System Overview
I designed and implemented a 3D compass interface in Unreal Engine for password entry in VR. It combines symbolic input and contextual cues to study authentication that fits an immersive environment.
Architecture
VR Client
The Unreal Engine prototype presents a spatial direction-entry interface in the headset.
Input and Interaction Layer
The user enters a password through symbolic choices, environmental context, or a combination of both, depending on the study condition.
Study and Feedback Layer
The prototype supports repeated authentication attempts for a within-subject, multi-session study. Researchers measured entry efficiency, recall, subjective memorability, and resistance to observational attacks.
Core Features
- Spatial compass: The interface presents 3D directional choices for password entry.
- Symbolic and contextual cues: The study could compare each cue type and their combination.
- Immersive entry: Input occurs inside VR rather than through a conventional text field.
- Study instrumentation: The prototype supported comparison across multiple sessions and conditions.
Interaction Pipeline
- The user sees the current VR authentication condition.
- The user selects a sequence using the spatial compass.
- The system records the entry and checks it against the assigned password.
- The study captures timing, recall, and subjective responses.
- Researchers compare performance across symbolic, contextual, and combined conditions.
Technical and Design Challenges
Balancing memorability and entry speed
The combined interface was designed to retain contextual cues while reducing the slower entry observed with context alone.
Studying observation resistance
The evaluation considered how visible entry actions could be interpreted by an observer, alongside ordinary usability measures.
Evaluation and Publications
A within-subject, multi-session study with 32 participants found faster entry for the combined interface than context-only authentication and higher subjective memorability than symbols alone. The full paper appeared at IEEE ISMAR 2024, with co-first authorship. An earlier IEEE VRW 2023 paper introduced the concept, with equal contribution.