NCP-OUSD key terms

The fundamental scene object in OpenUSD, identified by a path such as /World/Chair. A prim carries typed schemas, metadata, and properties, and can contain child prims to form a namespace hierarchy.

A slash-delimited absolute or relative identifier locating a prim in the namespace, for example /World/Group/Lamp. Absolute paths begin at the pseudo-root, while relative paths are resolved against a given anchor prim.

The invisible parent of all absolute prim paths, written as a single slash. It is not a real prim but serves as the root of the prim namespace and the anchor for absolute path resolution.

An in-memory composed view of one or more layers, presented as a single prim hierarchy. Queries against a stage return composed, resolved values rather than raw per-layer data.

A single serialized USD document, typically a .usd, .usda, .usdc, or .usdz file, that stores prim specs and property values. Layers are the units that composition arcs combine into a stage.

A composition arc that merges another layer into a layer's own opinions at lower strength, without introducing a namespace offset. Sublayers are listed in the layer metadata and compose in a defined order.

A composition arc that pulls the composed contents of another prim or layer into a prim, optionally under a namespace offset. References are the primary mechanism for reusing assets across scenes.

A reference-like composition arc whose contents can be loaded or unloaded on demand. Payloads let large assets stay unloaded until needed, keeping stage opening and traversal fast.

The set of operators, including sublayers, references, payloads, inherits, variants, and specializes, that combine layer opinions into a composed stage. Each arc has a defined strength and namespace behavior.

The ordered list of layers contributing to a single root layer, including the root layer itself and its recursive sublayers. Order within the stack determines opinion strength during value resolution.

The strength ordering of composition arcs: Local opinions, Inherits, Variants, References, Payloads, and Specializes. It predicts which opinion wins when multiple arcs author the same property.

A named group of alternatives, each a variant, from which exactly one selection is active at a time. Variant sets encode asset options such as model versions, looks, or levels of detail.

The stored choice indicating which variant of a variant set is active on a prim. Selections propagate down the hierarchy and may be authored in a different layer than the variant definitions.

A named, registered type describing the properties and behavior a prim or attribute may have. Schemas are applied to prims through the API and drive typing, validation, and tooling behavior.

A schema that defines a prim's primary type, such as Xform or Mesh, recorded in the prim's typeName metadata. A prim has at most one typed schema at a time.

A single- or multi-apply schema that adds capabilities to a prim without changing its primary type. Multi-apply schemas are instantiated with an instance name, such as MaterialBindingAPI on a mesh.

A named, typed data field on a prim, either an attribute holding values over time or a relationship pointing at other paths. Properties are addressed by namespaced names like points or material:binding.

A property that stores typed values, optionally varying over time samples. Attributes declare a type name and may carry metadata such as interpolation, variability, and custom flags.

A property that stores paths to other prims or properties instead of data values. Relationships express connections such as material bindings and do not support time samples.

A value authored for an attribute at a specific time code, enabling animation. Multiple time samples on one attribute define how the value changes over the stage's time range.

A unitless numeric value, often mapped to frames, used to index time samples and stage time metadata. Time codes are converted to seconds through the stage's time codes per second.

Attribute metadata declaring how values between time samples are computed, such as held or linear. It controls whether a value steps, blends, or is undefined between authored samples.

Structured, typed information attached to layers, prims, properties, or other objects that describes them without being scene data. Examples include kind, documentation, and asset info.

Model hierarchy metadata declaring a prim as a component, group, assembly, or subcomponent. Kinds let tools treat model hierarchies semantically, for example when selecting or instancing whole models.

A convention in which prims tagged with kinds form nested models, from assemblies down to components. It gives pipelines a predictable structure for traversal, selection, and asset boundaries.

Layer or prim metadata describing an asset's identity, such as its name, version, and identifier. Pipelines use it to track provenance and resolve dependencies between published assets.

The pluggable component that maps asset paths, such as @./textures/a.png@, to resolvable locations. Resolvers implement studio-specific search paths, versioning, and URI schemes.

A packaged, uncompressed zip archive containing USD layers and their referenced assets, designed for portable distribution. It is commonly used to ship assets to mobile and web viewers.

A composition feature where prims sharing the same instanceable reference or payload are drawn and stored once, with shared data. Instancing reduces memory and draw calls for repeated assets.

A prim metadata flag marking a prim as eligible for instancing when it has a direct reference or payload. Once instanced, its namespace becomes read-only and its children are shared.

The hidden, shared prim under /__Prototype_1 and similar paths that holds the composed contents of an instanced asset. Instances point at prototypes rather than owning their own children.

The OpenUSD rendering architecture that consumes a scene description and produces renderable output through pluggable backends. Hydra separates scene data from renderer-specific implementation.

A composable, read-only interface in Hydra that exposes scene data as a tree of prim and property entries. Scene indices can be chained to filter or transform data before rendering.

A Hydra component that pulls from a scene index and maintains the renderable state consumed by a render delegate. It tracks changes and synchronizes updates to the backend.

A Hydra plugin implementing a specific renderer, such as a rasterizer or path tracer. It receives scene data and draw calls from the render index to produce images.

A traversal utility that walks a prim hierarchy depth-first, with options to skip or prune subtrees. It is the standard way to iterate stage contents efficiently in C++.

The practice of checking a stage or layer against schema and pipeline rules to catch missing, mistyped, or inconsistent data. Validators typically report errors and warnings by path.

A debugging technique that temporarily excludes selected layers from composition, often via the stage's muted layer list. It isolates whether a problem originates in a particular layer.