Xerophobia started with a simple question: what if you couldn't see the monster? In VR, where presence is everything, removing sight seemed terrifying. The challenge was making audio alone convey enough information to let players navigate, survive, and feel hunted.
The Core Concept
Xerophobia places you in darkness, pursued by a giant centipede. You hear its legs scraping stone, its body sliding through tunnels, its breathing as it gets close. Your only tool? Echolocation through sound and spatial audio cues.
The design pillars were:
- Sound is navigation: Players must learn to "see" with their ears
- Tension through anticipation: Hearing the threat build is scarier than seeing it
- Spatial awareness matters: VR's 3D audio is essential, not optional
- Accessible difficulty: Audio clarity adjusts to player skill
Spatial Audio Implementation
Godot's audio system was key. The implementation uses Area3D nodes with custom audio zones that dynamically adjust based on player position. Each sound has distance attenuation, directional bias, and environment effects (reverb for caves, dampening for cloth).
The centipede's movement creates layered audio:
- Base loop: constant slithering ambience
- Leg sounds: individual taps that spatialize per leg position
- Breathing: proximity-based intensity
- Impact sounds: when it hits walls or moves quickly
Echolocation Mechanic
Players can emit echolocation clicks. The game uses raycasts in all directions, and when they hit objects, it triggers audio cues with distance-based timing. Close objects return quickly, distant ones take longer. Different materials have different "reflections"—stone rings clear, wood sounds muffled.
This was the hardest part to balance. Too much information and it's not scary. Too little and players get frustrated. Playtesting showed most players adapt after 10-15 minutes.
VR-Specific Challenges
VR horror is different from flat-screen horror:
- Motion sickness: Sudden movements or disorientation can make players sick
- Personal space: Things getting close to your face is genuinely unsettling
- Control limitations: No jump scares that take away control—players hate it
- Comfort options: Teleport vs smooth movement, vignettes, snap turning
The game includes comfort options but was designed assuming smooth locomotion for maximum immersion.
The AI System
The centipede needed to feel intelligent but not omniscient. It uses a sound-based detection system:
- Player movement creates noise proportional to speed
- Crouching reduces noise significantly
- Running is loud and attracts from far away
- The centipede investigates recent sounds, not current player position
This creates cat-and-mouse gameplay where standing still feels safe until the centipede's random patrol brings it near you anyway. Delicious tension.
Sound Design Process
The audio uses library assets and synthesized effects. The centipede's sound came from layering:
- Snake sliding sounds (pitched down)
- Wet leather creaking
- Horse breathing (pitched down and reverbed)
- Scuttling insects (layered and timed to movement)
The result is alien and organic. Several playtesters said the sound alone was enough to make them uncomfortable even before seeing the creature.
Performance Optimization
VR demands consistent frame rates. On Quest 1 and 2, targeting 72fps minimum required several optimizations:
- Baked lighting for all static geometry
- Aggressive LODs for the centipede model
- Limiting active audio sources to 8 max
- Occlusion culling for audio (sounds behind walls are quieter)
- Simple shader materials, no complex effects
The environments are deliberately claustrophobic partly for design, partly because smaller spaces are easier to optimize.
Player Psychology
Horror in VR is intense. Too intense can cause players to quit. I learned:
- Safe spaces matter: Small alcoves where players can catch their breath
- Pacing is everything: Constant tension leads to fatigue, not fear
- Agency reduces fear: Giving players tools makes them braver
- Lack of agency increases fear: But only if it's designed, not accidental
The echolocation mechanic gives players just enough control to feel capable while maintaining vulnerability.
Testing and Iteration
Playtesting VR is harder than flat games:
- You can't watch over someone's shoulder effectively
- Recording VR footage doesn't capture the experience
- Motion sickness means shorter test sessions
- You need a VR headset for each tester
I ran about 30 playtest sessions with friends and volunteers. Key insights:
- 15% of players couldn't handle darkness + VR movement (added brightness option)
- Most players initially move too fast (added tutorial that teaches stillness)
- The centipede needed to "win" occasionally or players lose respect for it
- Death should be quick—prolonged death sequences make players quit
Lessons Learned
Building a VR game taught me:
- Playtest early and often: VR comfort issues emerge in real use, not theory
- Audio is expensive: Budget for good sound or find a collaborator
- Accessibility options aren't optional: Motion sickness is real and varies by person
- Less is more: Simple, polished mechanics beat complex, janky ones
- VR presence is powerful: Small details like hand tracking elevate the experience
The Future of VR Horror
We're still figuring out what works in VR horror. Traditional jump scares are too intense. Psychological horror works well but is hard to design. Audio-driven experiences like Xerophobia are underexplored.
I believe the future of VR horror is in immersive systems—making players feel hunted through emergent AI, environmental storytelling, and sensory manipulation. Not cutscenes or scripted events, but dynamic, reactive experiences.
Xerophobia was my experiment in this direction. It's not perfect, but it proved that audio-first design can create genuine fear in VR. And honestly? That's terrifying to know.
← Back to Blog