From Panoramas to Presence: Stereoscopic Rendering and Virtual Staging for Immersive Indoor XR

Abstract

Immersive Extended Reality (XR) is redefining how we experience and interact with digital environments, with panoramic imaging playing a fundamental role in representing indoor spaces. However, achieving truly immersive panoramic experiences faces challenges at every step of the process. As a result, the stereo pairs created suffer from limitations such as low image quality, depth perception inconsistency, bandwidth inefficiency, and rendering latency, which would disrupt user immersion. This thesis aims to integrate new and existing techniques for stereoscopic generation, virtual staging, and user experience optimization into a single study to enable the seamless generation of immersive indoor environments. The research is structured around the lifecycle of a panoramic XR experience, from the capture and processing of panoramic images to their transformation into fully immersive environments. First, we explore the technological foundations of panoramic imaging, portraying its foundational importance in XR, along with the challenges and key limitations of current technologies. We then examine methods of generating true stereoscopic panoramic environments, comparing traditional image composition techniques with new deep-learning-driven solutions that significantly improve depth perception and visual realism. Beyond static images, the thesis also provides a review of virtual staging technologies, introducing VISPI (Virtual Staging Pipeline for Single Indoor Panoramic Images), an AI-based system capable of automatically decluttering, relighting, and inserting virtual objects into panoramic scenes. Finally, we turn to the human experience, evaluating immersion, visual comfort, and perceptual depth through extensive user studies in WebXR, using a Meta Quest 3 headset to measure how AI-driven enhancements impact usability and presence. The results indicate that integrating traditional methods for generating immersive environments with AI-driven techniques for stereoscopic rendering and virtual staging advances the field of Indoor Immersive Reality by enabling the generation of more immersive, comfortable, and realistic panoramic experiences across industries such as virtual real estate, interactive training, and metaverse applications. The thesis's contributions establish a foundation for the next generation of virtual environments—ones that are more engaging, visually better, and accessible to all.

Date of Award	2025
Original language	American English
Awarding Institution	HBKU College of Science and Engineering