Multi-resolution Meshes for Multiple Target, Single Content Adaptation within the MPEG-21 Framework HyungSeok Kim, Chris Joslin, Thomas DiGiacomo, Stephane Garchery, Nadia Magnenat-Thalmann MIRALab - University of Geneva {Kim, Joslin, Giacomo, Garchery, Thalmann}@miralab.unige.ch Abstract To present three-dimensional data both in heavy and light-weight clients, an adaptation scheme is required. Current state-of-the art research shows promising results for specific purpose but still it is not well adoptable for the light-weight clients such as mobile devices. In this research, we present a method for transmitting adapted 3D contents to multiple target devices. To accomplish this goal, we devised a clustered representation of a multi-resolution model that is flexible, simple, efficient, and works with the MPEG-21 adaptation mechanism. 1. Introduction 1.1 Targeting a Context Three-dimensional (3D) representation is one of the cornerstones of computer graphics and multimedia content. Advances in this domain, coupled with the highly fuelled progression of 3D graphics cards, has pushed the complexity of these representations into a new arena whereby a single real-time model can consist of a million polygons or more. Huge architectural buildings, everyday objects, even humans themselves can be represented using 3D graphics in such detail it is difficult to distinguish between real and virtual. Concurrently, and much towards the other end of the scale, many devices (Personal Digital Assistants (PDAs), Mobile Phones, Laptops, etc) are now “3D capable” to enhance a user’s experience and provide much more depth to the information presented. In many cases these devices are accessing the same content from the same service provider (providing virtual maps/guides, multi-user games etc) and it is this broadness of content and the heterogeneity of devices (in terms of performance, capability, network connection etc) that is the main concern in a market that is continuously expanding. It is also the concern of the user to obtain the best quality for their device, i.e. a general expectation of any device of higher performance is that overall the quality of the experience will be better. Two main constraints are the real target areas of MPEG-21 Digital Item Adaptation (DIA); the first to provide a range of content for a wide variety of devices, and secondly making sure that the content is of the highest quality according to the device’s capabilities. 1.2 Graphics DIA DIA offers a solution to these current industry woes and whilst most of the more common examples are in the area of video and audio, the popularization of applications shows that there is also a popular trend for 3D graphics. Whilst there are many concerns for 3D graphics as a whole, as it is a very large domain already, we focus our attention in this paper solely on representation, and emphasize on the constraints that need to be considered in the domain of virtual humans. In addition, whilst the research area of multi-resolution models [1] is one that has been significantly explored we introduce it into the domain of DIA in order to consider the constraints imposed by multiple targets, and the adaptation from a single content. Our main research goal was to devise a method for multi-resolution representation/adaptation to transfer and render a virtual human (face and body) on heterogeneous targets devices with the following considerations: • Range of Complexity - As the model will not be targeted specifically for a particular network or device, we needed to define a generic method. In essence this means the ability to represent a virtual human from its lowest possible representation towards its most complex • Device Capability - We defined a generic method for assessing the capability of the device to enable the closest approximation. As the target device is not specified, it should be assumed that this assessment should also be heterogeneous • Network Capacity - Based on an assessment of the network condition we used a generic method for adaptation that allows for both a specific bandwidth and maintains a certain level of Quality of Service (QoS)