stroke with the pen. The system creates a blob (i.e. a rotund object) that is displayed


29

instead of the input stroke. A smoothed and plane-approximated stroke that circulates the blob is displayed to serve as a handle for deformation. The user now either creates another blob or makes use of the deformation with the gripper. A change of the system mode from drawing to deformation is not necessary. Deformation is done by pulling a stroke with the gripper. To refine the deformation, additional strokes can be added by drawing an open stroke near a blob. The stroke is projected onto the blob and wrapped around. Figures 3.8 - 3.11 illustrate a usage scenario.

Figure 3.8: Input stroke and created virtual 3D object

Figure 3.9: Second input stroke and object, deformation of second object

Figure 3.10: Adding a stroke and wrapped stroke


30

Figure 3.11: Deformation at the added stroke

3.4.3 Virtual 3D Object Creation The mesh creation of FiberMesh is based on 2D input strokes. The 2D input coordinates required by the algorithm were gained by calculating an input plane that best approximates the original 3D coordinates. To assist users with the input of a plane stroke, the use of a plexiglas pad or the projection of the input onto a displayed virtual plane were considered. This notion was abandoned because experience in the CAVE showed that the result of the approximation of a free-hand input stroke in most cases is satisfying in the sense that it meets the expectations of the outcome from the specific stroke. Object Creation Procedure In the ImmersiveFiberMesh system, 3D coordinates of the stroke are collected until the user releases the pen. From this data the best fitting plane is approximated. The input is then projected onto the approximated plane and this new stroke is taken as input for the FiberMesh mesh-creation algorithm. The resulting mesh is converted to a node in the scenegraph along with the recalculated input stroke that FiberMesh created. This stroke is used as a


31

deformation handle. The scenegraph node is transformed back to the location of the original input. Approximation of the input plane for initial mesh creation The best fitting plane of the 3D input coordinates is found with least-squares approximation. The plane-equation in Hesse normal form is: nT x − d = 0 where n is the normal of the plane, d the distance of the plane to the origin and x is the respective position vector of all points that lie on the plane. At the same time nT p − d is the distance of a point p from a plane with normal n and distance d to the origin. The center of gravity of the input points is assumed to be a point of the wanted plane. First all input points are translated by the center of gravity. The distance of the plane to the origin is then 0. Therefore nT p is the distance of a point p from the translated plane. Let pi be the translated input points of the initial stroke. The goal is to calculate a plane that best approximates pi . To achieve this, a minimization of the sum of the squared distances of all pi from the wanted plane is done. minn {

(nT pi )(nT pi )} with ||n|| = 1

X i

With nT pi = pTi n it follows: X

minn {

nT (pi pTi )n} with ||n|| = 1 ⇔ minn {nT (

X

i

pi pTi )n} with ||n|| = 1

i

Let P = ( i pi pTi ). Using the method of Lagrange multipliers the constraint ||n|| = 1 can be included: P

minn,λ {nT P n + λ(||n|| − 1)}

⇔

minn,λ {nT P n + λ(nT n − 1)}


32

To find extreme values, differentiation of the term with respect to n is done and the equation is set to zero: 2P n + 2λn = 0

⇔

P n + λn = 0

The resulting equation describes the eigenvalue problem: ˜ P n = λn The 3 x 3 Matrix P has typically three eigenvalues that describe minimum, point of inflection and maximum. The eigenvector n that belongs to the eigenvalue with the smallest absolute value describes the minimum. It is the sought normal vector of the approximated plane. After calculating n, all points are projected onto the approximated plane and retranslated by the center of gravity of the initial input points. The points are then used as input for the FiberMesh mesh-creation algorithm.

3.4.4 Removal of Intersections and Loops The FiberMesh algorithm stops when input strokes are self-intersecting. Since the input stroke is projected onto a plane, there is no easy way for the user to estimate whether the 3D stroke is self-intersecting1 . Furthermore, users that are not used to the 3D input method often create loops. As a consequence intersections and loops were removed from the input stroke. The removal was done with regard to the approximated input plane. The FiberMesh algorithm tests whether the position of segments of a stroke are closer than a threshold value. A simple approach was taken in using this mechanism to remove the narrowest segment in succession until the stroke is without loops 1

In the first implementation this led to the input strategy of leaving a large gap in the stroke, because the FiberMesh algorithm closes the gap automatically without intersection.


33

and intersections.

3.4.5 Extension to Multiple Instances The FiberMesh application uses one single mesh and hence one object for the whole design process. This object can be extruded and deformed. To provide the opportunity to use the whole space of the CAVE, object creation in ImmersiveFiberMesh was extended to multiple objects. Limitation of the sketching process to only one object would have been to restrict the design process and the use of the application unnecessarily. Each object is a mesh of its own. The disadvantage of this approach is that deformation of one object does not propagate to connected objects.

3.4.6 Color Palette A color palette was added close to a wall of the CAVE. The user chooses a color in drawing through the colored field. The palette was added for convenient reasons, replacing the keyboard input of SketchApp. Placing the palette on a wall seemed sufficient, for the task of changing color is not a main task of the application.

3.4.7 Deformation A gripper is used for the deformation of objects. The embodiment of the gripper-TUI is borrowed from the everyday experience that grippers are used to bend objects. The gripper was part of the initial system. Figure 3.12 shows the gripper.


34

Figure 3.12: Tangible user interface gripper

Initial input stroke and further added strokes remain visible on the blobs and serve as handles for deformation. To deform a blob, the user pulls with the gripper at a vertex of one of these strokes. The system constantly checks whether the gripper is near a stroke of one of the displayed blobs and then gives a visual feedback. The vertex that can be used for deformation is highlighted with a red sphere. Figure 3.13 shows the visual feedback.

Figure 3.13: Visual feedback for deformation

The user adds a stroke by drawing an open line near a blob, starting and ending outside the blob. The additional stroke is lighter in color than the initial stroke. If the new line overlaps the initial stroke two times it is projected onto the initially approximated input plane and transferred to the mechanism of FiberMesh. Otherwise the initial new line stays visible to give a feedback to the user that the input failed. If a stroke was added, it is then wrapped around the whole object and can be used as a deformation handle.

3.5 Discussion regarding Sketching Properties

35

The FiberMesh algorithm for the propagation of deformation along the vertices of a stroke was slightly changed. Propagation was expanded to more vertices at the same time. When first picking a vertex of a stroke for deformation, FiberMesh generates a peak that is smoothed after further pulling. This behavior did not attract attention in the desktop application but was recognizable in the 3D environment. The concept of FiberMesh was maintained to add strokes to the blob and use only vertices of strokes as deformation handles. FiberMesh is based on the idea that the mesh is determined and constrained by these strokes only and deformation is propagated through the network of the input strokes. Another possible approach would have been to pull at any vertex of a blob and abandon the addition of further strokes. This approach would have simplified the interaction for the user but would have led to less expressiveness and level of detail because objects would not be constrained by additional strokes.

3.5 Discussion regarding Sketching Properties By means of the characteristic properties Buxton introduced for sketches [Bux07], this chapter addresses the question whether the characteristics of a sketch are maintained by the developed application. A sketch is quick This criterion applies to the initial object creation of the application. Objects can be produced rapidly and the initial creation is faster than line-based sketching, because only one input stroke is needed to create an object. If the object has to be adjusted with the provided deformation tool, handling time increases. But still the creation of objects can be regarded as quick. Timely: A sketch can be provided when needed This surely does not apply to sketching in an immersive 3D environment. To


36

some degree this also does not apply to any other computer-aided system. Nevertheless an immersive 3D environment is not easily provided when needed. Inexpensive: A sketch is cheap The employed technology does not fulfill this either. To some degree it does not apply to any technology-based approach. Sketches are disposable In the focus group interviews of Israel et al. [IWM+ 09] some designers contradicted this statement. There are sketches that are never disposed and sketches that designers appreciate more than the final drawing. Users of the application often wanted to save their sketch. This is for sure partly due to the novelty of the technology but may also indicate an emotional connection to the 3D creation. Since 3D sketching in immersive environments is not yet part of the everyday design process this criterion can not be assessed. Plentiful: Sketches tend to not exist in isolation Most users that were asked to create an object made different attempts at the task. In this sense the criterion is satisfied. Clear vocabulary: The style distinguishes a sketch from other types of renderings The application creates rotund objects from input strokes. The intended style is not that of an exact technical drawing or CAD model. But the impression resembles a finished object. The model does not have the usual feel of a sketch: The shading is too precise and the silhouettes are smooth and show no corrections, since corrections are done by pulling a line, not by adding lines. But corrections are what gives sketches their typical style. The objects in the immersive environment look as if cast in plastic. Figure 3.14 shows objects created with ImmersiveFiberMesh.


37

Figure 3.14: Two instances of a stool (created during the user study)

Distinct Gesture: Fluently, open and free. Not tight and precise in the sense of an engineering drawing The objects do not look like technical drawings. But nevertheless the impression is more precise than that of a line-based sketch. Minimal Detail: Only include what is required This property addresses the level of detail of a sketch. The application creates rotund objects with limited level of detail. This is a restriction of the application. The level of detail was intended to be appropriate to early stages of design with a small amount of required detail. Appropriate Degree of Refinement: Not beyond the level of the project being depicted A high degree of refinement can not be accomplished with the application. For the addressed early design phases considered here, the level of the project is mainly to clarify the problem and develop a frame conception of solutions. The degree of refinement could be seen as appropriate for this level. Suggest and explore rather than confirm The exploration of objects in their spatial dimension is a central advantage of sketching in 3D. To look at an object and the arrangement of objects from different points of view can give new insights and suggestions. This


38

can be seen as an inherent feature of the application. But as the created objects resemble finished objects, their own function to suggest rather than to confirm is limited. Sketches are intentionally ambiguous Buxton states: ”If you want to get the most out of a sketch, you need to leave big enough holes. Ambiguity creates the holes. It is what enables a sketch to be interpreted in different ways, even by the person who created it.” [Bux07] The created 3D objects make the impression of finished objects. This leaves no room for ambiguity. Leaving aside the criteria that concern the technical equipment, the resemblance to finished objects can be regarded as the main difference of the applied approach to Buxton’s criteria. This resemblance does not indicate a sketchy style and offers no ambiguity. Nevertheless one main criterion, the quickness of a sketch, is maintained and the approach can in a sense even be regarded as faster than line-based sketching. The influence of these differences on the sketching process could not be predicted. In order to gain information about the usefulness of the approach concerning the sketching process, a user study was conducted.

CHAPTER 4 User Study

A comparative user study was conducted to evaluate to what extent sketchbased object creation in an immersive 3D environment is supportive to the sketching process. This chapter describes the study. The intention of the study was to evaluate whether immersive 3D-media have advantages over 2D-media when using sketch-based object creation. Furthermore, the possible benefit of automatic object creation while sketching in contrast to sketching without object creation, both in immersive 3D-media, was evaluated. Twelve design students and professionals were invited. They were asked to accomplish tasks under three conditions: 2D input sketchbased modeling (FiberMesh1 [NISA07]), immersive 3D sketch-based modeling (ImmersiveFiberMesh2 , the development of this thesis) and immersive 3D line-based sketching (SketchApp3 [IWM+ 09]). The study was conducted in the Virtual Reality Laboratory of the FraunhoferInstitute for Production Systems and Design Technology (IPK), Berlin.

1 2 3

cp. section 3.3 cp. section 3.4 cp. section 3.2

4.1 Hypotheses

40

4.1 Hypotheses The user study was conducted in order to investigate two hypotheses. Hypothesis 1: Immersive 3D-media1 are better suited to externalize images of voluminous objects (e.g. inner images of products) than 2D-media2 . Hypothesis 2: The total workload3 of designing voluminous objects can be reduced by reducing the number of motoric process steps (e.g. movements) that are necessary to create these objects.

4.2 Evaluated conditions Three conditions were evaluated. 1. FiberMesh (F2) [NISA07] 2D input sketch-based modeling 2. ImmersiveFiberMesh (F3) Immersive 3D sketch-based modeling (the development of this thesis) 3. SketchApp (S3) [IWM+ 09] Immersive 3D line-based sketching The former two conditions were run in a CAVE with five rear-projected walls (2.5-meter edge length) and active stereo vision. Both glasses and interaction devices were tracked using magnetic tracking. The third condition was run

1 2 3

Immersive 3D-media are taken to be systems with an immersive 3D sensation of inand output (e.g. a CAVE). 2D-media are taken to be systems with 2D in- and output (e.g. a tablet-PC) The total workload is taken to be the sum of the motoric and mental workload of a task.

4.2 Evaluated conditions

41

on a tablet-PC, using a touchpen for interaction. The study focussed on the influence on the sketching process of space of interaction (i.e. immersive 3D-media vs. 2D-media) and interaction technique (i.e. sketch-based modeling vs. line-based sketching), respectively. In order to investigate the influence of the space of interaction, ImmersiveFiberMesh and FiberMesh were compared. These two conditions differ with regard to the space of interaction (i.e. immersive 3D-media vs. 2D-media). The interaction techniques of the two conditions are different on a physical level, but have the same conception of automatic shape creation from simple input strokes. The consistency of the interaction technique is limited by differences in input strategy (see further below). To investigate the influence of interaction technique, ImmersiveFiberMesh and SketchApp were compared. These two conditions have the same space of interaction (i.e. an immersive 3D environment), while the interaction techniques differ (i.e. sketch-based modeling vs. line-based sketching). FiberMesh and SketchApp differ in terms of both interaction technique and space of interaction. These two conditions were not compared with one another since it would not be possible to estimate the mutual influence of the differences. The different strategies to extend objects in FiberMesh and ImmersiveFiberMesh limit the constancy of the interaction technique. The concept of FiberMesh is to create one model that can be extruded. To create an extrusion the user draws a closed stroke onto the model, rotates the model and draws the silhouette of the extrusion. ImmersiveFiberMesh has no extrusion functionality. To extend an object the user creates another object at the appropriate location. The main difference can be seen in the drawing of the closed stroke onto the model. The drawing of the silhouette of the extrusion in the case of FiberMesh and the drawing of the second object in the case of ImmersiveFiberMesh can be regarded as similar tasks and the rotation of the object is inherent to the 2D input-method.

4.3 Method of collecting data

42

Furthermore, in ImmersiveFiberMesh parts of the scene could be moved. The movement of the models resulted from a restriction of the size of deformation in the FiberMesh algorithm. If the user tries to deform a model too expansively, the model ”snaps back” to its original shape and is just moved along the screen. This behavior was not removed in ImmersiveFiberMesh and participants of the study used it to move objects. Since FiberMesh does not provide movement of parts of the model this also limits the constancy of the interaction technique. With these reservations, the interaction techniques of both conditions can be regarded as analogous: Both conditions use analogous methods to achieve the basic functionality of initial object creation and object-deformation. Other features of FiberMesh were not used during the study. The color of objects in ImmersiveFiberMesh was fixed to white, as is the case in FiberMesh.

4.3 Method of collecting data Two validated questionnaires were used, the NASA-TLX (NASA Task Load Index) [NAS88] and AttrakDiff2 [HBK03], and two additional questions were posed. The NASA-TLX assesses the subjective workload of a human-machine system. ”The NASA-TLX is a multi-dimensional rating procedure that derives an overall workload score based on a weighted average of ratings on six subscales: Mental Demands, Physical Demands, Temporal Demands, Own Performance, Effort and Frustration.” [NAS88] According to Moroney et al., the weighted rating, consisting of pairwise comparison and weighting of the subscales done by the test persons, and the unweighted rating are highly correlated. They conclude that the time consuming use of weighting scales is not necessary [MBEM92]. The NASA-TLX was used without weights in this study. The scale of rating was interpreted as values between 0 and 10. The scale on the questionnaire had a length of 6.6 cm. The values were measured


43

in mm and scaled to a 10 cm scale. The polarity of the dimensions is from low to high, in the case of performance from good to poor. A German translation of the questionnaire was taken from [Sei02], the entire questionnaire is listed in appendix A. Table 4.1 lists the rating scale definitions.

Title

Endpoints

Descriptions

MENTAL DEMAND

Low/High

How much mental and perceptual activity was required (e.g., thinking, deciding, calculating, remembering, looking, searching, etc.)? Was the task easy or demanding, simple or complex, exacting or forgiving?

PHYSICAL DEMAND

Low/High

How much physical activity was required (e.g., pushing, pulling, turning, controlling, activating, etc.)? Was the task easy or demanding, slow or brisk, slack or strenuous, restful or laborious?

TEMPORAL DEMAND

Low/High

How much time pressure did you feel due to the rate or pace at which the tasks or task elements occurred? Was the pace slow and leisurely or rapid and frantic?

PERFORMANCE

Good/Poor

How successful do you think you were in accomplishing the goals of the task set by the experimenter (or yourself)? How satisfied were you with your performance in accomplishing these goals?

EFFORT

Low/High

How hard did you have to work (mentally and physically) to accomplish your level of performance?

FRUSTRATION

Low/High

How insecure, discouraged, irritated, stressed and annoyed versus secure, gratified, content, relaxed and complacent did you feel during the task?

Table 4.1: Rating Scale Definitions of NASA-TLX [NAS88]


44

In order to measure the user-perceived usability of the evaluated conditions the questionnaire AttrakDiff2 [HBK03] was used. This questionnaire analyzes the perceived usability of an interactive product (i.e. the pragmatic quality) and additionally measures the hedonic qualities of stimulation and identity and the overall attractiveness. Following Hassenzahl et al. an interactive product possesses perceived pragmatic quality if it allows for effective and efficient achievement of behavioral goals. But above this, human beings aspire toward personal development. Products may be able to support this development by offering new and interesting features that motivate and stimulate the users and thus may help to complete their tasks. Hassenzahl et al. continue that human beings express themselves through objects. They want to be seen by relevant others in a specific way. A product may support this notion in communicating the desired identity [HBK03]. The semantic differential of AttrakDiff2 lists 28 opposed pairs of adjectives in random order and random polarity. Test persons are asked to mark one out of seven boxes. Answers are interpreted as values between -3 and 3. The dimensions of the questionnaire are • pragmatic quality (PQ), • hedonic quality identity (HQI), • hedonic quality stimulation (HQS) and • attractiveness (ATTR) Each dimension of the questionnaire includes seven attributes. The dimensions are calculated by averaging the included attributes per participant. A high PQ score primarily implies a high perceived usability. A high rating for the HQI dimension implies that a high capability of communicating identity to others was perceived. The attributes that are used for this dimension are primarily social or outwards attributes. The HQS dimension describes the


45

perceived novelty, stimulation and challenge. The attributes for the HQS dimension are inwards attributes that are primarily related to personal growth [Has04]. The attributes of AttrakDiff2 with their respective polarity and the German questionnaire that was used in the study are listed in appendix A. Additionally, two questions were asked: 1. Sketching process The tool supported me well in expressing my idea of the object. 2. Sketch The completed sketch was in accordance with my idea of the object. Participants were asked to mark one out of five boxes of a Likert scale labeled ’strongly agree’ (5), ’agree’ (4), ’neither agree nor disagree’ (3), ’disagree’ (2) and ’strongly disagree’ (1) (corresponding values for the evaluation in brackets). The first question was intended to investigate the suitability of the tool concerning the sketching process. The second question concerns the quality of the completed sketch. Sketching process and completed sketch are two distinct qualities of a sketch that are both relevant in judging the suitability of a sketching condition. The German questions that were used are listed in appendix A. To avoid the influence of a given order, test conditions were permuted. This led to six groups, each consisting of two randomly assigned test persons. Nevertheless, sequence effects can not be ruled out and the size of the groups

4.4 Participants

46

has to be regarded as small.1 Table 4.2 lists the six groups and according permutation of the conditions.

Group

Permutation of conditions

G1

F2 S3 F3

G2

F2 F3 S3

G3

S3 F2 F3

G4

S3 F3 F2

G5

F3 F2 S3

G6

F3 S3 F2

Table 4.2: Groups and according permutation of the conditions FiberMesh (F2), ImmersiveFiberMesh (F3) and SketchApp (S3)

The order of the tasks were not permuted since the two tasks were not compared to each other.

4.4 Participants Participants were acquired via email. A mailing list of design students and professionals of the Fraunhofer IPK was used that had been compiled for 1

An additional one-way ANOVA with group as factor and all dimensions in combination with each condition as dependent variables was done. The Levene statistic for all dependent variables was significant, meaning that no homogeneity of variance was given. Therefore the results of the ANOVA are doubtful. Group 2 and 6 had significant results for HQI for task 1 and task 2 with regard to SketchApp, as well as for NASA-TLX physical demand for task 1 and task 2 with regard to SketchApp. In these two groups, ImmersiveFiberMesh is followed immediately by SketchApp. For the physical demand with regard to SketchApp for task 1, significant results were also found between group 3 and 5 as well as 5 and 6. For the performance with regard to FiberMesh for task 2 a significant result was found between group 1 and 4. This can be seen as an indication, that the order of the conditions had influence on the results.

4.5 Tasks

47

preceding studies. The invitation contained criteria to partake in the study. The participants were to be professionals or students in the field of design with experience in sketching. They were to have stereoscopic vision and should be aged between 25 and 55. The lower age limit was later reduced to 22. German language abilities were required due to the fact that questionnaires and descriptions were in German. Twelve test persons were invited, eleven male and one female, aged from 22 to 43, average age was 29.5 (SD = 6.4). Eight persons were students of product or communication design, four were professional product designers. The students had a mean duration of study of 2.75 years (SD = 1.28), the mean professional experience of the product-designers was 7.25 years (SD = 3.3). All stated that they regularly sketch on paper and via computer and regularly use 3D-CAD programs. Four had used Virtual Reality environments before, the others had no VR experience.

4.5 Tasks In each condition participants had to accomplish two tasks. Task 1 - Sketching an object from memory: A stool was shown to the participants. They were allowed to look at it and take it into their hands. Then the stool was taken away. The participants were asked to sketch the stool. The stool is shown in figure 4.1. This task was meant to be a less creative task. The purpose of the task was to let participants externalize a pre-existing inner image. Looking at the object, they were supposed to form an inner image. This mentally stored imagination of the object could be retrieved while sketching. Without the need to develop creative design ideas, this was supposed to lead to a concentration on the externalization of the image. Consequently, the task was intended to address the adequacy of a condition with respect to externalizing an inner image of an object.

4.6 Procedure of the study

48

Figure 4.1: Stool to be drawn in task 1

Task 2 - Designing an object: Participants were asked to design a comfortable armchair. This was intended to be a creative design task, since participants had to develop their own imagination of an object while sketching. It was meant to address the support offered by a condition to the creative sketching process. The objects were chosen from the category of rotund objects because FiberMesh and the developed ImmersiveFiberMesh are applications that support the creation of this kind of objects. This restriction is inherent to the approach used. Especially since ImmersiveFiberMesh had no cut functionality, and consequently cut was also not used in FiberMesh, the task to create an angular shape could not be accomplished. The German descriptions of the tasks are listed in appendix A.

4.6 Procedure of the study In a pretest the course of the study was tested and the material was checked. It turned out that a time limit for the handling of the tasks had to be set in

4.7 Evaluation

49

order to keep the duration at about two hours. At the beginning of each testing, participants received a description that contained the general procedure of the study. Participants were then asked to fill out a data security sheet and a questionnaire concerning personal data (i.e. age, gender, profession) and previous knowledge concerning VR-systems, 3D-CAD software and sketching respectively. For each new condition an instruction sheet was handed out. The test persons could practice for 5 to 10 minutes, verbal explanations of the use of the tool were given and questions were answered. After this time the description of the first task was handed out. 10 minutes were provided to complete the task. The participants were then asked to answer the AttrakDiff2 questionnaire followed by the NASA-TLX questionnaire as well as the two additional questions. Then the second task was conducted in the same manner, including the same questionnaires. All three conditions were conducted in the same procedure. The duration of the whole test per person was about two hours. During the study photos were taken. Comments of the participants were written down by the author. The material of the study is listed in appendix A.

4.7 Evaluation The questionnaires AttrakDiff2 and NASA-TLX use metric scales and can be evaluated by means of analysis of variance (ANOVA) ([HBK03], [Sei02]). A one-way ANOVA was conducted with the conditions FiberMesh, ImmersiveFiberMesh and SketchApp as three levels of the factor. The investigated dependent variables were the four dimensions of AttrakDiff2 (pragmatic quality (PQ), hedonic quality identity (HQI), hedonic quality stimulation (HQS) and attractiveness (ATTR)) and the six dimensions of NASA-TLX (mental demand, physical demand, temporal demand, own performance, effort and frustration) as well as the mean value of the six dimensions of NASA-TLX (mean work load).

4.8 Results

50

Only two comparisons were of interest (ImmersiveFiberMesh vs. FiberMesh and ImmersiveFiberMesh vs. SketchApp). A Scheffé post-hoc-test was conducted to investigate which pair of conditions reached a significant level. A Levene statistic on the values that were found to be significant showed no significance, that means that homogeneity of variance can be assumed. A post-hoc Bonferroni correction was conducted because multiple ANOVAs were done on the set of data. The correction did not change the results. Since the scale of the two additional questions is ordinal, a Friedman test was conducted. The test analyzes ranks within independent conditions. A significant level is reached if at least one order of ranks within a condition is significantly different from the order in the other conditons. To investigate which pair of conditions reached a significant level, the Friedman test was also conducted pairwise for the triples that reached significance level. PASW Statistics 18 (2009) was used for the statistical analysis.

4.8 Results AttrakDiff2 For the hedonic quality stimulation (HQS) for task 1, ImmersiveFiberMesh ranked significantly higher than FiberMesh (F (2,33) = 6.22; p < 0.05). ImmersiveFiberMesh: M = 2.02, SD = 0.78 FiberMesh: M = 1.17, SD = 0.89 The Scheffé post-hoc-test reached a significance of pF 3−F 2 < 0.05 between these two conditions. For the attractiveness (ATTR) for task 2, ImmersiveFiberMesh ranked significantly higher than FiberMesh (F (2,33) = 8.95; p < 0.05). ImmersiveFiberMesh: M = 1.21, SD = 1.06 FiberMesh: M = 0.21, SD = 1.05 The Scheffé post-hoc-test reached a significance of pF 3−F 2 < 0.05 between

4.8 Results

51

these two conditions. For the pragmatic qualtity (PQ) for task 2, SketchApp ranked significantly higher than ImmersiveFiberMesh (F (2,33) = 8,56; p < 0.01). SketchApp: M = 0.95, SD = 0.96 ImmersiveFiberMesh: M = −0.30, SD = 1.36 The Scheffé post-hoc-test reached a significance of pS3−F 3 < 0.05 between these two conditions. There was missing data for task 1: pragmatic quality (PQ) (attribute Confusing - clear), hedonic quality stimulation (HQS) (attribute Cautious courageous) and attractiveness (ATTR) (attribute Unpleasant - pleasant) and for task 2: pragmatic quality (PQ) (attribute Confusing - clear). Figure 4.2 and 4.3 show the mean values for task 1 and task 2. In appendix A, mean values and standard deviation are listed.

Figure 4.2: AttrakDiff2: Mean values for task 1

4.8 Results

52

Figure 4.3: AttrakDiff2: Mean values for task 2

NASA-TLX For the physical demand for task 1, ImmersiveFiberMesh was ranked significantly higher (i.e. more demanding) than FiberMesh (F (2,33) = 4.96; p < 0.05). ImmersiveFiberMesh: M = 5.54, SD = 2.57 FiberMesh: M = 2.30, SD = 2.89 The Scheffé post-hoc-test reached a significance of pF 3−F 2 < 0.05 between these two conditions. For the own performance for task 2, the comparison of ImmersiveFiberMesh and SketchApp almost reached a significant level, the Scheffé post-hoc-test reached a significance of pF 3−S3 = 0.058 between these two conditions. ImmersiveFiberMesh (M = 5.78, SD = 2.65) was ranked higher (i.e. poorer own performance) than SketchApp (M = 3.31, SD = 2.38). There was missing data of one sample for task 1, for the dimensions effort and frustration. Figure 4.4 and 4.5 show the mean values for task 1 and task 2. In appendix

4.8 Results

53

A, mean values and standard deviation are listed.

Figure 4.4: NASA TLX: Mean values for task 1

Figure 4.5: NASA TLX: Mean values for task 2

Additional questions The Friedman test for task 2 (designing an object), question 1 (concerning the sketching process) reached a significant level (χ2 (2) = 13,29; p < 0.05) and the additional pairwise Friedman test showed that the order of ranks between SketchApp and ImmersiveFiberMesh were significantly different (χ2F 3−S3 (1) = 7.364; p < 0.05). SketchApp was ranked higher (more supportive) than

4.9 Discussion of results

54

ImmersiveFiberMesh. SketchApp: median = 3.00, IQR = 3.25 − 4.00; ImmersiveFiberMesh: median = 3.00, IQR = 2.00 − 3.00; FiberMesh: median = 2.00, IQR = 1.25 − 2.75.

4.9 Discussion of results The analysis of the results of the study yielded few differences between the conditions, but some relevant findings were made. Hypothesis 1 is directed towards a comparison between 3D-media and 2Dmedia with respect to their suitability to externalize inner images of voluminous objects. Performing task 1, participants regarded ImmersiveFiberMesh as more stimulating than FiberMesh, even though at the same time, ImmersiveFiberMesh was perceived as more physically demanding. Task 1 was supposed to address the adequacy of a condition in externalizing an inner image of an object, in contrast to the creative development of an object without an external representation. The HQS dimension describes the perceived novelty, stimulation and challenge of an interactive application. The result could indicate a stimulating impact of the immersive 3D environment while externalizing inner images. Also, the attractiveness of ImmersiveFiberMesh with regard to the creative task (task 2, designing an armchair) was seen as higher than that of FiberMesh. This task was designed to investigate to what extent a condition supported the creative sketching process. The results could be regarded as supportive to hypothesis 1 in the sense that the immersive 3D medium seemed to have a stimulating effect and the 3D condition was perceived as more attractive than the 2D condition for the creative sketching process. But with regard to the pragmatic quality, indicating the perceived usability of an application, no differences were found and the additional questions did not show a preference for ImmersiveFiberMesh over FiberMesh.


55

Hypothesis 2 was directed towards a comparison between SketchApp and ImmersiveFiberMesh with regard to the subjective workload. No results were found that support this hypothesis. The presumption that the total workload declines if the system creates an object from an input stroke instead of the user drawing the whole object could not be supported since no significant effects were found in the dimensions of the NASA-TLX among these two conditions. Another result is that, with regard to the pragmatic quality, participants preferred SketchApp over ImmersiveFiberMesh to perform a design task (task 2, designing an armchair). The almost significant result for the own performance (NASA TLX) of task 2 between these two conditions is in accordance with this result, if the satisfaction with the own performance could be seen as related to the pragmatic quality. The pragmatic quality primarily implies a high perceived usability. The additional questions also showed a preference of SketchApp over ImmersiveFiberMesh concerning the sketching process for task 2. These results seem to underline the importance of line-based sketching with regard to a creative design task. Further investigation could be directed at a possible benefit of a combination of line-based sketching and sketch-based modeling in an immersive 3D environment. Generally it has to be taken into account that functionality of the conditions differed (see section 4.2: ’Evaluated Conditions’) and that the robustness of the applications applied for the conditions was also different. The FiberMesh version was the most unstable of the three applications and SketchApp was most stable. Since these differences between the conditions narrow comparability, and also taking into account that the number of participants was small, this study can only give a tendency for further, formalized studies. Figures 4.6 - 4.7 show some results of participants using ImmersiveFiberMesh.


Figure 4.6: Stools sketched by participants for task 1, using ImmersiveFiberMesh

Figure 4.7: Armchairs designed by participants for task 2, using ImmersiveFiberMesh

56

CHAPTER 5 Discussion and Outlook

The presented work provides shape modeling in an immersive 3D environment from sketched input strokes. The shape creation and deformation algorithms of FiberMesh [NISA07] were made available in a 3D environment and the integrated tool was evaluated in order to investigate whether this approach supports the design process in the early design phases. FiberMesh was designed as desktop application. Consequently, the algorithms work with 2D input strokes. The algorithms of FiberMesh were not changed in this thesis. This led to the approximation of an initial input plane and the recalculation of additional strokes with regard to this plane. The focus was on the task of transferring FiberMesh’s shape creation and deformation into a 3D environment. But the approach has the disadvantage that it does not exploit the capability of a 3D environment to provide 3D coordinates. The approach to use ’blobby inflation’ as object creation limited the class of shapes that could be created in the 3D environment to rotund objects. Additionally, the provided application lacked the feature of cutting objects. The possibility to create plane surfaces and sharp edges is certainly important for many design tasks. Even the round-shaped stool that participants were asked to draw in the user study had a plane seating and users had difficulties creating this shape. Further development of the application should provide

5 Discussion and Outlook

58

cutting (e.g. a ’flat iron’ tool). Nevertheless, with the approach to integrate blobby object creation from simple input strokes into an immersive 3D environment, a further step was taken in the investigation of 3D sketching tools. The tool should be further developed. Beside the implementation of cutting, translation of objects should be provided and the growth of a model from deformation should be restricted. Under certain conditions (e.g. intensive one-sided expansion) the model expands heavily on the opposed side of a pulled vertex or the model even ’explodes’. This lack of robustness and predictability was criticized by participants of the study for both FiberMesh and the developed application. In the conducted user study, users perceived the shape modeling in the immersive 3D environment as more stimulating and attractive than under 2D conditions. The stimulating effect and perceived attractiveness could be regarded as the impact of the immersive 3D environment. These two qualities are also likely to be supportive to the design process. Since users preferred line-based sketching in an immersive 3D environment over the provided sketch-based modeling in the same environment, a next step could be to investigate whether the combination of both, line-based sketching and sketch-based object-creation in a 3D environment, gives benefit to the sketching process. Line-based sketching could be compared to a condition that offers a choice of the interaction method. In the combined approach, the user might change the input technique by selecting another tip of the pen (e.g. a brush for shape creation versus a pencil for lines). The investigation could indicate whether the combination of both interaction techniques is an important feature to create a benefit to the early sketching process in 3D environments.

Bibliography

[Ale02] M. Alexa. Shape Spaces from Morphing. Dissertation, TU Darmstadt, 2002. (Cited on page 1) [Bux07] B. Buxton. Sketching User Experiences. Morgan Kaufmann, San Francisco, 2007. (Cited on pages 1, 6, 17, 35, and 38) [CA09] M. T. Cook and A. Agah. A survey of sketch-based 3-dimensional modeling techniques. Accepted Manuscript: Interacting with Computers, 2009. (Cited on pages 9, 13, 15, and 16) [CSSJ05] J. J. Cherlin, F. Samavati, M. C. Sousa, and J. A. Jorge. Sketch-based modeling with few strokes. Proceedings of the 21st spring conference on Computer graphics, 2005. (Cited on pages 10, 11, 12, and 13) [DM96] D. Drascic and P. Milgram. Perceptual issues in augmented reality. In SPIE Stereoscopic Displays and Applications VII, Virtual Reality Systems III, pages 123 – 134, 1996. (Cited on page 19) [Doe98] D. Doerner. Thought and design - Research strategies, single-case approach and methods of validation. In H. Birkhofer, P. BadkeSchaub and E. Frankenberger (Eds.), Designers - The key to successful Product Development, pages 3 – 11. London: Springer, 1998. (Cited on page 6) [GS08] E. Grinspun and A. Secord. Discrete differential geometry: An

Bibliography

60

applied introduction. SIGGRAPH ASIA 2008 COURSE NOTES, 2008. (Cited on page 23) [Has04] M. Hassenzahl. The interplay of beauty, goodness, and usability in interactive products. HUMAN-COMPUTER INTERACTION, Lawrence Erlbaum Associates, Inc., 19:319–349, 2004. (Cited on pages 45 and 70) [HBK03] M. Hassenzahl, M. Burmeister, and F. Koller. Attrakdiff: Ein Fragebogen zur Messung wahrgenommener hedonischer und pragmatischer Qualität. In Ziegler, J., Szwillus, G. (Hrsg.) Tagungsband Mensch und Computer 2003: Interaktion in Bewegung, pages 187–196. B. G. Teubner, Stuttgart, 2003. (Cited on pages 8, 42, 44, and 49) [HI07] J. Hurtienne and J. H. Israel. Image schemas and their metaphorical extensions - Intuitive patterns for tangible interaction. In Proceedings of the 1st international conference on Tangible and embedded interaction, Baton Rouge, Louisiana, pages 127 – 134, 2007. (Cited on page 18) [HSS98] W. Hacker, P. Sachse, and F. Schroda. Design thinking - Possible ways to successful solutions in product development. In H. Birkhofer, P. Badke-Schaub and E. Frankenberger (Eds.), Designers - The key to successful Product Development, pages 205 – 216. London: Springer, 1998. (Cited on pages 1, 5, 6, 7, and 16) [IH07] T. Igarashi and J. F. Hughes. A suggestive interface for 3d drawing. ACM SIGGRAPH courses, 2007. (Cited on pages 9 and 10) [IHP+ 09] J. H. Israel, J. Hurtienne, A. E. Pohlmeyer, C. Mohs, M. C. Kindsmüller, and A. Naumann. On intuitive use, physicality and tangible user interfaces. International Journal of Arts and Technology, 2(4):348 – 366, 2009. (Cited on page 19)

Bibliography

61

[IMT99] T. Igarashi, S. Matsuoka, and H. Tanaka. Teddy: A sketching interface for 3d freeform design. In Proceedings ACM SIGGRAPH 99, pages 409–416, 1999. (Cited on pages 13 and 14) [Ish08] H. Ishii. Tangible bits: Beyond pixels. In Proceedings of the 2nd international conference on Tangible and embedded interaction, New York, 2008. (Cited on page 18) [IWM+ 09] J. H. Israel, E. Wiese, M. Mateescu, Chr. Zöllner, and R. Stark. Investigating three-dimensional sketching for early conceptual design - Results from expert discussions and user studies. Computers and Graphics, 2009. (Cited on pages 2, 6, 7, 8, 16, 17, 36, 39, and 40) [IZ07] J. H. Israel and E. Zacharias. Möglichkeiten und Grenzen des Skizzierens im Dreidimensionalen Raum aus Designersicht. 7. Berliner Werkstatt Mensch-Maschine-Systeme, 2007. (Cited on page 7) [KFM+ 01] D. F. Keefe, D. A. Feliz, T. Moscovich, D. H. Laidlaw, and J. J. LaViola Jr. Cavepainting: A fully immersive 3d artistic medium and interactive experience. In Proceedings of the 2001 symposium on Interactive 3D graphics, pages 85 – 93, 2001. (Cited on page 12) [Kob09] L. Kobbelt. OpenMesh. http://openmesh.org, Computer Graphics Group RWTH Aachen, 2009. (Cited on pages 20, 25, and 26) [Kru94] R. A. Krueger. Focus Groups, A Practical Guide for Applied Research. SAGE Publications, London, 1994. (Cited on page 7) [MBEM92] W.F. Moroney, D.W. Biers, F.T. Eggemeier, and J.A. Mitchell. A comparison of two scoring procedures with the NASA task load index in a simulated flight task. In Proceedings of the 1992 IEEE National Aerospace Electronics Conference, Piscataway, NJ, pages 734–740, 1992. (Cited on page 42)

Bibliography

62

[NAS88] Human Performance Research Group NASA. TASK LOAD INDEX (NASA-TLX). NASA Ames Research Center, Moffett Field, California, 1988. (Cited on pages 42 and 43) [Nea07] A. Nealen. Interfaces and Algorithms for the Creation, Modification and Optimization of Surface Meshes. Dissertation, TU Berlin, 2007. (Cited on pages 23, 24, and 25) [NISA07] A. Nealen, T. Igarashi, O. Sorkine, and M. Alexa. Fibermesh: Designing freeform surfaces with 3d curves. In Proceedings of ACM SIGGRAPH 07, 2007. (Cited on pages i, ii, 2, 14, 16, 20, 21, 22, 27, 39, 40, and 57) [Ope09] OpenSG. http://www.opensg.org, 2009. (Cited on page 18) [PBFG07] G. Pahl, W. Beitz, J. Feldhusen, and K.-H. Grote. Konstruktionslehre - Grundlagen erfolgreicher Produktentwicklung - Methoden und Anwendung. Springer, 2007. (Cited on page 6) [Pet01] J. Petruschat. Gespräch. form+zweck. How to handle hands?, 18:3 – 9, 2001. (Cited on page 7) [PL03] M. Pache and U. Lindemann. Sketching in 3d. What should future tools for conceptual design look like? In Lindemann, U. (Hrsg.) Human Behaviour in Design: Individuals, Teams, Tools., pages 243–252. Springer-Verlag, Berlin, 2003. (Cited on page 19) [R¨02] A. Römer. Unterstützung des Design Problem Solving: Einsatz und Nutzen einfacher externer Hilfsmittel in den frühen Phasen des konstruktiven Entwurfsprozesses. Dissertation, TU Dresden, 2002. (Cited on pages 1, 4, and 5) [Sac01] P. Sachse. Idea materialis: Entwurfsdenken und Darstellungshandeln oder Über die allmähliche Verfertigung der Gedanken beim Skizzieren und Modellieren. Habilitation, TU Dresden, 2001. (Cited on page 5)

Bibliography

63

[SCOL+ 04] O. Sorkine, D. Cohen-Or, Y. Lipman, M. Alexa, C. Rössl, and H.-P. Seidel. Laplacian surface editing. In Proceedings of the Eurographics/ACM SIGGRAPH Symposium on Geometry Processing, pages 179 – 188. Eurographics Association, 2004. (Cited on page 24) [Sei02] K. Seifert. Evaluation multimodaler Computer-Systeme in frühen Entwicklungsphasen. Dissertation, TU Berlin, 2002. (Cited on pages 43 and 49) [Tve02] B. Tversky. What do sketches say about thinking? AAAI Technical Report SS-02-08, AAAI Press, Menlo Park, California, 2002. (Cited on pages 1, 5, and 6) [WS01] G. Wesche and H.-P. Seidel. Freedrawer - A free-form sketching system on the responsive workbench. In Proceedings of the ACM symposium on Virtual reality software and technology, pages 167 – 174, 2001. (Cited on page 13) [ZHH96] R. C. Zeleznik, K. P. Herndon, and J. F. Hughes. Sketch: An interface for sketching 3d scenes. In Proceedings of SIGGRAPH 96, pages 163–170, 1996. (Cited on pages 9 and 10)

List of Figures

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9

Sketching as Self-Communication . . . . . . . . . . . . . . . Two scenes created with SKETCH . . . . . . . . . . . . . . A screenshot of Chateau . . . . . . . . . . . . . . . . . . . . Rotational blending . . . . . . . . . . . . . . . . . . . . . . . Scribble method . . . . . . . . . . . . . . . . . . . . . . . . . A painting with CavePainting: Wedding Day - Daniel Keefe FreeDrawer: Curves and added surfaces . . . . . . . . . . . . Teddy user interface, run on a graphical tablet . . . . . . . . Teddy: Mesh construction . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

3.1 3.2 3.3 3.4

Tangible user interface pen . . . . . . . . . . . . . . . . . . . Operations of FiberMesh . . . . . . . . . . . . . . . . . . . . FiberMesh: Creation from a professional 2D animation artist Discrete first- and second-order differentials of a piecewise linear curve . . . . . . . . . . . . . . . . . . . . . . . . . . . Rotated local coordinate frames after curve deformation . . Halfedge data structure . . . . . . . . . . . . . . . . . . . . . Components of ImmersiveFiberMesh . . . . . . . . . . . . . Input stroke and created virtual 3D object . . . . . . . . . . Second input stroke and object, deformation of second object Adding a stroke and wrapped stroke . . . . . . . . . . . . . Deformation at the added stroke . . . . . . . . . . . . . . . . Tangible user interface gripper . . . . . . . . . . . . . . . . . Visual feedback for deformation . . . . . . . . . . . . . . . .

. 19 . 21 . 22

3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13

. . . . . . . . . .

6 10 10 11 12 12 13 14 14

24 25 26 27 29 29 29 30 34 34

List of Figures

65

3.14 Two instances of a stool (created during the user study) . . . . 37 4.1 4.2 4.3 4.4 4.5 4.6 4.7

Stool to be drawn in task 1 . . . . . . . . . . . . . . . . . . . AttrakDiff2: Mean values for task 1 . . . . . . . . . . . . . . . AttrakDiff2: Mean values for task 2 . . . . . . . . . . . . . . . NASA-TLX: Mean values for task 1 . . . . . . . . . . . . . . . NASA-TLX: Mean values for task 1 . . . . . . . . . . . . . . . Stools sketched by participants for task 1, using ImmersiveFiberMesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Armchairs designed by participants for task 2, using ImmersiveFiberMesh . . . . . . . . . . . . . . . . . . . . . . . . . . .

48 51 52 53 53 56 56

List of Tables

4.1 4.2

Rating Scale Definitions of NASA-TLX . . . . . . . . . . . . . 43 Groups and according permutation of the conditions FiberMesh (F2), ImmersiveFiberMesh (F3) and SketchApp (S3) . . . . . 46

A.1 A.2 A.3 A.4 A.5 A.6 A.7

AttrakDiff2: Arithmetic mean and standard deviation for task1 AttrakDiff2: Arithmetic mean and standard deviation for task2 NASA-TLX: Arithmetic mean and standard deviation for task1 NASA-TLX: Arithmetic mean and standard deviation for task2 Additional Questions: median and interquartile range for task1 Additional Questions: median and interquartile range for task2 Dimensions with respective attributes and polarity of AttrakDiff2

67 67 68 68 69 69 70

APPENDIX A User Study

MP Q

SDP Q

MHQI

SDHQI

MHQS

SDHQS

MAT T R

SDAT T R

F2

-0.55

1.20

0.10

0.92

1.17

0.89

0.27

1.38

F3

-0.23

1.37

0.44

0.80

2.02

0.78

1.06

1.40

S3

0.88

1.33

0.89

0.86

2.17

0.54

1.87

0.69

Table A.1: AttrakDiff2: Arithmetic mean and standard deviation for task1 Pragmatic quality (PQ), Hedonic quality identification (HQI), Hedonic quality stimulation (HQS), Attractiveness (ATTR) MP Q

SDP Q

MHQI

SDHQI

MHQS

SDHQS

MAT T R

SDAT T R

F2

-0.85

0.89

-0.06

0.92

1.31

1.07

0.21

1.05

F3

-0.30

1.36

0.67

0.82

1.99

0.73

1.21

1.06

S3

0.95

0.96

1.06

0.73

2.10

0.48

1.82

0.65

Table A.2: AttrakDiff2: Arithmetic mean and standard deviation for task2 Pragmatic quality (PQ), Hedonic quality identification (HQI), Hedonic quality stimulation (HQS), Attractiveness (ATTR)

A User Study

68

MM EN

SDM EN

MP HY

SDP HY

MT EM

SDT EM

MP ER

SDP ER

F2

5.38

2.44

2.30

2.89

4.47

1.8

6.62

2.44

F3

6.21

2.32

5.54

2.57

4.60

2.29

6.76

1.94

S3

6.16

2.99

4.95

2.59

2.95

2.30

4.81

3.18

MEF F

SDEF F

MF RU

SDF RU

Mmean

SDmean

F2

6.46

2.6

5.11

2.49

5.06

1.53

F3

6.00

2.35

4.62

2.57

5.62

1.38

S3

4.89

2.73

3.90

3.32

4.62

1.75

Table A.3: NASA-TLX: Arithmetic mean and standard deviation for task1 Mental Demand (MEN), Physical Demand (PHY), Temporal Demand (TEM), Performance (PER), Effort (EFF), Frustration (FRU) and Mean Work Load (mean) MM EN

SDM EN

MP HY

SDP HY

MT EM

SDT EM

MP ER

SDP ER

F2

6.88

1.89

2.5

2.72

5.45

2.17

7.68

2.23

F3

5.49

2.36

4.73

2.4

3.85

2.03

5.78

2.65

S3

4.92

2.32

5.43

2.21

3.04

1.78

3.31

2.38

MEF F

SDEF F

MF RU

SDF RU

Mmean

SDmean

F2

6.29

2.59

5.96

2.63

5.79

1.37

F3

5.20

2.53

4.71

2.33

4.96

1.39

S3

4.32

2.45

2.58

2.26

3.93

1.43

Table A.4: NASA-TLX: Arithmetic mean and standard deviation for task2 Mental Demand (MEN), Physical Demand (PHY), Temporal Demand (TEM), Performance (PER), Effort (EFF), Frustration (FRU) and Mean Work Load (mean)

A User Study

69

median

IQR

F2, question 1 (Sketching process)

2.00

1.00 - 3.50


2.00

2.00 - 3.00

S3, question 1 (Sketching process)

3.00

2.25 - 4.00

F2, question 2 (Sketch)

2.00

1.25 - 2.00


2.00

1.25 - 3.00

S3, question 2 (Sketch)

2.50

1.25 - 4.00

Table A.5: Additional Questions: median and interquartile range for task1 median

IQR


2.00

1.25 - 2.75


3.00

2.00 - 3.00

S3, question 1 (Sketching process)

4.00

3.25 - 4.00


2.00

1.00 - 2.75


2.00

1.25 - 3.75

S3, question 2 (Sketch)

4.00

3.00 - 4.00

Table A.6: Additional Questions: median and interquartile range for task2

A User Study

Scale

70

Attributes

Hedonic quality-identification (HQI) HQI_1

Isolating - integrating

HQI_2

Amateurish - professional

HQI_3

Gaudy - classy

HQI_4

Cheap - valuable

HQI_5

Noninclusive - inclusive

HQI_6

Takes me distant from people - brings me closer to people

HQI_7

Unpresentable - presentable

Hedonic quality-stimulation (HQS) HQS_1

Typical - original

HQS_2

Standard - creative

HQS_3

Cautious - courageous

HQS_4

Conservative - innovative

HQS_5

Lame - exciting

HQS_6

Easy - challenging

HQS_7

Commonplace - new

Pragmatic quality (PQ) PQ_1

Technical - human

PQ_2

Complicated - simple

PQ_3

Impractical - practical

PQ_4

Cumbersome - direct

PQ_5

Unpredictable - predictable

PQ_6

Confusing - clear

PQ_7

Unruly - manageable

Attractiveness (ATTR) ATTR_1

Unpleasant - pleasant

ATTR_2

Ugly - attractive

ATTR_3

Disagreeable - likeable

ATTR_4

Rejecting - inviting

ATTR_5

Bad - good

ATTR_6

Repelling - appealing

ATTR_7

Discouraging - motivating

Table A.7: Dimensions with respective attributes and polarity of AttrakDiff2 [Has04]

A User Study

71

Allgemeine Informationen zum Versuch Lieber Versuchsteilnehmer, liebe Versuchsteilnehmerin, vielen Dank, dass Sie an diesem Versuch teilnehmen. Der Versuch findet im Rahmen einer Diplomarbeit statt, die sich mit der Formerzeugung in virtuellen 3D-Umgebungen befasst. Im Versuch wird es darum gehen, Skizzier- und Modellieraufgaben in einer virtuellen 3DUmgebung (CAVE) und auf einem PC auszuführen. Für den Versuchsteil in der virtuellen 3D-Umgebung tragen Sie eine Brille, die die 3DDarstellung von virtuellen Objekten ermöglicht und ähnlich funktioniert wie „3D-Brillen“, die beispielsweise beim Betrachten von 3D-Filmen (I-Max etc.) eingesetzt werden. Für den anderen Versuchsteil wird ein Tablet-PC verwendet, bei dem die Eingabe mit einem Stift auf dem Bildschirm erfolgt. Die Software befindet sich im Entwicklungsstadium. Die Robustheit der Software ist daher nicht mit der von herkömmlicher Software zu vergleichen und es kann zu Abstürzen kommen. Versuchsablauf Zunächst möchten wir Sie bitten, einen Fragebogen zur Person und eine DatenschutzErklärung auszufüllen. Im Versuch selbst sollen Sie dann in drei verschiedenen Skizzieranwendungen jeweils zwei (gleichbleibende) Aufgaben ausführen. Zwei der Anwendungen laufen in der virtuellen 3DUmgebung und eine auf dem PC. Bei jeder Anwendung werden die folgenden Abschnitte durchlaufen: 1. Zuerst können Sie das Skizzieren mit der jeweiligen Anwendung ca. fünf bis zehn Minuten lang üben. 2. Im Anschluss daran bearbeiten Sie die erste Skizzieraufgabe. 3. Es folgen zwei kurze Fragebögen. 4. Dann bearbeiten Sie die zweite Skizzieraufgabe. 5. Es folgen die gleichen zwei Fragebögen. Für die Skizzieraufgaben haben Sie jeweils bis zu zehn Minuten Zeit. Insgesamt wird der Versuch etwa zwei Stunden dauern. Skizzen werden nicht gespeichert. Sie können jederzeit Fragen an die Versuchsleiterin stellen. Sollte Ihnen während des Versuchs schlecht werden oder sollten Sie sich unwohl fühlen, können Sie den Versuch jederzeit abbrechen.

A User Study

72

Erklärung zum Datenschutz Während des folgenden Versuches werden Fotos gemacht. (Wenn Sie das nicht möchten, streichen Sie bitte diesen Absatz). Alle Daten, die im Rahmen dieser Studie erhoben werden (Fragebögen, Mitschriften und Notizen) werden streng vertraulich behandelt. Schriftlich erhobene Daten und Informationen werden nur in anonymisierter Form gespeichert. Die Daten werden für die Dauer der Studie und für die für eine Diplomarbeit vorgeschriebene Zeit gespeichert. Danach werden alle erhobenen Daten gelöscht.

Ich erkläre mich damit einverstanden, dass Fotos, die während des Versuchs gemacht wurden, in wissenschaftlichen Publikationen verwendet werden. (Falls Sie dies nicht wünschen, streichen Sie bitte diesen Absatz) Ich erkläre mich damit einverstanden, dass Teile der Aufzeichnungen in wissenschaftlichen Publikationen verwendet werden. (Falls Sie dies nicht wünschen, streichen Sie bitte diesen Absatz) Hiermit bestätige ich, dass ich die Erklärung zum Datenschutz gelesen habe und damit einverstanden bin.

___________________________________________________________________ Ort, Datum und Unterschrift

___________________________________________________________________ Vor- und Nachname in Druckbuchstaben

A User Study

73

Vpnr:______

Fragebogen zur Person Alter:____________ Geschlecht:

männlich

Studiengang:

weiblich

_____________________ seit ____________ Jahren

oder Beruf: Händigkeit: Sehhilfe

_____________________ seit ____________ Jahren Rechtshänder/in Brille

Linkshänder/in Kontaktlinsen

beide Hände gleich keine

Wenn Sie im Studien- oder Arbeitsalltag skizzieren, wie oft skizzieren Sie dann ... .... mit Papier und Stift? täglich

mehrmals in der Woche

etwa 1 mal in der Woche

etwa 2-3 mal im Monat

etwa 1 mal im Monat

seltener als 1 mal im Monat

nie


etwa 1 mal im Monat


nie


nie

... mittels des Computers? täglich



Wie häufig nutzen Sie 3D-Konstruktionssoftware (3D CAD)? täglich




etwa 1 mal im Monat

Wie viel Erfahrung haben Sie mit virtuellen Umgebungen? („Holobench“, „Cave“, „Datenhelme“, andere VR-Systeme) nutze ich nutze ich habe ich schon regelmäßig gelegentlich einmal ausprobiert

habe ich schon einmal gesehen

gar keine

A User Study

74

Anleitung: Skizzenbasiertes Modellieren in 2D Sie sollen nun das 2D skizzenbasierte Modellieren auf dem Tablet-PC üben. Als Skizzierwerkzeug erhalten Sie einen Stift, mit dem Sie auf der Oberfläche des PCs skizzieren können. Um ein Objekt zu skizzieren, zeichnen Sie eine Silhouette. Aus dieser Silhouette wird automatisch ein Objekt erzeugt. Das Objekt kann gedreht werden, indem Sie den Stift mit gedrückter Stift-Taste auf dem Objekt positionieren und dann den Stift bewegen. Außerhalb des Objekts bewirkt dies eine Verschiebung des Objekts. Die Anwendung sieht die Erzeugung eines einzigen Objekts vor, das anschließend manipuliert werden kann. Folgende Funktionen stehen zur Verfügung: 1. Erweiterung: Um das Objekt zu erweitern, zeichnen Sie zunächst eine geschlossene Linie innerhalb des Objekts, z.B. einen Kreis. Drehen Sie dann das Objekt so, dass Sie die Silhouette der Erweiterung zeichnen können. Nach dem Zeichnen wird die Erweiterung automatisch erzeugt. 2. Formveränderung (Ziehen): Wenn Sie die Form eines Objekts verändern möchten, tippen Sie mit dem Stift auf den Button „pull“ an der unteren Seite des Eingabefensters. Nähern Sie sich mit dem Stift der Silhouette des Objekts. Die potentielle Zugriffsstelle wird visuell hervorgehoben. Bei gedrückter Stift-Taste können Sie dann das Objekt durch Ziehen verändern. Um wieder zu zeichnen, tippen Sie auf den Button „draw“. 3. Formveränderung mittels zusätzlicher Linie: Sie können ergänzende Linien auf das Objekt zeichnen. Das Objekt kann hierfür beliebig gedreht werden. Zeichnen Sie eine nicht geschlossene Linie quer über das gesamte Objekt. Die Linie muss ausserhalb des Objekts beginnen und enden. Die Linie wird dann automatisch um das gesamte Objekt gelegt. Tippen Sie zum Abschluss mit dem Stift auf das Objekt. Das Objekt läßt sich dann auch an diesen zusätzlichen Linien durch Ziehen verändern. Durch tippen auf den Button „undo“ an der unteren Seite des Eingabefensters kann die letzte Aktion rückgängig gemacht werden. Eine neue Skizze können Sie anlegen, indem Sie die Anwendung neu starten. Hierzu schließen Sie das Fenster der Anwendung und tippen mit dem Stift doppelt auf die Datei „FiberMesh“. Bitte benutzen Sie nur die beschriebenen Funktionen der Anwendung.

A User Study

75

Anleitung: Skizzenbasiertes Modellieren in 3D Sie sollen nun das 3D skizzenbasierte Modellieren in der CAVE üben. Als Skizzierwerkzeug erhalten Sie einen drucksensitiven Stift. Wenn sie auf den oberen Bügel drücken, können Sie damit „in die Luft“ zeichnen. Um ein Objekt zu skizzieren, zeichnen Sie in einem Zug eine Silhouette, die in etwa in einer Ebene im Raum liegt. Wenn Sie den Stift loslassen wird aus dieser Silhouette automatisch ein rundliches Objekt erzeugt. Auch die Silhouette des Objekts wird dargestellt. Sie können beliebig viele Objekte erzeugen. Als Formveränderungswerkzeug erhalten Sie eine Zange. Wenn Sie die Form eines Objekts verändern möchten, nähern Sie sich mit der Zange der Silhouette des Objekts. Die potentielle Zugriffsstelle wird durch eine rote Kugel visuell hervorgehoben. Wenn Sie mit der Zange zugreifen, können Sie das Objekt durch Ziehen verändern. Sie können ergänzende Linien auf das Objekt zeichnen. Hierzu zeichnen Sie eine nicht geschlossene Linie quer über die ursprüngliche Eingabesilhouette des Objekts. Die Linie muss ausserhalb der Eingabesilhouette beginnen und enden. Die Linie wird dann automatisch um das gesamte Objekt gelegt. Das Objekt läßt sich dann auch an diesen zusätzlichen Linien mit der Zange verändern. Auf dem Tisch vor sich sehen sie einen Schieberegler. Durch Verschieben des Reglers können Sie sich den Entstehungsprozess Ihrer Skizze ansehen. Sie können ihn auch benutzen, um gezeichnete Objekte zu löschen. Bewegen Sie den Regler dazu so lange, bis alle zu löschenden Teile Ihrer Skizze verschwunden sind, und fahren Sie dann mit dem Skizzieren fort.

A User Study

76

Anleitung: Linienbasiertes Skizzieren in 3D Sie sollen nun das 3D linienbasierte Skizzieren in der CAVE üben. Als Skizzierwerkzeug erhalten Sie einen drucksensitiven Stift. Wenn sie auf den oberen Bügel drücken, können Sie damit eine Linie „in die Luft“ zeichnen. Je stärker Sie drücken, desto breiter wird der Strich gezeichnet. Auf dem Tisch vor sich sehen sie einen Schieberegler. Durch Verschieben des Reglers können Sie sich den Entstehungsprozess Ihrer Skizze ansehen. Sie können ihn auch benutzen, um Aktionen rückgängig zu machen. Bewegen Sie den Regler dazu so lange, bis alle zu löschenden Teile Ihrer Skizze verschwunden sind, und fahren Sie dann mit dem Skizzieren fort.

A User Study

77

Aufgabe 1: Abzeichnen eines Objekts Ihre Aufgabe ist es, einen realen Hocker abzuzeichnen. Dieser Hocker wird Ihnen von der Versuchsleiterin gezeigt. Sie können sich den Hocker genau ansehen, danach wird er wieder entfernt. Bitte skizzieren Sie dann den Hocker. Entscheiden Sie bitte selbst, wie viele Skizzen Sie anlegen wollen und teilen Sie der Versuchsleiterin mit, wenn Sie mit der Bearbeitung der Aufgabe fertig sind.

A User Study

78

Aufgabe 2: Entwerfen eines Objekts Ihre Aufgabe ist es jetzt, einen bequemen Sessel nach Ihren eigenen Vorstellungen zu entwerfen. Wenn Sie es wünschen, können Sie sich vorher etwas Zeit nehmen, um über die Lösung nachzudenken. Bitte skizzieren Sie dann den Sessel. Entscheiden Sie bitte selbst, wie viele Skizzen Sie anlegen wollen und teilen Sie der Versuchsleiterin mit, wenn Sie mit der Bearbeitung der Aufgabe fertig sind.

A User Study

79

Perm:______

Vpnr:______

Beurteilung 3D skizzenbasiertes Modellieren Nachfolgend finden Sie Wortpaare, mit deren Hilfe Sie die Beurteilung des 3D skizzenbasierten Modellierens vornehmen können, das Sie gerade kennen gelernt haben. Diese Wortpaare stellen jeweils extreme Gegensätze dar, zwischen denen eine Abstufung möglich ist. Denken Sie nicht lange über die Wortpaare nach, sondern geben Sie bitte die Einschätzung ab, die Ihnen spontan in den Sinn kommt. Es gibt keine richtigen oder falschen Antworten. Kreuzen Sie je Zeile bitte immer ein Feld an. Danach wird die Beanspruchung durch die Skizziertechnik auf zwei weiteren Blättern erfragt. Abschließend folgen zwei Fragen zu Skizzierprozess und Skizze. Die Fragebögen werden nach jeder Aufgabe ausgefüllt. Berücksichtigen Sie bitte beim Ausfüllen jeweils nur die Eindrücke, die Sie bei der entsprechenden Aufgabe hatten.

Bitte umblättern!

A User Study

80

Vpnr:______

Beurteilung „3D skizzenbasiertes Modellieren“ Aufgabe 1 Bitte geben Sie mit Hilfe der folgenden Wortpaare Ihren Eindruck zur Skizziertechnik „3D skizzenbasiertes Modellieren“ für Aufgabe 1 wieder. menschlich isolierend

technisch verbindend

angenehm

unangenehm

originell

konventionell

einfach

kompliziert

fachmännisch hässlich praktisch

laienhaft schön unpraktisch

sympathisch

unsympathisch

umständlich

direkt

stilvoll

stillos

voraussagbar

unberechenbar

minderwertig

wertvoll

ausgrenzend

einbeziehend

bringt mich den Leuten näher nicht vorzeigbar zurückweisend phantasielos gut

trennt mich von Leuten vorzeigbar einladend kreativ schlecht

verwirrend

übersichtlich

abstoßend

anziehend

mutig innovativ lahm harmlos

vorsichtig konservativ fesselnd herausfordernd

motivierend

entmutigend

neuartig

herkömmlich

widerspenstig

handhabbar

A User Study

81

Vpnr:______

Beurteilung „3D skizzenbasiertes Modellieren“ Aufgabe 2 Bitte geben Sie mit Hilfe der folgenden Wortpaare Ihren Eindruck zur Skizziertechnik „3D skizzenbasiertes Modellieren“ für Aufgabe 2 wieder. menschlich isolierend

technisch verbindend

angenehm

unangenehm

originell

konventionell

einfach

kompliziert

fachmännisch hässlich praktisch

laienhaft schön unpraktisch

sympathisch

unsympathisch

umständlich

direkt

stilvoll

stillos

voraussagbar

unberechenbar

minderwertig

wertvoll

ausgrenzend

einbeziehend

bringt mich den Leuten näher nicht vorzeigbar zurückweisend phantasielos gut

trennt mich von Leuten vorzeigbar einladend kreativ schlecht

verwirrend

übersichtlich

abstoßend

anziehend

mutig innovativ lahm harmlos

vorsichtig konservativ fesselnd herausfordernd

motivierend

entmutigend

neuartig

herkömmlich

widerspenstig

handhabbar

A User Study

82

A User Study

83

A User Study

84

Vpnr:______

Zusätzliche Fragen Bitte schätzen Sie den Skizzierprozess und die Skizze anhand der beiden nachfolgenden Aussagen ein. Kreuzen Sie dafür bitte ein Feld pro Aussage an. trifft voll zu

trifft eher zu

teils/ teils

trifft eher nicht zu

trifft gar nicht zu

□

□

□

□

□

□

□

□

□

□

Skizzierprozess: 1

Das Werkzeug hat mich gut dabei unterstützt, meine Vorstellung vom Objekt auszudrücken. Skizze:

2

Die fertige Skizze entsprach meiner Vorstellung vom Objekt.

Shape Modeling with Sketched Feature Lines in ... - helen perkunder

Shape Modeling with Sketched Feature Lines in ... - helen perkunder

Suggest Documents

Feature and Feature Interaction Modeling with Feature-Solution Graphs

Reconciling Feature Modeling: A Feature Modeling Metamodel

Feature Scoping for Product Lines

Web-based Shape Modeling with HyperFun

Integrating Feature Modeling with the RSEB

Shape Modeling with Front Propagation: A Level

Modeling and Mining Complex Networks with Feature

Feature extraction of separation and attachment lines

Qualitative Evaluation of Feature Lines on

Feature Classification for Robust Shape-Based ...

Feature-based shape transformation for polyhedral

Feature-Based Shape Recognition by Support

Feature-based shape transformation for polyhedral objects

Shape Descriptor/Feature Extraction Techniques - UCI Math

Shape of atomic lines emitted by cryoplasma in Helium

Flexible feature binding in software product lines - Infosun

Optimum Feature Selection in Software Product Lines ... - Tim Menzies

Code Clones in Feature-Oriented Software Product Lines - CiteSeerX

Feature Cohesion in Software Product Lines: An Exploratory ... - Infosun

Feature Selection for troubleshooting in complex assembly lines

Considering Feature Interactions in Product Lines - Semantic Scholar

Coupling Transmission Lines for Wave Shape Adjust in High-Voltage ...

Modeling Software Product Lines with AoURN - jUCMNav - uOttawa

Modeling of Transmission Lines with EM Wave ...