You are here : 
SUMMARY AND CONCLUSIONS OF THE AR/MR WORKSHOP
The main aim of this workshop was mainly to collect and arrange a collection of MR authoring requirements from industrial end-users and from other currently running or recently finished IST projects. The first session was devoted to present an overview on MR authoring in the AMIRE project as well as the authoring requirements coming from one of the project demonstrators (OMV petrochemical plant).
During the second session statements regarding MR authoring requirements from selected projects were presented. The aim of the ORIGAMI project is the development of an integrated environment for the mixing of real and virtual content in TV/film production. Two application scenarios were described: off-line virtualisation of 3D environments and objects and television productions with dynamic contents. Good examples where the content creation itself is done in an MR environment were showed and explained. ORIGAMI distinguishes different authoring roles at different abstraction levels. The objective of the CREATE project is to develop a mixed reality framework to enable highly interactive real-time construction and manipulation of realistic virtual worlds based on real sources. Prototypes on the Cultural Heritage and architecture/urban planning are expected. According to CREATE, one of the main problems is that content creation is still expensive (3d is complex) and cost should be reduced. For some MR applications 2D images, schemas and additional textual information could be enough and 3D have to be considered just as an extension. The STAR project, services and training through augmented reality, also distinguishes different authoring tools at different abstraction levels and suggested one meta-authoring tool that represents the workflow. They stated that authoring should be independent of the end user device. An scene editor tool, based on AR-XML, is used for describing the AR/MR scenes.
According to ARVIKA, authoring is neither modeling nor programming. Content creation is modeling plus authoring. It's needed a 'Redaction Tool' for production of the single animations 'Power- Point like' based on a style guide, offering a set of basic objects (library), a set of basic animations that can be parameterised (templates) and timelines (duration). Additional functionalities should be calibration (virtual - real world), scene preview and import/export (e.g.VRML). From the point of view of DAIMLER-CHRYSLER, authoring should be independent of and adaptable to the underlying VR/AR system. AR should be used in complex tasks, infrequent tasks or tasks performed for the first time. VR/AR system are based on tracking devices, rendering engine and interaction metaphors/devices. They identified the following general authoring requirements for AR systems: integration with existing data and process chain (interaction with legacy applications), consideration of current repair and documentation practices (extension and coexistence with them) and support of industrial grade VR/AR systems (ARToolkit is a too low common denominator).
Finally, in the DWARF (Distributed Wearable Augmented Reality Framework) project authoring is made currently on source code level. XML-based service descriptions allow high-level distributed description of the overall system
Several components use data formats that can be generated with established authoring tools: viewer component processes VRML files, taskflow engine uses XML description of state machine and the user interface controller uses XML description of Petri Nets. DWARF is a dynamic system in which the combination of components happen at run time, Visualization of running system is possible. A tool for assembling a network of components at design time would be useful (visual programming). Some components could be greatly enhanced by authoring support.
Content production was recognised as a serious problem which has often been neglected (e.g. CREATE). Some MR applications assume an extensive support of the end-user author by technical experts (e.g. ORIGAMI: the director is supported by a whole team of experts), some MR applications don't (e.g. DaimlerChrysler: the writer of technical documentation). In the first case, the need for good authoring tools is not that apparent. Contents needs to be specified technology independent. Otherwise, advances in MR technology cannot utilised without serious penalties in adapting the content. In the current situation there is not much concern about it since there is not much content available.
In summary, some of main authoring requirements mentioned during the workshop were the following ones:
- Authoring tool should compensate for errors in the underlying MR technology (e.g. marker not detected).
- Authoring process should support a cross-platform delivery (e.g. standalone application or AR-plugin for a browser).
- Authoring tool should explicitly support an automatic user interface generation.
- Authoring tool should support templates (e.g. for animations)
- Authoring tool should support a timeline.
- Authoring tool should support calibration tasks.
- Authoring tool should support creation of multimedia annotations (which should be described in XML and stored in a database) and configuration of their presentation (e.g. static or dynamic, location, time).
- Authoring tool should allow the author to specify navigation.
- Authoring process should take different profiles into account (e.g. in a training lab vs. in the field).
- Authoring process should be integrated in current production workflow
- Authoring process should take transition and co-existence of MR content with traditional content into account.
- Authoring system should support industrial grade VR systems
- Authoring system should support the integration of many legacy systems
- Authoring tool should support modification of content (in contrast to creation from scratch)
- Authoring process and authoring system should take testing into account.
- Authoring framework should support multiple layers with functionalities like that provided in a game engine (e.g. physically-based simulation).
- Authoring framework should provide defaults (e.g. default renderer).
- Authoring tools should provide means for specifying the interaction behavior.
The third session was to discuss about possible scenarios for the production and authoring process of the future, the expected technological and conceptual developments and how it will affect the way AR/MR applications will be authored and produced. Initially, a summary of results from the ACM SIGGRAPH and Eurographics Campfire: 'Production Process of 3D Computer Graphics Applications - Structures, Roles and Tools' and the workshop at the Web3D 2003. Symposium: 'Adopting and Augmenting X3D for Efficient 3D Content Production: Concepts and Tools' was presented to set the scene. For some of the attendants there is not a need for a specific description format of MR content. They propose instead the usage of several formats for the different sub-parts of the MR content production: 3D model, video, tracking, etc. Another view was that forming an XML standard to exchange data between different authoring platforms would be useful.
Finally, the last session was to present the possible schemas of collaboration with external projects. Siemens Business Systems C-LAB presented the AMIRE market analysis, the questionnaires, the AMIRE Web Forum and the 'legal framework'. Some of the attendants expressed concerns and doubts regarding the market feasibility of open source. In order to start the collaboration activities with other projects, the participants were asked to fill-in the AMIRE questionnaire as a starting point for a wider analysis for MR/AR. The participants were also asked to discuss and comment, when the first version of the AMIRE authoring tool will be available on the AMIRE Web Forum.
In conclusion, we got, in general terms, a good feedback from the participants and many requirements from the project presented. It also become quite clear that authoring and content production is a very important issue for a more widespread use of mixed reality but there is no general agreement of how this authoring should be done.
According to its objective, the AMIRE project will continue to promote an open approach to authoring for Mixed-Reality applications. The project will make use of the many contacts established on the occasion of this workshop to look for concrete and pragmatic solutions that could benefit the MR community. Important aspects are the preparation of a components and gems survey, the establishment of agreed interfaces between components and if appropriate the development of format converters. More information about the AMIRE project can be obtained from http://www.amire.net/
Jose Luis Los Arcos
LABEIN
March 20, 2003