Beautification

It is often, particularly in engineering and architectural design, that computer-aided tools are used at a later stage in design to create three-dimensional shapes and objects (van Dick & Mayer, 1997), so designers can explore the three-dimensional entity of objects. Especially with arguments that commercial CAD systems for the manipulation of digital models are still not adequately suited to support the design process (Cheutet et al., 2005) – especially conceptual thinking (Oh, Stuerzlinger & Danahy, 2006), 2-D sketching-based systems are now evolving into systems that offer design environment in a 3-D manner, and support additional functionalities such as 3-D manipulation of sketched objects.

In 2005, Cheutet and his colleagues addressed the way designers express an object shape in 2-D sketches through character lines and how these lines form a basis for sketching shapes in 3-D; hence, the authors proposed a free-form deformation features modeling method to enable 3-D objects handling extended from 2-D sketches.

Turner, Chapman, and Penn (2000) looked at developing a broader 3-D system called Stilton, in which the user is presented with a 2-D drawing plane and drawings may be reconstructed to form solid objects, and thus, allowing a user to navigate 3-D model in different perspectives, such as panoramic photographs (i.e. “sketching out volume around the user”). Turner et al. suggested that such feature of Stilton can greatly assist with some aspects of familiarization and qualitative assessment of sites – particularly useful in construction and architectural design. However, no end-user trials has yet been undertaken to suggest the usefulness of such tool.

Although with the increasing number of systems with 3D sketching environment, such systems require more sophisticated design and implementation to support better usability and accessibility; thus, more work is still needed to address technical issues such as visualization, mathematical and programming issues, as well as recognition and interpretation of 3D sketches.
1.5.3. On Improving Computer-Supported sketch tools

While many researchers have focused on developing new types of computer-supported sketch tools, whether 2-D or 3D, domain-specific or domain-independent, knowledge-based or gesture-based, with different design support capabilities etc, others have been motivated by the critical need to improve on the design, functionalities and implementation techniques, and hence, effectiveness (and usability), of such systems. Examples of current improvement work include, Arvo and Novins’ (2005) success on improving ways of manipulating sketches by preserving sketch appearance, while Nealen (2005), examined editing aspects of sketch (i.e. selection and moving a handle) in which complex shape modeling tasks were greatly simplified. In his sketch-based system (“Sketch Now”), Jonson (2005) attempted to stimulate the feel of using different kinds of paper; however, as predicted, many difficult issues had to be addressed. Oh (2004) made further developments (improvements) based on the Electronic Cocktail Napkin system (Gross, 1996; Gross & Do, 1996), resulting a new sketch-based tool – Design Evaluator – to support architectural design reasoning, but in a three-dimensional manner. Also in the 3-D sketching scene, Oh, Stuerzlinger and Danahy (2006) took a better approach towards 3-D conceptual design systems and developed SESAME (Sketch, Extrude, Sculpt, and Manipulate Easily). Furthermore, the comparison of SESAME with a conventional CAD package demonstrated the effectiveness of SESAME. Dijk and Mayer (1997) argued that CAD systems lack flexibility to modify 3-D design objects – a major obstacle for effective computer support during conceptual modeling; therefore, a new algorithm was developed to allow faster and easier conversion from 2-D to 3-D model space, in which the 3D surfaces can also be directly modified by sketching. Likewise, also with a mathematical approach, Liu, Tang and Joneja (2005) focused more on the technical aspects and presented an alternative method for sketch-based free-form shape modeling, and demonstrated its usefulness and effectiveness by testing the technique with two design applications.

On the whole, one can observe that the future of computer-supported “natural design” can be hopeful with the on-going, rapid development and expansion of existing digital-sketching systems and its related functionalities.
1.5.4. Bridging the Gap: A closer look at ‘Beautification’ (‘Formalization’)

Since nearly two decades ago, Thomassen, Teulings, and Schomaker (1988) had already suggested that it could be an advantage if a single system could be designed that is capable of dealing with different forms of graphemic, symbolic, and graphic output: i.e. a system that would not only be able to read handwritten letters, digits, words, and text-editing comments, but would also read non-alphanumeric characters and symbols, that would interpret formulae and roughly sketched graphics, and that would render these in a neat, orderly fashion, properly grouped and aligned. Maarse, Schomaker and Teulings (1988) went on to further to give a hypothetical example of the automatic processing of combined graphic and cursively written material entered on a graphic tablet. Lin and Pun (1978) also proposed a similar interactive system for the interpretation of hand-sketched line figures such as flow charts and logic circuit diagrams, in which hand-drawn diagrams and hand-writing, are converted into computerized, standardized diagrams and hand-writing.

Future development recommended over twenty years ago is now happening with the emerging trend in sketch-based interfaces for design, and such computer-supported sketch systems are now becoming more accessible, whilst being further explored and improved. Not only do current computer-supported sketch tools bridge the gap between paper and pen and computer-aided design tools, to support natural ways of designing (i.e. sketching on paper), whilst providing the advantages of computer-based tools (e.g. design transformation, editing support and version control); they also offer an interface for processing sketched objects in a useful manner (e.g. Forbus, Ferguson & Usher, 2001; Gross, 1996; Oh, 2004).

One of the fundamental approaches toward sketch processing is the recognition of sketched shapes and objects, for example as predefined components – e.g. as Tang (2005) noted, in the context of Web interface design, hand-written letters can be recognized as “labels” and a hand-drawn circle can be recognized as a radio button (where as in the mechanical engineering domain, a circle may be recognized as a wheel of an automobile). Moreover, the increasing demand and research on digital sketch tools further points out to the need for better recognition of sketched objects and hand-writing (e.g. Agar & Novins, 2003; Bo Yu, 2003, Rubine, 1991), to cater for other useful functionalities, especially for sketch transformation and rendering (Bolz, 1993; Chung, Mirica & Plimmer, 2005; Hong & Landay, 2006; Hse & Newton, 2005; Landay, 1996; Lin & Pun, 1978; Maarse et al., 1988; Newman et al., 2003; Pomm & Werlen, 2004; Plimmer & Grundy, 2005; Schweikardt & Gross, 2000; Shesh & Chen, 2004; Wang, Sun & Plimmer, 2005).

Thus, one of the major challenges embedded in the development of computer-supported sketch-based design tools is the need to “beautify” sketched content (i.e. beautification), both during sketching-based input and during conversion to computer-rendered form (Plimmer & Apperley, 2004; Plimmer and Grundy, 2005).
1.5.4.1. ‘Beautification’ versus ‘Formality’

The term ‘beautification’ was originally used in urban planning, and now, in the context of computer-supported sketch tools, beautification generally refers to the process of tidying up a hand-drawn diagram (Plimmer & Grundy, 2005). In the wider context of the design process, designers typically sketch on paper and later transfer the design onto CAD tools, which can be viewed as a beautification process; however, it was criticized that such transferring process from paper to the computer is error-prone and unproductive (Tang, 2005; Young, 2005). In contrast, in terms of computer-supported sketch-based design tools (with a fundamental goal to bridge the gap), such beautification process is downsized into requiring only one design medium, and at the same time, eliminating the transferring process – i.e. rough, untidy, sketched input can be transformed to appear neater, tidier and more formal, by using a single design tool. The idea behind beautification is similar to text recognition where hand-written input is recognized as text form and could be transformed into computer fonts; hence, with sketched objects, hand-drawn input is recognized and rendered into a higher level(s) of representation (e.g. morphing of hand-written characters into fonts, straightening of lines, 3-D rendering from 2-D objects) – better referred to as ‘formality’ in the context of this study.

‘Formality’ can be described as the outcome of beautification, and incorporates both concepts of precision and fidelity (as described in the section on prototypes). Formality is defined, in this study, as the level of tidiness and professionalism conveyed in the appearance of a design (prototype). This definition of formality is based on previous studies on website design practices in which web site designers were observed to design sites at different levels of refinement – site map, storyboard, and individual page – and that designers sketch at all levels during the early stages of design, ranging from rough sketches to tidied-up diagrams for (re)presentation to clients, to computer-rendered designs (Lin et al., 2000; Newman et al. 2003). This concept can be illustrated in the context of web interface design: when sketching a text box, the designer specifies the element size (rough and uneven or same), position of each field (roughly spread or aligned), and titles for each label and shape of each field (neatness of the hand-writing and lines) all of which constitutes the overall appearance (tidiness/neatness) of a design. A low-formality design would appear rough and sketchy (e.g. wiggly shape of a object), with imprecision in the design (e.g. elements and labels unaligned) and would not look like the final version of the design (low-fidelity); where as, a high-formality design would appear tidy and professional (e.g. with straight smooth lines, alignment of labels and elements), and with precision (e.g. location of elements) and would closely resemble the final version of the design (high-fidelity). In addition to precision and fidelity, the definition of formality also includes aspects of legibility and readability. According to Watzman’s definition (2003), readability refers to the ability to find what you need on the page while legibility refers to the ability to read it when you get there. Furthermore, topography could make a design look more or less formal, by being more or less readable and legible. Legibility is determined by many characteristics including typeface; letter spacing, word spacing, line spacing; justified vs ragged columns; movement; colour; viewing environment etc; thus, in the context of this study, a low formality design consist of untidy hand-writing, as opposed to standardized, rendered text (e.g. fonts) in a high formality design. As elements in the design can be tidied-up, based on principles of Gestalt psychology (Koffka, 1935) of perceptual organization (e.g. grouping, proximity, similarity among others), Brink et al. (2002) suggested that it can help an individual towards a higher-level of comprehension of the display; hence, the increase in legibility and readability – i.e. formality.

Overall, the level of formality of a design’s appearance is affected by the level (extent) of beautification applied to the original design. In other words, one can increase the formality of a design through beautification, so the beautified design appears tidier, neater and comparatively more aesthetically pleasing (e.g. aligned elements and straightened lines) than the non-beautified, original hand-drawn design.

In summary, beautification is a process, and formality is the outcome of beautification; both of which are important concepts in computer-supported sketch-based design tools.
1.5.4.2. Practicality of Beautification in the design process

Although the practicality of beautification in sketch-based interfaces is still yet to be scientifically assessed in professional design settings, based on qualitative and anecdotal evidence from design industry professionals, one can argue that beautification in a computer-supported sketch-based environment can be useful.

A typical design situation, suggested by several authors (e.g. Brink et al., 2002; Newman et al., 2003), is during the early stages in the design process of project development when clients are involved – particularly, the presentation of ideas and conceptual design to clients. Brinck, Gergle, Wood, (2002) pointed out that paper mock-ups and/or prototyping may elicit clients’ negative responses such as “unprofessional” look and feel of early design examples, and also, because early designs (prototypes) are deliberately incomplete, clients can sometimes be distracted by the fact that not every project requirements have been fully implemented, and may be led to conclude that important elements have been forgotten. With this in mind, designs for client presentation should appear processional, informal, but not too rough and sketchy.

In addition, the fact that designers do want clients to feel that a design finished, and are reluctant to show their sketches (Newman et al., 2003), beautification can play an important role in situations involving client presentation (as well as group collaboration). For example, it can help tailor the needs for the presentation of designs with the desired level of formality. In addition, efficiency is increased during the nearing of the deadline by eliminating the needed to transfer paper design onto the computer (e.g. by scanning or recreating the design from scratch), because hand-drawn-sketches are in the computer (tablet) i.e. digitized; and in a few taps/clicks on the beautification icons in the sketch system, rough sketches are then beautified (to the extent required).

It has become apparent that recently, there has been an increasing interest in developing tools and techniques to support the recognition and interpretation of 2-D sketched shapes, which is then rendered and expanded into 3-D objects/models – for example, Zenka and Slavik (2003) developed a system for creating ‘smart’ sketches of free form 3D objects. With the help of the system, a 2-D sketch of an object is enriched with information about its 3-D structure through rendering, and the user can further rotate the created sketch of the object and view it from various angles. Zenka and Slavik argued that their system is very easy to use since most of the user input is a traditional 2-D sketch, and that it doesn’t limit the user’s creativity – based on the notion that sketching facilitates creativity (e.g. Goel, 1995; Verstijnen, et al, 1998). A comparable sketch-based application, SKETCH, was developed by Zeleznik, Herndon, and Hughes (2006) to support an environment for sketching and editing 3D scenes. SKETCH uses simple non- non-photorealistic rendering based on primitive line drawings, in which, due to the gestural interface, operations can also be specified within the 3D environment. Other recent example of 2-D to 3-D rendering techniques include 3-D botanical tree modeling (e.g. Okabe, 2005); conversion of 2-D sketched objects into 3-D intended geometric shapes/objects in engineering design (e.g. Kim & Kin, 2006; Shesh & Chen, 2004) based on fuzzy logic (e.g. Qin, Wright, & Jordanov 2000; 2001); and 3-D models derived from 2-D free-hand drawings in architectural design (e.g. Schweikardt & Gross, 2000). However, as many issues and problems are yet to be solved (e.g. accurate rendering, efficient underlying mathematical algorithms, and requirements for sketch-based 3-D environment); and thus, such work can be seen as the beginning for beautification techniques that render 2-D sketches into 3-D models, and systems that support it.

Furthermore, a number of studies on the holistic development of computer-supported sketch systems that undertake different forms of sketch beautification have also shown some promising results. For example, earlier in 1993, Bolz focused on implementing a tool, AssistenzComputer, that supports (semi-) automatic beautification of drawings (Bolz, 1993), which was an extension to a conventional graphics editor. AssistenzComputer analyses diagrams that consist of straight line segments using a user-defined knowledge-based approach, and it also looks for flaws such as gaps and miss-alignments, which are smoothed out in the beautified version of the sketch content. Hse and Newton (2005) developed a complete sketch system for recognizing and beautifying symbols, with its aimed to be simple and convenient by allowing the user to sketch the desired shape directly and then replace it with a beautified symbol with the correct transformation, all in one step. Interestingly Hse and Newton implemented such system as an interface to the Microsoft Powerpoint application which makes it possible for a user to sketch symbols directly onto a presentation slide.

SILK (Landay, 1996; Landay & Myers, 2001), a well-known sketch system, is a tool aimed at the early stages of design, particularly for user interface design, when designs are typically sketched rather than prototyped in software. Even in its sketched form, the user interface is functional; for example, sketched buttons can be pressed and tools can be selected in a toolbar. The sketch can also be transformed into an actual interface where the design’s formality (appearance) is increased, and become a higher-fidelity prototype (resembling the ‘look-and-feel’ of the final system (e.g. a user interface, a HTML page etc). Thus, SILK enables the designer to test the design at any point, which enables iterative design. In a similar manner as SILK, SATIN (Hong & Landay, 2006) is a Java-based toolkit designed to support natural sketching in the design process. While enabling beautification manipulation and rendering of objects by using specific libraries, recognizers, interpreters, and multi-interpreters to handle pen-input, SATIN also provide support for zooming and rotating objects, as well as switching between multiple views of an object. Ideogramic (Damm & Hansen, 2002), a sketch-based case-tool, allows the user to choose the level of beautification desired. The hand-drawn input can be left unaltered or recognized and transformed into formal UML (Unified Modelling Language) shapes such as classes (rectangle shapes with lines and text). Informal (hand-drawn) and formal (computer-rendered, beautified) elements can coexist in the same diagram (design). Additionally, the system also supports layout beautification (e.g. alignment).

In an extensive study, Newman et al. (2003) developed DENIM – an informal sketch tool for Website design, based on the findings in their study to investigate on Web site design practice (see discussion in Newman et al. 2003 for more details on their findings). Overall, it supports sketching input, allows a designer to view and edit designs at different levels, for example an individual Web page versus a whole Web site map, through zooming. Designers are also able to interact with their sketched design as if in a Web browser, thus allowing rapid creation and exploration of interactive prototypes. Regarding beautification, DENIM recognizes simple symbols such as rectangles and lines, which forms the basis for beautification – which varies depending on the level the user is working at, for example in storyboard view, a line drawn to indicate navigation between pages is smoothed and has a dot added to the source point and an arrow at the destination point. Moreover, Newman and his colleagues had made various improvements in DENIM based on an evaluation study with professional designers in which valuable feedback and data was collected and analyzed.

InkKit, an informal sketch-based design tool was developed by Young (2005) and Tang (2005) in collaboration – Young (2005) focused on the flexible and accurate sketch recognition, while Tang (2005) examined beautification aspects of the system. According to Tang and Young, the process of transferring a design from paper and pen to the computer is error-prone and unproductive, and therefore the main aim of InkKit is to address such issue. Additionally, beautification was referred to, by Tang, as “design transformation” – a function which automates such transferring process by converting the sketched input to produce the computer-equivalent output diagram for the appropriate domain (e.g. interface design, floor plans, electrical circuits, flow diagrams). Usability testing on InkKit showed positive results, showing that users were very satisfied with the application based on learnability, understandability and user satisfaction. Chung, Mirica and Plimmer (2005) further explored InkKit by adding more editing support and beautification functions. In their version of InkKit, additional beautification functions include: standardization of recognized components into predefined sizes based on taxonomies specified by the user; vertical and horizontal alignment of elements in groups; and components can be snapped to the grid. By default, all components of a sketch are recognized and beautified as a whole; however, as noted by Chung et al., future development of beautification techniques could include the beautification of selected components only, by adjusting related algorithms. With its accessibility and much of the basics done, hence, InkKit had form basis for exploring beautification this study.

On the whole, many computer-supported sketch systems have been developed to bridge the gap between the use of paper and pen, and computer-based tools. Results from Bailey and Konstan’s (2003) study further points out to the great future by showing that the use of their informal design tool (DEMAIS) is more effective in comparison to using pencil and paper and and Authorwware (CAD tool) for exploration and communication behavior in early multimedia design. Also, sketch-based tools do not only support the development of the final product, but it also aids the whole design process (e.g. needless for the time-consuming, error-prone process of transferring paper designs onto the computer); thus, beautification has become an important aspect to include in sketch-based tools. It can be concluded that different types of sketch-based applications may require different beautification techniques, which can be used at different phases during the design process (Plimmer & Grundy, 2005). In other words, the design process can be enhanced by tailoring the type of beautification and the time it is used according to the specific nature of the design task.

4. 
   Related studies: Interaction with hand-drawn versus computer-rendered diagrams
   

Much of the research on informal sketch-based design tools has been on the development and implementation of such systems to bridge the gap between the paper and pencil and computer paradigms; and thus, important aspects of sketch-based systems like recognition and beautification have also been frequently examined. However, there has been a lack of empirical research on designers’ interaction with such sketch tools and almost none on the effects of design formality (beautification outcome) on designers’ cognition and performance in the design process. Current research on such area predominantly consists of anecdotal and non-empirical work that are descriptive and prescriptive in nature; however, most worryingly, only two directly related (empirical) studies were found in the literature that fit into the topic of research on the effects of design formality (i.e. Black, 1996; Plimmer & Apperley, 2004).

Research has shown that designers interact differently with hand-drawn diagrams (low formality design) compared to tidy, formal, computer-rendered diagrams (high formality design). As noted in the previous section on paper and digital prototypes, anecdotal evidence from professional designers (e.g. Beaudouin-Lafon & Mackay, 2003; Brink et al, 2002; Newman et al, 2003) suggested that paper prototypes (low formality) are viewed as a conceptual, unfinished a product – thus, implying that paper prototypes tend to provoke more comments (e.g. suggestions for changes, perceived inadequacies). In contrast, digital prototypes (high formality) are typically perceived as a completed, final, unchangeable product, and thus, clients/reviewers tend to question less and make fewer suggestions for changes, especially on functional aspects of the product, and tend to focus on the details related to the final product (e.g. aesthetics). Likewise, as Plimmer and Apperley (2003) noted, a tidy design implies, perhaps incorrectly, that the design is committed and complete. In addition, Wong (1992) and Wagner (1990) also noted that informal, hand-drawn designs are better for eliciting comments from others, compared to formal, computer-rendered designs.

Empirical evidence to support such anecdotal evidence is very limited – only two relevant empirical studies were found that examined the effects of interaction with low formality designs and high formality designs (i.e. Black, 1990; Plimmer and Apperley, 2004). The two studies formed the basis for the present study.

Results from Black’s (1990) study questionnaire indirectly confirmed that the appearance (formality) of a design may affect designer’s performance, hence, design outcome. In her study, Black compared visible planning on paper and on screen in twenty-nine first year students (novice graphic designers), with a focus on students’ assessment of the experience and outcomes of their work. Following the completion of a training session on relevant aspects of text manipulation as well as basic word processing using specific Mac text editors, students completed a text design exercise in which students planned the same text twice, working once on paper and once on screen. Half the students worked on paper first and then moved on to screen, and the other half worked in reverse order to control for order effects. The questionnaire was completed twice by students: once immediately after the design exercise when students had had no formal feedback from staff about the final drafts; and once after a class review in which students had displayed their final drafts for discussion by their course tutor. One of the main findings was that before the review of designs by the tutor, students were significantly more satisfied with their final drafts on screen than their final drafts on paper, however, after the review, students’ satisfaction with the paper design was higher than the screen design. Such finding suggested that students may have perceived the final draft on screen as a finished design that required no more improvements, as opposed to the paper design drafts which appeared rough and informal, which may have been perceived as non-finished designs that required more exploration and modification, thus students perceived it as a less satisfying design compared to the design on screen. Furthermore, students’ exploration of ideas may have been limited by the ‘finished’ feedback (appearance of the screen) they got from their screen draft.

However, a major limitation of the study was that it provided only an indirect measure of the impact of medium (i.e. formality of designs) on visible planning (i.e. designing; the design process). Future research to tackle such limitation might be to objectively address the processes and products of designing on paper and screen and computer, and more specifically, to examine and compare the number and quality of solutions proposed in each medium, and maybe, the assessment of end product of designing in each medium – such aspects will help unravel the effects of formality of designs on designers’ performance in design (e.g. design decision making and design quality). However, one must note that the process of designing in the two media may not always be informative, as tasks performed during the use of the two media differ. Furthermore, the problem of subjective definitions of what constitutes a design solution, may affect the validity of the comparisons of the number and quality of design solutions produced in each medium. Other minor limitations of the Black’s study include the lack of experimental procedures to control for extraneous variables and the simplistic questionnaire with only six questions that made one cautious about its validity. Moreover, in the context of questionnaire studies, the number of subjects in Black’s study (n=25) was small, thus, sufficient power for the study may not have been achieved, which in turn, further limits the conclusiveness of such study.

Similarly, yet differently, another study examined and compared designers’ interaction with rough, sketchy informal diagrams and computer-rendered formal diagrams (Plimmer & Apperley, 2004). Unlike Black (1990) who focused on traditional media (pen and paper versus computer text editor), Plimmer and Apperley concentrated on different design tools. In their study, interface designs were created for two applications: a form for a book catalogue and a credit card application form. Each of the design was presented in two ways: 1) a rough, hand-drawn version (low formality) of the design created in Freeform – an informal sketch-based design tool; and 2) a formal, computer-rendered version (high formality) of the design created with Visual Basic (VB) form designer – a computer-based interface design tool. According to Plimmer and Apperley, six small groups of subjects (two or three people all from a second year university programming course) participated were asked to review the designs on the medium in which it was presented on (either Freeform or the VB form design). For each application design, subjects were given a brief problem statement (description and requirements of the design) and two scenarios (what end-users might fill in on the two forms). Subjects were asked to check the design given against the application description and requirements, and to use the scenarios provided to fill in the form, as if they were the form-filler (subjects were instructed that modifications to the designs were to be made within the same medium the design was presented on). In addition, subjects’ skills were comparable – as noted, all subjects were familiar with VB but not with Freeform (thus, received five minutes of on the use the basic sketching features).

The applications (book catalogue and credit card forms) were carefully specified by Plimmer and Apperley so that the number and type of elements in it were comparable in each application. For example, each application included an option pair typically represented by radio buttons along with their associated labels (i.e. the book catalog included the specification of book binding as hardback or paper back, while the credit card application form included applicant gender). Each application also included a selection from a list set typically represented by a dropdown list (i.e. the book catalog specified a book’s genre by selecting from a dropdown list, while the credit application specified applicant income described in a set of specified ranges.

Finally, the main results found was that the number of changes made to the designs was significantly different when using Freeform compared to VB – five out of six groups made significantly more changes in Freeform than in VB, regardless of the application or the order of the design exercise. Plimmer and Apperley further added that that nearly all changes were improvements of the design. Another important finding in Plimmer and Apperley’s study was that the time spent on different types of activity differed significantly between the use of Freeform and VB, which was particularly true in terms of the speed of making changes to the design. Although more skilled in using VB, it took subjects longer to complete the design exercise in VB. Plimmer suggested that this may be due to the disproportionate effort on keeping the design tidy by aligning, sizing elements, as opposed to the focus on the design task (i.e. the design of the application) during the use of FreeForm (informal-sketch based design tool).

However, although the design and manipulation of stimulus (i.e. the designs and their specifications) was controlled, the design task was relatively simplistic, especially in the context of group design, as the changeable aspects, particularly the number of ‘errors’ in designs presented to the subjects, were both little. Hence, the slight difference in the number of change may indicate a significant real effect – plus the question of external validity. Therefore, for future replications of Plimmer and Apperley’s study, it maybe more effective to include a few more elements in the designs (stimulus) to create a more stimulating design task, which also enables differences to be observed in a clearer and more accurate manner. Another limitation of Plimmer’s study was the uncontrolled group situations (groups of two or three people) when working on design tasks, which reflects to the little attempt made to account for the group processes in the design tasks – perhaps studying individual design process could provide more robust, reliable and interpretable data.

5. 
   The Present study: Aims and hypotheses
   

The main purposes of this study were to use an experimental approach to: 1) further explore the concept of beautification in the context of sketch-based design tools by examining the dimensionality of beautification and its techniques; and 2) to also investigate the effects of design formality (beautification outcome) – from rough (hand-drawn, non-beautified) sketches to formal (beautified) diagrams – on design performance. More specifically, this study explores the effects of different formality level of designs on design performance, by measuring the number of changes made: total changes, quality changes and expected changes made (described in Section 2: Method). Also taken into account are factors that may play a role in affecting the relationships between formality level of designs and design outcome, including expertise (e.g. design experience, education, domain-specific knowledge), design perception and design medium. In addition, the present study has a particular focus on the early stages of the design process because of their determinant influence on design costs and design outcomes.

1. 
   That the number of functional changes made (total, quality and expected changes) will differ as levels of formality of a design increase (or decrease).
   
2. 
   That the number of functional changes made (total, quality and expected changes) will differ between ‘experts’ and ‘novices’ as levels of formality of a design increase (or decrease). More specifically:
   
   1. 
      That there is a difference in design performance between subjects with task related design experience and subjects with no design experience to some non-task related design experience.
      
   2. 
      That there is a difference in design performance between subjects with task-related domain-specific knowledge (i.e. study major/specialization) and subjects with no task-related domain-specific knowledge to some task-related domain specific knowledge.
      
   3. 
      That there is a difference in design performance between subjects with higher education (study) level (university graduates/postgraduates) and subjects with lower education level in comparison (university undergraduates).
      
3. 
   That subjects will enjoy working on designs that appear more formal (higher formality – i.e. more beautified) more than designs that appear less formal, rougher with a sketchy look-and-feel (lower formality – i.e. less beautified).
   
4. 
   That there is no difference in preference between designing on paper compared to designing on the tablet PC (InkKit).
   
5. 
   That design medium/tool preference in real world design situations would be more diverse than design medium/tool preference during the experiment.