sdd:depicts
sdd:depicts grounds a scene instance in the fact it depicts. On a sdd:Node the object is a world resource; on an sdd:Edge the object is a quoted triple (RDF-star) naming the relationship the edge visualizes, not a plain resource.
sdd:depicts is the groundedness predicate that separates structural-visual quads from aesthetic ones (spec §3.1): an element that depicts something is derived cache, V = derive(F, O), verified by re-derivation; an element with no sdd:depicts is purely aesthetic content. This is also what makes an edge's connection verifiable as a single unit — the quoted triple it depicts is exactly what the source is checked against, rather than the edge's two endpoints being checked separately.