This guide primarily uses Houdini to explain concepts, as it is one of the most accessible and easiest ways to learn the ways of Usd with. You can install a non-commercial version for free from their website. It is highly recommended to use it when following this guide.

If this is your first time working with Usd, we recommend watching this 20 minute video from Apple:

Understand USD fundamentals (Frm WWDC 2022)

It covers the basic terminology in a very succinct manner.

from pxr import Sdf, Usd
## Stages
# Native USD
stage = Usd.Stage.CreateInMemory()
# Houdini Python LOP
stage = hou.pwd().editableStage()
## Layers
# Native USD
layer = Sdf.Layer.CreateAnonymous()
# Houdini Python LOP
layer = hou.pwd().editableLayer()

We'd actually recommend starting out with the lower level API as soon as you can, as it will force you to write performant code from the start.

You'll be using these two API levels all the time when working with Usd. The high level API is often used with read ops, the low level with write ops or inspecting the underlying caches of the high level API.

This level always operates on the composed state of the stage. This means as soon as you are working stages, you'll be using the higher level API. It also takes care of validation data/setting common data, whereas the lower level API often leaves parts up to the user.

We try to stay terminology agnostic as much as we can, but some vocab you just have to learn to use USd. We compiled a small cheat sheet here, that can assist you with all those weird Usd words.

Anything that is path related in your hierarchy, use Sdf.Path objects. It will make your life a lot easier than if you were to use strings.

When using Usd, we'll rarely run into the relationship [] encoded targets paths. Instead we use the Usd.Relationship/Sdf.RelationshipSpec methods to set the path connections. Therefore it is just good to know they exist.

In production, these are the classes you'll have the most contact with. They handle data creation/storage/modification. They are the heart of what makes Usd be Usd.

The job of specifiers is mainly to define if a prim should be visible to hierarchy traversals. More info about traversals in our Layer & Stage section.

By default stage traversals will skip over over only prims. Prims that only have an over also do not get forwarded to Hydra render delegates.

• By default stage traversals will skip over class prims.
• Usd refers to class prims as "abstract", as they never directly contribute to the hierarchy.
• We target these class prims via inherits/internal references and specialize composition arcs

As you can see on this page, most of the prim functionality is actually done via metadata, except path, composition and property related functions/attributes.

The 'IsA' check is a very valueable check to see if something is an instance of a (base) class. It is similar to Python's isinstance method.

The prim's prim.IsA(<typeName>) checks are highly performant, you should use them as often as possible when traversing stages to filter what prims you want to edit. Doing property based queries to determine if a prim is of interest to you, is a lot slower.

Technically properties are also stored as metadata on the Sdf.PrimSpec. So later on when we look at composition, keep in mind that the prim stack therefore also drives the property stack. That's why the prim index is on the prim level and not on the property level.

...
print(prim_spec.properties, prim_spec.attributes, prim_spec.relationships)
print(prim_spec.GetInfo("properties"))
...

Since the lower level API doesn't see the schema properties, these commands will only return what is actually in the layer, in Usd speak authored.

For attributes that don't need to be accessed by Hydra (USD's render abstraction interface), we don't need to set the interpolation. In order for an attribute, that does not derive from a schema, to be accessible for the Hydra, we need to namespace it with primvars:, more info below at primvars. If the attribute element count for non detail (constant) attributes doesn't match the corresponding prim/vertex/point count, it will be ignored by the renderer (or crash it).

When we set schema attributes, we don't need to set the interpolation, as it is provided from the schema.

Attributes are the only part of USD than can encode time varying data.

Attribute connections are encoded from target attribute to source attribute. The USD file syntax is: <data type> <attribute name>.connect = </path/to/other/prim.<attribute name>

def Scope "materials"
{
    def Material "karmamtlxsubnet" (
    )
    {
        token outputs:mtlx:surface.connect = </materials/karmamtlxsubnet/mtlxsurface.outputs:out>

        def Shader "mtlxsurface" ()
        {
            uniform token info:id = "ND_surface"
            string inputs:edf.connect = </materials/karmamtlxsubnet/mtlxuniform_edf.outputs:out>
            token outputs:out
        }

        def Shader "mtlxuniform_edf"
        {
            uniform token info:id = "ND_uniform_edf"
            color3f inputs:color.connect = </materials/karmamtlxsubnet/mtlx_constant.outputs:out>
            token outputs:out
        }

        def Shader "mtlx_constant"
        {
            uniform token info:id = "ND_constant_float"
            float outputs:out
        }
    }
}

• An attribute with the primvars: can be accessed at render time by your render delegate for things like settings, materials and AOVs
• detail (UsdGeom.Tokens.constant interpolation) primvars are inherited down the hierarchy, ideal to apply a constant value per USD prim, e.g. for render geometry settings or instance variation.

As a best practice you should build your hierarchies in such a way that you don't have to write a purpose value per prim. A typical setup is to have a <asset root>/GEO, <asset root>/PROXY, ... hierarchy, where you can then tag the GEO, PROXY, ... prims with the purpose. That way all child prims receive the purpose and you have a single point where you can override the purpose.

This is useful, if you for example want to load a whole scene in proxy purpose and a specific asset in render purpose. You then just have to edit a single prim to make it work.

### High Level ###
from pxr import Sdf, Usd, UsdGeom
stage = Usd.Stage.CreateInMemory()
cube_prim = stage.DefinePrim(Sdf.Path("/bicycle/RENDER/cube"), "Cube")
render_prim = cube_prim.GetParent()
render_prim.SetTypeName("Xform")
UsdGeom.Imageable(render_prim).GetPurposeAttr().Set(UsdGeom.Tokens.render)
sphere_prim = stage.DefinePrim(Sdf.Path("/bicycle/PROXY/sphere"), "Sphere")
proxy_prim = sphere_prim.GetParent()
proxy_prim.SetTypeName("Xform")
UsdGeom.Imageable(proxy_prim).GetPurposeAttr().Set(UsdGeom.Tokens.proxy)
# We can also query the inherited purpose:
imageable_api = UsdGeom.Imageable(cube_prim)
print(imageable_api.ComputePurpose()) # Returns: 'render'

### Low Level ###
from pxr import Sdf, UsdGeom
layer = Sdf.Layer.CreateAnonymous()
prim_path = Sdf.Path("/bicycle")
prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
prim_spec.specifier = Sdf.SpecifierDef
prim_spec.typeName = "Cube"
attr_spec = Sdf.AttributeSpec(prim_spec, "purpose", Sdf.ValueTypeNames.Token)
attr_spec.default = UsdGeom.Tokens.render

### High Level ###
# UsdGeom.Imageable()
# Get: 'ComputeVisibility'
# Set: 'MakeVisible', 'MakeInvisible'
from pxr import Sdf, Usd, UsdGeom
stage = Usd.Stage.CreateInMemory()
cube_prim = stage.DefinePrim(Sdf.Path("/set/yard/bicycle"), "Cube")
sphere_prim = stage.DefinePrim(Sdf.Path("/set/garage/bicycle"), "Sphere")
set_prim = cube_prim.GetParent().GetParent()
set_prim.SetTypeName("Xform")
cube_prim.GetParent().SetTypeName("Xform")
sphere_prim.GetParent().SetTypeName("Xform")
UsdGeom.Imageable(set_prim).GetVisibilityAttr().Set(UsdGeom.Tokens.invisible)
# We can also query the inherited visibility:
# ComputeEffectiveVisibility -> This handles per purpose visibility
imageable_api = UsdGeom.Imageable(cube_prim)
print(imageable_api.ComputeVisibility()) # Returns: 'invisible'
# Make only the cube visible. Notice how this automatically sparsely
# selects only the needed parent prims (garage) and makes them invisible.
# How cool is that!
imageable_api.MakeVisible()

### Low Level ###
from pxr import Sdf, UsdGeom
layer = Sdf.Layer.CreateAnonymous()
bicycle_prim_path = Sdf.Path("/set/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
bicycle_prim_spec.typeName = "Cube"
bicycle_vis_attr_spec = Sdf.AttributeSpec(prim_spec, "visibility", Sdf.ValueTypeNames.Token)
bicycle_vis_attr_spec.default = UsdGeom.Tokens.inherited

We cover how to use the a bounding box cache in detail in our stage API query caches for optimized bounding box calculation and extent writing.

### High Level ###
# UsdGeom.Boundable()
# Get: 'GetExtentAttr', 'CreateExtentAttr'
# Set: 'ComputeExtent '
from pxr import Sdf, Usd, UsdGeom
stage = Usd.Stage.CreateInMemory()
cube_prim = stage.DefinePrim(Sdf.Path("/bicycle/cube"), "Cube")
bicycle_prim = cube_prim.GetParent()
bicycle_prim.SetTypeName("Xform")
# If we change the size, we have to re-compute the bounds
cube_prim.GetAttribute("size").Set(10)
boundable_api = UsdGeom.Boundable(cube_prim)
print(boundable_api.GetExtentAttr().Get()) # Returns:  [(-1, -1, -1), (1, 1, 1)]
extent = boundable_api.ComputeExtent(Usd.TimeCode.Default())
boundable_api.GetExtentAttr().Set(extent)
print(boundable_api.GetExtentAttr().Get()) # Returns: [(-5, -5, -5), (5, 5, 5)]
# Author extentsHint
# The bbox cache has to be specified with what frame and purpose to query
bbox_cache = UsdGeom.BBoxCache(1001, [UsdGeom.Tokens.default_, UsdGeom.Tokens.render])
model_api = UsdGeom.ModelAPI(bicycle_prim)
extentsHint = model_api.ComputeExtentsHint(bbox_cache)
model_api.SetExtentsHint(extentsHint)
# Or model_api.SetExtentsHint(extentsHint, <frame>)
### Low Level ###
from pxr import Sdf, UsdGeom, Vt
layer = Sdf.Layer.CreateAnonymous()
cube_prim_path = Sdf.Path("/bicycle/cube")
cube_prim_spec = Sdf.CreatePrimInLayer(layer, cube_prim_path)
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, cube_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
bicycle_prim_spec.typeName = "Xform"
# The querying should be done via the high level API.
extent_attr_spec = Sdf.AttributeSpec(cube_prim_spec, "extent", Sdf.ValueTypeNames.Vector3fArray)
extent_attr_spec.default = Vt.Vec3fArray([(-1, -1, -1), (1, 1, 1)])
site_attr_spec = Sdf.AttributeSpec(cube_prim_spec, "size", Sdf.ValueTypeNames.Float)
site_attr_spec.default = 10
extent_attr_spec.default = Vt.Vec3fArray([(-5, -5, -5), (5, 5, 5)])
# Author extentsHint
extents_hint_attr_spec = Sdf.AttributeSpec(bicycle_prim_spec, "extentsHint", Sdf.ValueTypeNames.Vector3fArray)
extents_hint_attr_spec.default = Vt.Vec3fArray([(-5, -5, -5), (5, 5, 5)])

### High Level ###
# Get: 'GetForwardedTargets', 'GetTargets',
# Set: 'AddTarget', 'SetTargets'
# Clear: 'RemoveTarget', 'ClearTargets'
from pxr import Sdf, Usd, UsdGeom
stage = Usd.Stage.CreateInMemory()
cube_prim = stage.DefinePrim(Sdf.Path("/cube_prim"), "Cube")
sphere_prim = stage.DefinePrim(Sdf.Path("/sphere_prim"), "Sphere")
myFavoriteSphere_rel = cube_prim.CreateRelationship("myFavoriteSphere")
myFavoriteSphere_rel.AddTarget(sphere_prim.GetPath())
print(myFavoriteSphere_rel.GetForwardedTargets()) # Returns:[Sdf.Path('/sphere_prim')]
# myFavoriteSphere_rel.ClearTargets()
# We can also forward relationships to other relationships.
cylinder_prim = stage.DefinePrim(Sdf.Path("/sphere_prim"), "Cylinder")
myFavoriteSphereForward_rel = cylinder_prim.CreateRelationship("myFavoriteSphereForward")
myFavoriteSphereForward_rel.AddTarget(myFavoriteSphere_rel.GetPath())
# GetForwardedTargets: This gives us the final fowarded paths. We'll use this most of the time.
# GetTargets: Gives us the paths set on the relationship, forwarded paths are not baked down.
print(myFavoriteSphereForward_rel.GetForwardedTargets()) # Returns:[Sdf.Path('/sphere_prim')]
print(myFavoriteSphereForward_rel.GetTargets()) # Returns: [Sdf.Path('/cube_prim.myFavoriteSphere')]

### Low Level ###
from pxr import Sdf, UsdGeom
layer = Sdf.Layer.CreateAnonymous()
cube_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/cube_prim"))
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
sphere_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/sphere_prim"))
sphere_prim_spec.specifier = Sdf.SpecifierDef
sphere_prim_spec.typeName = "Cube"
rel_spec = Sdf.RelationshipSpec(cube_prim_spec, "proxyPrim")
rel_spec.targetPathList.Append(sphere_prim_spec.path)
# The targetPathList is a list editable Sdf.PathListOp.
# Forwarded rels can only be calculated via the high level API.

One of the biggest bottlenecks in USD is creating enormous hierarchies as you then have a lot of prims that need to be considered as value sources. When creating proxy purpose prims/meshes, we should try to keep it as low-res as possible. Best case we only have a single proxy prim per asset.

### High Level ###
from pxr import Sdf, Usd, UsdGeom
stage = Usd.Stage.CreateInMemory()
render_prim = stage.DefinePrim(Sdf.Path("/bicycle/RENDER/render"), "Cube")
proxy_prim = stage.DefinePrim(Sdf.Path("/bicycle/PROXY/proxy"), "Sphere")
bicycle_prim = render_prim.GetParent().GetParent()
bicycle_prim.SetTypeName("Xform")
render_prim.GetParent().SetTypeName("Xform")
proxy_prim.GetParent().SetTypeName("Xform")
imageable_api = UsdGeom.Imageable(render_prim)
imageable_api.SetProxyPrim(proxy_prim)
# Query the proxy prim
print(imageable_api.ComputeProxyPrim()) # Returns: None
# Why does this not work? We have to set the purpose!
UsdGeom.Imageable(render_prim).GetPurposeAttr().Set(UsdGeom.Tokens.render)
UsdGeom.Imageable(proxy_prim).GetPurposeAttr().Set(UsdGeom.Tokens.proxy)
print(imageable_api.ComputeProxyPrim()) # Returns: (Usd.Prim(</bicycle/PROXY/proxy>), Usd.Prim(</bicycle/RENDER/render>))

### Low Level ###
from pxr import Sdf, UsdGeom
layer = Sdf.Layer.CreateAnonymous()
render_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/render"))
render_prim_spec.specifier = Sdf.SpecifierDef
render_prim_spec.typeName = "Cube"
proxy_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/proxy"))
proxy_prim_spec.specifier = Sdf.SpecifierDef
proxy_prim_spec.typeName = "Cube"
proxyPrim_rel_spec = Sdf.RelationshipSpec(render_prim_spec, "proxyPrim")
proxyPrim_rel_spec.targetPathList.Append(Sdf.Path("/proxy"))

The constructors for arrays allow us to input tuples/lists instead of the explicit base classes.

# Explicit
pxr.Vt.Vec3dArray([Gf.Vec3d(1,2,3)])
# Implicit
pxr.Vt.Vec3dArray([(1,2,3)])

Instead of searching for the corresponding Gf or Vt array class, we can get it from the type name and instantiate it:

vec3h_array = Sdf.ValueTypeNames.Point3hArray.type.pythonClass([(0,1,2)])
# Or: Usd.Attribute.GetTypeName().type.pythonClass([(0,1,2)]) / Sdf.AttributeSpec.typeName.type.pythonClass([(0,1,2)])
print(type(vec3h_array)) # Returns: <class 'pxr.Vt.Vec3hArray'>

As Vt.Array support the buffer protocol, we can map the arrays to numpy without data duplication and perform high performance value editing.

from pxr import Vt
import numpy as np
from array import array
# Python Arrays
vt_array = Vt.Vec3hArray.FromBuffer(array("f", [1,2,3,4,5,6])) # Returns: Vt.Vec3hArray(2, (Gf.Vec3h(1.0, 2.0, 3.0),Gf.Vec3h(4.0, 5.0, 6.0),))
# From Numpy Arrays 
Vt.Vec3hArray.FromNumpy(np.ones((10, 3)))
Vt.Vec3hArray.FromBuffer(np.ones((10, 3)))
# To Numpy arrays
np.array(vt_array)

When creating attributes we always have to specify a Sdf.ValueTypeName, which defines the data type & role. USD ships with great computer graphics related classes in the Gf module, which we can use for calculations. The Vt.Array wrapper around the Gf data base classes implements the buffer protocol, so we can easily map these arrays to Numpy arrays and perform high performance value edits. This is great for adjusting geometry related attributes.

The constructors for arrays allow us to input tuples/lists instead of the explicit base classes.

# Explicit
pxr.Vt.Vec3dArray([Gf.Vec3d(1,2,3)])
# Implicit
pxr.Vt.Vec3dArray([(1,2,3)])

Instead of searching for the corresponding Gf or Vt array class, we can get it from the type name:

vec3h_array =  Sdf.ValueTypeNames.Point3hArray.type.pythonClass((0,1,2))
print(type(vec3h_array)) # Returns: <class 'pxr.Vt.Vec3hArray'>

We'll be using schema classes a lot in production, so we recommend familiarizing yourself with the below examples.

They are the main interface for your prims in the high level API that gives you getters/setters for all the standard properties that ship with Usd.

### Typed Schemas ###
# From type name
prim_type_name = prim.GetTypeName()
prim_typed_schema = Usd.SchemaRegistry.GetTypeFromName(prim_type_name).pythonClass(prim)
# From prim type info
prim_typed_schema = prim.GetPrimTypeInfo().GetSchemaType().pythonClass(prim)

### API Schemas ###
# Non-Applied API Schemas
non_applied_api_schema = Usd.ModelAPI(prim)
# Applied API Schemas
applied_api_schema = UsdGeom.MotionAPI.Apply(prim)

The schema classes then give you access to all of the schemas Get/Set methods and utility functions.

As covered in our prim section, Usd has a PrimDefinition/PrimTypeInfo classes we can use to inspect all properties and metadata given through applied and typed schemas on a given prim. This prim definition/type info carry the full type signature of a given prim.

To get the class from the prim, we can run:

# From type name
prim_type_name = prim.GetTypeName()
prim_typed_schema = Usd.SchemaRegistry.GetTypeFromName(prim_type_name).pythonClass(prim)
# From prim type info
prim_typed_schema = prim.GetPrimTypeInfo().GetSchemaType().pythonClass(prim)

This way we don't have to find and import the right class ourselves.

### Typed Schemas ###
# From type name
prim_type_name = prim.GetTypeName()
prim_typed_schema = Usd.SchemaRegistry.GetTypeFromName(prim_type_name).pythonClass(prim)
# From prim type info
prim_typed_schema = prim.GetPrimTypeInfo().GetSchemaType().pythonClass(prim)

### API Schemas ###
# Non-Applied API Schemas
non_applied_api_schema = Usd.ModelAPI(prim)
# Applied API Schemas
applied_api_schema = UsdGeom.MotionAPI.Apply(prim)

The 'IsA' check is a very valuable check to see if something is an instance of a (base) class. It is similar to Python's isinstance method.

In production, you'll use the assetInfo/customData prim metadata fields to track any production related data. You can also use metadata to edit composition arcs, though the high level API offers nice class wrappers that wrap this for you.

from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
prim_path = Sdf.Path("/bicycle")
prim = stage.DefinePrim(prim_path, "Xform")
prim.SetMetadata("assetInfo", {"version": "1"})
prim.SetAssetInfoByKey("identifier", Sdf.AssetPath("bicycler.usd"))
prim.SetMetadata("customData", {"sizeUnit": "meter"})
prim.SetCustomDataByKey("nested:shape", "round")

We go into more detail over in the schema section on how to create or lookup registered schemas.

The variability actually only declares the intent of time capabilities of the property. You can still write time samples for uniform variability attributes, there are chances though that something else somewhere in the API/Hydra won't work then though. So as a best practice don't write animated attributes to Sdf.VariabilityUniform declared schema attributes. Relationships always have the Sdf.VariabilityUniform intent as they can't be animated.

This is often used to track pipeline relevant data in DCCs. For node based DCCs, this is a convenient way to pass general data down through the node network. For layer based DCCs, this can be used to tag layers (for example to anonymous layers that carry specific pipeline data).

Stages and layers are what make USD work, it is our entry point to accessing our hierarchies.

There are also utility methods available, that are not in the Sdf.Layer/Usd.Stage namespace. We cover these in our advanced concepts in production section.

• Layers are managed via a singleton pattern: Each layer is only opened once in memory and is identified by the layer identifier.
• Layers identifiers can have two different formats:
  • Standard identifiers: Sdf.Layer.CreateNew("/file/path/or/URI/identifier.<ext(.usd/.usdc/.usda)>")
  • Anonymous identifiers: Sdf.Find('anon:<someHash(MemoryLocation)>:<customName>'
• Layers store our prim and property specs, they are the data container for all USD data that gets persistently written to file. When we want to edit layers directly, we have to use the low-level API, the high level API edits the stage, which in return forwards the edits to the layer that is set by the active edit target.
• The Sdf.FileFormat plugin interface allows us to implement plugins that convert the content of (custom) file format's to the USD's prim/property/metadata data model. This is how USD manages the USD crate (binary), alembic and vdb formats.
• USD's crate (binary) format allows layers to be lazily read and written to. Calling layer.Save() multiple times, flushes the in-memory content to disk by appending it to the .usd file, which allows us to efficiently write large layer files. This format can also read in hierarchy data without loading property value data. This way we have low IO when opening files, as the property data gets lazy loaded on demand. This is similar to how we can parse image metadata without reading the image content.

for layer in Sdf.Layer.GetLoadedLayers():
    print(layer.identifier)
# Skip anonymous layers
for layer in Sdf.Layer.GetLoadedLayers():
    if layer.anonymous:
        continue
    print(layer.identifier)

from pxr import Sdf
flippy_layer = Sdf.Layer.FindOrOpen("/opt/hfs19.5/houdini/usd/assets/rubbertoy/geo.usdc")
pig_layer = Sdf.Layer.FindOrOpen("/opt/hfs19.5/houdini/usd/assets/pig/geo.usdc")
flippy_layer.TransferContent(pig_layer)

# Add code to modify the stage.
# Use drop down menu to select examples.
#### Low Level ####
# Get: 'identifier', 'resolvedPath', 'realPath', 'fileExtension' 
# Set: 'identifier'
## Helper functions:
# Get: 'GetFileFormat', 'GetFileFormatArguments', 'ComputeAbsolutePath'
# Create: 'CreateIdentifier', 'SplitIdentifier'
### Anoymous identifiers
# Get: 'anonymous'
## Helper functions:
# Get: 'IsAnonymousLayerIdentifier'
import os
from pxr import Sdf
## Anonymous layers
layer = Sdf.Layer.CreateAnonymous() 
print(layer.identifier) # Returns: anon:0x7f8a1040ba80
layer = Sdf.Layer.CreateAnonymous("myCustomAnonLayer")
print(layer.identifier) # Returns: anon:0x7f8a10498500:myCustomAnonLayer
print(layer.anonymous, layer.resolvedPath or "-", layer.realPath or "-", layer.fileExtension) # Returns: True, "-", "-", "sdf"
print(Sdf.Layer.IsAnonymousLayerIdentifier(layer.identifier)) # Returns True
## Standard layers
layer.identifier = "/my/cool/file/path/example.usd"
print(layer.anonymous, layer.resolvedPath or "-", layer.realPath or "-", layer.fileExtension)
# Returns: False, "/my/cool/file/path/example.usd", "/my/cool/file/path/example.usd", "usd"
# When accesing an identifier string, we should always split it for args to get the URI
layer_uri, layer_args = layer.SplitIdentifier(layer.identifier)
print(layer_uri, layer_args) # Returns: "/my/cool/file/path/example.usd", {}
layer_identifier = layer.CreateIdentifier("/dir/file.usd", {"argA": "1", "argB": "test"})
print(layer_identifier) # Returns: "/dir/file.usd:SDF_FORMAT_ARGS:argA=1&argB=test"
layer_uri, layer_args = layer.SplitIdentifier(layer_identifier)
print(layer_uri, layer_args) # Returns: "/dir/file.usd", {'argA': '1', 'argB': 'test'}
# CreateNew requires the file path to be writable
layer_file_path = os.path.expanduser("~/Desktop/layer_identifier_example.usd")
layer = Sdf.Layer.CreateNew(layer_file_path, {"myCoolArg": "test"})
print(layer.GetFileFormat()) # Returns: Instance of Sdf.FileFormat
print(layer.GetFileFormatArguments()) # Returns: {'myCoolArg': 'test'}
# Get the actuall resolved path (from our asset resolver):
print(layer.identifier, layer.realPath, layer.fileExtension) # Returns: "~/Desktop/layer_identifier_example.usd" "~/Desktop/layer_identifier_example.usd" "usd"
# This is the same as:
print(layer.identifier, layer.resolvedPath.GetPathString())
# Compute a file relative to the directory of the layer identifier
print(layer.ComputeAbsolutePath("./some/other/file.usd")) # Returns: ~/Desktop/some/other/file.usd

### Low Level ###
# Create: 'New', 'CreateNew', 'CreateAnonymous', 
# Get: 'Find','FindOrOpen', 'OpenAsAnonymous',  'FindOrOpenRelativeToLayer', 'FindRelativeToLayer',
# Set: 'Save', 'TransferContent', 'Import', 'ImportFromString', 'Export', 'ExportToString'
# Clear: 'Clear', 'Reload', 'ReloadLayers'
# See all open layers: 'GetLoadedLayers'
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
## The .CreateNew command will check if the layer is saveable at the file location and create an empty file.
layer_file_path = os.path.expanduser("~/Desktop/layer_identifier_example.usd")
layer = Sdf.Layer.CreateNew(layer_file_path)
print(layer.dirty) # Returns: False
## Our layers are marked as "dirty" (edited) as soon as we make an edit.
prin_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/pig"))
print(layer.dirty) # Returns: True
layer.Save()
# Only edited (dirty) layers are saved, when layer.Save() is called
# Only edited (dirty) layers are reloaded, when layer.Reload(force=False) is called.
# Our layer.Save() and layer.Reload() methods also take an optional "force" arg.
# This forces the layer to be saved. We can also call layer.Save() multiple times,
# with the USD binary format (.usd/.usdc). This will then dump the content from memory
# to disk in "append" mode. This avoids building up huge memory footprints when
# creating large layers.
## We can also transfer layer contents:
other_layer = Sdf.Layer.CreateAnonymous()
layer.TransferContent(other_layer)
# Or we import the content from another layer
layer.Import(layer_file_path)
# This is the same as:
layer.TransferContent(layer.FindOrOpen((layer_file_path)))
# We can also import/export to USD ascii representations,
# this is quite usefull for debugging and inspecting the active layer.
# layer.ImportFromString(other_layer.ExportAsString())
layer = Sdf.Layer.CreateAnonymous()
prin_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/pig"))
print(layer.ExportToString())
# Returns:
"""
#sdf 1.4.32
over "pig"
{
}
"""
# We can also remove all prims that don't have properties/metadata
# layer.RemoveInertSceneDescription()

The most important methods are:

• layer.GetCompositionAssetDependencies(): This gets layer identifiers of sublayer/reference/payload composition arcs. This is only for the active layer, it does not run recursively.
• layer.UpdateCompositionAssetDependency("oldIdentifier", "newIdentifier"): The allows us to remap any sublayer/reference/payload identifier in the active layer, without having to edit the list-editable ops ourselves. Calling layer.UpdateCompositionAssetDependency("oldIdentifier", "") removes a layer.

### Low Level ###
# Get: 'GetCompositionAssetDependencies', 'GetAssetInfo', 'GetAssetName', 'GetExternalAssetDependencies',
# Set: 'UpdateCompositionAssetDependency', 'UpdateExternalReference', 'UpdateAssetInfo'
import os
from pxr import Sdf
HFS_env = os.environ["HFS"]
layer = Sdf.Layer.FindOrOpen(os.path.join(HFS_env, "houdini","usd","assets","pig","payload.usdc"))
# Get all sublayer, reference and payload files (These are the only arcs that can load files)
print(layer.GetCompositionAssetDependencies()) # Returns: ['./geo.usdc', './mtl.usdc']
# print(layer.GetExternalReferences(), layer.externalReferences) # The same thing, deprecated method.
# Get external dependencies for non USD file formats. We don't use this with USD files.
print(layer.GetExternalAssetDependencies()) # Returns: [] 
# Get layer asset info. Our asset resolver has to custom implement this.
# A common use case might be to return database related side car data.
print(layer.GetAssetName()) # Returns: None
print(layer.GetAssetInfo()) # Returns: None
layer.UpdateAssetInfo() # Re-resolve/refresh the asset info. This just force requeries our asset resolver query.
# The perhaps most powerful method for dependencies is:
layer.UpdateCompositionAssetDependency("oldIdentifier", "newIdentifier")
# This allows us to repath any composition arc (sublayer/reference/payload) to a new file in the active layer.
# Calling layer.UpdateCompositionAssetDependency("oldIdentifier", ""), will remove the identifier from the
# list-editable composition arc ops.

### Low Level ###
# Get: 'permissionToEdit', 'SetPermissionToEdit'
# Set: 'permissionToSave', 'SetPermissionToSave'
import os
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
print("Can edit layer", layer.permissionToEdit) # Returns: True
Sdf.CreatePrimInLayer(layer, Sdf.Path("/bicycle"))
# Edit permission
layer.SetPermissionToEdit(False)
try: 
    # This will now raise an error
    Sdf.CreatePrimInLayer(layer, Sdf.Path("/car"))
except Exception as e:
    print(e)
layer.SetPermissionToEdit(True)
# Save permission
print("Can save layer", layer.permissionToSave) # Returns: False
try:
    # This fails as we can't save anoymous layers
    layer.Save()
except Exception as e:
    print(e)
# Changing this on anoymous layers doesn't work
layer.SetPermissionToSave(True)
print("Can save layer", layer.permissionToSave) # Returns: False
# If we change the identifer to not be an anonymous identifer, we can save it.
layer.identifier = os.path.expanduser("~/Desktop/layerPermission.usd")
print("Can save layer", layer.permissionToSave) # Returns: True

### Low Level ###
# Get: 'IsMuted', 'GetMutedLayers'
# Set: 'AddToMutedLayers', 'RemoveFromMutedLayers'
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
print(Sdf.Layer.IsMuted(layer)) # Returns: False
Sdf.Layer.AddToMutedLayers(layer.identifier)
print(Sdf.Layer.GetMutedLayers()) # Returns: ['anon:0x7f8a1098f100']
print(Sdf.Layer.IsMuted(layer)) # Returns: True
Sdf.Layer.RemoveFromMutedLayers(layer.identifier)
print(Sdf.Layer.IsMuted(layer)) # Returns: False

# For sublayering we modify the .subLayerPaths attribute on a layer.
# This is the same for both the high and low level API.
### High Level & Low Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Layer onto root layer
layer_a = Sdf.Layer.CreateAnonymous()
layer_b = Sdf.Layer.CreateAnonymous()
root_layer = stage.GetRootLayer()
# Here we pass in the file paths (=layer identifiers).
root_layer.subLayerPaths.append(layer_a.identifier)
root_layer.subLayerPaths.append(layer_b.identifier)
# Once we have added the sublayers, we can also access their layer offsets:
print(root_layer.subLayerOffsets) # Returns: [Sdf.LayerOffset(), Sdf.LayerOffset()]
# Since layer offsets are ready only copies, we need to assign a newly created 
# layer offset if we want to modify them. We also can't replace the whole list, as
# it needs to keep a pointer to the array.
layer_offset_a = root_layer.subLayerOffsets[0]
root_layer.subLayerOffsets[0] = Sdf.LayerOffset(offset=layer_offset_a.offset + 10, 
                                                scale=layer_offset_a.scale * 2)
layer_offset_b = root_layer.subLayerOffsets[1]
root_layer.subLayerOffsets[1] = Sdf.LayerOffset(offset=layer_offset_b.offset - 10, 
                                                scale=layer_offset_b.scale * 0.5)
print(root_layer.subLayerOffsets) # Returns: [Sdf.LayerOffset(10, 2), Sdf.LayerOffset(-10, 0.5)]

# If we want to sublayer on the active layer, we just add it there.
layer_c = Sdf.Layer.CreateAnonymous()
active_layer = stage.GetEditTarget().GetLayer()
root_layer.subLayerPaths.append(layer_c.identifier)

### Low Level ###
# Properties: 'pseudoRoot', 'rootPrims', 'empty'
# Get: 'GetObjectAtPath', 'GetPrimAtPath', 'GetPropertyAtPath', 'GetAttributeAtPath', 'GetRelationshipAtPath',
# Traversal: 'Traverse',
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
# Check if a layer actually has any content:
print(layer.empty) # Returns: True
print(layer.pseudoRoot) # The same as layer.GetPrimAtPath("/")
# Define prims
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/set/yard/bicycle"))
person_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/characters/mike"))
print(layer.rootPrims) # Returns: {'set': Sdf.Find('anon:0x7ff9f8ad7980', '/set'),
                       #           'characters': Sdf.Find('anon:0x7ff9f8ad7980', '/characters')}
# The GetObjectAtPath method gives us prim/attribute/relationship specs, based on what is at the path
attr_spec = Sdf.AttributeSpec(bicycle_prim_spec, "tire:size", Sdf.ValueTypeNames.Float)
attr_spec.default = 10
rel_sec = Sdf.RelationshipSpec(person_prim_spec, "vehicle")
rel_sec.targetPathList.Append(Sdf.Path(bicycle_prim_spec.path))
print(type(layer.GetObjectAtPath(attr_spec.path))) # Returns: <class 'pxr.Sdf.AttributeSpec'>
print(type(layer.GetObjectAtPath(rel_sec.path))) # Returns: <class 'pxr.Sdf.RelationshipSpec'>
# Traversals work differently compared to stages.
def traversal_kernel(path):
    print(path)
layer.Traverse(layer.pseudoRoot.path, traversal_kernel)
print("---")
# Returns:
"""
/set/yard/bicycle.tire:size
/set/yard/bicycle
/set/yard
/set
/characters/mike.vehicle[/set/yard/bicycle]
/characters/mike.vehicle
/characters/mike
/characters
/
"""
# As we can see, it traverses all path related fields, even relationships, as these map Sdf.Paths.
# The Sdf.Path object is used as a "filter", rather than the Usd.Prim object.
def traversal_kernel(path):
    if path.IsPrimPath():
        print(path) 
layer.Traverse(layer.pseudoRoot.path, traversal_kernel)
print("---")
""" Returns:
/set/yard/bicycle
/set/yard
/set
/characters/mike
/characters
"""
def traversal_kernel(path):
    if path.IsPrimPropertyPath():
        print(path) 
layer.Traverse(layer.pseudoRoot.path, traversal_kernel)
print("---")
""" Returns:
/set/yard/bicycle.tire:size
/characters/mike.vehicle
"""
tire_size_attr_spec = attr_spec
tire_diameter_attr_spec = Sdf.AttributeSpec(bicycle_prim_spec, "tire:diameter", Sdf.ValueTypeNames.Float)
tire_diameter_attr_spec.connectionPathList.explicitItems = [tire_size_attr_spec.path]
def traversal_kernel(path):
    if path.IsTargetPath():
        print(">> IsTargetPath", path) 
layer.Traverse(layer.pseudoRoot.path, traversal_kernel)
""" Returns:
IsTargetPath /set/yard/bicycle.tire:diameter[/set/yard/bicycle.tire:size]
IsTargetPath /characters/mike.vehicle[/set/yard/bicycle]
"""

from pxr import Usd
stage = Usd.Stage.CreateInMemory()
print(stage.GetInterpolationType()) # Returns: Usd.InterpolationTypeLinear
stage.SetInterpolationType(Usd.InterpolationTypeHeld)

# Get: 'GetColorConfiguration', 'GetColorManagementSystem', 'GetColorConfigFallbacks'
# Set: 'SetColorConfiguration', 'SetColorManagementSystem', 'SetColorConfigFallbacks'

from pxr import Usd, Sdf
stage = Usd.Stage.CreateInMemory()
bicycle_prim = stage.DefinePrim("/bicycle")
stage.SetMetadata("customLayerData", {"myCustomStageData": 1})
# Is the same as:
layer = stage.GetRootLayer()
metadata = layer.customLayerData
metadata["myCustomRootData"] = 1
layer.customLayerData = metadata
# As with layers, we can also set the default prim
stage.SetDefaultPrim(bicycle_prim)
# Is the same as:
layer.defaultPrim = "bicycle"

stage.SetEditTarget(layer)
# We can also explicitly create the edit target:
# Or
stage.SetEditTarget(Usd.EditTarget(layer))
# Or
stage.SetEditTarget(stage.GetEditTargetForLocalLayer(layer))
# These all have the same effect.

import os
from pxr import Sdf, Usd, UsdUtils
stage = Usd.Stage.CreateInMemory()
root_layer = stage.GetRootLayer()
# Create sublayers with references
bottom_layer_file_path = os.path.expanduser("~/Desktop/layer_bottom.usda")
bottom_layer = Sdf.Layer.CreateNew(bottom_layer_file_path)
top_layer_file_path = os.path.expanduser("~/Desktop/layer_top.usda")
top_layer = Sdf.Layer.CreateNew(top_layer_file_path)
root_layer.subLayerPaths.append(top_layer_file_path)
top_layer.subLayerPaths.append(bottom_layer_file_path)
stage.SetEditTarget(top_layer)
prim = stage.DefinePrim(Sdf.Path("/pig_1"), "Xform")
prim.GetReferences().AddReference("/opt/hfs19.5/houdini/usd/assets/pig/pig.usd", "/pig")
stage.SetEditTarget(bottom_layer)
prim = stage.DefinePrim(Sdf.Path("/pig_1"), "Xform")
prim.GetReferences().AddReference("/opt/hfs19.5/houdini/usd/assets/rubbertoy/rubbertoy.usd", "/rubbertoy")
# Save
stage.Save()
# Layer stack
print(stage.GetLayerStack(includeSessionLayers=False)) 
""" Returns:
[Sdf.Find('anon:0x7ff9f47b9600:tmp.usda'),
Sdf.Find('/home/lucsch/Desktop/layer_top.usda'),
Sdf.Find('/home/lucsch/Desktop/layer_bottom.usda')]
""" 
layers = set()
for layer in stage.GetLayerStack(includeSessionLayers=False):
    layers.add(layer)
    layers.update([Sdf.Layer.FindOrOpen(i) for i in layer.GetCompositionAssetDependencies()])
print(list(layers))
""" Returns:
[Sdf.Find('/opt/hfs19.5/houdini/usd/assets/rubbertoy/rubbertoy.usd'), 
Sdf.Find('/home/lucsch/Desktop/layer_top.usda'),
Sdf.Find('anon:0x7ff9f5677180:tmp.usda'),
Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/pig.usd'),
Sdf.Find('/home/lucsch/Desktop/layer_bottom.usda')]
"""

import os
from pxr import Sdf, Usd, UsdUtils
stage = Usd.Stage.CreateInMemory()
root_layer = stage.GetRootLayer()
# Create sublayers with references
bottom_layer_file_path = os.path.expanduser("~/Desktop/layer_bottom.usda")
bottom_layer = Sdf.Layer.CreateNew(bottom_layer_file_path)
top_layer_file_path = os.path.expanduser("~/Desktop/layer_top.usda")
top_layer = Sdf.Layer.CreateNew(top_layer_file_path)
root_layer.subLayerPaths.append(top_layer_file_path)
top_layer.subLayerPaths.append(bottom_layer_file_path)
stage.SetEditTarget(top_layer)
prim = stage.DefinePrim(Sdf.Path("/pig_1"), "Xform")
prim.GetReferences().AddReference("/opt/hfs19.5/houdini/usd/assets/pig/pig.usd", "/pig")
stage.SetEditTarget(bottom_layer)
prim = stage.DefinePrim(Sdf.Path("/pig_1"), "Xform")
prim.GetReferences().AddReference("/opt/hfs19.5/houdini/usd/assets/rubbertoy/rubbertoy.usd", "/rubbertoy")
# Save
stage.Save()
# Layer stack
print(stage.GetLayerStack(includeSessionLayers=False)) 
""" Returns:
[Sdf.Find('anon:0x7ff9f47b9600:tmp.usda'),
Sdf.Find('/home/lucsch/Desktop/layer_top.usda'),
Sdf.Find('/home/lucsch/Desktop/layer_bottom.usda')]
""" 
layers = set()
for layer in stage.GetLayerStack(includeSessionLayers=False):
    layers.add(layer)
    layers.update([Sdf.Layer.FindOrOpen(i) for i in layer.GetCompositionAssetDependencies()])
print(list(layers))
""" Returns:
[Sdf.Find('/opt/hfs19.5/houdini/usd/assets/rubbertoy/rubbertoy.usd'), 
Sdf.Find('/home/lucsch/Desktop/layer_top.usda'),
Sdf.Find('anon:0x7ff9f5677180:tmp.usda'),
Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/pig.usd'),
Sdf.Find('/home/lucsch/Desktop/layer_bottom.usda')]
"""

### High Level ###
# Create: 'CreateNew', 'CreateInMemory', 'Open', 'IsSupportedFile', 
# Set: 'Save', 'Export', 'Flatten', 'ExportToString', 'SaveSessionLayers',
# Clear: 'Reload'
import os
from pxr import Sdf, Usd
# The stage.CreateNew has multiple method signatures, these take:
# - stage root layer identifier: The stage.GetRootLayer().identifier, this is where your stage get's saved to.
# - session layer (optional): We can pass in an existing layer to use as an session layer.
# - asset resolver context (optional): We can pass in a resolver context to aid path resolution. If not given, it will call
#                                      ArResolver::CreateDefaultContextForAsset() on our registered resolvers.
# - The initial payload loading mode: Either Usd.Stage.LoadAll or Usd.Stage.LoadNone
stage_file_path = os.path.expanduser("~/Desktop/stage_identifier_example.usda")
stage = Usd.Stage.CreateNew(stage_file_path)
# The stage creation will create an empty USD file at the specified path.
print(stage.GetRootLayer().identifier) # Returns: /home/lucsch/Desktop/stage_identifier_example.usd
prim = stage.DefinePrim(Sdf.Path("/bicycle"), "Xform")
# We can also create a stage in memory, this is the same as Sdf.Layer.CreateAnonymous() and using it as a root layer
# stage = Usd.Stage.CreateInMemory("test")
# Or:
layer = Sdf.Layer.CreateAnonymous()
stage.Open(layer.identifier)
## Saving
# Calling stage.Save(), calls layer.Save() on all dirty layers that contribute to the stage.
stage.Save()
# The same as:
for layer in stage.GetLayerStack(includeSessionLayers=False):
    if not layer.anonymous and layer.dirty:
        layer.Save()
# Calling stage.SaveSessionLayers() will also save all session layers, that are not anonymous
## Flatten
# We can also flatten our layers of the stage. This merges all the data, so it should be used with care,
# as it will likely flood your RAM with large scenes. It removes all composition arcs and returns a single layer
# with the combined result
stage = Usd.Stage.CreateInMemory()
root_layer = stage.GetRootLayer()
prim = stage.DefinePrim(Sdf.Path("/bicycle"), "Xform")
sublayer = Sdf.Layer.CreateAnonymous()
root_layer.subLayerPaths.append(sublayer.identifier)
stage.SetEditTarget(sublayer)
prim = stage.DefinePrim(Sdf.Path("/car"), "Xform")
print(root_layer.ExportToString())
"""Returns:
#usda 1.0
(
    subLayers = [
        @anon:0x7ff9f5fed300@
    ]
)

def Xform "bicycle"
{
}
"""
flattened_result = stage.Flatten()
print(flattened_result.ExportToString())
"""Returns:
#usda 1.0
(
)

def Xform "car"
{
}

def Xform "bicycle"
{
}
"""
## Export:
# The export command calls the same thing we just did
# layer = stage.Flatten()
# layer.Export("/myFilePath.usd")
print(stage.ExportToString()) # Returns: The same thing as above
## Reload:
stage.Reload()
# The same as:
for layer in stage.GetUsedLayers():
    # !IMPORTANT! This does not check if the layer is anonymous,
    # so you will lose all your anon layer content.
    layer.Reload()
    # Here is a saver way:
    if not layer.anonymous:
        layer.Reload()

### High Level ###
# Get: 'GetPseudoRoot', 'GetObjectAtPath',
#      'GetPrimAtPath', 'GetPropertyAtPath','GetAttributeAtPath', 'GetRelationshipAtPath',
# Set: 'DefinePrim', 'OverridePrim', 'CreateClassPrim', 'RemovePrim'
# Traversal: 'Traverse','TraverseAll'
from pxr import Sdf, Usd, UsdUtils
stage = Usd.Stage.CreateInMemory()
# Define and change specifier
stage.DefinePrim("/changedSpecifier/definedCube", "Cube").SetSpecifier(Sdf.SpecifierDef)
stage.DefinePrim("/changedSpecifier/overCube", "Cube").SetSpecifier(Sdf.SpecifierOver)
stage.DefinePrim("/changedSpecifier/classCube", "Cube").SetSpecifier(Sdf.SpecifierClass)
# Or create with specifier
stage.DefinePrim("/createdSpecifier/definedCube", "Cube")
stage.OverridePrim("/createdSpecifier/overCube")
stage.CreateClassPrim("/createdSpecifier/classCube")
# Create attribute
prim = stage.DefinePrim("/bicycle")
prim.CreateAttribute("tire:size", Sdf.ValueTypeNames.Float)
# Get the pseudo root prim at "/"
pseudo_root_prim = stage.GetPseudoRoot()
# Or:
pseudo_root_prim = stage.GetPrimAtPath("/")
# Traverse:
for prim in stage.TraverseAll():
    print(prim)
""" Returns:
Usd.Prim(</changedSpecifier>)
Usd.Prim(</changedSpecifier/definedCube>)
Usd.Prim(</changedSpecifier/overCube>)
Usd.Prim(</changedSpecifier/classCube>)
Usd.Prim(</createdSpecifier>)
Usd.Prim(</createdSpecifier/definedCube>)
Usd.Prim(</createdSpecifier/overCube>)
Usd.Prim(</createdSpecifier/classCube>)
"""
# Get Prims/Properties
# The GetObjectAtPath returns the entity requested by the path (prim/attribute/relationship)
prim = stage.GetObjectAtPath(Sdf.Path("/createdSpecifier"))
prim = stage.GetPrimAtPath(Sdf.Path("/changedSpecifier"))
attr = stage.GetObjectAtPath(Sdf.Path("/bicycle.tire:size"))
attr = stage.GetAttributeAtPath(Sdf.Path("/bicycle.tire:size"))

# Standard
start_prim = stage.GetPrimAtPath("/") # Or stage.GetPseudoRoot(), this is the same as stage.Traverse()
iterator = iter(Usd.PrimRange(start_prim))
for prim in iterator:
    if prim.IsA(UsdGeom.Imageable): # Some condition as listed above or custom property/metadata checks
        # Don't traverse into the child prims
        iterator.PruneChildren()
# Pre and post visit:
start_prim = stage.GetPrimAtPath("/") # Or stage.GetPseudoRoot(), this is the same as stage.Traverse()
iterator = iter(Usd.PrimRange.PreAndPostVisit(start_prim))
for prim in iterator:
    if not iterator.IsPostVisit():
        if prim.IsA(UsdGeom.Imageable): # Some condition as listed above or custom property/metadata checks
            # Don't traverse into the child prims
            iterator.PruneChildren()
# Custom Predicate
predicate = Usd.PrimIsActive & Usd.PrimIsLoaded # All prims, even class and over prims.
start_prim = stage.GetPrimAtPath("/") # Or stage.GetPseudoRoot(), this is the same as stage.Traverse()
iterator = iter(Usd.PrimRange.PrimRange(start_prim, predicate=predicate))
for prim in iterator:
    if not iterator.IsPostVisit():
        if prim.IsA(UsdGeom.Imageable): # Some condition as listed above or custom property/metadata checks
            # Don't traverse into the child prims
            iterator.PruneChildren()

prim_paths = []
variant_prim_paths = []
property_paths = [] # The Sdf.Path class doesn't distinguish between attributes and relationships
property_relationship_target_paths = []
def traversal_kernel(path):
    print(path)
    if path.IsPrimPath():
        prim_paths.append(path)
    elif path.IsPrimVariantSelectionPath():
        variant_prim_paths.append(path)
    elif path.IsPropertyPath():
        property_paths.append(path)
    elif path.IsTargetPath():
        property_relationship_target_paths.append(path)
layer.Traverse(layer.pseudoRoot.path, traversal_kernel)

We'll be using loading mechanisms to optimize loading only what is relevant for the current task at hand.

from pxr import Sdf
layer = Sdf.Layer.FindOrOpen("/my/layer/identifier")
Sdf.Layer.AddToMutedLayers(layer.identifier)
Sdf.Layer.RemoveFromMutedLayers(layer.identifier)

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
layer_A = Sdf.Layer.CreateAnonymous("Layer_A")
layer_B = Sdf.Layer.CreateAnonymous("Layer_B")
layer_C = Sdf.Layer.CreateAnonymous("Layer_C")
stage.GetRootLayer().subLayerPaths.append(layer_A.identifier)
stage.GetRootLayer().subLayerPaths.append(layer_B.identifier)
stage.GetRootLayer().subLayerPaths.append(layer_C.identifier)
# Mute layer
stage.MuteLayer(layer_A.identifier)
# Unmute layer
stage.UnmuteLayer(layer_A.identifier)
# Or both MuteAndUnmuteLayers([<layers to mute>], [<layers to unmute>])
stage.MuteAndUnmuteLayers([layer_A.identifier, layer_B.identifier], [layer_C.identifier])
# Check what layers are muted
print(stage.GetMutedLayers()) # Returns: [layerA.identifier, layerB.identifier]
print(stage.IsLayerMuted(layer_C.identifier)) # Returns: False

We use layer muting in production for two things:

• Artists can opt-in to load layers that are relevant to them. For example in a shot, a animator doesn't have to load the background set or fx layers.
• Pipeline-wise we have to ensure that artists add shot layers in a specific order (For example: lighting > fx > animation > layout >). Let's say a layout artist is working in a shot, we only want to display the layout and camera layers. All the other layers can (should) be muted, because A. performance, B. there might be edits in higher layers, that the layout artist is not interested in seeing yet. If we were to display them, some of these edits might block ours, because they are higher in the layer stack.

Similar to prim activation, the prim population mask controls what prims (and their child prims) are even considered for being loaded into the stage. Unlike activation, the prim population mask does not get stored in a USD layer. It is therefore a pre-filtering mechanism, rather than an artist facing "what do I want to hide from my scene" mechanism.

## Stage
# Create: 'OpenMasked',
# Get: 'GetPopulationMask',
# Set: 'SetPopulationMask', 'ExpandPopulationMask'
## Population Mask
# Usd.StagePopulationMask()
# Has: 'IsEmpty',  'Includes', 'IncludesSubtree'
# Get: 'GetIncludedChildNames', 'GetIntersection', 'GetPaths', 'GetUnion'
# Set:  'Add', 'Intersection',  'Union'
# Constants: 'All'
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Create hierarchy
prim_paths = [
    "/set/yard/biycle",
    "/set/yard/shed/shovel",
    "/set/yard/shed/flower_pot",
    "/set/yard/shed/lawnmower",
    "/set/yard/shed/soil",
    "/set/yard/shed/wood",
    "/set/garage/car",
    "/set/garage/tractor",
    "/set/garage/helicopter",
    "/set/garage/boat",
    "/set/garage/key_box",
    "/set/garage/key_box/red",
    "/set/garage/key_box/blue",
    "/set/garage/key_box/green",
    "/set/people/mike",
    "/set/people/charolotte"
]
for prim_path in prim_paths:
    prim = stage.DefinePrim(prim_path, "Cube")
    
population_mask = Usd.StagePopulationMask()
print(population_mask.GetPaths())
print(population_mask.All()) # Same as: Usd.StagePopulationMask([Sdf.Path("/")])
# Or stage.GetPopulationMask()
population_mask.Add(Sdf.Path("/set/yard/shed/lawnmower"))
population_mask.Add(Sdf.Path("/set/garage/key_box"))
stage.SetPopulationMask(population_mask)
print("<< hierarchy >>")
for prim in stage.Traverse():
    print(prim.GetPath())
"""Returns:
/set
/set/yard
/set/yard/shed
/set/yard/shed/lawnmower
/set/garage
/set/garage/key_box
/set/garage/key_box/red
/set/garage/key_box/blue
/set/garage/key_box/green
"""
# Intersections tests
print(population_mask.Includes("/set/yard/shed")) # Returns: True
print(population_mask.IncludesSubtree("/set/yard/shed")) # Returns: False (As not all child prims are included)
print(population_mask.IncludesSubtree("/set/garage/key_box")) # Returns: True (As all child prims are included)

stage.ExpandPopulationMask(relationshipPredicate=lambda r: r.GetName() == 'material:binding',
                           attributePredicate=lambda a: False)

Payloads are USD's mechanism of disabling the load of heavy data and instead leaving us with a bounding box representation (or texture card representation, if you set it up). We can configure our stages to not load payloads at all or to only load payloads at specific prims.

## Stage
# Has: 'FindLoadable', 
# Get: 'GetLoadRules', 'GetLoadSet'
# Set: 'Load', 'Unload', 'LoadAndUnload', 'LoadAll', 'LoadNone', 'SetLoadRules'
# Constants: 'InitialLoadSet.LoadAll', 'InitialLoadSet.LoadNone' 
## Stage Load Rules
# Has: 'IsLoaded', 'IsLoadedWithAllDescendants', 'IsLoadedWithNoDescendants'
# Get: 'GetRules', 'GetEffectiveRuleForPath',
# Set: 'LoadWithDescendants', 'LoadWithoutDescendants'
#      'LoadAll',  'LoadNone', 'Unload', 'LoadAndUnload', 'AddRule',  'SetRules'
# Clear: 'Minimize'
# Constants: StageLoadRules.AllRule, StageLoadRules.OnlyRule, StageLoadRules.NoneRule
from pxr import Sdf, Usd
# Spawn example data, this would be a file on disk
layer = Sdf.Layer.CreateAnonymous()
prim_path = Sdf.Path("/bicycle")
prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
prim_spec.specifier = Sdf.SpecifierDef
prim_spec.typeName = "Cube"
attr_spec = Sdf.AttributeSpec(prim_spec, "size", Sdf.ValueTypeNames.Double)
for frame in range(1001, 1010):
    value = float(frame - 1001)
    layer.SetTimeSample(attr_spec.path, frame, value)
# Payload data
stage = Usd.Stage.CreateInMemory()
ref = Sdf.Payload(layer.identifier, "/bicycle")
prim_path = Sdf.Path("/set/yard/bicycle")
prim = stage.DefinePrim(prim_path)
ref_api = prim.GetPayloads()
ref_api.AddPayload(ref)
# Check for what can be payloaded
print(stage.FindLoadable()) # Returns: [Sdf.Path('/set/yard/bicycle')]
# Check what prim paths are payloaded
print(stage.GetLoadSet()) # Returns: [Sdf.Path('/set/yard/bicycle')]
# Unload payload
stage.Unload(prim_path)
print(stage.GetLoadSet()) # Returns: []
# Please consult the official docs for how the rule system works.
# Basically we can flag primpaths to recursively load their nested child payloads or to only load the top most payload.

The draw mode can be used to tell our Hydra render delegates to not render a prim and its child hierarchy. Instead it will only display a preview representation.

The preview representation can be one of:

• Full Geometry
• Origin Axes
• Bounding Box
• Texture Cards

Like visibility, the draw mode is inherited downwards to its child prims. We can also set a draw mode color, to better differentiate the non full geometry draw modes, this is not inherited though and must be set per prim.

from pxr import Gf, Sdf, Usd, UsdGeom, UsdShade
stage = Usd.Stage.CreateInMemory()
cone_prim = stage.DefinePrim(Sdf.Path("/set/yard/cone"), "Cone")
cone_prim.GetAttribute("radius").Set(4)
Usd.ModelAPI(cone_prim).SetKind("component")
sphere_prim = stage.DefinePrim(Sdf.Path("/set/yard/sphere"), "Sphere")
Usd.ModelAPI(sphere_prim).SetKind("component")
for ancestor_prim_path in sphere_prim.GetParent().GetPath().GetAncestorsRange():
    ancestor_prim = stage.GetPrimAtPath(ancestor_prim_path)
    ancestor_prim.SetTypeName("Xform")
    Usd.ModelAPI(ancestor_prim).SetKind("group")
# Enable on parent
set_prim = stage.GetPrimAtPath("/set")
set_geom_model_API = UsdGeom.ModelAPI.Apply(set_prim)
set_geom_model_API.GetModelDrawModeAttr().Set(UsdGeom.Tokens.bounds)
set_geom_model_API.GetModelDrawModeColorAttr().Set(Gf.Vec3h([1,0,0]))
# If we enable "apply" on the parent, children will not be drawn anymore,
# instead just a single combined bbox is drawn for all child prims.
# set_geom_model_API.GetModelApplyDrawModeAttr().Set(1)
# Enable on child
sphere_geom_model_API = UsdGeom.ModelAPI.Apply(sphere_prim)
# sphere_geom_model_API.GetModelDrawModeAttr().Set(UsdGeom.Tokens.default_)
sphere_geom_model_API.GetModelDrawModeAttr().Set(UsdGeom.Tokens.cards)
sphere_geom_model_API.GetModelDrawModeColorAttr().Set(Gf.Vec3h([0,1,0]))
# For "component" (sub-)kinds, this is True by default
# sphere_geom_model_API.GetModelApplyDrawModeAttr().Set(0)

When traversing, using the above "filters" to narrow down your selection well help keep your traversals fast, even with hierarchies with millions of prims. We recommend first filtering based on metadata, as this is a lot faster than trying to access attributes and their values.

Another important feature is stopping traversal into child hierarchies. This can be done by calling ìterator.PruneChildren():

from pxr import Sdf, UsdShade
root_prim = stage.GetPseudoRoot()
# We have to cast it as an iterator to gain access to the .PruneChildren() method.
iterator = iter(Usd.PrimRange(root_prim))
for prim in iterator:
    if prim.IsA(UsdShade.Material):
        # Don't traverse into the shader network prims
        iterator.PruneChildren()

Here is the most common syntax you'll be using:

# Standard
start_prim = stage.GetPrimAtPath("/") # Or stage.GetPseudoRoot(), this is the same as stage.Traverse()
iterator = iter(Usd.PrimRange(start_prim))
for prim in iterator:
    if prim.IsA(UsdGeom.Imageable): # Some condition as listed above or custom property/metadata checks
        # Don't traverse into the child prims
        iterator.PruneChildren()
# Pre and post visit:
start_prim = stage.GetPrimAtPath("/") # Or stage.GetPseudoRoot(), this is the same as stage.Traverse()
iterator = iter(Usd.PrimRange.PreAndPostVisit(start_prim))
for prim in iterator:
    if not iterator.IsPostVisit():
        if prim.IsA(UsdGeom.Imageable): # Some condition as listed above or custom property/metadata checks
            # Don't traverse into the child prims
            iterator.PruneChildren()
# Custom Predicate
predicate = Usd.PrimIsActive & Usd.PrimIsLoaded # All prims, even class and over prims.
start_prim = stage.GetPrimAtPath("/") # Or stage.GetPseudoRoot(), this is the same as stage.Traverse()
iterator = iter(Usd.PrimRange.PrimRange(start_prim, predicate=predicate))
for prim in iterator:
    if not iterator.IsPostVisit():
        if prim.IsA(UsdGeom.Imageable): # Some condition as listed above or custom property/metadata checks
            # Don't traverse into the child prims
            iterator.PruneChildren()

# Check if the active prim is marked as instanceable:
# The prim.IsInstance() checks if it is actually instanced, this
# just checks if the 'instanceable' metadata is set.
prim.IsInstanceable()
# Check if the active prim is an instanced prim:
prim.IsInstance()
# Check if we are inside an instanceable prim:
prim.IsInstanceProxy()
# Check if the active prim is a prototype root prim with the following format /__Prototype_<idx>
prim.IsPrototype()
# For these type of prototype root prims, we can get the instances via:
prim.GetInstances()
# From each instance we can get back to the prototype via
prim.GetPrototype()
# Check if we are in the /__Prototype_<idx> prim:
prim.IsInPrototype()

# When we are within an instance, we can get the prototype via:
if prim.IsInstanceProxy():
    for ancestor_prim_path in prim.GetAncestorsRange():
        ancestor_prim = stage.GetPrimAtPath(ancestor_prim_path)
        if ancestor_prim.IsInstance():
            prototype = ancestor_prim.GetPrototype()
            print(list(prototype.GetInstances()))
            break

The traversal for layers works differently. Instead of an iterator, we have to provide a "kernel" like function, that gets an Sdf.Path as an input. Here is the most common syntax you'll be using:

prim_paths = []
variant_prim_paths = []
property_paths = [] # The Sdf.Path class doesn't distinguish between attributes and relationships
property_relationship_target_paths = []
def traversal_kernel(path):
    print(path)
    if path.IsPrimPath():
        prim_paths.append(path)
    elif path.IsPrimVariantSelectionPath():
        variant_prim_paths.append(path)
    elif path.IsPropertyPath():
        property_paths.append(path)
    elif path.IsTargetPath():
        property_relationship_target_paths.append(path)
layer.Traverse(layer.pseudoRoot.path, traversal_kernel)

import random
from pxr import Sdf, Usd, UsdGeom,UsdShade, Tf
stage = Usd.Stage.CreateInMemory()
layer = stage.GetEditTarget().GetLayer()

leaf_prim_types = ("Cube", "Cylinder", "Sphere", "Mesh", "Points", "RectLight", "Camera")
leaf_prim_types_count = len(leaf_prim_types)

def create_hierarchy(layer, root_prim_path, max_levels):
    def generate_hierarchy(layer, root_prim_path, leaf_prim_counter, max_levels):
        levels = random.randint(1, max_levels)
        for level in range(levels):
            level_depth = root_prim_path.pathElementCount + 1
            prim_path = root_prim_path.AppendChild(f"level_{level_depth}_child_{level}")
            prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
            prim_spec.specifier = Sdf.SpecifierDef
            # Type
            prim_spec.typeName = "Xform"
            # Kind
            prim_spec.SetInfo("kind", "group")
            # Seed parent prim specs
            hiearchy_seed_state = random.getstate()
            # Active
            random.seed(level_depth)
            if random.random() < 0.1:
                prim_spec.SetInfo("active", False)
            random.setstate(hiearchy_seed_state)
            if levels == 1:
                # Parent prim
                # Kind
                prim_spec.nameParent.SetInfo("kind", "component")
                # Purpose
                purpose_attr_spec = Sdf.AttributeSpec(prim_spec.nameParent, "purpose", Sdf.ValueTypeNames.Token)
                if random.random() < .9:
                    purpose_attr_spec.default = UsdGeom.Tokens.render
                else:
                    purpose_attr_spec.default = UsdGeom.Tokens.proxy
                # Seed leaf prim specs
                leaf_prim_counter[0] += 1
                hiearchy_seed_state = random.getstate()
                random.seed(leaf_prim_counter[0])
                # Custom Leaf Prim attribute
                prim_spec.typeName = leaf_prim_types[random.randint(0, leaf_prim_types_count -1)]
                prim_spec.assetInfo["is_leaf"] = True
                prim_spec.ClearInfo("kind")
                is_leaf_attr_spec = Sdf.AttributeSpec(prim_spec, "is_leaf", Sdf.ValueTypeNames.Bool)
                is_leaf_attr_spec.default = True
                # Active
                if random.random() < 0.1:
                    prim_spec.SetInfo("active", False)
                # Visibility
                visibility_attr_spec = Sdf.AttributeSpec(prim_spec, "visibility", Sdf.ValueTypeNames.Token)
                if random.random() < .5:
                    visibility_attr_spec.default = UsdGeom.Tokens.inherited
                else:
                    visibility_attr_spec.default = UsdGeom.Tokens.invisible
                    
                random.setstate(hiearchy_seed_state)
            else:
                generate_hierarchy(layer, prim_path, leaf_prim_counter, max_levels -1)
    random.seed(0)
    leaf_prim_counter = [0] # Make sure this is a pointer
    generate_hierarchy(layer, root_prim_path, leaf_prim_counter, max_levels)
                
with Sdf.ChangeBlock():
    prim_path = Sdf.Path("/profiling_grp")
    prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
    prim_spec.specifier = Sdf.SpecifierDef
    prim_spec.typeName = "Xform"
    prim_spec.SetInfo("kind", "group")
    create_hierarchy(layer, prim_path, 9)

# We assume we are running on the stage from the previous example.
root_prim = stage.GetPrimAtPath("/profiling_grp")
leaf_prim_types = ("Cube", "Cylinder", "Sphere", "Mesh", "Points", "RectLight", "Camera")

def profile(func, label, root_prim):
    # The changeblock doesn't do much here as we are only querying data, but
    # let's keep it in there anyway.
    with Sdf.ChangeBlock():
        runs = 3
        sw = Tf.Stopwatch()
        time_delta = 0.0
        for run in range(runs):
            sw.Reset()
            sw.Start()
            matched_prims = []
            for prim in iter(Usd.PrimRange(root_prim)):
                if func(prim):
                    matched_prims.append(prim)
            sw.Stop()
            time_delta += sw.seconds
        print("{:.5f} Seconds | {} | Match {}".format(time_delta / runs, label, len(matched_prims)))

print("----")

def profile_boundable(prim):
    return prim.IsA(UsdGeom.Boundable)

profile(profile_boundable, "IsA(Boundable)", root_prim)
    
def profile_GetTypeName(prim):
    return prim.GetTypeName() in leaf_prim_types

profile(profile_GetTypeName, "GetTypeName", root_prim)

def profile_kind(prim):
    model_api = Usd.ModelAPI(prim)
    return model_api.GetKind() != Kind.Tokens.group
    
profile(profile_kind, "Kind", root_prim)
 
def profile_assetInfo_is_leaf(prim):
    asset_info = prim.GetAssetInfo()
    return asset_info.get("is_leaf", False)

profile(profile_assetInfo_is_leaf, "IsLeaf AssetInfo ", root_prim)

def profile_attribute_has_is_leaf(prim):
    if prim.HasAttribute("is_leaf"):
        return True
    return False

profile(profile_attribute_has_is_leaf, "IsLeaf Attribute Has", root_prim)

def profile_attribute_is_leaf(prim):
    is_leaf_attr = prim.GetAttribute("is_leaf")
    if is_leaf_attr:
        if is_leaf_attr.Get():
            return True
    return False

profile(profile_attribute_is_leaf, "IsLeaf Attribute ", root_prim)

def profile_attribute_extra_validation_is_leaf(prim):
    if prim.HasAttribute("is_leaf"):
        is_leaf_attr = prim.GetAttribute("is_leaf")
        if is_leaf_attr.Get():
            return True
    return False

profile(profile_attribute_extra_validation_is_leaf, "IsLeaf Attribute (Validation)", root_prim)

0.17678 Seconds | IsA(Boundable) | Match 38166
0.17222 Seconds | GetTypeName | Match 44294
0.42160 Seconds | Kind | Match 93298
0.38575 Seconds | IsLeaf AssetInfo  | Match 44294
0.27142 Seconds | IsLeaf Attribute Has | Match 44294
0.38036 Seconds | IsLeaf Attribute  | Match 44294
0.37459 Seconds | IsLeaf Attribute (Validation) | Match 44294

• Terminology: A single time/value pair is called time sample, if an attribute doesn't have time samples, it has default value (Which just means it has a single static value).
• Time samples are encoded in a simple {<time(frame)>: <value>} dict.
• If a frame is requested where no time samples exist, it will be interpolated if the data type allows it and non changing array lengths in neighbour time samples exist. Value queries before/after the first/last time sample will be clamped to these time samples.
• Time samples are encoded unitless/per frame and not in time units. This means they have to be shifted depending on the current frames per second.
• Only attributes can carry time sample data. (So composition arcs can not be time animated, only offset/scaled (see the LayerOffset section on this page)).

Reading and writing is quite straight forward:

from pxr import Sdf, Usd
### High Level ###
stage = Usd.Stage.CreateInMemory()
prim_path = Sdf.Path("/bicycle")
prim = stage.DefinePrim(prim_path, "Cube")
size_attr = prim.GetAttribute("size")
for frame in range(1001, 1005):
    time_code = Usd.TimeCode(float(frame - 1001))
    # .Set() takes args in the .Set(<value>, <frame>) format
    size_attr.Set(frame, time_code)
print(size_attr.Get(1005)) # Returns: 4

### Low Level ###
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
prim_path = Sdf.Path("/bicycle")
prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
prim_spec.specifier = Sdf.SpecifierDef
prim_spec.typeName = "Cube"
attr_spec = Sdf.AttributeSpec(prim_spec, "size", Sdf.ValueTypeNames.Double)
for frame in range(1001, 1005):
    value = float(frame - 1001)
    # .SetTimeSample() takes args in the .SetTimeSample(<path>, <frame>, <value>) format
    layer.SetTimeSample(attr_spec.path, frame, value)
print(layer.QueryTimeSample(attr_spec.path, 1005)) # Returns: 4

Anything that has time varying data will be written as time samples. Usually DCCs handle the time sample creation for you, but there are situations where we need to write them ourselves. For example if you want to efficiently combine time samples from two different value sources or if we want to turn a time sample into a default sampled value (a value without animation).

A single frame(time)/value pair is called time sample, if an attribute doesn't have time samples, it has default value (Which just means it has a single static value). The time value is the active frame where data is being exported on. It is not time/FPS based. This means that depending on frames per second set on the stage, different time samples are read. This means if you have a cache that was written in 25 FPS, you'll have to shift/scale the time samples if you want to use the cache in 24 FPS and you want to have the same result. More about this in the examples below.

When writing a large amount of samples, you should use the low level API as it is a lot faster.

If you need to check if an attribute is time sampled, run the following:

# !Danger! For value clipped (per frame loaded layers),
# this will look into all layers, which is quite expensive.
print(size_attr.GetNumTimeSamples())
# You should rather use:
# This does a check for time sample found > 2.
# So it stops looking for more samples after the second sample.
print(size_attr.ValueMightBeTimeVarying())

If you know the whole layer is in memory, then running GetNumTimeSamples() is fine, as it doesn't have t open any files.

Stitching multiple files to a single file is usually used for small per frame USD files. If you have large (1 GB > ) files per frame, then we recommend using values clips. During stitching all data has to be loaded into memory, so your RAM specs have to be high enough to handle all the files combined.

Composition wise, value clips (or the layer that specifies the value clip metadata) is right under the local arc strength and over inherit arcs.

Technically you can remove all default attributes from the per frame files after running the topology layer generation. This can save a lot of disk space,but you can't partially re-render specific frames of the cache though. So only do this if you know the cache is "done".

Since we attach the value clips to asset root prims, our value clipped caches can't have values above asset root prims.

For more info on how value clips affect instancing, check out our composition section. There you will also find an example with multiple asset roots re-using the same value clipped cache.

With value clips it can be very expensive to call attribute.GetTimesamples() as this will open all layers to get the samples in the interval that is specified in the metadata. It does not only read the value clip metadata. If possible use attribute.GetTimeSamplesInInterval() as this only opens the layers in the interested interval range.

Materials themselves are usually generated by the DCC you are working in, so we usually don't have to create them ourselves. What we do use the UsdShade module for is editing material bindings and overriding individual nodes and their connections in a material node graph.

Material bindings are a special kind of relationship. Here are a few important things to know:

• When looking up material bindings, USD also looks at parent prims if it can't find a written binding on the prim directly. This means you can create the binding on any parent prim and just as with primvars, it will be inherited downwards to its children.
• The "binding strength" can be adjusted, so that a child prim assignment can also be override from a binding higher up the hierarchy.
• Material bindings can also be written per purpose, if not then they bind to all purposes. (Technically it is not called purpose, the token names are UsdShade.MaterialBindingAPI.GetMaterialPurposes() -> ['', 'preview', 'full']). The 'preview' is usually bound to the 'UsdGeom.Tokens.proxy' purpose, the 'full' to the 'UsdGeom.Tokens.render' purpose.
• The material binding can be written in two ways:
  • Direct Binding: A relationship that points directly to a material prim
  • Collection Based Binding: A relationship that points to another collection, that then stores the actual binding paths) and to a material prim to bind.

## UsdShade.MaterialBindingAPI(<boundable prim>)
# This handles all the binding get and setting
## These classes can inspect an existing binding
# UsdShade.MaterialBindingAPI.DirectBinding()
# UsdShade.MaterialBindingAPI.CollectionBinding()
### High Level ###
from pxr import Sdf, Usd, UsdGeom, UsdShade
stage = Usd.Stage.CreateInMemory()
render_prim = stage.DefinePrim(Sdf.Path("/bicycle/RENDER/render"), "Cube")
material_prim = stage.DefinePrim(Sdf.Path("/bicycle/MATERIALS/example_material"), "Material")
bicycle_prim = render_prim.GetParent().GetParent()
bicycle_prim.SetTypeName("Xform")
render_prim.GetParent().SetTypeName("Xform")
material_prim.GetParent().SetTypeName("Xform")
# Bind materials via direct binding
material = UsdShade.Material(material_prim)
mat_bind_api = UsdShade.MaterialBindingAPI.Apply(render_prim)
mat_bind_api.Bind(material)
# Unbind all
mat_bind_api.UnbindAllBindings()
# Bind via collection
collection_name = "material_example"
collection_api = Usd.CollectionAPI.Apply(bicycle_prim, collection_name)
collection_api.GetIncludesRel().AddTarget(material_prim.GetPath())
collection_api.GetExpansionRuleAttr().Set(Usd.Tokens.expandPrims)
mat_bind_api.Bind(collection_api, material, "material_example")

### Low Level ###
from pxr import Sdf, UsdGeom
layer = Sdf.Layer.CreateAnonymous()
render_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/render"))
render_prim_spec.specifier = Sdf.SpecifierDef
render_prim_spec.typeName = "Cube"
material_prim_spec = Sdf.CreatePrimInLayer(layer, Sdf.Path("/material"))
material_prim_spec.specifier = Sdf.SpecifierDef
material_prim_spec.typeName = "Material"
## Direct binding
material_bind_rel_spec = Sdf.RelationshipSpec(render_prim_spec, "material:binding")
material_bind_rel_spec.targetPathList.Append(Sdf.Path("/render"))
# Applied Schemas
schemas = Sdf.TokenListOp.Create(
    prependedItems=["MaterialBindingAPI"]
)
render_prim_spec.SetInfo("apiSchemas", schemas)

def Scope "materials"
{
    def Material "karmamtlxsubnet" (
    )
    {
        token outputs:mtlx:surface.connect = </materials/karmamtlxsubnet/mtlxsurface.outputs:out>

        def Shader "mtlxsurface" ()
        {
            uniform token info:id = "ND_surface"
            string inputs:edf.connect = </materials/karmamtlxsubnet/mtlxuniform_edf.outputs:out>
            token outputs:out
        }

        def Shader "mtlxuniform_edf"
        {
            uniform token info:id = "ND_uniform_edf"
            color3f inputs:color.connect = </materials/karmamtlxsubnet/mtlx_constant.outputs:out>
            token outputs:out
        }

        def Shader "mtlx_constant"
        {
            uniform token info:id = "ND_constant_float"
            float outputs:out
        }
    }
}

We rarely write the initial transforms ourselves, this is something our DCCs excel at. We do query transforms though for different scenarios:

• We can bake down the transform to a single prim. This can then be referenced or inherited and used as a parent constraint like mechanism.
• When merging hierarchies, we often want to preserve the world transform of leaf prims. Let's say we have two stages: We can simply get the parent xform of stage A and then apply it in inverse to our leaf prim in stage B. That way the leaf prim in stage B is now in local space and merging the stages returns the expected result. We show an example of this below.

We use collections for multiple things:

• Creating a group of target paths, that are of interest to other departments, e.g. mark prims that are useful for FX/layout/lighting selections (for example character vs. background). Another common thing to use them for is storing render layer selections, that then get applied in our final render USD file.
• Faster navigation of hierarchies by isolating to collections that interest us.
• As collections can contain other collections, they are a powerful mechanism of forwarding and aggregating selections.

# Usd.CollectionAPI.Apply(prim, collection_name)
# collection_api = Usd.CollectionAPI(prim, collection_nam)
# collection_query = collection_api.ComputeMembershipQuery()
### High Level ###
from pxr import Sdf, Usd, UsdUtils
stage = Usd.Stage.CreateInMemory()
bicycle_prim = stage.DefinePrim(Sdf.Path("/set/yard/biycle"), "Cube")
car_prim = stage.DefinePrim(Sdf.Path("/set/garage/car"), "Sphere")
tractor_prim = stage.DefinePrim(Sdf.Path("/set/garage/tractor"), "Cylinder")
helicopter_prim = stage.DefinePrim(Sdf.Path("/set/garage/helicopter"), "Cube")
boat_prim = stage.DefinePrim(Sdf.Path("/set/garage/boat"), "Cube")
set_prim = bicycle_prim.GetParent().GetParent()
set_prim.SetTypeName("Xform")
bicycle_prim.GetParent().SetTypeName("Xform")
car_prim.GetParent().SetTypeName("Xform")
# Create collection
collection_name = "vehicles"
collection_api = Usd.CollectionAPI.Apply(set_prim, collection_name)
collection_api.GetIncludesRel().AddTarget(set_prim.GetPath())
collection_api.GetExcludesRel().AddTarget(bicycle_prim.GetPath())
collection_api.GetExpansionRuleAttr().Set(Usd.Tokens.expandPrims)
print(Usd.CollectionAPI.GetAllCollections(set_prim)) # Returns: [Usd.CollectionAPI(Usd.Prim(</set>), 'vehicles')]
print(Usd.CollectionAPI.GetCollection(set_prim, "vehicles")) # Returns: Usd.CollectionAPI(Usd.Prim(</set>), 'vehicles')
collection_query = collection_api.ComputeMembershipQuery()
print(collection_api.ComputeIncludedPaths(collection_query, stage))
# Returns: [Sdf.Path('/set'), Sdf.Path('/set/garage'), Sdf.Path('/set/garage/car'), Sdf.Path('/set/yard')]
# Set it to explicit only
collection_api.GetExpansionRuleAttr().Set(Usd.Tokens.explicitOnly)
collection_query = collection_api.ComputeMembershipQuery()
print(collection_api.ComputeIncludedPaths(collection_query, stage))
# Returns: [Sdf.Path('/set')]

# To help speed up collection creation, USD also ships with util functions:
# UsdUtils.AuthorCollection(<collectionName>, prim, [<includePathList>], [<excludePathList>])
collection_api = UsdUtils.AuthorCollection("two_wheels", set_prim, [set_prim.GetPath()], [car_prim.GetPath()])
collection_query = collection_api.ComputeMembershipQuery()
print(collection_api.ComputeIncludedPaths(collection_query, stage))
# Returns:
# [Sdf.Path('/set'), Sdf.Path('/set/garage'), Sdf.Path('/set/yard'), Sdf.Path('/set/yard/biycle')]
# UsdUtils.ComputeCollectionIncludesAndExcludes() gives us the possibility to author 
# collections more sparse, that the include to exclude ratio is kept at an optimal size.
# The Python signature differs from the C++ signature:
"""
include_paths, exclude_paths = UsdUtils.ComputeCollectionIncludesAndExcludes(
    target_paths,
    stage,
    minInclusionRatio = 0.75,
    maxNumExcludesBelowInclude = 5,
	minIncludeExcludeCollectionSize = 3,
    pathsToIgnore = [] # This ignores paths from computation (this is not the exclude list)
)		
"""
target_paths = [tractor_prim.GetPath(), car_prim.GetPath(), helicopter_prim.GetPrimPath()]
include_paths, exclude_paths = UsdUtils.ComputeCollectionIncludesAndExcludes(target_paths,stage, minInclusionRatio=.9)
print(include_paths, exclude_paths)
# Returns:
# [Sdf.Path('/set/garage/car'), Sdf.Path('/set/garage/tractor'), Sdf.Path('/set/garage/helicopter')] []
include_paths, exclude_paths = UsdUtils.ComputeCollectionIncludesAndExcludes(target_paths,stage, minInclusionRatio=.1)
print(include_paths, exclude_paths)
# Returns: [Sdf.Path('/set/garage')] [Sdf.Path('/set/garage/boat')]
# Create a collection from the result
collection_api = UsdUtils.AuthorCollection("optimized", set_prim, include_paths, exclude_paths)

from pxr import Sdf, Usd, UsdUtils
stage = Usd.Stage.CreateInMemory()
# Create hierarchy
prim_paths = [
    "/set/yard/biycle",
    "/set/yard/shed/shovel",
    "/set/yard/shed/flower_pot",
    "/set/yard/shed/lawnmower",
    "/set/yard/shed/soil",
    "/set/yard/shed/wood",
    "/set/garage/car",
    "/set/garage/tractor",
    "/set/garage/helicopter",
    "/set/garage/boat",
    "/set/garage/key_box",
    "/set/garage/key_box/red",
    "/set/garage/key_box/blue",
    "/set/garage/key_box/green",
    "/set/people/mike",
    "/set/people/charolotte"
]
for prim_path in prim_paths:
    prim = stage.DefinePrim(prim_path, "Cube")
print("<< hierarchy >>")
for prim in stage.Traverse():
    print(prim.GetPath())
    parent_prim = prim.GetParent()
    while True:
        if parent_prim.IsPseudoRoot():
            break
        parent_prim.SetTypeName("Xform")
        parent_prim = parent_prim.GetParent()
# Returns:
"""
<< hierarchy >>
/HoudiniLayerInfo
/set
/set/yard
/set/yard/biycle
/set/yard/shed
/set/yard/shed/shovel
/set/yard/shed/flower_pot
/set/yard/shed/lawnmower
/set/yard/shed/soil
/set/yard/shed/wood
/set/garage
/set/garage/car
/set/garage/tractor
/set/garage/helicopter
/set/garage/boat
/set/garage/key_box
/set/garage/key_box/red
/set/garage/key_box/blue
/set/garage/key_box/green
/set/people
/set/people/mike
/set/people/charolotte
"""
# Collections
collection_prim = stage.DefinePrim("/collections")
storage_include_prim_paths = ["/set/garage/key_box", "/set/yard/shed"]
storage_exclude_prim_paths = ["/set/yard/shed/flower_pot"]
collection_api = UsdUtils.AuthorCollection("storage", collection_prim, storage_include_prim_paths, storage_exclude_prim_paths)
collection_query = collection_api.ComputeMembershipQuery()
included_paths = collection_api.ComputeIncludedPaths(collection_query, stage)
# print(included_paths)
# Prune inverse:
print("<< hierarchy pruned >>")
iterator = iter(Usd.PrimRange(stage.GetPseudoRoot()))
for prim in iterator:
    if prim.IsPseudoRoot():
        continue
    if prim.GetPath() not in included_paths and not len(prim.GetAllChildrenNames()):
        iterator.PruneChildren()
        prim.SetActive(False)
    else:    
        print(prim.GetPath())
# Returns:
"""
<< hierarchy pruned >>
/set
/set/yard
/set/yard/shed
/set/yard/shed/shovel
/set/yard/shed/lawnmower
/set/yard/shed/soil
/set/yard/shed/wood
/set/garage
/set/garage/key_box
/set/garage/key_box/red
/set/garage/key_box/blue
/set/garage/key_box/green
/set/people
"""

In production, the mose common use case you'll use the notification system for is changes in the stage. You can use these notifications to track user interaction and to trigger UI refreshes.

The following are a must have for anyone using Usd in production:

• usdmanager: Your go-to Usd ascii editing app
• usdview: Your go-to standalone Usd 3d viewing/debugging app
  • TheGrill/Grill: A Usd View plugin to view composition arcs

This section houses terminology essentials and a detailed explanation of how the underlying mechanism of editing/composing arcs works. Some may consider it a deep dive topic, we'd recommend starting out with it first though, as it saves time later on when you don't understand why something might not work.

from pxr import Sdf
# Sdf.ReferenceListOp, Sdf.PayloadListOp, Sdf.PathListOp,
path_list_op = Sdf.PathListOp()
# There are multiple sub-lists, which are just normal Python lists.
# 'prependedItems', 'appendedItems', 'deletedItems', 'explicitItems',
# Legacy sub-lists (do not use these anymore): 'addedItems', 'orderedItems'
# Currently the way these are exposed to Python, you have to re-assign the list, instead of editing it in place.
# So this won't work:
path_list_op.prependedItems.append(Sdf.Path("/cube"))
path_list_op.appendedItems.append(Sdf.Path("/sphere"))
# Instead do this:
path_list_op.prependedItems = [Sdf.Path("/cube")]
path_list_op.appendedItems = [Sdf.Path("/sphere")]
# To clear the list op:
print(path_list_op) # Returns: SdfPathListOp(Prepended Items: [/cube], Appended Items: [/sphere])
path_list_op.Clear()
print(path_list_op) # Returns: SdfPathListOp()
# Repopulate via constructor
path_list_op = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])
print(path_list_op) # Returns: SdfPathListOp(Prepended Items: [/cube], Appended Items: [/sphere])
# Add remove items
path_list_op.deletedItems = [Sdf.Path("/sphere")]
print(path_list_op) # Returns: SdfPathListOp(Deleted Items: [/sphere], Prepended Items: [/cube], Appended Items: [/sphere])
# Notice how it just stores lists, it doesn't actually apply them. We'll have a look at that next.

# In the high level API, all the function signatures that work on list-editable ops
# usually take a position kwarg which corresponds to what list to edit and the position (front/back)
Usd.ListPositionFrontOfAppendList
Usd.ListPositionBackOfAppendList
Usd.ListPositionFrontOfPrependList
Usd.ListPositionBackOfPrependList
# We cover how to use this is our 'Composition Arcs' section.

from pxr import Sdf
### Merging basics ###
path_list_op_layer_top = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/disc"), Sdf.Path("/cone")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/cone"),Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
# Notice how on merge it makes sure that each sublist does not have the values of the other sublists, just like a Python set()
print(result) # Returns: SdfPathListOp(Deleted Items: [/cube], Prepended Items: [/disc, /cone], Appended Items: [/sphere])
# Get the flattened result. This does not apply the deleteItems, only ApplyOperations does that. 
print(result.GetAddedOrExplicitItems()) # Returns: [Sdf.Path('/disc'), Sdf.Path('/cone'), Sdf.Path('/sphere')]

### Deleted and added items ###
path_list_op_layer_top = Sdf.PathListOp.Create(appendedItems=[Sdf.Path("/disc"), Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
print(result) # Returns: SdfPathListOp(Appended Items: [/sphere, /disc, /cube])
# Since it now was in the explicit list, it got removed.

### Explicit mode ###
# There is also an "explicit" mode. This clears all previous values on merge and marks the list as explicit.
# Once explicit and can't be un-explicited. An explicit list is like a reset, it 
# doesn't know anything about the previous values anymore. All lists that are merged
# after combine the result to be explicit.
path_list_op_layer_top = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.CreateExplicit([Sdf.Path("/disc")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
print(result, result.isExplicit) # Returns: SdfPathListOp(Explicit Items: [/disc]), True
# Notice how the deletedItems had no effect, as "/cube" is not in the explicit list.

path_list_op_layer_top = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.CreateExplicit([Sdf.Path("/disc"), Sdf.Path("/cube")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
print(result, result.isExplicit) # Returns: SdfPathListOp(Explicit Items: [/disc]), True
# Since it now was in the explicit list, it got removed.

You might be wondering now, if encapsulation forces the content of Reference/Payload to be self contained, in the sense that the composition arcs within that file do not "look" outside the file. The answer is: It depends on the composition arc.

For Inherits and Specializes the arcs still evaluate relative to the composed scene. E.g. that means if you have an inherit somewhere in a referenced in layer stack, that inherit will still be live. So if you edit a property in the active stage, that gets inherited somewhere in the file, it will still propagate all the changes from the inherit source to all the inherit targets. The only thing that is "locked" is the composition arcs structure, not the way the composition arc evaluates. This extra "live" lookup has a performance penalty, so be careful with using Inherits and Specializes when nesting layers stacks via References and Payloads.

For Internal References this does not work though. They can only see the encapsulated layer stack and not the "live" composed stage. This makes composition faster for internal references.

We show some interactive examples in Houdini in our LIVRPS section, as this is hard to describe in words.

• Composition arcs target the layer stack, not individual layers. They recursively target the composed result (aka the result of all layers combined via the composition arc rules) of each layer they load in.
• We can only list edit composition arcs via list editable ops in the active layer stack. The active layer stack is usually the active stage (unless when we "hack" around it via edit targets, which you 99% of the time don't do).

A edit target defines, what layer all calls in the high level API should write to.

stage.SetEditTarget(layer)
# We can also explicitly create the edit target:
# Or
stage.SetEditTarget(Usd.EditTarget(layer))
# Or
stage.SetEditTarget(stage.GetEditTargetForLocalLayer(layer))
# These all have the same effect.

from pxr import Sdf, Usd
## Standard way of using edit targets
stage = Usd.Stage.CreateInMemory()
root_layer = stage.GetRootLayer()
a_layer = Sdf.Layer.CreateAnonymous("LayerA")
b_layer = Sdf.Layer.CreateAnonymous("LayerB")
root_layer.subLayerPaths.append(a_layer.identifier)
root_layer.subLayerPaths.append(b_layer.identifier)
# Direct edits to different layers
stage.SetEditTarget(Usd.EditTarget(a_layer))
bicycle_prim = stage.DefinePrim(Sdf.Path("/bicycle"), "Xform")
stage.SetEditTarget(Usd.EditTarget(b_layer))
car_prim = stage.DefinePrim(Sdf.Path("/car"), "Xform")
print(b_layer.ExportToString())
"""Returns:
#sdf 1.4.32
def Xform "car"
{
}
"""
## Reference/Payload Edit Targets
asset_stage = Usd.Stage.CreateInMemory()
cube_prim = asset_stage.DefinePrim(Sdf.Path("/root/RENDER/cube"), "Xform")
asset_layer = asset_stage.GetRootLayer()
shot_stage = Usd.Stage.CreateInMemory()
car_prim = shot_stage.DefinePrim(Sdf.Path("/set/garage/car"), "Xform")
car_prim.GetReferences().AddReference(asset_layer.identifier, Sdf.Path("/root"), Sdf.LayerOffset(10))
# We can't construct edit targets to layers that are not sublayers in the active layer stack.
# Is this a bug? According to the docs https://openusd.org/dev/api/class_usd_edit_target.html it should work.
# shot_stage.SetEditTarget(Usd.EditTarget(asset_layer))
## Variant Edit Targets
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path, "Xform")
variant_sets_api = bicycle_prim.GetVariantSets()
variant_set_api = variant_sets_api.AddVariantSet("color", position=Usd.ListPositionBackOfPrependList)
variant_set_api.AddVariant("colorA")
variant_set_api.SetVariantSelection("colorA")
with variant_set_api.GetVariantEditContext():
    # Anything we write in this edit target context, goes into the variant.
    cube_prim_path = bicycle_prim_path.AppendChild("cube")
    cube_prim = stage.DefinePrim(cube_prim_path, "Cube")
print(stage.GetEditTarget().GetLayer().ExportToString())
"""Returns:
#usda 1.0

def Xform "bicycle" (
    variants = {
        string color = "colorA"
    }
    prepend variantSets = "color"
)
{
    variantSet "color" = {
        "colorA" {
            def Cube "cube"
            {
            }

        }
    }
}
"""

from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
root_layer = stage.GetRootLayer()
a_layer = Sdf.Layer.CreateAnonymous("LayerA")
b_layer = Sdf.Layer.CreateAnonymous("LayerB")
root_layer.subLayerPaths.append(a_layer.identifier)
root_layer.subLayerPaths.append(b_layer.identifier)
# Set edit target to a_layer
stage.SetEditTarget(a_layer)
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path, "Xform")
edit_target = Usd.EditTarget(b_layer)
with Usd.EditContext(stage, edit_target):
    print("Edit Target Layer:", stage.GetEditTarget().GetLayer()) # Edit Target Layer: Sdf.Find('anon:0x7ff9f4391580:LayerB')
    car_prim_path = Sdf.Path("/car")
    car_prim = stage.DefinePrim(car_prim_path, "Xform")
print("Edit Target Layer:", stage.GetEditTarget().GetLayer()) # Edit Target Layer: Sdf.Find('anon:0x7ff9f4391580:LayerA')
# Verify result
print(a_layer.ExportToString())
"""Returns:
#sdf 1.4.32
def Xform "bicycle"
{
}
"""
print(b_layer.ExportToString())
"""Returns:
#sdf 1.4.32
def Xform "car"
{
}
"""

We'll be using the below code, whenever we write composition arcs in USD.

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what the sublayer arc is used for.

# For sublayering we modify the .subLayerPaths attribute on a layer.
# This is the same for both the high and low level API.
### High Level & Low Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Layer onto root layer
layer_a = Sdf.Layer.CreateAnonymous()
layer_b = Sdf.Layer.CreateAnonymous()
root_layer = stage.GetRootLayer()
# Here we pass in the file paths (=layer identifiers).
root_layer.subLayerPaths.append(layer_a.identifier)
root_layer.subLayerPaths.append(layer_b.identifier)
# Once we have added the sublayers, we can also access their layer offsets:
print(root_layer.subLayerOffsets) # Returns: [Sdf.LayerOffset(), Sdf.LayerOffset()]
# Since layer offsets are ready only copies, we need to assign a newly created 
# layer offset if we want to modify them. We also can't replace the whole list, as
# it needs to keep a pointer to the array.
layer_offset_a = root_layer.subLayerOffsets[0]
root_layer.subLayerOffsets[0] = Sdf.LayerOffset(offset=layer_offset_a.offset + 10, 
                                                scale=layer_offset_a.scale * 2)
layer_offset_b = root_layer.subLayerOffsets[1]
root_layer.subLayerOffsets[1] = Sdf.LayerOffset(offset=layer_offset_b.offset - 10, 
                                                scale=layer_offset_b.scale * 0.5)
print(root_layer.subLayerOffsets) # Returns: [Sdf.LayerOffset(10, 2), Sdf.LayerOffset(-10, 0.5)]

# If we want to sublayer on the active layer, we just add it there.
layer_c = Sdf.Layer.CreateAnonymous()
active_layer = stage.GetEditTarget().GetLayer()
root_layer.subLayerPaths.append(layer_c.identifier)

### High Level & Low Level ###
import loputils
from pxr import Sdf
# Hou LOP Node https://www.sidefx.com/docs/houdini/hom/hou/LopNode.html
# See $HFS/houdini/python3.9libs/loputils.py
"""
def createPythonLayer(node, savepath=None):
    # Tag the layer as "LOP" so we know it was created by LOPs.
    layer = Sdf.Layer.CreateAnonymous('LOP')
    # Add a Houdini Layer Info prim where we can put the save path.
    p = Sdf.CreatePrimInLayer(layer, '/HoudiniLayerInfo')
    p.specifier = Sdf.SpecifierDef
    p.typeName = 'HoudiniLayerInfo'/stage/list_editable_ops/pythonscript6
    if savepath:
        p.customData['HoudiniSavePath'] = hou.text.expandString(savepath)
        p.customData['HoudiniSaveControl'] = 'Explicit'
    # Let everyone know what node created this layer.
    p.customData['HoudiniCreatorNode'] = node.sessionId()
    p.customData['HoudiniEditorNodes'] = Vt.IntArray([node.sessionId()])
    node.addHeldLayer(layer.identifier)
    return layer
"""
# Sublayer onto root layer via Python LOP node
node = hou.pwd()
stage = node.editableStage()
layer = loputils.createPythonLayer(node, '$HIP/myfile.usda')
node.addSubLayer(layer.identifier)
# This doesn't seem to work at the moment, as Houdini does some custom root layer handeling
# print(root_layer.subLayerPaths) # Our added layer does not show up. So we have to use the `sublayer` node.
# root_layer = stage.GetRootLayer()
# root_layer.subLayerOffsets[0] = Sdf.LayerOffset(offset=10, scale=1)

# Sublayer onto active layer via Python LOP node, here we can do the usual.
node = hou.pwd()
layer = node.editableLayer()
layer_a = loputils.createPythonLayer(node, '$HIP/myfile.usda')
layer.subLayerPaths.append(layer_a.identifier)
layer.subLayerOffsets[0] = Sdf.LayerOffset(offset=10, scale=1)

# Since layers are automatically garbage collected once they go out of scope,
# we can tag them to keep them persistently in memory for the active session.
layer_b = Sdf.Layer.CreateAnonymous()
node.addHeldLayer(layer_b.identifier)
# This can then be re-used via the standard anywhere in Houdini.
layer_b = Sdf.Layer.FindOrOpen(layer_b.identifier)

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what value clips are used for.

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what the inherit arc is used for.

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path)
cube_prim_path = Sdf.Path("/cube")
cube_prim = stage.DefinePrim(cube_prim_path, "Cube")
inherits_api = bicycle_prim.GetInherits()
inherits_api.AddInherit(cube_prim_path, position=Usd.ListPositionFrontOfAppendList)
# inherits_api.SetInherits() # Clears the list editable ops and authors an Sdf.PathListOp.CreateExplicit([])
# inherits_api.RemoveInherit(cube_prim_path)
# inherits_api.ClearInherits() # Sdf.PathListOp.Clear()
# inherits_api.GetAllDirectInherits() # Returns all inherits generated in the active layer stack
### Low Level ###
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_path = Sdf.Path("/cube")
cube_prim_spec = Sdf.CreatePrimInLayer(layer, cube_prim_path)
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
bicycle_prim_spec.inheritPathList.appendedItems = [cube_prim_path]

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what the variant arc is used for.

def Xform "car" (
    variants = {
        string color = "colorA"
    }
    prepend variantSets = "color"
)
{
    variantSet "color" = {
        "colorA" {
            def Cube "cube"
            {
            }
        }
        "colorB" {
            def Sphere "sphere"
            {
            }
        }
    }
}

def Xform "car" (
    variants = {
        string color = "colorA"
    }
)
{
    variantSet "color" = {
        "colorA" {
            def Cube "cube"
            {
            }
        }
        "colorB" {
            def Sphere "sphere"
            {
            }
        }
    }
}

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()

bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path, "Xform")
## Methods of Usd.VariantSets
# Has: 'HasVariantSet'
# Get: 'GetNames', 'GetVariantSet', 'GetVariantSelection', 'GetAllVariantSelections'
# Set: 'AddVariantSet', 'SetSelection'
variant_sets_api = bicycle_prim.GetVariantSets()
## Methods of Usd.VariantSet
# Has: 'HasAuthoredVariant', 'HasAuthoredVariantSelection'
# Get: 'GetName', 'GetVariantNames', 'GetVariantSelection', 'GetVariantEditContext', 'GetVariantEditTarget'
# Set: 'AddVariant', 'SetVariantSelection'
# Clear: 'BlockVariantSelection', 'ClearVariantSelection'
variant_set_api = variant_sets_api.AddVariantSet("color", position=Usd.ListPositionBackOfPrependList)
variant_set_api.AddVariant("colorA")
# If we want to author on the selected variant, we have to select it first
variant_set_api.SetVariantSelection("colorA")
with variant_set_api.GetVariantEditContext():
    # Anything we write in the context, goes into the variant (prims and properties)
    cube_prim_path = bicycle_prim_path.AppendChild("cube")
    cube_prim = stage.DefinePrim(cube_prim_path, "Cube")
# We can also generate the edit target ourselves, but we still need to set the
# variant selection, seems like a bug. Changing variants is a heavy op ...
variant_set_api.AddVariant("colorB")
variant_set_api.SetVariantSelection("colorB")
variant_prim_path = bicycle_prim_path.AppendVariantSelection("color", "colorB") 
layer = stage.GetEditTarget().GetLayer()
edit_target = Usd.EditTarget.ForLocalDirectVariant(layer, variant_prim_path)
# Or
edit_target = variant_set_api.GetVariantEditTarget()
edit_context = Usd.EditContext(stage, edit_target) 
with edit_context as ctx:
    sphere_prim_path = bicycle_prim_path.AppendChild("sphere")
    sphere_prim = stage.DefinePrim("/bicycle/sphere", "Sphere")
### Low Level ###
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
bicycle_prim_spec.typeName = "Xform"
# Variants
cube_prim_path = bicycle_prim_path.AppendVariantSelection("color", "colorA").AppendChild("cube")
cube_prim_spec = Sdf.CreatePrimInLayer(layer, cube_prim_path)
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
sphere_prim_path = bicycle_prim_path.AppendVariantSelection("color", "colorB").AppendChild("sphere")
sphere_prim_spec = Sdf.CreatePrimInLayer(layer, sphere_prim_path)
sphere_prim_spec.specifier = Sdf.SpecifierDef
sphere_prim_spec.typeName = "Sphere"
# Variant Selection
bicycle_prim_spec.variantSelections["color"] = "colorA"

# We can also author the variants via variant specs
layer = Sdf.Layer.CreateAnonymous()
car_prim_path = Sdf.Path("/car")
car_prim_spec = Sdf.CreatePrimInLayer(layer, car_prim_path)
car_prim_spec.specifier = Sdf.SpecifierDef
car_prim_spec.typeName = "Xform"
# Variants
variant_set_spec = Sdf.VariantSetSpec(car_prim_spec, "color")
variant_spec = Sdf.VariantSpec(variant_set_spec, "colorA")
cube_prim_spec = Sdf.PrimSpec(variant_spec.primSpec, "cube", Sdf.SpecifierDef)
cube_prim_spec.typeName = "Cube"
variant_spec = Sdf.VariantSpec(variant_set_spec, "colorB")
cube_prim_spec = Sdf.PrimSpec(variant_spec.primSpec, "sphere", Sdf.SpecifierDef)
cube_prim_spec.typeName = "Sphere"
# Ironically this does not setup the variant set names metadata, so we have to author it ourselves.
car_prim_spec.SetInfo("variantSetNames", Sdf.StringListOp.Create(prependedItems=["color"]))
# Variant Selection
car_prim_spec.variantSelections["color"] = "colorA"

When editing variants, we can also move layer content into a (nested) variant very easily via the Sdf.CopySpec command. This is a very powerful feature!

from pxr import Sdf
# Spawn other layer, this usually comes from other stages, that your DCC creates/owns.
some_other_layer = Sdf.Layer.CreateAnonymous()
root_prim_path = Sdf.Path("/root")
cube_prim_path = Sdf.Path("/root/cube")
cube_prim_spec = Sdf.CreatePrimInLayer(some_other_layer, cube_prim_path)
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
# Create demo layer
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
bicycle_prim_spec.typeName = "Xform"
# Copy content into variant
variant_set_spec = Sdf.VariantSetSpec(bicycle_prim_spec, "color")
variant_spec = Sdf.VariantSpec(variant_set_spec, "colorA")
variant_prim_path = bicycle_prim_path.AppendVariantSelection("color", "colorA")
Sdf.CopySpec(some_other_layer, root_prim_path, layer, variant_prim_path)
# Variant selection
bicycle_prim_spec.SetInfo("variantSetNames", Sdf.StringListOp.Create(prependedItems=["color"]))
bicycle_prim_spec.variantSelections["color"] = "colorA"

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()

def variant_nested_edit_context(prim, variant_selections, position=Usd.ListPositionBackOfPrependList):
    """Author nested variants
    Args:
        prim (Usd.Prim): The prim to author on.
        variant_selections (list): A list of tuples with [('variant_set_name', 'variant_name')] data.
        position (Usd.ListPosition): The list position of the variant set.
    Returns:
        Usd.EditContext: An edit context manager.
    """
    
    if not variant_selections:
        raise Exception("No valid variant selections defined!")
        
    def _recursive_variant_context(prim, variant_selections, position):
        variant_sets_api = bicycle_prim.GetVariantSets()
        
        variant_selection = variant_selections.pop(-1)
        variant_set_name, variant_name = variant_selection
        
        variant_set_api = variant_sets_api.AddVariantSet(variant_set_name, position=position)
        variant_set_api.AddVariant(variant_name)
        # Be aware, this authors the selection in the variant
        # ToDo make this a context manager that cleans up the selection authoring.
        variant_set_api.SetVariantSelection(variant_name)
        
        if not variant_selections:
            return variant_set_api.GetVariantEditContext()
        else:
            with variant_set_api.GetVariantEditContext():
                return _recursive_variant_context(prim, variant_selections, position)
    
    variant_selections = variant_selections[::-1]
    return _recursive_variant_context(prim, variant_selections, position)

bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path, "Xform")
# Variants
variant_selections = [("model", "old"), ("LOD", "lowRes")]
edit_context = variant_nested_edit_context(bicycle_prim, variant_selections)
with edit_context as ctx:
    sphere_prim_path = bicycle_prim_path.AppendChild("sphere")
    sphere_prim = stage.DefinePrim("/bicycle/sphere", "Sphere")
variant_selections = [("model", "old"), ("LOD", "highRes")]
edit_context = variant_nested_edit_context(bicycle_prim, variant_selections)
with edit_context as ctx:
    sphere_prim_path = bicycle_prim_path.AppendChild("cube")
    sphere_prim = stage.DefinePrim("/bicycle/cube", "Cube")
variant_selections = [("model", "new"), ("LOD", "lowRes")]
edit_context = variant_nested_edit_context(bicycle_prim, variant_selections)
with edit_context as ctx:
    sphere_prim_path = bicycle_prim_path.AppendChild("cylinder")
    sphere_prim = stage.DefinePrim("/bicycle/cube", "Cylinder")
# Variant selections
# Be sure to explicitly set the overall selection, otherwise if will derive from,
# the nested variant selections.
variant_sets_api = bicycle_prim.GetVariantSets()
variant_sets_api.SetSelection("model", "old")
variant_sets_api.SetSelection("LOD", "lowRes")

### Low Level ###
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
def variant_nested_prim_path(prim_path, variant_selections):
    variant_prim_path = prim_path
    for variant_set_name, variant_name in variant_selections:
        variant_prim_path = variant_prim_path.AppendVariantSelection(variant_set_name, variant_name)
    return variant_prim_path
    
def define_prim_spec(layer, prim_path, type_name):
    prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
    prim_spec.specifier = Sdf.SpecifierDef
    prim_spec.typeName = type_name

bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
bicycle_prim_spec.typeName = "Xform"
# Variants
variant_selections = [("model", "old"), ("LOD", "lowRes")]
variant_prim_path = variant_nested_prim_path(bicycle_prim_path, variant_selections)
define_prim_spec(layer, variant_prim_path.AppendChild("sphere"), "Sphere")
variant_selections = [("model", "old"), ("LOD", "highRes")]
variant_prim_path = variant_nested_prim_path(bicycle_prim_path, variant_selections)
define_prim_spec(layer, variant_prim_path.AppendChild("cube"), "Cube")
variant_selections = [("model", "new"), ("LOD", "lowRes")]
variant_prim_path = variant_nested_prim_path(bicycle_prim_path, variant_selections)
define_prim_spec(layer, variant_prim_path.AppendChild("cylinder"), "Cylinder")
# Variant selections
# The low level API has the benefit of not setting variant selections
# in the nested variants.
bicycle_prim_spec.variantSelections["model"] = "old"
bicycle_prim_spec.variantSelections["LOD"] = "highRes"

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what the reference arc is used for.

from pxr import Sdf
ref = Sdf.Reference("/file/path.usd", "/prim/path", Sdf.LayerOffset(offset=10, scale=1))
# The reference object is a read only instance.
print(ref.assetPath) # Returns: "/file/path.usd"
print(ref.primPath) # Returns: "/prim/path"
print(ref.layerOffset) # Returns: Sdf.LayerOffset(offset=10, scale=1)
try: 
    ref.assetPath = "/some/other/file/path.usd"
except Exception:
    print("Read only Sdf.Reference!")

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Spawn temp layer
reference_layer = Sdf.Layer.CreateAnonymous("ReferenceExample")
reference_bicycle_prim_path = Sdf.Path("/bicycle")
reference_bicycle_prim_spec = Sdf.CreatePrimInLayer(reference_layer, reference_bicycle_prim_path)
reference_bicycle_prim_spec.specifier = Sdf.SpecifierDef
reference_bicycle_prim_spec.typeName = "Cube"
# Set the default prim to use when we specify no primpath. It can't be a prim path, it must be a root prim.
reference_layer.defaultPrim = reference_bicycle_prim_path.name
# Reference
reference_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
reference = Sdf.Reference(reference_layer.identifier, reference_bicycle_prim_path, reference_layer_offset)
# Or: If we don't specify a prim, the default prim will get used, as set above
reference = Sdf.Reference(reference_layer.identifier, layerOffset=reference_layer_offset)
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path)
references_api = bicycle_prim.GetReferences()
references_api.AddReference(reference, position=Usd.ListPositionFrontOfAppendList)
# references_api.SetReferences() # Clears the list editable ops and authors an Sdf.ReferenceListOp.CreateExplicit([])
# references_api.RemoveReference(cube_prim_path)
# references_api.ClearReferences() # Sdf.ReferenceListOp.Clear()
### Low Level ###
from pxr import Sdf
# Spawn temp layer
reference_layer = Sdf.Layer.CreateAnonymous("ReferenceExample")
reference_bicycle_prim_path = Sdf.Path("/bicycle")
reference_bicycle_prim_spec = Sdf.CreatePrimInLayer(reference_layer, reference_bicycle_prim_path)
reference_bicycle_prim_spec.specifier = Sdf.SpecifierDef
reference_bicycle_prim_spec.typeName = "Cube"
reference_layer.defaultPrim = reference_bicycle_prim_path.name
# In Houdini add, otherwise the layer will be garbage collected.
# node.addHeldLayer(reference_layer.identifier)
# Reference
layer = Sdf.Layer.CreateAnonymous()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
reference_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
reference = Sdf.Reference(reference_layer.identifier, reference_bicycle_prim_path, reference_layer_offset)
# Or: If we don't specify a prim, the default prim will get used, as set above
reference = Sdf.Reference(reference_layer.identifier, layerOffset=reference_layer_offset)
bicycle_prim_spec.referenceList.appendedItems = [reference]

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Spawn hierarchy
cube_prim_path = Sdf.Path("/cube")
cube_prim = stage.DefinePrim(cube_prim_path, "Cube")
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path)
# Reference
reference_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
reference = Sdf.Reference("", cube_prim_path, reference_layer_offset)
references_api = bicycle_prim.GetReferences()
references_api.AddReference(reference, position=Usd.ListPositionFrontOfAppendList)
# Or:
references_api.AddInternalReference(cube_prim_path, reference_layer_offset, position=Usd.ListPositionFrontOfAppendList)
### Low Level ###
from pxr import Sdf
# Spawn hierarchy
layer = Sdf.Layer.CreateAnonymous()
cube_prim_path = Sdf.Path("/cube")
cube_prim_spec = Sdf.CreatePrimInLayer(layer, cube_prim_path)
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
# Reference
reference_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
reference = Sdf.Reference("", cube_prim_path, reference_layer_offset)
bicycle_prim_spec.referenceList.appendedItems = [reference]

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what the payload arc is used for.

from pxr import Sdf
payload = Sdf.Payload("/file/path.usd", "/prim/path", Sdf.LayerOffset(offset=10, scale=1))
# The reference object is a read only instance.
print(payload.assetPath) # Returns: "/file/path.usd"
print(payload.primPath) # Returns: "/prim/path"
print(payload.layerOffset) # Returns: Sdf.LayerOffset(offset=10, scale=1)
try: 
    payload.assetPath = "/some/other/file/path.usd"
except Exception:
    print("Read only Sdf.Payload!")

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Spawn temp layer
payload_layer = Sdf.Layer.CreateAnonymous("PayloadExample")
payload_bicycle_prim_path = Sdf.Path("/bicycle")
payload_bicycle_prim_spec = Sdf.CreatePrimInLayer(payload_layer, payload_bicycle_prim_path)
payload_bicycle_prim_spec.specifier = Sdf.SpecifierDef
payload_bicycle_prim_spec.typeName = "Cube"
# Set the default prim to use when we specify no primpath. It can't be a prim path, it must be a root prim.
payload_layer.defaultPrim = payload_bicycle_prim_path.name
# Payload
payload_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
payload = Sdf.Payload(payload_layer.identifier, payload_bicycle_prim_path, payload_layer_offset)
# Or: If we don't specify a prim, the default prim will get used, as set above
payload = Sdf.Payload(payload_layer.identifier, layerOffset=payload_layer_offset)
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path)
payloads_api = bicycle_prim.GetPayloads()
payloads_api.AddPayload(payload, position=Usd.ListPositionFrontOfAppendList)
# payloads_api.SetPayloads() # Clears the list editable ops and authors an Sdf.PayloadListOp.CreateExplicit([])
# payloads_api.RemovePayload(cube_prim_path)
# payloads_api.ClearPayloads() # Sdf.PayloadListOp.Clear()
### Low Level ###
from pxr import Sdf
# Spawn temp layer
payload_layer = Sdf.Layer.CreateAnonymous("PayLoadExample")
payload_bicycle_prim_path = Sdf.Path("/bicycle")
payload_bicycle_prim_spec = Sdf.CreatePrimInLayer(payload_layer, payload_bicycle_prim_path)
payload_bicycle_prim_spec.specifier = Sdf.SpecifierDef
payload_bicycle_prim_spec.typeName = "Cube"
payload_layer.defaultPrim = payload_bicycle_prim_path.name
# In Houdini add, otherwise the layer will be garbage collected.
# node.addHeldLayer(payload_layer.identifier)
# Payload
layer = Sdf.Layer.CreateAnonymous()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
payload_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
payload = Sdf.Payload(payload_layer.identifier, payload_bicycle_prim_path, payload_layer_offset)
# Or: If we don't specify a prim, the default prim will get used, as set above
payload = Sdf.Payload(payload_layer.identifier, layerOffset=payload_layer_offset)
bicycle_prim_spec.payloadList.appendedItems = [payload]

In our Composition Strength Ordering (LIVRPS) section we cover in detail, with production related examples, what the specialize arc is used for.

### High Level ###
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path)
cube_prim_path = Sdf.Path("/cube")
cube_prim = stage.DefinePrim(cube_prim_path, "Cube")
specializes_api = bicycle_prim.GetSpecializes()
specializes_api.AddSpecialize(cube_prim_path, position=Usd.ListPositionFrontOfAppendList)
# inherits_api.SetSpecializes() # Clears the list editable ops and authors an Sdf.PathListOp.CreateExplicit([])
# inherits_api.RemoveSpecialize(cube_prim_path)
# inherits_api.ClearSpecializes() # Sdf.PathListOp.Clear()
### Low Level ###
from pxr import Sdf
layer = Sdf.Layer.CreateAnonymous()
bicycle_prim_path = Sdf.Path("/bicycle")
bicycle_prim_spec = Sdf.CreatePrimInLayer(layer, bicycle_prim_path)
bicycle_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_path = Sdf.Path("/cube")
cube_prim_spec = Sdf.CreatePrimInLayer(layer, cube_prim_path)
cube_prim_spec.specifier = Sdf.SpecifierDef
cube_prim_spec.typeName = "Cube"
bicycle_prim_spec.specializesList.appendedItems = [cube_prim_path]

We have a supplementary Houdini scene, that you can follow along with, available in this site's repository. All the examples below will walk through this file as it easier to prototype and showcase arcs in Houdini via nodes, than writing it all in code.

We'll be using composition arcs to load data from different files and hierarchy locations. This is the core mechanism that makes USD powerful, as we can layer/combine our layers in meaningful ways.

For USD to be able to scale well, we can also "lock" the composition on prims with the same arcs, so that they can use the same data source. This allows us to create instances, which keep our memory footprint low.

• Composition arcs target layer stacks, not individual layers. This just means that they recursively load what is in a layer.
• When arcs target a non root prim, they do not receive parent data that usually "flows" down the hierarchy. This means that primvars, material bindings or transforms from ancestor prims do not get "inherited" (we don't mean the inherited arc here). They do see the composition result though. So for example if your file reference targets a prim inside a variant, it can't change the variant as the variant is not in the stage it was referenced into to.
• Internal composition arcs (inherit/internal references/specialize) cannot target ancestor or child arcs. We can only target sibling prims or prims that are at/under a different "/" stage root prim.
• Composition arcs only look in the active layer stack and in "higher" layer stacks (layer stacks that reference/payload the active layer stack).

When resolving nested composition arcs and value clips, the arc/value clip metadata, that is authored on the closest ancestor parent prim or the prim itself, wins. In short to quote from the USD glossary “ancestral arcs” are weaker than “direct arcs”. To make our lives easier, we recommend having predefined locations where you author composition arcs. A typical location is your asset root prim and a set/assembly root prim.

flowchart LR
    userRoot(["I want to load"])
    userExternal(["a file (with a time offset/scale)"])
    userExternalWholeLayer(["with the whole content (that is light weight and links to other files)"])
    userExternalHierarchyInLayer(["with only a specific part of its hierarchy (to a specific path in the stage)"])
    userExternalHierarchyInLayerHeavyData(["thats contains a lot of heavy data"])
    userExternalHierarchyInLayerLightData(["thats contains light weight data"])
    userInternal(["an existing sub-hierarchy in the active layer stack/stage"])
    userInternalTimeOffset(["into a new hierarchy location (with a time offset/scale)"])
    userInternalOverride(["to add overrides to multiple (instanced) prims"])
    userInternalBaseValues(["to act as the base-line values that can be overriden by higher layers and arcs"])
    userInternalVariation(["as a variation of the hierarchy"])

    compositionArcSublayer(["Sublayer"])
    compositionArcInherit(["Inherit"])
    compositionArcVariant(["Variant"])
    compositionArcReferenceFile(["Reference"])
    compositionArcReferenceInternal(["Reference"])
    compositionArcPayload(["Payload"])
    compositionArcSpecialize(["Specialize"])
    style compositionArcSublayer fill:#63beff
    style compositionArcInherit fill:#63beff
    style compositionArcVariant fill:#63beff
    style compositionArcReferenceFile fill:#63beff
    style compositionArcReferenceInternal fill:#63beff
    style compositionArcPayload fill:#63beff
    style compositionArcSpecialize fill:#63beff

    userRoot --> userExternal
    userExternal --> userExternalWholeLayer
    userExternal --> userExternalHierarchyInLayer
    userExternalWholeLayer --> compositionArcSublayer
    userExternalHierarchyInLayer --> userExternalHierarchyInLayerHeavyData
    userExternalHierarchyInLayer --> userExternalHierarchyInLayerLightData
    userExternalHierarchyInLayerLightData --> compositionArcReferenceFile
    userExternalHierarchyInLayerHeavyData --> compositionArcPayload
    userRoot --> userInternal
    userInternal --> userInternalTimeOffset
    userInternalTimeOffset --> compositionArcReferenceInternal
    userInternal --> userInternalOverride
    userInternalOverride --> compositionArcInherit
    userInternal --> userInternalBaseValues
    userInternalBaseValues --> compositionArcSpecialize
    userInternal --> userInternalVariation
    userInternalVariation --> compositionArcVariant

flowchart LR
    userRoot(["I want to"])
    userRootTimeOffset(["time offset/scale my hierarchy (via a `Sdf.LayerOffset`)"])
    userRootNoTimeOffset(["not time offset/scale my hierarchy"])

    compositionArcSublayer(["Sublayer"])
    compositionArcInherit(["Inherit"])
    compositionArcVariant(["Variant"])
    compositionArcReference(["Reference (External/Internal)"])
    compositionArcPayload(["Payload"])
    compositionArcSpecialize(["Specialize"])
    style compositionArcSublayer fill:#63beff
    style compositionArcInherit fill:#63beff
    style compositionArcVariant fill:#63beff
    style compositionArcReference fill:#63beff
    style compositionArcPayload fill:#63beff
    style compositionArcSpecialize fill:#63beff

    userRoot --> userRootTimeOffset
    userRootTimeOffset --> compositionArcSublayer
    userRootTimeOffset --> compositionArcReference
    userRootTimeOffset --> compositionArcPayload
    userRoot --> userRootNoTimeOffset
    userRootNoTimeOffset --> compositionArcInherit
    userRootNoTimeOffset --> compositionArcVariant
    userRootNoTimeOffset --> compositionArcSpecialize

flowchart TD
    compositionArcSublayer(["Sublayers (Direct Opinions)"])
    compositionArcValueClip(["Value Clips (Lower than Direct Opinions)"])
    compositionArcInherit(["Inherits"])
    compositionArcVariant(["Variants"])
    compositionArcReferenceFile(["References"])
    compositionArcReferenceInternal(["References"])
    compositionArcPayload(["Payloads"])
    compositionArcSpecialize(["Specialize"])
    compositionArcInternal(["Internal Arcs (Target Hierarchy)"])
    compositionArcExternal(["File Arcs (Target File (+ Hierarchy))"])
    style compositionArcSublayer fill:#63beff
    style compositionArcInherit fill:#63beff
    style compositionArcVariant fill:#63beff
    style compositionArcReferenceFile fill:#63beff
    style compositionArcReferenceInternal fill:#63beff
    style compositionArcPayload fill:#63beff
    style compositionArcSpecialize fill:#63beff    

    compositionArcInternal --> compositionArcInherit
    compositionArcInternal --> compositionArcVariant
    compositionArcInternal --> compositionArcReferenceInternal
    compositionArcInternal --> compositionArcSpecialize
    compositionArcExternal --> compositionArcSublayer
    compositionArcExternal --> compositionArcReferenceFile
    compositionArcExternal --> compositionArcPayload
    compositionArcSublayer --> compositionArcValueClip

We have a supplementary Houdini scene, that you can follow along with, available in this site's repository. All the examples below will walk through this file as it easier to prototype and showcase arcs in Houdini via nodes, than writing it all in code.

Typically we'll be using sublayers for mainly these things:

• As a mechanism to separate data when working in your DCCs. On file write we usually flatten layers to a single flattened output(s)(if you have multiple save paths set). Why not put everything on the same layer? We can use the layer order as a form of control to A. allow/block edits (or rather seeing them have an effect because of weaker opinion strength) B. Sort data from temporary data.
• To load in references and payloads. That way all the heavy lifting is not done (and should not be done) by the sublayer arc.
• In shot workflows to load different shot layers. Why don't we do this via references or payloads you might be asking yourself? As covered in our fundamentals section, anything your reference or payload in will be encapsulated. In shot workflows we want to keep having access to list editable ops. For example if we have a layout and a lighting layer, the lighting layer should still be able to remove a reference, that was created in the layout layer.

If you want to create/edit sublayer arcs via code, see our Composition Arc - Code API section.

Value clips are written via metadata entries on prims. If you want to create/edit value clips via code, see our value clips section.

We cover a production based example of how to load value clips in our Composition for production section. Here are some import things to keep in mind:

• When making prims instanceable, the value clip metadata has to be under the instanceable prim, as the value clip metadata can't be read from outside of the instance (as it would then mean each instance could load different clips, which would defeat the purpose of instanceable prims).
• Value clip metadata can't be inherited/internally referenced/specialized in. It must reside on the prim as a direct opinion.

See the production examples for how to best load value clips.

• We use inherit arcs as a "broadcast" operator for overrides: When we want to apply an edit to our hierarchy in multiple places, we typically create a class prim, whose child prims contain the properties we want to modify. After that we create an inherit arc on all prims that should receive the edit. As it is the second highest arc behind direct opinions, it will always have the highest composition strength, when applied to instanceable prims, as instanceable prims can't have direct opinions.
• The inherit arc lookup is never encapsulated, the inherit arc list-editable op is. This means that any layer stack, that re-creates the prims that that the inherit targets, gets used by the inherit. This does come at a performance cost, as the composition engine needs to check all layer stacks from where the arc was authored and higher for the hierarchy that the inherit targets.
• The inherit arc commonly gets used together with the class prim specifier. The class prim specifier is specifically there to get ignored by default traversals and to provide template hierarchies that can then get inherited (or internally referenced/specialized) to have a "single source to multiple targets" effect.
• Depending on if we are working on shots or assets are common practices:
  • Assets: When creating assets, we can author a /__CLASS__/<assetName> inherit. When we use the asset in shots, we can then easily add overrides to all assets of this type, by creating prims and properties under that specific class prim hierarchy. While this sounds great in theory, artists often want to only selectively apply an override to an asset. Therefore having the additional performance cost of this arc in assets is something might not worth doing. See the next bullet point.
  • Shots: This is where inherits shine! We usually create inherits to:
    • Batch apply render geometry settings to (instanceable) prims. This is a great way of having a single control point to editing render settings per different areas of interest in your scene.
    • Batch apply activation/visibility to instanceable prims. This way we don't increase the prototype count.

If you want to create/edit inherit arcs via code, see our Composition Arc - Code API section.

Here is a typical code pattern we'll use when creating inherits:

from pxr import Sdf
...
# Inspect prototype and collect what to override
prototype = prim.GetPrototype()
...
# Create overrides
class_prim = stage.CreateClassPrim(Sdf.Path("/__CLASS__/myCoolIdentifier"))
edit_prim = stage.DefinePrim(class_prim.GetPath().AppendChild("leaf_prim"))
edit_prim.CreateAttribute("size", Sdf.ValueTypeNames.Float).Set(5)
...
# Add inherits
instance_prims = prototype.GetInstances()
for instance_prim in instance_prims:
    inherits_api = instance_prim.GetInherits()
    inherits_api.AddInherit(class_prim.GetPath(), position=Usd.ListPositionFrontOfAppendList)    

When you've worked a while in USD, you sometimes wonder why we need all these different layering rules. Why can't life be simple for once?

Head over to our Composition in production section for a more production related view on composition. There we discuss how to get the best out of each arc, without making it overly complicated.

• We use it as a mechanism to swap between (wait for it ...) variations of a hierarchy. The main applications are:
  • Assets: Instead of having multiple assets with variations based of off a single base asset, we can store one asset and add variants. That way it is cleaner to track throughout the pipeline.
  • Shots: Variants in shots are typically used when a sequence based hierarchy needs to be replaced by a per shot variation. While you could also solve this by just deactivating the prims your need per shot or via sublayer ordering (if you overwrite the same prims), variants offer a cleaner workflow. This way we can keep the hierarchy at the same place and all our lighting department needs to do is target the same hierarchy, when picking what to render. Since variants swap out a whole sub section of a hierarchy, we also ensure that the geometry is not getting any unwanted attributes from other layers.
• We can have any number of nested variants. A typical example is having multiple model variants, which each in return have different LOD (level of detail) variants.
• We can also use it as a mechanism to share mesh data. For example if we have a car asset, we can write the car without a variant and then add all car accessories (which don't overlap hierarchy-wise with the car meshes) as variants. That way the artists can swap through what they need on top of the "base" model.
• We don't need to have a variant selection. If we block or unset the selection, no variant will be selected/loaded, which results in an empty hierarchy. Fallbacks for variant set selections can be configured via the USD API or a USD plugin (API Docs, search for 'Variant Management')
• How are variants structured? Variants are written "inline", unlike the inherit/reference/payload/specialize arcs, they do not point to another hierarchy path. Instead they are more similar to child prims (specs). We usually then write other arcs, like payloads, into the variants, that do the actual heavy data loading.
• We can also use variants as the mechanism to "variant away" arcs that have been encapsulated. More info in our fundamentals section.

If you want to create/edit variant arcs via code, see our Composition Arc - Code API section.

def Xform "car" (
    variants = {
        string color = "colorA"
    }
    prepend variantSets = "color"
)
{
    variantSet "color" = {
        "colorA" {
            def Cube "cube"
            {
            }
        }
        "colorB" {
            def Sphere "sphere"
            {
            }
        }
    }
}

def Xform "car" (
    variants = {
        string color = ""
    }
    prepend variantSets = "color"
)
{
    variantSet "color" = {
        "colorA" {
            def Cube "cube"
            {
            }
        }
        "colorB" {
            def Sphere "sphere"
            {
            }
        }
    }
}

When working with nested variants in production, we recommend locking down the naming convention for the variant set names as well as the nested order. We also recommend not creating nested variants that only exists for a specific parent variant. This way, variant sets don't "randomly" come into existence based on other nested variant selections.

That way all your code knows where to look when authoring variants and authoring variants can be automated.

from pxr import Sdf, Usd

node = hou.pwd()
layer = node.editableLayer()

source_node = node.parm("spare_input0").evalAsNode()
source_stage = source_node.stage()
source_layer = source_node.activeLayer()

with Sdf.ChangeBlock():
    edit = Sdf.BatchNamespaceEdit()

    iterator = iter(Usd.PrimRange(source_stage.GetPseudoRoot()))
    for prim in iterator:
        if "GEO" not in prim.GetChildrenNames():
            continue
        iterator.PruneChildren()
        prim_path = prim.GetPath()
        prim_spec = layer.GetPrimAtPath(prim_path)
        # Move content into variant
        variant_set_spec = Sdf.VariantSetSpec(prim_spec, "model")
        variant_spec = Sdf.VariantSpec(variant_set_spec, "myCoolVariant")
        variant_prim_path = prim_path.AppendVariantSelection("model", "myCoolVariant")
        edit.Add(prim_path.AppendChild("GEO"), variant_prim_path.AppendChild("GEO"))
        # Variant selection
        prim_spec.SetInfo("variantSetNames", Sdf.StringListOp.Create(prependedItems=["model"]))
        prim_spec.variantSelections["model"] = "myCoolVariant"
        
    if not layer.Apply(edit):
        raise Exception("Failed to apply layer edit!")

from pxr import Sdf, Usd

node = hou.pwd()
layer = node.editableLayer()

source_node = node.parm("spare_input0").evalAsNode()
source_stage = source_node.stage()
source_layer = source_node.activeLayer()

with Sdf.ChangeBlock():
    iterator = iter(Usd.PrimRange(source_stage.GetPseudoRoot()))
    for prim in iterator:
        if "GEO" not in prim.GetChildrenNames():
            continue
        iterator.PruneChildren()
        prim_path = prim.GetPath()
        prim_spec = Sdf.CreatePrimInLayer(layer, prim_path)
        prim_spec.specifier = Sdf.SpecifierDef
        prim_spec.typeName = "Xform"
        parent_prim_spec = prim_spec.nameParent
        while parent_prim_spec:
            parent_prim_spec.specifier = Sdf.SpecifierDef
            parent_prim_spec.typeName = "Xform"
            parent_prim_spec = parent_prim_spec.nameParent
                    
        # Copy content into variant
        variant_set_spec = Sdf.VariantSetSpec(prim_spec, "model")
        variant_spec = Sdf.VariantSpec(variant_set_spec, "myCoolVariant")
        variant_prim_path = prim_path.AppendVariantSelection("model", "myCoolVariant")
        Sdf.CopySpec(source_layer, prim_path, layer, variant_prim_path)
        # Variant selection
        prim_spec.SetInfo("variantSetNames", Sdf.StringListOp.Create(prependedItems=["model"]))
        prim_spec.variantSelections["model"] = "myCoolVariant"

• The reference arc is one of the most used arcs in USD: Its main use case is to combine smaller pieces of scene description into larger aggregates. Its main applications are:
  • Assets: For assets we mainly stick to payloading in the individual layers (model/fx/materials) or we create a single payload that then references all the individual layers. So depending on how you build your asset structure, you might end up not using references, except for when building assembly type assets. Assemblies are USD files that reference other USD files for a logical grouping of assets, for example a film set or city.
  • Shots: This is the most common place of usage for references. We use references to bring in our assets or assemblies. We also use it to time offset assets or existing hierarchies, as the reference arc can point to another prim path in the active layer stack. This makes it a powerful tool to drive (time) variations of assets in shots.
• As written in the previous bullet point, the reference arc should aggregate data. That means the files it is loading should not be heavy, but contain only (lofted) metadata and references/payloads to other files.
• The reference arc targets a specific prim in the hierarchy. When using references to load a file, we either point to a specific prim we want to load from that file or we don't specify a prim and then the value of the "defaultPrim" layer metadata gets used. The default prim has to be a direct child prim of the pseudo root prim "/".
• The reference arc (as the payload arc) uses the principle of encapsulation. This means once a file is referenced in, the composition arcs in the file can't be list-edited any more.

If you want to create/edit reference arcs via code, see our Composition Arc - Code API section.

• You might be wondering why we should use payloads when we have references? The answer is USD's hierarchy loading mechanisms. Payloads are special, in that we can tell USD to not load any payloads or only specific hierarchies with payloads when working with stages. We cover it in more detail in our Loading Mechanisms section. This makes payload's main purpose loading heavy data.
• The payload arc is therefore also one of the most used arcs in USD, its main applications are:
  • Assets: Any layer in your asset, that contains geometry/renderable data, should be behind a payload. We can also first aggregate different files via references and then load this aggregated file via a payload. The important thing is that when we load the asset into the shot, we can opt-out of having to load the heavy data directly.
  • Shots: In shots we use payloads to bring in shot relevant caches, most importantly animation and FX caches. Now you might have noticed that payloads are lower than references. This means if we want to load an fx cache over an asset that was brought in as a reference, we either have to first import the payload somewhere else and then link to it via an inherit or variant, or we don't load it as a payload and bring it in as a reference. More on how this affects performance below.
• The payload arc targets a specific prim in the hierarchy. When using payloads to load a file, we either point to a specific prim we want to load from that file or we don't specify a prim and then the value of the "defaultPrim" layer metadata gets used. The default prim has to be a direct child prim of the pseudo root prim "/".
• The payload arc (as the reference arc) uses the principle of encapsulation. This means once a file is payloaded in, the composition arcs in the file can't be list-edited any more. See the reference section above for more info. Now with payloads this isn't an issue that much, because typically we use payloads to point to a cache file that carries the raw data and not other cache files.
• Payloads can also be time offset via an Sdf.LayerOffset.

If you want to create/edit payload arcs via code, see our Composition Arc - Code API section.

• We use specialize arcs as a "broadcast" operator for supplying a template-like hierarchy: When we want to supply a "base" hierarchy to multiple places, we typically create a class prim, whose child prims contain the properties we want to modify. After that we create a specialize arc on all prims that should receive the hierarchy.
• The specialize arc lookup is never encapsulated, the specialize arc list-editable op is. This means that any layer stack, that re-creates the prims that that the specialize targets, gets used by the specialize. This does come at a performance cost, as the composition engine needs to check all layer stacks from where the arc was authored and higher for the hierarchy that the specialize targets.
• The specialize arc commonly gets used together with the class prim specifier. The class prim specifier is specifically there to get ignored by default traversals and to provide template hierarchies that can then get inherited (or internally referenced/specialized) to have a "single source to multiple targets" effect.
• Depending on if we are working on shots or assets are common practices:
  • Assets: As assets provide the data for our shot content, specializes are more typical here.
    • When creating assets, we can author a /__CLASS__/<assetName> specialize. When we use the asset in shots, we can then easily add "underrides" to all assets of this type, by creating prims and properties under that specific class prim hierarchy. Since specializes have the lowest strength, any other composition arc that provides data will win in any higher layer. While this sounds great in theory, artists often want to only selectively apply an override to an asset. Therefore having the additional performance cost of this arc in assets is something might not worth doing.
    • The other common case is to use them for materials:
      • We put all of our materials under a /__CLASS__ and then specialize it to the asset materials. This way asset (material) variants can add overrides to the materials. We could also use an internal reference arc to achieve this, depending on how you structure your composition though it would defeat the purpose of even a non direction opinion style authoring. For example if we payload individual asset layers together (e.g. fx_layer.usd, model_layer.usd), then the internal reference would be encapsulated and not live anymore. Whereas a specialize would be.
      • We specialize materials with each other. You can find the example from the USD glossary below that shows how this works.
  • Shots: In shots, specializes are more rare, as the shot layer stack is the "final" layer stack that gets rendered. Some use cases might be:
    • Batch apply render geometry settings to (instanceable) prims. This is a great way of having a single control point to editing render settings per different areas of interest in your scene. This has the same intent as inherits, the difference is that existing overrides are kept in place, as with inherits they are not (unless they are direct opinions on sublayers).
    • Building a template hierarchy for new content hierarchies that don't 'over' over any existing prims.

If you want to create/edit specialize arcs via code, see our Composition Arc - Code API section.

Instancing is the multi re-use of a part of the hierarchy, so that we don't have to load it into memory multiple times. In USD speak the term for the "base" copy, all instances refer to, is Prototype.

# Check if the active prim is marked as instanceable:
# The prim.IsInstance() checks if it is actually instanced, this
# just checks if the 'instanceable' metadata is set.
prim.IsInstanceable()
# Check if the active prim is an instanced prim:
prim.IsInstance()
# Check if we are inside an instanceable prim:
prim.IsInstanceProxy()
# Check if the active prim is a prototype root prim with the following format /__Prototype_<idx>
prim.IsPrototype()
# For these type of prototype root prims, we can get the instances via:
prim.GetInstances()
# From each instance we can get back to the prototype via
prim.GetPrototype()
# Check if we are in the /__Prototype_<idx> prim:
prim.IsInPrototype()

# When we are within an instance, we can get the prototype via:
if prim.IsInstanceProxy():
    for ancestor_prim_path in prim.GetAncestorsRange():
        ancestor_prim = stage.GetPrimAtPath(ancestor_prim_path)
        if ancestor_prim.IsInstance():
            prototype = ancestor_prim.GetPrototype()
            print(list(prototype.GetInstances()))
            break

Using list editable ops in non composition arc scenarios is rare, as we often want a more attribute like value resolution behavior. It is good to know though that the mechanism is there. A good production use case is making metadata that tracks asset dependencies list editable, that way all layers can contribute to the sidecar data.

from pxr import Sdf
path_list_op = Sdf.PathListOp()
# There are multiple sub-lists, which are just normal Python lists.
# 'prependedItems', 'appendedItems', 'deletedItems', 'explicitItems',
# Legacy sub-lists (do not use these anymore): 'addedItems', 'orderedItems'
# Currently the way these are exposed to Python, you have to re-assign the list, instead of editing it in place.
# So this won't work:
path_list_op.prependedItems.append(Sdf.Path("/cube"))
path_list_op.appendedItems.append(Sdf.Path("/sphere"))
# Instead do this:
path_list_op.prependedItems = [Sdf.Path("/cube")]
path_list_op.appendedItems = [Sdf.Path("/sphere")]
# To clear the list op:
print(path_list_op) # Returns: SdfPathListOp(Prepended Items: [/cube], Appended Items: [/sphere])
path_list_op.Clear()
print(path_list_op) # Returns: SdfPathListOp()
# Repopulate via constructor
path_list_op = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])
print(path_list_op) # Returns: SdfPathListOp(Prepended Items: [/cube], Appended Items: [/sphere])
# Add remove items
path_list_op.deletedItems = [Sdf.Path("/sphere")]
print(path_list_op) # Returns: SdfPathListOp(Deleted Items: [/sphere], Prepended Items: [/cube], Appended Items: [/sphere])
# Notice how it just stores lists, it doesn't actually apply them. We'll have a look at that next.

# For example when editing a relationship:
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
yard_prim = stage.DefinePrim("/yard")
car_prim = stage.DefinePrim("/car")
rel = car_prim.CreateRelationship("locationsOfInterest")
rel.AddTarget(yard_prim.GetPath(), position=Usd.ListPositionFrontOfAppendList)
# Result:
"""
def "car"
{
    custom rel locationsOfInterest
    append rel locationsOfInterest = </yard>
}
"""
# The "Set<Function>" signatures write an explicit list:
rel.SetTargets([yard_prim.GetPath()])
# Result:
"""
def "car"
{
    custom rel locationsOfInterest = </yard>
}
"""

Again it is very important, that composition arc related list editable ops get combined with a different rule set. We cover this extensively in our fundamentals section.

Non-composition related list editable ops do not make use of encapsulation. This means that any layer can contribute to the result, meaning any layer can add/remove/set explicit. When getting the value of the list op for non-composition arc list ops, we get the absolute result, in the form of an explicit list editable item list.

In contrast: When looking at composition list editable ops, we only get the value of the last layer that edited the value, and we have to use composition queries to get the actual result.

This makes non-composition list editable ops a great mechanism to store averaged side car data. Checkout our Houdini example below, to see this in action.

from pxr import Sdf
### Merging basics ###
path_list_op_layer_top = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/disc"), Sdf.Path("/cone")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/cone"),Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
# Notice how on merge it makes sure that each sublist does not have the values of the other sublists, just like a Python set()
print(result) # Returns: SdfPathListOp(Deleted Items: [/cube], Prepended Items: [/disc, /cone], Appended Items: [/sphere])
# Get the flattened result. This does not apply the deleteItems, only ApplyOperations does that. 
print(result.GetAddedOrExplicitItems()) # Returns: [Sdf.Path('/disc'), Sdf.Path('/cone'), Sdf.Path('/sphere')]

### Deleted and added items ###
path_list_op_layer_top = Sdf.PathListOp.Create(appendedItems=[Sdf.Path("/disc"), Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
print(result) # Returns: SdfPathListOp(Appended Items: [/sphere, /disc, /cube])
# Since it now was in the explicit list, it got removed.

### Explicit mode ###
# There is also an "explicit" mode. This clears all previous values on merge and marks the list as explicit.
# Once explicit and can't be un-explicited. An explicit list is like a reset, it 
# doesn't know anything about the previous values anymore. All lists that are merged
# after combine the result to be explicit.
path_list_op_layer_top = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.CreateExplicit([Sdf.Path("/disc")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
print(result, result.isExplicit) # Returns: SdfPathListOp(Explicit Items: [/disc]), True
# Notice how the deletedItems had no effect, as "/cube" is not in the explicit list.

path_list_op_layer_top = Sdf.PathListOp.Create(deletedItems = [Sdf.Path("/cube")])
path_list_op_layer_middle = Sdf.PathListOp.CreateExplicit([Sdf.Path("/disc"), Sdf.Path("/cube")])
path_list_op_layer_bottom = Sdf.PathListOp.Create(prependedItems = [Sdf.Path("/cube")], appendedItems = [Sdf.Path("/sphere")])

result = Sdf.PathListOp()
result = result.ApplyOperations(path_list_op_layer_top)
result = result.ApplyOperations(path_list_op_layer_middle)
result = result.ApplyOperations(path_list_op_layer_bottom)
print(result, result.isExplicit) # Returns: SdfPathListOp(Explicit Items: [/disc]), True
# Since it now was in the explicit list, it got removed.

Since collections are made up of relationships, we can technically list edit them too. Most DCCs set the collection relationships as an explicit item list though, as layering paths can be confusing.

To summarize: Composition (the process of calculating the value sources) is cached, value resolution is not, to allow random access data loading.

We'll be using the prim cache population module for inspecting composition. This is more of a deep dive topic, but you may at some point run into this in production.

An example scenario might be, that when we want to author a new composition arc, we first need to check if there are existing strong arcs, than the arc we intend on authoring. For example if a composition query detects a variant, we must also author at least a variant or a higher composition arc in order for our edits to come through.

from pxr import Sdf, Usd
# Create stage with two different layers
stage = Usd.Stage.CreateInMemory()
root_layer = stage.GetRootLayer()
layer_top = Sdf.Layer.CreateAnonymous("exampleTopLayer")
layer_bottom = Sdf.Layer.CreateAnonymous("exampleBottomLayer")
root_layer.subLayerPaths.append(layer_top.identifier)
root_layer.subLayerPaths.append(layer_bottom.identifier)
# Define specs in two different layers
prim_path = Sdf.Path("/cube")
stage.SetEditTarget(layer_top)
prim = stage.DefinePrim(prim_path, "Xform")
prim.SetTypeName("Cube")
stage.SetEditTarget(layer_bottom)
prim = stage.DefinePrim(prim_path, "Xform")
prim.SetTypeName("Cube")
attr = prim.CreateAttribute("debug", Sdf.ValueTypeNames.Float)
attr.Set(5, 10)
# Print the stack (set of layers that contribute data to this prim)
# For prims this returns all the Sdf.PrimSpec objects that contribute to the prim.
print(prim.GetPrimStack()) # Returns: [Sdf.Find('anon:0x7f6e590dc300:exampleTopLayer', '/cube'), 
                           #           Sdf.Find('anon:0x7f6e590dc580:exampleBottomLayer', '/cube')]
# For attributes this returns all the Sdf.AttributeSpec objects that contribute to the attribute.
# If we pass a non default time code value clips will be included in the result.
# This type of function signature is very unique and can't be found anywhere else in USD.
time_code = Usd.TimeCode.Default()
print(attr.GetPropertyStack(1001)) # Returns: [Sdf.Find('anon:0x7f9eade0ae00:exampleBottomLayer', '/cube.debug')]
print(attr.GetPropertyStack(time_code)) # Returns: [Sdf.Find('anon:0x7f9eade0ae00:exampleBottomLayer', '/cube.debug')]

import os
from subprocess import call
from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
# Spawn temp layer
reference_layer = Sdf.Layer.CreateAnonymous("ReferenceExample")
reference_bicycle_prim_path = Sdf.Path("/bicycle")
reference_bicycle_prim_spec = Sdf.CreatePrimInLayer(reference_layer, reference_bicycle_prim_path)
reference_bicycle_prim_spec.specifier = Sdf.SpecifierDef
reference_bicycle_prim_spec.typeName = "Cube"
reference_layer_offset = Sdf.LayerOffset(offset=10, scale=1)
reference = Sdf.Reference(reference_layer.identifier, reference_bicycle_prim_path, reference_layer_offset)
bicycle_prim_path = Sdf.Path("/red_bicycle")
bicycle_prim = stage.DefinePrim(bicycle_prim_path)
references_api = bicycle_prim.GetReferences()
references_api.AddReference(reference, position=Usd.ListPositionFrontOfAppendList)
# You'll always want to use the expanded method,
# otherwise you might miss some data sources!
# This is also what the UIs use.
prim = bicycle_prim
print(prim.GetPrimIndex()) 
print(prim.ComputeExpandedPrimIndex())
# Dump the index representation to the dot format and render it to a .svg/.png image.
prim_index = prim.ComputeExpandedPrimIndex()
print(prim_index.DumpToString())
graph_file_path = os.path.expanduser("~/Desktop/usdSurvivalGuide_prim_index.txt")
graph_viz_png_file_path = graph_file_path.replace(".txt", ".png")
graph_viz_svg_file_path = graph_file_path.replace(".txt", ".svg")
prim_index.DumpToDotGraph(graph_file_path, includeMaps=False)
call(["dot", "-Tpng", graph_file_path, "-o", graph_viz_png_file_path])
call(["dot", "-Tsvg", graph_file_path, "-o", graph_viz_svg_file_path])

def iterator_child_nodes(root_node):
    yield root_node
    for child_node in root_node.children:
        for child_child_node in iterator_child_nodes(child_node):
            yield child_child_node

def iterator_parent_nodes(root_node):
    iter_node = root_node
    while iter_node:
        yield iter_node
        iter_node = iter_node.parent
            
print("Pcp Node Refs", dir(prim_index.rootNode))
for child in list(iterator_child_nodes(prim_index.rootNode))[::1]:
    print(child, child.arcType, child.path, child.mapToRoot.MapSourceToTarget(child.path))
""" The arc type will one one of:
Pcp.ArcTypeRoot
Pcp.ArcTypeInherit
Pcp.ArcTypeVariant
Pcp.ArcTypeReference
Pcp.ArcTypeRelocate
Pcp.ArcTypePayload
Pcp.ArcTypeSpecialize
"""

We can also access the Pcp.Cache of the stage via: pcp_cache = stage._GetPcpCache()

from pxr import Sdf, Usd
stage = Usd.Stage.CreateInMemory()
prim = stage.DefinePrim("/pig")
refs_API = prim.GetReferences()
refs_API.AddReference("/opt/hfs19.5/houdini/usd/assets/pig/pig.usd")
print("----")
def _repr(arc):
    print(arc.GetArcType(), 
          "| Introducing Prim Path", arc.GetIntroducingPrimPath() or "-",
          "| Introducing Layer", arc.GetIntroducingLayer() or "-",
          "| Is ancestral", arc.IsAncestral(),
          "| In Root Layer Stack", arc.IsIntroducedInRootLayerStack())
print(">-> Direct Root Layer Arcs")
query = Usd.PrimCompositionQuery.GetDirectRootLayerArcs(prim)
for arc in query.GetCompositionArcs():
    _repr(arc)
print(">-> Direct Inherits")
query = Usd.PrimCompositionQuery.GetDirectInherits(prim)
for arc in query.GetCompositionArcs():
    _repr(arc)
print(">-> Direct References")
query = Usd.PrimCompositionQuery.GetDirectReferences(prim)
for arc in query.GetCompositionArcs():
    _repr(arc)
"""Returns:
>-> Direct Root Layer Arcs
Pcp.ArcTypeRoot | Introducing Prim Path - | Introducing Layer - | Is ancestral False | In Root Layer Stack True
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('anon:0x7f9b60d56b00:tmp.usda') | Is ancestral False | In Root Layer Stack True
>-> Direct Inherits
Pcp.ArcTypeInherit | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/pig.usd') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeInherit | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/pig.usd') | Is ancestral False | In Root Layer Stack False
>-> Direct References
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('anon:0x7f9b60d56b00:tmp.usda') | Is ancestral False | In Root Layer Stack True
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/payload.usdc') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeReference | Introducing Prim Path /pig{geo=medium} | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/mtl.usdc') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeReference | Introducing Prim Path /ASSET_geo_variant_1/ASSET | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/mtl.usdc') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/payload.usdc') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/geo.usdc') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/geo.usdc') | Is ancestral False | In Root Layer Stack False
Pcp.ArcTypeReference | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/geo.usdc') | Is ancestral False | In Root Layer Stack False
"""
# Custom filter
# For example let's get all direct payloads, that were not introduced in the active root layer stack.
query_filter = Usd.PrimCompositionQuery.Filter()
query_filter.arcTypeFilter = Usd.PrimCompositionQuery.ArcTypeFilter.Payload
query_filter.dependencyTypeFilter = Usd.PrimCompositionQuery.DependencyTypeFilter.Direct
query_filter.arcIntroducedFilter = Usd.PrimCompositionQuery.ArcIntroducedFilter.All
query_filter.hasSpecsFilter = Usd.PrimCompositionQuery.HasSpecsFilter.HasSpecs
print(">-> Custom Query (Direct payloads not in root layer that have specs)")
query = Usd.PrimCompositionQuery(prim)
query.filter = query_filter
for arc in query.GetCompositionArcs():
    _repr(arc)
"""Returns:
>-> Custom Query (Direct payloads not in root layer that have specs)
Pcp.ArcTypePayload | Introducing Prim Path /pig | Introducing Layer Sdf.Find('/opt/hfs19.5/houdini/usd/assets/pig/pig.usd') | Is ancestral False | In Root Layer Stack False
"""

As listed above, some plugins need to be compiled. Most DCCs ship with a customized USD build, where most vendors adhere to the VFX Reference Platform and only change USD with major version software releases. They do backport important production patches though from time to time. That's why we recommend using the USD build from the DCC instead of trying to self compile and link it to the DCC, as this guarantees the most stability. This does mean though, that you have to compile all plugins against each (major version) releases of each individual DCC.

The kind metadata is mainly used to for two things:

• Traversal: You can quickly detect (and stop traversal to children) where the assets are in your hierarchy by marking them as a a model subtype like component. A typical use case would be "find me all fx assets": In your pipeline you would define a 'fx' model kind subtype and then you can traverse for all prims that have the 'fx' kind set.
• DCCs use this to drive user selection in UIs. This way we can quickly select non-leaf prims that are of interest. For example to transform an asset in the hierarchy, we only want to select the asset root prim and not its children, so we can tell our DCC to select only model kind prims. This will limit the selection to all sub-kinds of model.

If you have created custom kinds, you can place icons (png/svg) in your $HOUDINI_PATH/config/Icons/SCENEGRAPH_kind_<name>.<ext> folder and they will be shown in your scene graph tree panel.

In production, you'll use kinds as a way to filter prims when looking for data. As kind data is written in prim metadata, it is very fast to access and suited for high performance queries/traversals. For more info on traversals, take a look at the traversal section.

You should always tag all prims in the hierarchy with kinds at least to the asset level. Some Usd methods as well as DCCs will otherwise not traverse into the child hierarchy of a prim if they come across a prim without a kind being set. So this means you should have at least group kind metadata set for all parent prims of model sub-kind prims.

You should always tag all prims with kinds at least to the asset level in the hierarchy. Some DCCs will otherwise not traverse into the hierarchy if they come across a prim without a hierarchy. So this means you should have group kinds set for all parent prims of model prims.

Make sure that your whole hierarchy has kinds defined (to the prim you want to search for), otherwise your prim.IsModel and prim.IsGroup checks will fail. This also affects how DCCs implement traversal: For example when using Houdini's LOPs selection rules with the %kind:component syntax, the selection rule will not traverse into the prim children and will stop at the parent prim without a kind. Manually checking via registry.IsA(prim.GetKind(), "component") still works though, as this does not include the parents in the check. (See examples below) #ToDo Add icon

In production, most of the sidecar metadata should be tracked via the assetInfo and customData metadata entries. It does make sense to extend the functionality with own metadata keys for:

• Doing high performance lookups. Metadata is fast to read, as it follows simpler composition rules, so we can use it as a IsA replacement mechanism we can tag our prims/properties with.
• We can add list editable ops metadata fields, this can be used as a way to layer together different array sidecar data. For an example see our Reading/Writing the installed custom metadata section.

This page only covers how to compile/install custom schemas, as we cover what schemas are in our schemas basic building blocks of USD section.

Codeless schemas are ideal for smaller studios or when you need to prototype a schema. The result only consists of a plugInfo.json and generatedSchema.usda file and is instantly created without any need for compiling.

Most DCCs ship with a customized USD build, where most vendors adhere to the VFX Reference Platform and only change USD with major version software releases. They do backport important production patches though from time to time. That's why we recommend using the USD build from the DCC instead of trying to self compile and link it to the DCC, as this guarantees the most stability. This does mean though, that you have to compile all plugins against each (major version) releases of each individual DCC.

We'll usually want to generate custom schemas, when we want to have a set of properties/metadata that should always exist (with a fallback value) on certain prims. A typical use case for creating an own typed/API schema is storing common render farm settings or shot related data.

$HFS/bin/usdGenSchema
# For example on Linux:
/opt/hfs19.5/bin/usdGenSchema

Codeless schemas are ideal for smaller studios or when you need to prototype a schema. The result only consists of a plugInfo.json and generatedSchema.usda file.

over "GLOBAL" (
    customData = {
        bool skipCodeGeneration = true
    }
) {
}

over "GLOBAL" (
    customData = {
        string libraryName       = "usdSchemaExamples"
        string libraryPath       = "."
        string libraryPrefix     = "UsdSchemaExamples"
    }
) {
}

over "GLOBAL" (
    customData = {
        string libraryName       = "usdSchemaExamples"
        string libraryPath       = "."
        string libraryPrefix     = "UsdSchemaExamples"
        bool skipCodeGeneration = true
    }
) {
}

On Windows you can also run hython usdGenSchema schema.usda dst to avoid having to source the env yourself.

cd /path/to/your/schema # In our case: .../VFX-UsdSurvivalGuide/files/plugins/schemas/codelessSchema
usdGenSchema schema.usda dst

cd ./files/plugins/schemas/codelessTypedSchema/
chmod +x build.sh # Add execute rights
source ./build.sh # Run usdGenSchema and source the env vars for the plugin path

// Linux
export PXR_PLUGINPATH_NAME=/Enter/Path/To/dist:${PXR_PLUGINPATH_NAME}
// Windows
set PXR_PLUGINPATH_NAME=/Enter/Path/To/dist;%PXR_PLUGINPATH_NAME%

Currently these instructions are only tested for Linux. We might add Windows support in the near future. (We use CMake, so in theory it should be possible to run the same steps in Windows too.)

cd .../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema
chmod +x build.sh # Add execute rights
source ./build.sh # Run usdGenSchema and source the env vars for the plugin path

On Windows you can also run hython usdGenSchema schema.usda dst to avoid having to source the env yourself.

cd /path/to/your/schema # In our case: ../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema
rm -R src
usdGenSchema schema.usda src

If you adjust your own schemas, you will have edit the following in these files:

• module.cpp: Per user define schema you need to add a line consisting of TF_WRAP(<SchemaClassName>);
• moduleDeps.cpp: If you add C++ methods, you will need to declare any dependencies what your schemas have. This file also contains the namespace for C++/Python where the class modules will be accessible. We change RegisterLibrary(TfToken("usdSchemaExamples"), TfToken("pxr.UsdSchemaExamples"), reqs); to RegisterLibrary(TfToken("usdSchemaExamples"), TfToken("UsdSchemaExamples"), reqs); as we don't want to inject into the default pxr namespace for this demo.

# Clear build & install dirs
rm -R build
rm -R dist
# Build
cmake . -B build
cmake --build build --clean-first # make clean all
cmake --install build             # make install

// Linux
export PYTHONPATH=..../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema/dist/lib/python:/opt/hfs19.5/python/lib/python3.9/site-packages:$PYTHONPATH
export PXR_PLUGINPATH_NAME=.../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema/dist/resources:$PXR_PLUGINPATH_NAME
export LD_LIBRARY_PATH=.../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema/dist/lib:/python/lib:/dsolib:$LD_LIBRARY_PATH
// Windows
set PYTHONPATH=..../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema/dist/lib/python;/opt/hfs19.5/python/lib/python3.9/site-packages;%PYTHON_PATH%
set PXR_PLUGINPATH_NAME=.../VFX-UsdSurvivalGuide/files/plugins/schemas/compiledSchema/dist/resources;%PXR_PLUGINPATH_NAME%

For Windows, specifying the linked .dll search path is different. We'll add more info in the future.

flowchart LR
    assetIdentifier(["@Asset Identifier@"]) --> assetResolver([Asset Resolver])
    assetResolver --> assetResolvedPath([Resolved Asset Path])

def "bicycle" (
    assetInfo = {
        dictionary UsdSurvivalGuide = {
            asset Asset = @assetMetadata@
            asset[] AssetDependencyPaths = [@dependencyPathA@, @dependencyPathB@]
        }
    }
    prepend references = @/opt/hfs19.5/houdini/usd/assets/pig/pig.usd@
){
    custom asset Asset = @someAsset@
    custom asset[] AssetArray = [@someAssetA@, @someAssetB@]
}

flowchart LR
    assetIdentifier(["@Asset Identifier@"]) --> assetResolver([Asset Resolver])
    assetResolver --> assetResolvedPath([Resolved Asset Path])
    assetResolverContext([Asset Resolver Context]) --> assetResolver  

resolved_path = stage.ResolveIdentifierToEditTarget("someAssetIdentifier")
# Get the Python string
resolved_path_str = resolved_path.GetPathString() # Or str(resolved_path)

from pxr import Ar
resolver = Ar.GetResolver()
resolved_path = resolver.Resolve("someAssetIdentifier")
# Get the Python string
resolved_path_str = resolved_path.GetPathString() # Or str(resolved_path)

from pxr import Sdf
asset_path = Sdf.AssetPath("someAssetIdentifier", "/some/Resolved/Path.usd")
print(asset_path.path)         # Returns: "someAssetIdentifier"
print(asset_path.resolvedPath) # Returns: "/some/Resolved/Path.usd"

Most DCCs ship with a customized USD build, where most vendors adhere to the VFX Reference Platform and only change USD with major version software releases. They do backport important production patches though from time to time. That's why we recommend using the USD build from the DCC instead of trying to self compile and link it to the DCC, as this guarantees the most stability. This does mean though, that you have to compile all plugins against each (major version) releases of each individual DCC.

We'll be using the asset resolver to redirect file paths for different reasons, mainly:

• Pinning: As USD pipelines are often based on "the latest asset/shot gets automatically refreshed/loaded" principle, we use the asset resolver to pin a USD stage to a specific version/state, so that it doesn't receive any asset/shot updates anymore.
• Redirecting the path from a custom identifier to an actual data resource. You'll only be using this aspect, if your resolver is not file based and instead URI (myCustomURIPrefix:<identifier>) based.

from pxr import Ar
...
resolver = Ar.GetResolver()
# The resolver context is actually a list, as there can be multiple resolvers 
# running at the same time. In this example we only have a single non-URI resolver
# running, therefore we only have a single element in the list.
context_collection = stage.GetPathResolverContext()
activeResolver_context = context_collection.Get()[0]
# Your asset resolver has to Python expose methods to modify the context.
activeResolver_context.ModifySomething()
# Trigger Refresh (Some DCCs, like Houdini, additionally require node re-cooks.)
resolver.RefreshContext(context_collection)
...

context_collection = stage.GetPathResolverContext()
activeResolver_context = context_collection.Get()[0]

export TF_DEBUG=AR_RESOLVER_INIT

ArGetResolver(): Using asset resolver FileResolver from plugin ../dist/fileResolver/lib/fileResolver.so for primary resolver
ArGetResolver(): Found URI resolver ArDefaultResolver
ArGetResolver(): Found URI resolver FS_ArResolver
ArGetResolver(): Using FS_ArResolver for URI scheme(s) ["op", "opdef", "oplib", "opdatablock"]
ArGetResolver(): Found URI resolver FileResolver
ArGetResolver(): Found package resolver USD_NcPackageResolver
ArGetResolver(): Using package resolver USD_NcPackageResolver for usdlc from plugin usdNc
ArGetResolver(): Using package resolver USD_NcPackageResolver for usdnc from plugin usdNc
ArGetResolver(): Found package resolver Usd_UsdzResolver
ArGetResolver(): Using package resolver Usd_UsdzResolver for usdz from plugin usd

from pxr import Ar
from usdAssetResolver import FileResolver
print(Ar.GetResolver())
print(Ar.GetUnderlyingResolver()) # Returns: <usdAssetResolver.FileResolver.Resolver object at <address>>

from pxr import Ar
with Ar.ResolverScopedCache() as scope:
    resolver = Ar.GetResolver()
    path = resolver.Resolve("box.usda")

from pxr import Ar, Usd
from usdAssetResolver import FileResolver
resolver = Ar.GetUnderlyingResolver()
context_collection = resolver.CreateDefaultContext() # Returns: Ar.ResolverContext(FileResolver.ResolverContext())
context = context_collection.Get()[0]
context.ModifySomething() # Call specific functions of your resolver.
# Create a stage that uses the context
stage = Usd.Stage.CreateInMemory("/output/stage/filePath.usd", pathResolverContext=context)
# Or
stage = Usd.Stage.Open("/Existing/filePath/to/UsdFile.usd", pathResolverContext=context)

resolved_path = stage.ResolveIdentifierToEditTarget("someAssetIdentifier")
# Get the Python string
resolved_path_str = resolved_path.GetPathString() # Or str(resolved_path)

from pxr import Ar
resolver = Ar.GetResolver()
resolved_path = resolver.Resolve("someAssetIdentifier")
# Get the Python string
resolved_path_str = resolved_path.GetPathString() # Or str(resolved_path)

from pxr import Sdf
asset_path = Sdf.AssetPath("someAssetIdentifier", "/some/Resolved/Path.usd")
print(asset_path.path)         # Returns: "someAssetIdentifier"
print(asset_path.resolvedPath) # Returns: "/some/Resolved/Path.usd"

Enabling debug symbols, allows you to inspect specific log outputs of Usd. Typical use cases are to check if plugins are loaded correctly or if data is properly being refreshed.

Activating the log output via bash:

export TF_DEBUG=AR_RESOLVER_INIT

Then launch your app.

activated_symbols = pxr.Tf.Debug.SetDebugSymbolsByName("AR_RESOLVER_INIT", True) # Returns: ["AR_RESOLVER_INIT"]
activated_symbols = pxr.Tf.Debug.SetDebugSymbolsByName("AR_*", True)

If you want to benchmark you Usd stage operations, the profiling module offers a fast and easy way to visualize performance.