Archive of Past Events

October 30th 2007: Ian Shaw, Electronic Arts
Open Worlds in Video Games

Abstract
An informal talk about supporting open worlds in video games. The trend in many recent games has been to do away with a level-based structure, and to allow the user to roam freely about the world. The game engine must thus deal with the issues of streaming and simulating a world that is much larger than can fit in to main memory. These issues will be talked about in the context of the recent Harry Potter game which allowed the player to walk freely around Hogwarts and its environment.
October 23rd 2007: Verónica Costa Orvalho, Face In Motion
Fast and Reusable Facial Rigging and Animation

Abstract
Reproducing the subtleties of a face through animation requires developing a sophisticated character rig. But, creating by hand the inner structure and controls of each character is a very labor-intensive and time-consuming task. We developed an application that is 90--99% faster than traditional manual rigging. The application automatically transfers the rig and animations from the source to a target model. Unlike prior work related to morphing and re-targeting that focus on transferring animations, we transfer the complete facial setup in addition to animations. Our method is general, so artists can define their own rig and then quickly apply it to different models, even with disparate proportions and appearance (human, cartoon or fantastic). This gives artists complete freedom to manipulate the characters: they can create new animations and not be limited by pre-generated ones.
February 19th 2007: Adrian Hilton, University of Surrey

Abstract
Capture and representation of a persons appearance during movement in a form that can be manipulated for highly realistic computer animation in games and film is an open research problem. This talk will present a number of approaches that have been introduced to capture people from multiple view video using both model-based and model-free computer vision methodologies. Surface Motion Capture (SurfCap) will be introduced which allows representation and animation control of people with the captured dynamics of clothing during movement. SurfCap will be presented as an analogous technology to skeletal human motion capture using markers (MoCap) which has become a standard production tool. Surface motion graphs are used to animate people from multiple captured surface sequences allowing control of movement and action. Surface matching methods based on geometry image sequences using spherical parameterisation are used to transition between captured motion sequences and reconstruct skeletal movement. SurfCap's potential as a future technology for production in games and film will be discussed.
January 30th 2007: Andrea Cavallero, Queen Mary University
Visual Content Analysis

Abstract
In this talk I will introduce our activities in multimedia signal processing that include visual content analysis with application in advanced surveillance and multi-sensor systems. In particular, I will discuss in details appearance-based target tracking and present our recent results in this area, including examples of applications such as face tracking, control of PTZ cameras and event detection.
December 14th 2006: Eshan Rajabally, Loughborough University/BAE Systems
Multi-player Immersive Modelling and Simulation, Realistic Environments, Display Technologies and Operator Control

Abstract
With a growth in capability and a push towards networked enabled operations, in-situ air and ground-crew training on our core assets (Typhoon, Hawk, Tornado....) is becoming prohibitively expensive. As such, there is a drive towards training through virtual/immersive/synthetic environments (SEs) as well as mission planning, briefing, rehearsal and debriefing. Consequently the BAE SYSTEMS SEIC is seeking to leverage research into SE technology and its exploitation including varied areas such as mixed reality, datalinks, interoperability, visualisation, environment databases, human factors, VV&A etc.
December 13th 2006: Andrew Fitzgibbon, Microsoft Research
Practical Image-based Priors

Abstract
Novel view synthesis using image-based priors has recently been shown to provide high quality renderings of complex 3D scenes. However, current
methods have two problems: they are very slow, and are prone to fall into local minima. I shall talk about two investigations into these problems.

First is a coarse-to-fine method that can be used to reduce Computation significantly. In contrast to traditional multiple- View stereo methods, devising a coarse-to-fine strategy for this problem is complicated by the fact that image-based priors are strongly tied to the scale at which rendering is performed. We show how a hierarchical decomposition of the tecture patch database both allows multiple-scale analysis, and speeds up the imposition of the priors. Examples are shown on a number of challenging sequences, and illustrate that the new method yields comparable results to the previous method, with significant gains in speed.

I also discuss an alternative formulations of the prior based on Roth and Black's "Fields of experts" formulation. This formulation allows better minima to be found using an algorithm we have adapted from the statistical physics literature.

December 8th 2006: Chris Doran, Geomerics
New Techniques for Lighting: From Theory to Practical Implementation

Abstract
Few, if any, problems in gaming have attracted as much attention over the years as that of accurately lighting a scene given arbitrary light sources. The complexity of the problem is immense; the lighting equation governing all computer graphics describes a massively information-heavy scenario where every element of a scene can act as an emitter and potentially contribute to lighting the scene. Formulating the problem directly from this equation would involve vast matrices of transfer coefficients that are impractical to compute at real-time rates. But as computational power increases new techniques become available that had previously been dismissed as too expensive. In this talk I review recent developments from the fields of spherical wavelets, pre-computed radiance transfer, and radiosity. The talk includes practical examples of these algorithms, as well as a discussion of the future directions that the games industry should be investigating.

December 7th 2006: Oliver Grau, BBC Research and Development

Abstract
The 'production magic' team of the BBC R&D department develops new innovative production techniques using computer vision and graphics. The work started some years back with a target-based active tracking system for virtual studios. The presentation gives an overview of the more recent work including real-time object and camera tracking and 3D scene reconstruction.

This page was last modified on 30 October, 2007