📽️ Project Video
Here is the official project video for Depthbound, showcasing the gameplay and social interaction within the Mixed Reality prototype:
✨ Abstract
Analog board games enable social interaction through physical co-presence, shared artifacts, and nonverbal communication. Mixed Reality (MR) transfers these principles into hybrid environments but poses challenges regarding social isolation caused by Head-Mounted Displays (HMDs).
This thesis investigated how design principles can promote social interaction and shared presence in co-located MR. The study involved the development of Depthbound, a cooperative maritime-themed card game for three players, and an empirical evaluation comparing analog play against different MR configurations.
🎮 The Project: Depthbound
Depthbound is a co-located MR adaptation of the cooperative card game Hanabi.
- Core Mechanic: Asymmetric information—players see everyone's cards but their own. This forces communication and joint attention.
- Setting: A maritime dark-fantasy world featuring a "ghost captain" and an interactive parrot companion ("Flint") for diegetic feedback.
- Architecture: The game uses Shared Spatial Anchors to ensure the virtual table exists at the exact same physical position for all three players, creating a consistent shared reality.
🛠️ Tech Stack
- Hardware: Meta Quest 3
- Engine: Unity 6
- Networking: Photon Fusion 2 (Host-Client architecture with State Authority)
- XR Tools: Meta XR Core SDK, Interaction SDK, Meta Avatars SDK (with Lip Sync)
🔬 Methodology & Evaluation
A within-subject study ($N=9$) was conducted to compare three experimental conditions:
- Condition A (Analog): Classic physical card game (Baseline).
- Condition B (MR-Controller): MR version using controllers; no avatars.
- Condition C (MR-Avatars): MR version using Hand-Tracking and full-body Avatars.
📊 Key Findings
The empirical analysis yielded several critical insights for XR designers:
- Analog is still the Gold Standard: Condition A achieved the highest Social Presence scores ($M=6.49$). While MR is immersive, physical face-to-face interaction remains superior for reading subtle non-verbal cues.
- Avatars vs. "Extended Presence": Contrary to the initial hypothesis, full-body avatars did not significantly increase social presence compared to no avatars ($p=0.910$). In co-located passthrough, avatars can create a "visual conflict" with the visible physical body of the user. The study suggests using "Extended Presence"—minimalist augmentations—rather than full virtual replacements.
- Input Preference: While usability scores (SUS) were nearly identical for Controllers (77.2) and Hand-Tracking (78.9), 77.8% of participants preferred Hand-Tracking. It was perceived as more natural and lowered the barrier to entry for social play.
📘 Design Guidelines derived
Based on the study, a design guide was formulated for co-located MR:
- Prioritize Extended Presence: Don't obscure the physical user with an avatar unless you have perfect face tracking. Enhance what is already there.
- Shared Spatial Anchors are Mandatory: A shared coordinate system is the foundation of co-located trust.
- Center Attention: Due to limited FOV, critical game info must be on the table, not floating in menus, to maintain eye contact.
- Multimodal Feedback: Compensate for the lack of tactile feedback in Hand-Tracking with exaggerated audio-visual cues (spatial audio, particle effects).