Jarkko M. Vatjus-Anttila

Power Consumption Model of a Mobile GPU Based on Rendering Complexity

This paper presents a mathematical model for predicting power consumption of a mobile device when it is rendering 3D graphics. The model is based on 3D primitives (triangles, render batches, texels), and hence is hardware agnostic. With the model, a complexity of any given 3D scene can be predicted already at a production phase without access to the actual target hardware. This paper describes how the power consumption model is derived. The model is verified with measurements of real-world content and hardware. With the given hardware, 3D data and given verification scenarios, the model is able to predict the total power consumption with an error ranging from 0.3% to 3.2%.

A Virtual World Web Client Utilizing an Entity-Component Model

The popularity of virtual worlds has increased considerably in recent years. Currently, many service providers are trying to make it easier for users to access their virtual worlds. However, accessing a virtual world typically requires a client application that needs to be versioned for each device platform and operating system configuration. Using WebGL, an interactive 3D environment can be used on a cross-platform supported web browser. In this paper, we present a system architecture that utilizes an entity-component model, and a prototype implementation of a WebGL-based virtual world client to provide a plug-in free, extensible and open source web client for 3D virtual worlds. The performance of the web client was evaluated in terms of frame rate, CPU load, memory consumption and scene processing speed. Based on the results, the performance of the web client was good on a desktop PC, but mobile hardware specific optimizations are required to provide a good user experience on mobile devices.

Communication Architecture for a P2P-Enhanced Virtual Environment Client in a Web Browser

In this paper, communication architecture for a P2P-enhanced virtual environment (VE) client in a web browser is proposed. The communication architecture relies on recent and ongoing standardization efforts as well as novel research work in defining the building blocks that are required for implementing adaptive and scalable solutions for VEs. In addition, potential paths for future research are provided by identifying open research questions that are crucial in solving the technical challenges hindering the more wide-spread adoption of VEs. Special emphasis is on mobile devices of which capabilities for presenting 3D graphics have evolved fast during the last years.

Effect of 3D Content Simplification on Mobile Device Energy Consumption

Extensible 3D virtual spaces and their services are often too heavy for a mobile device to handle. The burden caused by such services is divided between extensive amounts of content, which need to be downloaded prior to using the service, and the complexity of the graphical reproduction process (rendering). In this paper, it is shown how texture optimization, texture compression and geometry optimization of such a content will affect the overall energy consumption of a mobile device during rendering. Results show that careful optimization of the 3D content for the mobile device gain energy savings of up to 40%.

On the Edge of a Virtual World Investigating Users' Preferences and Different Visualization Techniques

This paper investigates how the edges of a limited size 3D virtual world model can be visualized. We compared five alternative visualization techniques by conducting downloading, post-processing and rendering measurements and evaluating the designs with a user study. Recommendations for UI designers working especially in the mobile computing domain are presented.

Occlusion Based Message Dissemination Method in Networked Virtual Environments

Virtual spaces that host hundreds or even thousands of users create enormous amounts of traffic, which usually is broadcasted to the entire environment and congests the network medium very quickly. In this paper, a novel interest management (IM) algorithm called EA³ is presented and evaluated. EA³ is based on an existing IM algorithm, called A³ which uses euclidean distance to discover entities of interest. A³ was improved by utilizing a ray visibility algorithm, which helps to filter out traffic originating from entities behind obstacles. The evaluation of EA³ and A³ was conducted using an extensible networked virtual environment called real tend. The results show that, when either EA³ or A³ is used, the overall network throughput can be lowered significantly. In a city-like test scene, the drop was approximately 50 per cent with A³ and 62 per cent with EA³ compared to the situation when no IM algorithm was used. In terms of CPU load EA³ is slightly heavier to use when there is a lot of open space inside the virtual environment.

Adaptive content management for collaborative 3D virtual spaces

Collaborative 3D virtual spaces and their services are often too heavy for a mobile device to handle. The burden of such services is divided between extensive amounts of data, which need to be downloaded prior to using the service, and the complexity of the resulting graphical rendering process. In this paper, a proxy based architecture for collaborative virtual spaces is used to manipulate graphical data demand-time to favor both network bandwidth usage and graphical rendering process. In addition, a proof of concept test is shown, regarding how the simplification process gains savings for different client device profiles, including laptops, tablets and mobile devices.

3D object reconstruction processing chain for extensible virtual spaces

3D reconstruction of real world objects is an important content creation tool for the 3D internet. This poster shows how cheap commercial sensors can be used to reconstruct real 3D objects and import those objects within a 3D virtual world. This poster describes the implementation of a 3D content capturing processing chain using Microsoft Kinect. The capturing process includes a method for filtering and segmenting an object from the unwanted background data, a method for registering multiple scans together and a method for creating a solid surface presentation for the captured point cloud. The resulting 3D asset is imported and rendered in RealXtend Tundra 2.

A mobile client implementation for extensible network virtual worlds

This poster describes the implementation of a mobile client, called Chiru, for the realXtend extensible Network Virtual Environment. The paper describes the chiru client which can cope with network connection status and failures, multiple world connections and User Interface elements based on Qt/QML. The prototype is demonstrated on a mobile tablet device. Examples of power management profiles are presented and performance issues between Qt and OGRE are briefly described.