Even though revision control has been successfully deployed for text-based files for many years, it does not efficiently map to 3D assets. In this presentation we propose a non-linear concurrent revision control system for 3D assets. Scenes are represented as collections of polymorphic objects alongside their corresponding history in a database, which map existing revision paradigms to a novel scene manipulation. Our open source implementation leverages a NoSQL database and a 3D asset import library.

Similarly to existing version control systems it works independently from editing software and does not require any knowledge of user edits.

A study was conducted to evaluate the feasibility of using an artificial neural network (ANN) to forecast building energy use based on building architectural and energy use parameters. The objectives were to ascertain the prediction performance of the ANN and the causal strengths of each parameter on energy use.

Electricity and heat energy use data from the Display Energy Certificate (DEC) scheme together with building parameters were collected for London university buildings. 148 datasets were compiled covering 97 different buildings.

For electricity use the ANN mean absolute percentage error reduced to 34%, a 30% reduction relative to a theoretical benchmark-based approach; for heat use it reduced to 25%, a 49% reduction against the benchmark. Prediction performance appeared to be restricted however, perhaps owing to the limited number of training patterns.

Building activity, material, environment and glazing type had the greatest causal strengths for both types of energy use. Height was a strong determinant for heat use and summer sun hours and aspect ratio were significant for electricity use.

Various measures to develop the ANN methodology are presented, a principal measure being expanding the training dataset. A more extensive follow-up study is planned accordingly.

With the emergence of generative environmental simulation in architecture there is a growing need to search through a large number of potential solutions towards an explicit objective. It is argued that in fact the objectives are usually ill-defined, transient and numerous. And so the research focuses on integrating learning techniques to improve both the stability of optimised solutions and search efficiency, in light of the need for multi-objective and multi-modal exploration. The work is coordinated with Bentley's integration of EnergyPlus, STAAD structural analysis, Darwin multi-objective genetic algorithm and other custom simulation tools into the Generative Components software.

Jason Bruges Studio is a multi disciplinary design practice which includes architects, lighting designers, electrical engineers and industrial designers. The studios work is often achieved through the innovative application or reinterpretations of technologies appropriated from other fields. Work in our field could be considered as divided into three categories, automata, reactive and interactive. My research with the firm is concerned with developing work which could be considered interactive. I will show a selection of projects by the studio and by myself, and discuss how technology is appropriated and employed in the artworks the studio produces.

Given a sequence of images from a moving camera, depth mapping algorithms aim to calculate a distance value at each pixel. I will give an overview of computer vision techniques for depth map creation, and their application to film and media production.

Deformed 2d surfaces are commonly found in the field of cultural heritage acquisition, often in the form of damaged parchment, papyri, cloth, etc. The types of damage seen on these surfaces wide-ranging, including natural aging, fire-damage, moisture-damage, shrinking, and tearing, and the damage is almost always non-uniform.

The diversity and irregularity of this damage means that repairing such surfaces in a completely automated way is unfeasible. Instead we propose a user-guided system comprising various interactive tools, each of which alter or constrain the shape of the surface in some way, which allow the user to oversee the repair process.

Concept designs are key to the direction and ultimate success of a project. Major decisions which fundamentally affect the sustainability of a building are made in the early stages of design. However, in much the same way that 2D CAD drafting tools cannot be practically used for sketching out design ideas – it is often found that building simulation software cannot be used as a 'design' tool, instead they are evaluation tools that assess various design iterations. This process is often found by designers to be ill-suited for the very early design stages, when the design moves at a rapid rate in an often erratic manner.

This research explores computational techniques at the early stages of design, utilising operational performance data and environmental feedback as architectural design drivers. With a focus on school design, the aim is to develop a computational process which allows performance indicators to be communicated to the user in real time as early form and fabric parameters are altered interactively – allowing the designer to sketch performance as well as form.

The study focused on the response of pupils in a secondary school to their built environment, in particular the location of windows within classrooms. Two effects of the window were investigated, the effect of sitting in a directly day lit area and the pupil's access to a view. In order to understand the pupils' response to their environment, body motion analysis techniques have been employed where the subject's movement is used quantify there current state. This study looked for moments where learning engagement dropped, highlighted by increased movement (characterised by fidgeting).

To monitor this movement, GeneActiv bracelets were used that quantify the acceleration in three axes and relatively unobtrusive due to their small size. This data was then analysed against; distance from window, whether the pupil was directly day lit, age of the pupil, and the time elapsed since the start of the lesson. To analyse this data, an artificial neural network was utilised over different time periods to establish which input had the greatest influence on learner engagement.

Radiotherapy is an important tool in the treatment of cancer. Latest methods attempt to deliver a larger dose of radiation to the tumour during each treatment session, making accurate knowledge of tumour whereabouts crucial. Presented is a method which only uses existing clinical scans to build a surrogate dependent motion model of the tumour, to be used to drive next generation radiotherapy treatments.