semantics-supported collaborative creation of interactive 3d...

17
Semantics-Supported Collaborative Creation of Interactive 3D Content Krzysztof Walczak (B ) Pozna´ n University of Economics and Business, Niepodleglo´ sci 10, 61-875 Pozna´ n, Poland [email protected] http://www.kti.ue.poznan.pl/ Abstract. Interactive 3D techniques offer a new quality in the design of user interfaces of computer applications, in particular when 3D applica- tions are accessible remotely over the internet. A critical element for building practical 3D applications is adequate 3D synthetic content. However, designing high-quality interactive 3D content is a complex and challenging task. This difficulty is one of the major obstacles preventing widespread use of 3D techniques on the web in many promising appli- cation domains. In this paper, a new approach to 3D content creation is presented. The approach employs a range of techniques for simplification of the content creation process, but its main novelty lies in the fact that properly structured content can be created by communities of design- ers, regardless of whether they explicitly collaborate or not. Appropriate organization and consistency of the user-contributed content is enforced by the use of semantic web techniques with external domain ontologies. The presented approach is consistent both with the existing content cre- ation pipelines and business models, aiming at enhancing and not replac- ing the current practices in digital content creation. A prototype of a tool for 3D content design implemented as an extension to the Unity IDE is also presented. Keywords: 3D content · 3D web · Semantic web · Content creation · Unity IDE 1 Introduction Increasing use of interactive three-dimensional techniques, such as virtual reality (VR) and augmented reality (AR), for building rich multimedia human-computer interfaces has been made possible by significant advances in the performance of computing equipment (including mobile devices), steadily increasing throughput and availability of computer networks and increasingly sophisticated forms of content presentation and interaction with the content on end-users’ equipment. The development of graphics hardware and applications based on three- dimensional interactive user interfaces is largely driven by the rich and rapidly growing market of computer games. However, these techniques can be – and c Springer International Publishing AG 2017 L.T. De Paolis et al. (Eds.): AVR 2017, Part II, LNCS 10325, pp. 385–401, 2017. DOI: 10.1007/978-3-319-60928-7 33 Walczak K. (2017) Semantics-Supported Collaborative Creation of Interactive 3D Content. In: De Paolis L., Bourdot P., Mongelli A. (eds) Augmented Reality, Virtual Reality, and Computer Graphics. AVR 2017. Lecture Notes in Computer Science, vol 10325. Springer, Cham

Upload: others

Post on 21-Jul-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creationof Interactive 3D Content

Krzysztof Walczak(B)

Poznan University of Economics and Business,Niepodleg�losci 10, 61-875 Poznan, Poland

[email protected]

http://www.kti.ue.poznan.pl/

Abstract. Interactive 3D techniques offer a new quality in the design ofuser interfaces of computer applications, in particular when 3D applica-tions are accessible remotely over the internet. A critical element forbuilding practical 3D applications is adequate 3D synthetic content.However, designing high-quality interactive 3D content is a complex andchallenging task. This difficulty is one of the major obstacles preventingwidespread use of 3D techniques on the web in many promising appli-cation domains. In this paper, a new approach to 3D content creation ispresented. The approach employs a range of techniques for simplificationof the content creation process, but its main novelty lies in the fact thatproperly structured content can be created by communities of design-ers, regardless of whether they explicitly collaborate or not. Appropriateorganization and consistency of the user-contributed content is enforcedby the use of semantic web techniques with external domain ontologies.The presented approach is consistent both with the existing content cre-ation pipelines and business models, aiming at enhancing and not replac-ing the current practices in digital content creation. A prototype of a toolfor 3D content design implemented as an extension to the Unity IDE isalso presented.

Keywords: 3D content · 3D web · Semantic web · Content creation ·Unity IDE

1 Introduction

Increasing use of interactive three-dimensional techniques, such as virtual reality(VR) and augmented reality (AR), for building rich multimedia human-computerinterfaces has been made possible by significant advances in the performance ofcomputing equipment (including mobile devices), steadily increasing throughputand availability of computer networks and increasingly sophisticated forms ofcontent presentation and interaction with the content on end-users’ equipment.

The development of graphics hardware and applications based on three-dimensional interactive user interfaces is largely driven by the rich and rapidlygrowing market of computer games. However, these techniques can be – andc© Springer International Publishing AG 2017L.T. De Paolis et al. (Eds.): AVR 2017, Part II, LNCS 10325, pp. 385–401, 2017.DOI: 10.1007/978-3-319-60928-7 33

Walczak K. (2017) Semantics-Supported Collaborative Creation of Interactive 3D Content. In: De Paolis L., Bourdot P., Mongelli A. (eds) Augmented Reality, Virtual Reality, and Computer Graphics. AVR 2017. Lecture Notes in Computer Science, vol 10325. Springer, Cham

Page 2: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

386 K. Walczak

in fact are – successfully used also in other areas, such as rapid prototypingof products, architectural design and visualization, as well as simulation andtraining.

An entirely new level of applicability of the three-dimensional techniques isachievable by the development of standards that enable describing interactivesynthetic multimedia content in a way which is independent of hardware andsoftware, and publishing such content on the web. Therefore, extensive researchand standardization effort has been recently conducted aiming at the devel-opment of universal data formats for describing interactive three-dimensionalVR/AR content available on the web. These works led to the development ofstandards, such as VRML/X3D [45,46], MPEG-4 [18], COLLADA [19], andU3D [10]. With these standards it is possible to encode, transmit over the net-work, and present on the end-user’s equipment three-dimensional high-qualityinteractive multimedia content. A number of imperative programming librariesemploying web standards, such as HTML5, JavaScript, jQuery, SVG and CSShave been also designed. Prominent examples include WebGL [20], Three.js [31],Sprite3D.js [29], Phoria.js [26], Voxel [37] and Photon [15]. Also, X3DOM [14] isa notable example of the use of such imperative programming libraries to enablepresentation of declarative content (X3D) in web browsers.

Users can use virtual environments available over the network in the sameway as they can use their local counterparts. The importance of this technol-ogy is difficult to overestimate. Remote access to three-dimensional interactivemultimedia content and services enables creation of a wide range of services andapplications, such as access to geographically dispersed cultural content, travelguides, e-learning systems, e-commerce applications, social and entertainmentservices, and tele-work environments. The use of VR/AR techniques not onlycan lead to the implementation of better versions of already available services,but can also enable the implementation of entirely new services, which are notpossible without this kind of technology. It has been proven that in many casesviable business models accompany such developments.

Availability of universal standards enabling content description has much fur-ther reaching implications than just the ability to access pre-designed contentover the network. In conjunction with the availability of easy to use and inexpen-sive tools for content creation, it may open the possibility of social 3D contentco-creation by users that are both producers and consumers (prosumers), as itis possible today in the classical “two dimensional” web, leading to the develop-ment of Social 3D Web or “3D Web 2.0”. Content creation is understood hereas the design and programming of basic content elements as well as combiningelements of previously created content – including those created by other users.The possibility of involving large groups of prosumers in content creation is crit-ical for the development of content, which is interesting for users, up-to-date andof adequate quality.

However, creation, searching and combining distributed three-dimensionalinteractive content is much more complex and challenging than in the case ofstandard web pages. The relationships between components of an interactive

Page 3: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 387

three-dimensional virtual scene may include, in addition to its basic meaningand presentation form, also spatial, temporal, structural, logical, and behavioralaspects. This problem can be – at least to some extent – tackled by the use ofsemantic web techniques.

Research on the Semantic Web has been initiated by T. Berners-Lee andthe W3C (World-Wide Web Consortium) in 2001 and led to the developmentof standards such as RDF/RDF(S), OWL and SPARQL. The goal is to trans-form the current web into a distributed semantic database, linking structuredcontent and documents (Linked Data). Semantic descriptions enable achievinga new quality in building web applications that can “understand” the meaningof particular components of content and services as well as their relationships,leading to better methods of creating, recombining and presenting content.

Even for content with relatively low structural complexity, such as HTMLpages, the use of semantic techniques provides a significant increase in the abil-ity to automatically process and integrate the content. In the case of contentwith a high degree of structural complexity, such as interactive 3D componentsand virtual scenes (e.g., encoded in X3D), where the variety of types of relation-ships between different content components is much higher, the use of semantictechniques can help to develop an entirely new class of distributed multimediaapplications, which can be described as 3D Semantic Web.

In this paper, a new approach to 3D content creation, called SemFlex, ispresented. The approach employs a range of techniques for simplification of thecontent creation process, but its main novelty lies in the fact that highly struc-tured 3D content can be created by communities of developers, regardless ofwhether they explicitly collaborate or not. Appropriate organization and con-sistency of the user-contributed content is enforced by the use of semantic webtechniques with external domain ontologies. The presented approach is consis-tent with the existing content creation pipelines and business models, aimingat enhancing and not replacing the current practices in digital content creation.The overall architecture of the Semantic Content Design Environment based onSemFlex is also presented, together with a prototype of a tool for semantic 3Dcontent design implemented as an extension to the Unity IDE [34].

The rest of this paper is organized as follows. In Sect. 2, a review of the stateof the art in the field of 3D content creation is presented. In Sect. 3, the con-cept, the requirements and the overall architecture of a distributed environmentfor semantic 3D content creation are discussed. Section 4 describes implementa-tion of the Semantic Content Design Environment and provides a step-by-stepcontent design example. Finally, Sect. 5 concludes the paper.

2 State of the Art

Different approaches to the simplification of the 3D content creation processhave been elaborated. All these methods aim at reducing the amount and thecomplexity of information that designers need to provide to create 3D content.This can be achieved in several ways, as presented below.

Page 4: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

388 K. Walczak

2.1 3D Scanning

Automatic or semi-automatic 3D scanners can be used to capture the geome-try, the appearance and the movement of objects [2]. Specialized active scannersbased on laser ToF, triangulation, photogrammetry and structured light analysisare available [4]. More affordable – but less precise – techniques enable 3D recon-struction from series of images – often performed in the cloud, e.g., Autodesk123D and 3DSOM. Scanners largely simplify content creation and produce high-quality naturally-looking 3D models. Scanning is useful for creating digital rep-resentations of real objects and places. However, scanning cannot be used formodeling objects and places that do not exist, e.g., because they were destroyedor are simply a product of designer’s imagination. Also, scanning does not helpin modeling interactivity and behavior of objects. As a result, 3D scanning isoften used in combination with other methods of content creation, which allowdesigners to have more influence on the process and the created content.

2.2 Visual Content Modeling

A number of visual environments have been developed for 3D content modeling.Examples of software designated for traditional modeling or sculpting of 3Dobjects include: 3ds Max, Blender, Maya, Modo, 3D-Coat and Zbrush.

Advanced environments – intended for professional users – offer rich capabili-ties of designing various content elements, but their complexity requires expertisein 3D modeling. 3D content creation may be facilitated by narrowing the domainof application and the set of available operations. Environments of this kind(e.g., Sweet Home 3D, AutoCAD Civil 3D and Ghost Productions), designedfor domain experts, provide tools enabling relatively fast and efficient model-ing, without requiring users’ extensive experience in 3D content creation. This,however, significantly reduces the generality of the content creation process.

2.3 Parameterized and Procedural Modeling

Creation of 3D content can be simplified by using content templates that enableautomatic generation of content on demand [41,44]. Templates can be parame-terized, so that content designers can influence the content creation process. Theuse of templates reduces flexibility of content design, but it enables creating highquality content, while keeping the amount of information required at the contentcreation stage relatively low. There is an obvious trade-off between the flexibilityand the simplicity of content creation with templates. Templates are also usefulfor visualization of various types of data in 3D. Templates often employ elementscreated with the use of 3D scanners and visual modeling environments, and areoften used in combination with other content creation methods.

Sophisticated general-purpose procedural modeling software packages enableemploying parametrization in the creation of various types of objects, e.g.,Houdini [27], City Engine [11] and Vue [9]. Specialized procedural softwareenables employing parametrization in the creation of selected types of objects,

Page 5: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 389

e.g., SpeedTree [17], Plant Factory [8], Terragen [24], Acropora [38], Meta-Elements [22].

2.4 Content Models and Patterns

Increasing number and structural complexity of content generation parame-ters require the development of content models – well defined structures, whichdescribe how content is organized [40]. Based on such models, content generationsoftware can automatically create final form of the content. The main benefits ofusing content models are better organized and easier to understand and maintaincontent structures, automatically verifiable data consistency, and elimination ofredundancy in content representation. Content models can be stored in differ-ent types of repositories, depending on particular system requirements. Contentpatterns add another conceptual layer on top of the content models, providingwell-defined roles for specific elements of a content model [23,25].

2.5 Content Componentization

Instead of using fixed content models, rules of combining different types of con-tent elements can be defined, allowing designers to flexibly set up content frompredefined building blocks – components [5,36,39,40]. Components may includegeometrical objects, sounds, scenarios, interaction elements and others. Althoughcreation of content based on configuration of components constrains possiblefinal forms of the created content, the process is simpler and more efficient thancreating content from scratch, while usually still providing more flexibility thancontent templates and content models.

2.6 Separation of Concerns

Further simplification of content modeling requires separation of concernsbetween users with different expertise, who are equipped with different mod-eling tools. For example, limitations of component-based content creation canbe mitigated by providing means of extending the library of the available com-ponents. This can be achieved by assigning different roles to different contentdesigners. A non-expert content creator may build virtual scenes by assemblingready-to-use components. It is relatively easy to compose a scene, but the processis constrained. New content creation capabilities can be provided by introducingnew types of components. This task can be performed by programmers or 3Ddesigners.

The presented solutions significantly facilitate modeling of 3D content, butthey still have important limitations, which result from their orientation on themodeling of content instead of the modeling of concepts. As a consequence, 3Dcontent modeling experts do not use methods specific to the particular applica-tion domain, but methods specific to 3D graphics. It is also necessary – at somestage – to describe all components of the created 3D content.

Page 6: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

390 K. Walczak

2.7 Semantic Modeling

Further progress in modeling of 3D content is possible through the use of seman-tic web techniques [1,28,35,47]. Semantic web techniques enable modeling 3Dcontent using high-level concepts from a particular application domain, instead oflow-level concepts specific to 3D graphics. Inference can be used to enhance andsimplify the content creation process. Also, semantic modeling enables design-ing content in a platform independent way. A number of approaches have beenproposed to enable modeling of synthetic 3D content with the use semanticweb techniques [3,7,12,21,42,43]. A comprehensive study of the use of semanticweb techniques for 3D content representation and modeling has been presentedin [13].

2.8 Social Content Creation

Availability of easy to use methods of content creation, in particular methodsbased on the semantic web, opens the possibility of social 3D content co-creationby users that are both producers and consumers (prosumers), as it is currentlypossible in the “two-dimensional” Web 2.0. There are number of on-line 3D con-tent sharing sites with large and active communities of users. Examples includeTurbosquid [32], Unity Asset Store [33], 3D ContentCentral [6], Highend3D [16]and many others (e.g., Creative Crash, CG People Network, Falling Pixel, 3dExport, The 3D Studio, Evermotion, Renderosity, 3D Ocean, Daz 3D).

In some cases, 3D object library websites enable obtaining the required 3Dmodels in a quick and easy way. However, this is typically only true, if theuser’s requirements are modest and flexible. Even if the number of models onsuch websites is typically high, the models greatly differ in style, quality, for-mat and functionality, which makes it difficult to find objects of strictly definedfeatures. Moreover, the process of searching through such on-line databases istime-consuming because of the large number of objects and only roughly struc-tured organization of the libraries. Often it is also impossible to verify the models’quality before downloading (or purchasing) them.

2.9 Summary

Great structural and conceptual complexity of 3D content and multitude ofaspects that must be taken into account in the content design process make the3D content creation an inherently complex task. Several methods suitable indifferent contexts have been proposed to simplify this process. Practical applica-tions must combine different methods to provide the necessary trade-off betweenthe conflicting requirements of generality and simplicity of content creation andhigh quality of the created content.

Page 7: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 391

3 Semantic Social Content Creation

3.1 Requirements

In Sect. 2, a number of approaches aiming at the simplification of the contentcreation process have been discussed. To provide further progress in this field,these approaches need to be combined into a user-friendly and efficient designecosystem in such a way that the complementary benefits offered by particu-lar approaches are preserved and employed in a synergistic manner. The mainfunctional requirements for such a content creation method are discussed below.

1. The method should enable the use of content created with scanning systemsand modeling packages as well as the use of existing libraries of content. Itis important that the content elements can be imported in various formats,including formats which are currently popular (e.g., FBX, OBJ, 3DS), butnot limiting further extensibility of the supported set of file formats.

2. The method should enable parameterization of content. The use of parameterswith well-defined meaning and range of values greatly decreases the amount(entropy) of information that designers need to pass to the system whendesigning content.

3. The method should enable the use of components. Components enable cre-ation of higher-level building blocks, which are easier and quicker to use.Although by themselves constraining, used in connection with parameteriza-tion, components provide a very powerful tool.

4. The method should enable the use of high-level content models, which cangreatly simplify understanding and management of complex content struc-tures.

5. The method should enable separation of concerns, thus enabling specializa-tion of designers and development of more efficient tools providing particularfunctions.

6. The method should enable the use of semantics as a formal way of describingthe role and the structure of virtual scene components, thus defining theirpossible usage and simplifying search and parameterization.

7. Finally, the method should enable creation of content by groups of users eitherexplicitly or implicitly collaborating in the content design process.

Requirements 1 and 3 are met by currently available VR/AR authoring envi-ronments, such as the Unity Game Engine IDE [34]. Requirement 2 – the use ofparameters – is commonly used for built-in elements of editing tools, but rarelyavailable for elements created by designers. Requirement 4 – the use of high-levelmodels – is also available, through the use of scene models, scene graphs and pre-defined complex elements, such as Prefabs in Unity. Requirement 5 is achievedby good practices in the current content creation pipelines, but not explicitlysupported by content design tools. Requirement 6 is currently not supported,while requirement 7 only partially – by the use of content sharing websites.

Page 8: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

392 K. Walczak

3.2 The SemFlex Approach

To enable the use of parameterized user-contributed content elements, the useof semantics in content design, social creation of content by multiple distributedusers, as well as separation of concerns in the design process, a new approachto collaborative semantic content creation, called SemFlex, is proposed in thispaper.

In the SemFlex approach, the 3D content creation process is performed inan open distributed architecture, which enables sharing semantically describedcontent. Content creation is understood here as both designing basic contentelements and combining elements of previously created content – including thosecreated by other users – thus enabling designers to contribute content in a trulysocial manner. The possibility of involving large groups of prosumers in the 3Dcontent creation process is a necessity for the development of content, which isinteresting for users, up-to-date and of adequate quality.

The basic unit of content in the SemFlex approach is a semantically describedparameterized content element, called Flex Object (FO). Semantic editing tools(IDEs – Integrated Design Environments) are used by content designers both tocreate 3D virtual scenes based on FOs retrieved from shared content reposito-ries and to create new FOs and upload them to a shared repository. Differenttypes of FOs can be created, depending on their primary content (e.g., sceneobjects, materials, animations). FOs may represent simple independent genericcomponents intended for publication in a library (e.g., furniture, plants, cars,people, equipment) or complex ready-to-use environments that can form a basisfor building games, animations and videos. FOs may be offered in the sharedrepositories using various licensing and pricing models.

Each of the FOs may be parameterized – regardless of whether it is simpleor complex. Parameterization enables producing more generic components withwider possibilities of reuse, and enable users of these components to adjust themto their particular requirements. Default values of parameters permit designersto use FOs without setting all parameters.

Parameters are not intended to replace basic functionality of the used IDEs,but to enhance them with additional functionality. For example, characteristicssuch as position, orientation and scale of objects are well handled by exitingIDEs, so they do not need to be represented as parameters (although they could).However composition of objects from sub-objects (e.g., objects to put onto ashelf), identification of materials (e.g., material to be used for a table), physicalproperties or objects, animations and behavior are typically much more difficultto handle. For example, a query in the Unity Asset Store “table material” returns1540 results (as of Feb 28th, 2017), but no materials for the use in table designare really found. Instead, there is a variety of different tables and furnituresets in different prices (Fig. 1). A designer has to use a specific texture library(e.g., [30]) to find an appropriate image and build a material from scratch. Morethan one texture is typically needed to create a realistic material, which makesthe process even more complex and time consuming.

Page 9: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 393

Fig. 1. Searching for a table material in the Unity Asset Store.

Parameterization in SemFlex encompasses different types of features of theFlex Objects (Fig. 2). The list is extensible – new types of parameters may beadded in particular implementations. Basic types of FO parameters include:

– Sub-Object – this type of parameters is used for composition of objects. Anobject may consist of several geometrical elements (e.g., a building may con-sist of a facade, a number of identical windows positioned in specific places anda door – these would be three different Sub-Object parameters). It is impor-tant that a Sub-Object may be left empty, and that the same Sub-Object maybe used multiple times in an FO (e.g., windows, people in a crowd, cars instreets).

– Visual Material – this type of parameters is used for defining visual appear-ance of geometrical objects. Each of the materials used in an FO may bedeclared as a parameter. Visual material encompasses such elements as color,texture, smoothness, normal map, height map, occlusion, emission, etc. Thesame Visual Material parameter can be assigned to several components ofan FO (e.g., parameter Furnishing Textile could set a material for severalobjects: a sofa, an armchair, and a number of chairs).

– Physical Material – this type of parameters is used for defining physical prop-erties of objects in a virtual scene. This includes such elements as static fric-tion, dynamic friction and bounciness. Again, the same Physical Material canbe assigned to several components of an FO to ease and speed up the virtualscene design process.

– Animation – this type of parameters is used for establishing animation prop-erties of objects. This includes the animator (i.e., a state machine used for ani-mation) and a number of animation clips associated with the state machine.

Page 10: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

394 K. Walczak

– Animation Clip – this type of parameters is used when an animator is fixedin the object or selected by the Animation parameter, and there are stillindependent animation clips to be assigned to the object.

– Behavior – this type of parameters is used for assigning scripts describingbehavior of objects in the scene. This can include both individual behav-ior of objects (e.g., a window handle rotates when pushed) or interactionbetween components (e.g., window opens when the handle is rotated). Whenthe behavior scripts describe internal behavior of an FO, they can be auto-matically assigned to sub-objects through a name or a tag. When behaviorscripts are to govern interaction between elements created within differentFO parameterization sessions, the corresponding objects may have to be setlater by the designer.

Flex Object

Sub-Object

Visual Material

Physical Material

Anima on

Behavior

Scene Object FO

Material FO

Physical Material FO

Anima on FO

Script FO

RangeName

RangeName

RangeName

RangeName

RangeName

DSO: Cross(Union(Class))

Anima on Clip Anima on Clip FO

RangeName

Fig. 2. Types of Flex Object parameters.

4 The Semantic Content Design Environment

4.1 Implementation

The SemFlex approach has been implemented within the Semantic ContentDesign Environment – an open distributed architecture consisting of three kindsof communicating nodes: Semantic IDEs, Shared Content Repositories (SCRs)and Domain Ontology Services (DOSs), as presented in Fig. 3.

Page 11: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 395

Domain Ontology Service A

Content Designer

Domain Ontology A

Content Designer

Shared Content Repository I

Concept

Domain Ontology Service B

Content Store I

Ontology Designer

Content Store II

Seman c IDE 1 Seman c IDE 2

Shared Content Repository II

Ontology Designer

Domain Ontology BConcept

Internet

Fig. 3. Architecture of the semantic content design environment.

The Semantic IDEs, implemented as an extension to the Unity IDE (Fig. 4),communicate with SCRs and DOSs implemented as REST services running onweb servers. The Unity IDE has been selected because it is a popular cross-platform and feature-rich game engine and IDE [34]. Unity can be extended withcustom menus, editor windows, custom component editors and property drawersusing JavaScript and C#. The SemFlex Unity extension enables both creatingvirtual scenes using remote semantically described resources (FOs) and creatingand uploading such resources in the form of parameterized FOs to remote SCRs.

Fig. 4. Semantic IDE implemented as an extension to Unity IDE

To achieve maximum uniformity in the use of SemFlex content, the Unitypackage file format is used for all kinds of SemFlex resources. To keep the for-mat of resources intact, the parameterization information, which is necessaryfor SemFlex, is added as external metadata files. These files are encoded inJSON and contain information about the name and the type of the Flex Object(Scene Object, Material, Animation, etc.), the name of the asset folder where theresource should be stored (to maintain compatibility with non-SemFlex contentcreation pipelines) and SemFlex parameters. An example of a SemFlex metadatafile is provided in Listing 1.1 below.

Page 12: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

396 K. Walczak

Listing 1.1. Flex Object metadata file.

{"name ":" Table FO","type ":" Scene Object","folder ":" Interior/Furniture","parameters ":[{"type ":" Visual Material","displayName ":" Material","target ":[" Table Top",

"Table Leg 1","Table Leg 2","Table Leg 3","Table Leg 4"]

"range ":"( Material) & (Table)","empty ":" false"}

{"type ":"Sub -Object","displayName ":" Left Object","target ":"Sub -Object 1",

"range ":"( Office equipment , Computers) & (Desk)","empty ":" true"}

{"type ":"Sub -Object","displayName ":" Right Object","target ":"Sub -Object 2",

"range ":"( Office equipment , Computers) & (Desk)","empty ":" true"}

]}

4.2 Content Design Example

Instantiation of a Flex Object starts with opening the Flex Creator windowavailable in the Semantic IDE menu (Fig. 5). Top-left part of the figure is thescene hierarchy panel – initially containing only a light source, a camera and asingle object – Floor. Bottom-left part is the Flex Creator tool. The right partcontains a preview of the created 3D scene. The Flex Creator editor providesa field for entering the URI of an SCR and the button “Start Building NewFlex Object”. When the button is pressed, a list of available categories andobjects is displayed as a pop-up menu. The list of categories is taken from aDSO used by the repository (cf. Fig. 7). After selection of the object to create,a corresponding “Create:” button appears. When a user confirms, the actualobject creation process starts.

The object creation process consists of several phases. First, the object pack-age (.unitypackage) containing the FO is retrieved from the server together withthe SemFlex metadata file (.meta). After the package has been imported andstored in a temporary directory, the SemFlex editor installs the content of thepackage in the original location retrieved from the metadata file. Preservingthe original package location is important to guarantee the reuse of alreadydownloaded packages and compatibility with non-SemFlex content used by thedesigner. Regardless of whether a particular element has been downloaded asa SemFlex FO or as an independent asset – it can be used in the IDE in thesame way. After installing the package content, the Flex Object is created in thescene and attached to the current parameterization context. Next, based on theparameters described in the metadata file, further parameterization elements arecreated in the Flex Creator window, enabling continuation of the instantiationand parameterization process.

Page 13: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 397

Fig. 5. Flex Creator – selection of a Flex Object

(a) FO created without the Material pa-rameter set.

(b) FO with assigned material Oak”.

(c) FO with assigned material Plastic”. (d) Selection of a Sub-Object FO by a se-mantic category.

(e) Sub-Object Lamp” instantiated in aplaceholder associated with a sub-objectparameter.

(f) Second sub-object parameter Lap-top” assigned.

” ”

Fig. 6. Flex Creator – instantiating a Flex Object.

Page 14: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

398 K. Walczak

In the presented example (Fig. 6), the Table FO is associated with an MaterialFO, whose selection is indicated as the Material parameter. The parameter hasclearly indicated domain (table top and legs) and a range of possible values.The range identification is performed through the use of DSO concepts. In thepresented example, the texture needs to be an instance of two classes (or theirsubclasses) from the domain ontology: Material and Table. A generalizationcould be achieved, when instead the Table class, the Furniture class were used.This would provide more results, but not necessarily all of them would be usablein the given context.

Such a fine-grained approach to content componentization and parameteri-zation has two important advantages: (1) it simplifies the content creation byenabling the designer to select only suitable elements in the given context, and (2)it contributes to the establishment of large collections of reusable content (FOs).Subsequent steps of instantiating the Table FO are presented in Fig. 6(a)–(f).

Fig. 7. Fragment of a Domain Specific Ontology used for categorization of interiorfurniture objects specified in OWL.

5 Conclusions

In this paper, a new approach to collaborative content creation in a distributedenvironment has been presented. The approach, called SemFlex, is based onthe concept of the Flex Object – a semantically-described shared parameterizedobject, which is a unit of content reuse and extension. The method is flexible inthat it enables both transition from non-parameterized to parametrized contentand vice versa, i.e., providing the ability to use existing content libraries to createnew parameterized content and use parameterized content to create final virtualscenes. Therefore a mechanism is provided to reuse user-generated content tocreate new user-generated content in a structured way, which forms a foundationfor social content creation in 3D.

The content creation process is performed in a distributed architecture withmultiple collaborating designers. Collaborating designers can work with a single

Page 15: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 399

repository (SCR), but multiple SCRs can be set up for different teams or subjectdomains. Content design is simplified by the use semantics. A taxonomy is usedto describe classes of objects. These classes are used to define ranges of FOparameter values, therefore reducing the target object sets when parameterizinga particular feature of an object.

Currently, the Semantic IDE – implemented as an editor extension to theUnity IDE – provides a wizard-like instantiation of semantic FOs, import of con-tent from SCRs and creation of parameterized FO packages from existing contentelements (Fig. 4). Parameterization of the created FOs is performed manually,by editing the generated metadata file. In future work, we plan to extend thesemantic FO exporter to support selection of parameterized components andindication of their semantic ranges based on DSOs. This will further simplifythe FO design process.

The presented method of building semantic parameterized content elementsand their instantiation has been tested with limited number of relatively simpledesign scenarios. In future work, we plan to perform larger design trials, whichwill involve higher number of designers building more complex 3D environments.

Acknowledgments. This work was supported by the Polish National Science Centre(NCN) under Grant No. DEC-2012/07/B/ST6/01523.

References

1. Alpcan, T., Bauckhage, C., Kotsovinos, E.: Towards 3D internet: why, what, andhow? In: International Conference on Cyberworlds, CW 2007, pp. 95–99. IEEE(2007)

2. Berger, M., Tagliasacchi, A., Seversky, L., Alliez, P., Levine, J., Sharf, A., Silva, C.:State of the art in surface reconstruction from point clouds. In: Eurographics 2014 -State of the Art Reports. EUROGRAPHICS star report, Strasbourg, France, vol.1, pp. 161–185, April 2014. https://hal.inria.fr/hal-01017700

3. Chaudhuri, S., Kalogerakis, E., Giguere, S., Funkhouser, T.: Attribit: content cre-ation with semantic attributes. In: Proceedings of the 26th Annual ACM Sympo-sium on User Interface Software and Technology, pp. 193–202. ACM (2013)

4. Creaform: Metrology Products (2017). https://www.creaform3d.com/en/metrology-solutions/creaform-product-overview

5. Dachselt, R., Hinz, M., Meissner, K.: Contigra: An XML-based architecture forcomponent-oriented 3D applications. In: Proceedings of the Seventh InternationalConference on 3D Web Technology, Web3D 2002, NY, USA, pp. 155–163. ACM,New York (2002)

6. Dassault Systemes: 3D ContentCentral (2017). https://www.3dcontentcentral.com/

7. De Troyer, O., Bille, W., Romero, R., Stuer, P.: On generating virtual worlds fromdomain ontologies. In: Proceedings of the 9th International Conference on Multi-Media Modeling, Taipei, Taiwan, pp. 279–294 (2003)

8. E-on software Inc.: PlantFactory (2017). http://www.plantfactory-tech.com/9. E-on software Inc.: VUE - Solutions for Digital Nature (2017). http://www.

e-onsoftware.com/

Page 16: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

400 K. Walczak

10. ECMA International: Universal 3D File Format (2007). http://www.ecma-international.org/

11. Esri: Esri CityEngine (2017). http://www.esri.com/software/cityengine12. Flotynski, J., Walczak, K.: Ontology-based creation of 3D content in a service-

oriented environment. In: Abramowicz, W. (ed.) BIS 2015. LNBIP, vol. 208, pp.77–89. Springer, Cham (2015). doi:10.1007/978-3-319-19027-3 7

13. Flotynski, J., Walczak, K.: Ontology-based representation and modelling of syn-thetic 3D content: a state-of-the-art review. In: Computer Graphics Forum (2017).http://dx.doi.org/10.1111/cgf.13083

14. Fraunhofer I.G.D: X3DOM (2017). http://www.x3dom.org/15. Giannattasio, T.: PHOTON - CSS 3D Lighting Engine (2017). http://photon.

attasi.com/16. Highend3D: High Quality 3D Models, Scripts, Plugins and More! (2017). https://

www.highend3d.com/17. Interactive Data Visualization Inc.: Speedtree (2017). http://www.speedtree.com/18. ISO/IEC: MPEG-4 Information technology - Coding of audio-visual objects (2017).

http://mpeg.chiariglione.org/standards/mpeg-419. Khronos Group: COLLADA - 3D Asset Exchange Schema (2017). https://www.

khronos.org/collada/20. Khronos Group: WebGL - OpenGL ES for the Web (2017). https://www.khronos.

org/webgl/21. Latoschik, M.E., Blach, R., Iao, F.: Semantic modelling for virtual worlds a novel

paradigm for realtime interactive systems? In: VRST, pp. 17–20 (2008)22. META-Elements: META-Elements (2017). https://www.meta-elements.com/23. Pellens, B., De Troyer, O., Kleinermann, F.: CoDePA: a conceptual design pattern

approach to model behavior for X3D worlds. In: Proceedings of the 13th Inter-national Symposium on 3D Web Technology, Web3D 2008, NY, USA, pp. 91–99ACM, New York (2008). http://doi.acm.org/10.1145/1394209.1394229

24. Planetside Software LLC: Terragen 4 (2017). http://planetside.co.uk/25. Polys, N., Visamsetty, S., Battarechee, P., Tilevich, E.: Design patterns in compo-

nentized scenegraphs. In: Proceedings of SEARIS. Shaker Verlag (2009)26. Roast, K.: Phoria (2014). http://www.kevs3d.co.uk/dev/phoria/27. SideFX: Houdini VFX and Animation Tools (2017). https://www.sidefx.com/28. Spagnuolo, M., Falcidieno, B.: 3D media and the semantic web. IEEE Intell. Syst.

24(2), 90–96 (2009)29. Sprite3D.js: A small library for generating and manipulating CSS 3D transforms

(2017). http://minimal.be/lab/Sprite3D/30. Textures.com: Textures for 3D, graphic design and Photoshop (2017). http://www.

textures.com/31. Three.js: Javascript 3D library (2017). https://threejs.org/32. TurboSquid Inc.: 3D Models for Professionals (2017). https://www.turbosquid.

com/33. Unity Technologies: Asset Store (2017). https://www.assetstore.unity3d.com/en/34. Unity Technologies: Unity Game Engine v. 5.5.2 (2017). https://unity3d.com/

unity35. Van Gool, L., Leibe, B., Muller, P., Vergauwen, M., Weise, T.: 3D challenges and

a non-in-depth overview of recent progress. In: 3DIM, pp. 118–132 (2007)36. Visamsetty, S.S.S., Bhattacharjee, P., Polys, N.: Design patterns in X3D toolk-

its. In: Proceedings of the 13th International Symposium on 3D Web Technology,Web3D 2008, NY, USA, pp. 101–104. ACM, New York (2008). http://doi.acm.org/10.1145/1394209.1394230

Page 17: Semantics-Supported Collaborative Creation of Interactive 3D …semantic3d.org/wp-content/uploads/2017/06/AVR2017... · 2017-06-14 · DOI: 10.1007/978-3-319-60928-7 33 Walczak K

Semantics-Supported Collaborative Creation of Interactive 3D Content 401

37. Voxel.js: An open source voxel game building toolkit for modern web browsers(2014). http://voxeljs.com/

38. Voxelogic: Acropora (2017). http://www.voxelogic.com/39. Walczak, K.: Flex-VR: configurable 3D web applications. In: Proceedings of the

Conference on Human System Interactions, pp. 135–140. IEEE (2008)40. Walczak, K.: Structured design of interactive vr applications. In: Proceedings of the

13th International Symposium on 3D Web Technology, Web3D 2008, NY, USA, pp.105–113. ACM, New York (2008). http://doi.acm.org/10.1145/1394209.1394231

41. Walczak, K., Cellary, W.: X-VRML for advanced virtual reality applications. Com-put. 36(3), 89–92 (2003)

42. Walczak, K., Flotynski, J.: On-demand generation of 3D content based on semanticmeta-scenes. In: De Paolis, L.T., Mongelli, A. (eds.) AVR 2014. LNCS, vol. 8853,pp. 313–332. Springer, Cham (2014). doi:10.1007/978-3-319-13969-2 24

43. Walczak, K., Flotynski, J.: Semantic query-based generation of customized 3Dscenes. In: Proceedings of the 20th International Conference on 3D Web Technol-ogy, Web3D 2015, NY, USA, pp. 123–131. ACM, New york (2015). http://doi.acm.org/10.1145/2775292.2775311

44. Walczak, K., Wojciechowski, R., Wojtowicz, A.: Interactive production of dynamic3D sceneries for virtual television studio. In: The 7th Virtual Reality IC VRIC -Laval Virtual 2005, pp. 167–177, April 2005. http://www.laval-virtual.org/en/pres-colloque.php

45. Web3D: The Virtual Reality Modeling Language - ISO/IEC 14772–1 (2006).http://www.web3d.org/standards/all

46. Web 3D: X3D Specifications - ISO/IEC-19775-1 (2013). http://web3d.org/x3d/specifications/

47. Zahariadis, T., Daras, P., Laso-Ballesteros, I.: Towards future 3D media internet.In: NEM Summit, pp. 13–15 (2008)