Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.790902
Title: Improved real-time rendering for mixed reality
Author: Walton, David R.
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
The research presented in this thesis explores methods for rendering realistic virtual content in mixed reality applications. Rendering content for mixed reality presents unique challenges, as it is critical that the virtual content is not only realistic, but also consistent with its real surroundings. This thesis is concerned with methods to achieve this goal, focusing on those which require minimal prior information about the real scene and work in real time. In particular, methods are presented which address two problems in this area. The first problem is correct handling of occlusion between virtual and real objects. When adding virtual content to a real scene, it is challenging to determine where real objects should occlude the added virtual content. An approach was developed to combine the noisy colour and depth outputs of RGBD cameras to determine accurate real-virtual occlusions. A method was also developed to quantitatively assess the quality of such approaches. The second problem is capturing a detailed lighting model of a real environment quickly, and updating it in real time. The appearance of objects is greatly dependent on the surrounding lighting environment, so a detailed lighting model is invaluable when attempting to render realistic virtual content into the scene. A number of novel approaches to capture this lighting information and use it are developed, and applications for these approaches are also explored, including rendering virtual objects which reflect the changing real world around them. In contrast to previous approaches which require external light probes or infer lighting indirectly from the real scene, the presented approaches use a self-contained two-camera system and use the extra information to infer the lighting at the virtual object location.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.790902  DOI: Not available
Share: