Positive maximum distance, or null only for the final tier.
Model group id the population design names for this tier.
Canonical digest of the referenced model group, including parts, skeleton and materials, or the producer's explicit missing/unencodable reference marker digest.
Compiler-owned runtime model id derived from model; viewers and render gates look the runtime model up by it.
Semantic near-to-far tier.
One camera-selected runtime representation for a compiled population. A resolvable canonically encodable model group digest covers its parts, skeleton and materials; a missing or unencodable group instead hashes an explicit identity marker.