Real-Time Scent Generation for Video Games

Abstract

This thesis presents a versatile software framework designed to incorporate olfactory stimuli into interactive gaming environments, exemplified through its application in Minecraft. Our approach leverages recent advancements in wearable scent-release technology and cross-modal machine learning, focusing on an adaptable scent predicting and releasing framework with a customizable pipeline, allowing precise configuration and troubleshooting tailored to different gaming contexts. A quantitative user study was conducted to assess the system’s impact, comparing an optimized scent integration against a randomized scent-integrated experience. Results indicate a significant enhancement in the immersive gaming experience under the optimized configuration, emphasizing the importance of contextually aligned olfactory stimuli, confirming the effectiveness of our system.