Use this URL to cite or link to this record in EThOS:
Title: Interactive generation of user-editable virtual landscapes : from image-guided to model-based approaches
Author: Bradbury, Gwyneth Anne
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
In the context of virtual reality applications such as games, films and training exercises, the virtual environment is fundamental to the viewer’s experience. Improvements in computer graphics, particularly in rendering and artificial intelligence, continue to make virtual worlds look and feel more believable and more natural. This thesis focuses on virtual landscapes and the industry requirement for increasingly large digital assets, such as artificial landscapes extending as far as entire planets. This, however, requires high financial investment in artists and designers to carry out tedious, detailed tasks. As a result, tools which reduce the artists’ workload are valued, facilitating both faster work-flow and artistic control of the final visual result. Commercial production tools commonly incorporate procedural and stochastic methods. Although they can, in theory, provide a great deal of freedom, their use is limited by the fact that they often limit artistic control by employing non-intuitive parametric controls. This can make it hard for users to edit a generated scene or add data-based assets later. Alternative tools give the artist complete control over the content created at the expense of a heavy work-flow. This thesis addresses the problems involved in production of virtual landscapes by creating a series of tools which improve production workflow. These tools are built to simultaneously exploit real-world data as well as traditional algorithmic models, whilst also providing means of artist interaction and guidance. Contribution is made in the following three contexts: aerial footage is explored to create environments which incorporate the statistical distributions of different types of vegetation in real-world landscapes; stochastic methods are applied to real-world elevation maps to facilitate height-field manipulation using frequency-based controls; and finally, landscape ecology is investigated along with the application of ecological modelling to landscape synthesis, resulting in a toolset encouraging early user-involvement in the creative process. The presented systems run interactively, allowing user-involvement throughout, and together contribute to a fluid workflow for landscape production whereby the artist may define a scene in terms of a layout map of ecosystems (or object classes) as well as a customisable terrain elevation map. An agent-based simulation then ‘plants’ and ‘grows’ vegetation instances on the landscape. This simulation can be paused and all inputs adjusted at any point. The work presented represents an important contribution to the problem of virtual landscape contribution, addressing some of its more considerable bottlenecks.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available