Decoding Perceptual Realism for Real-Time Virtual and Augmented Realities

You are cordially invited to attend the PhD Dissertation Defence of Taimoor Tariq on Tuesday 28 October 2025 at 15:00 in room D1.13.

Abstract:
Imagine a day when memories unfold before us in videos and photos so vividly that we are not just watching them, but stepping back into them, immersed in every sound, every glance and every breath. In that moment, the screen becomes more than a display; it becomes the nearest thing we have to turning back time itself. In order to fully realize this dream of AR/VR, computer generated realities need to be visually indistinguishable from our physical reality. This goal has popularly been defined as passing the Visual Turing Test. I approach this challenge with a realization that visual reality is a fickle concept, not so much related to accurate/expected simulation of light matter interactions happening in the real-world, but more to the mysterious compressive encoding and processing done by our visual system. Our eyes play tricks on us, and in these tricks lie secrets that not only open the door for rethinking content synthesis for AR/VR, but realizations that fundamentally alter our subconsciously resolute beliefs of what visual realism means. During my doctoral research, I have strived to decode the fascinating processes of human visual perception with the goal of designing rendering techniques that help enable the real-time reproduction of perceptual realism in AR/VR, with specific focus on the core constituent cues of realism such as spatial quality, motion, dynamic range and depth. My thesis is that if we can effectively decode the different mysterious aspects of the human visual processing, we can design computer graphics techniques focused on minimalistically synthesizing our visual cognitive reality, as opposed to the complex physical reality. This would allow future techniques to omit processing of unnecessary visual information, enabling the potential for truly realistic real-time AR/VR.

Dissertation Committee:
- Prof. Piotr Krzysztof Didyk, Università della Svizzera italiana, Switzerland (Research Advisor)
- Prof. Kai Hormann, Università della Svizzera italiana, Switzerland (Internal Member)
- Prof. Michael Multerer, Università della Svizzera italiana, Switzerland (Internal Member)
- Prof. David Luebke, NVIDIA, US (External Member)
- Prof. Tobias Ritschel, UCL, UK (External Member)
- Prof. Ruth Rosenholtz, MIT CSAIL, US (External Member)