Extend DELUGE to variable-count, topology-changing, and correspondence-free point clouds

Extend DELUGE, the particle-streaming compression architecture, to data with variable point counts, topology changes such as particle splitting and merging, and captured point clouds lacking simulator-provided particle IDs by developing keyframe re-initialization, inter-frame correspondence search, and null-point handling.

Background

DELUGE is designed for interactive simulation streams in which particle IDs are preserved across frames, providing stable point correspondence and enabling efficient delta-frame coding. Its current design assumes a fixed number of particles and does not directly support data in which points appear, disappear, split, merge, or lack persistent identifiers.

The paper identifies ID assignment—not the compression algorithm—as the bottleneck when applying DELUGE to captured point clouds. Addressing this unresolved limitation would require keyframe re-initialization, correspondence estimation between frames, and explicit handling of null points; the authors suggest that the codec’s Global ID and Morton ordering could provide a starting point for accelerating correspondence search.