Every tree in the forest. Clustered spatial audio.
summary
Claims a runnable system that turns per-object sounds into dynamic clusters using FOD and middleware like fode or y's. Implements world space clustering for 2D/short impacts and perspective clustering for 3D distance via an angle-scaled capture radius, plus point weight based interpolation so emitters never teleport. Handles particle-scale counts (tested with ~1,000,000 points at 30 FPS on an older laptop), supports custom point parameters, but warns it fails for distinct speech or tightly timed footsteps.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:07Volumetric audio challenges
- 6:05World space clustering in Vessel
- 9:35How clustering works
- 12:25Spatialization terms
- 20:07Demo: perspective clustering
- 20:50What is perspective clustering
- 26:25Method changes
- 29:14Point weight interpolation
- 33:47Culling distance
- 40:51Audio parameters and point data
- 43:20Q&A
