USD Variable Expressions

Variable expressions are specially formatted strings that contain expressions that USD will evaluate at runtime. The expressions can contain variable data and logic used to dynamically create or change values in your USD scene. Use variable expressions to easily configure your scene without having to make extensive scene data modifications.

Variable expressions use named expression variables that can be easily defined and changed in layer metadata. Expression variables are defined in a “expressionVariables” dictionary metadata field in a layer.

Variable expressions can currently be used in:

  • Asset paths (including sublayer, reference, and payload asset paths as well as asset-valued attributes and metadata).

  • Variant selections.

The following simple example demonstrates using variable expressions to specify a reference path and a variant selection. An ASSET_PATH expression variable is set to a directory, and then used in a reference variable expression. Similarly, a VARIANT_CHOICE expression variable is defined and then used in a variable expression for selecting a variant.

#usda 1.0
(
    expressionVariables = {
        # Define path to reference that can be changed as needed
        string ASSET_PATH = "/finalAssets/SetA"
        # Define the expression variable that can be changed as needed
        string VARIANT_CHOICE = "variantA"
    }
)
def "asset0" (
    # Use a variable expression to complete the reference asset path
    references = @`"${ASSET_PATH}/extraAssets.usda"`@</asset0>
)
{
}

def Xform "asset1" (
    variants = {
        # Use a variable expression for specifying the variant
        string displayVariantSet = "`${VARIANT_CHOICE}`"
    }
    prepend variantSets = "displayVariantSet"
)
{
    ...displayVariantSet definition omitted...
}

Defining Expression Variables in a Layer

Add your named expression variables to the “expressionVariables” dictionary metadata field in your layer. Expression variables must be authored on either the root or session layer of a stage or in the root layer of a referenced layer stack.

#usda 1.0
(
    expressionVariables = {
        string VARIANT_CHOICE = "variantA"
        bool UseProxyModel = false
        int64 IDENTIFIER = 3254
        string[] renderPassList = ["foreground", "background", "FX"]
    }
)

Expression variables can currently be set to the following types.

string

A string, which in USD’s utf-8 text format, usda, is delimited by double or single quotes. Quote characters can be escaped using Python-style escaping (”"). A string value itself can contain a variable expression contained in backticks (“`”), that will be evaluated when the variable is used in a variable expression.

Examples:

string exampleStr = "I am a string"
string exampleStrEscaped = "A \"quote escaped\" string"
string exampleStrEscaped2 = """Another "escaped" string"""
string exampleStrBoolExpression = "`True`"
string exampleStrIfExpression = '`if(${USE_RED}, "red", "blue")`'
string exampleStrExpanded = "`charA_${MODEL_CATEGORY}_${MODEL_VERSION}`"

bool

A boolean value. Accepted values are True, true, False, or false.

Examples:

bool exampleBool = False
bool exampleBool2 = true

int64

A 64-bit integer.

Examples:

int64 exampleInt = 43
int64 exampleIntNeg = -500

<type>[]

A list of one of the previously mentioned types. The list can only contain values of one type. Lists with mixed types will result in an error.

Examples:

string[] exampleStringList = ["red", "green", "blue"]
int64[] exampleIntList = [4, 5, 6]
bool[] exampleBoolList = [True, False, False, True]

None

None is a special case value. Currently, you can’t set a variable directly to None. The None value can only be used within an expression, and is useful for expression functions, like the if() function, detailed in Expression Function Reference.

None is also not allowed in a list within an expression. The following examples will result in an error when evaluated:

`[1,2,3,None]` # Lists with None values are not allowed
`[if(False, 1)]` # The if() function will evaluate to None, so this is also not allowed in a list

Expression Variables and Composition

Like all other stage metadata, expression variables do not compose across sublayers in a layer stack except for the session and root layers of a stage. This avoids possible circular or order-dependent evaluations that would result in inconsistent behavior.

Expression variables will compose across references and payloads. If a referenced layer stack contains expressions that refer to expression variables, the expression variables from the chain of referencing layer stacks will be composed together and used to evaluate those expressions. However, note that the sublayer restriction from above still applies, so only expression variables authored on the root layer of the layer stacks will be considered.

Composing across references and payloads ensures that assets that internally use expressions referring to variables can be referenced by another asset and still work as expected. This also gives the referencing asset the ability to override the expression variables used by the referenced asset. For example, if you had the following defined in model.usda:

#usda 1.0
(
    expressionVariables = {
        string MODEL_VARIANT = "blue"
    }
    defaultPrim = "Model_1"
)

def Xform "Model_1" (
    variants = {
        string modelVariant = "`${MODEL_VARIANT}`"
    }
    prepend variantSets = "modelVariant"
)
{
    ...modelVariant definitions omitted...
}

And then referenced model.usda</Model_1> in your stage:

#usda 1.0
(
    expressionVariables = {
        string MODEL_VARIANT = "red"
    }
    defaultPrim = "Model_1"
)

over Xform "Model_1" (
    # reference model layer, but apply stage-level
    # MODEL_VARIANT variable value
    references = @model.usda@</Model_1>
)
{
}

The stage’s root layer MODEL_VARIANT variable will be applied to the referenced object, and the “red” variant will be used. If you omitted MODEL_VARIANT in your stage, the model.usda root layer’s MODEL_VARIANT would be applied instead, and the “blue” variant would be used.

Authoring Variable Expressions

Expressions are represented in USD scene data as strings that are surrounded with backticks (“`”). Variables used in an expression must be included using the ${variableName} format.

Expressions can use expression variables to produce dynamic values when evaluated. The following example uses two variables in an expression to produce a reference path.

references = @`${ASSET_DIRECTORY}/${CHARACTERS_DIRECTORY}/CharSetA.usda`@

When mixing expression variables and string content, you might need to escape special characters. The following examples show different ways of escaping characters.

# The following expression evaluates to "C:\USD\test.usd" or "D:\USD\test.usd"
# (depending on if COND evaluates to true or false) due to escaped backslashes
`if(${COND}, "C:\\USD\\test.usd", "D:\\USD\\test.usd")`

# The following evaluates to "escaped_var_${X}". No substitutions occur due
# to the escaped "$".
`"escaped_var_\${X}"`

Expressions can also use expression functions to add conditional logic. The following example uses the if() function to conditionally pick between two different reference paths, based on whether USE_LOW_POLY_ASSETS evaluates to true or false.

references = @`if(${USE_LOW_POLY_ASSETS}, "assetsLow/extraAsset.usda", "assetsNormal/extraAsset.usda")`@

The full list of expression functions is described in Expression Function Reference.

Expression Function Reference

The following functions are supported for expressions. For all functions, the function arguments can be values (of a type listed in Defining Expression Variables in a Layer), variables, or functions.

defined(<variable name>, …)

If <variable name> is defined in the composed set of expression variables visible to the layer where the expression is authored, return true, otherwise return false. This function can take multiple variable names. All listed variable names must be defined for this function to return true. Note that variable names must be provided as simple strings – you cannot use a string that contains a variable expression with this function.

Examples:

defined("VARIABLE1", "VARIABLE2") # returns true if VARIABLE1 and VARIABLE2 are both defined
if( defined("VARIABLE1"), ${VARIABLE1}, None) # if VARIABLE1 is defined, returns whatever VARIABLE1 evaluates to, otherwise None

if(<condition>, <true-value>, <false-value>)

If <condition> evaluates to true, returns <true-value>, otherwise returns <false-value>. An error results if <condition> does not evaluate to a boolean value. Note that <true-value> and <false-value> must evaluate to the same type, or None.

Examples:

# If the USE_OLD_ASSETS_BOOLEAN_VAR variable is set to true, return "/archived/Assets",
# otherwise return whatever string value CURRENT_ASSETS_PATH evaluates to.
if(${USE_OLD_ASSETS_BOOLEAN_VAR}, "/archived/Assets", ${CURRENT_ASSETS_PATH})

if(<condition>, <true-value>)

If <condition> evaluates to true, returns <true-value>, otherwise returns usda:None. An error results if <condition> does not evaluate to a boolean value.

Examples:

if(${USE_OLD_ASSETS_BOOLEAN_VAR}, "/archived/Assets")

and(<x>, <y>, …)

Returns the result of the logical and operation of all arguments. If all arguments evaluate to true, this will return true, otherwise this returns false. An error results if any argument does not evaluate to a boolean value.

Examples:

and( ${BOOL_VAR}, True )  # if BOOL_VAR evaluates to true, this will return true, otherwise false

or(<x>, <y>, …)

Returns the result of the logical or operation of all arguments. If any argument evaluates to true, this will return true. If all arguments evaluate to false, this will return false. An error results if any argument does not evaluate to a boolean value.

Examples:

or( false, true, false )  # returns true

not(<x>)

Returns the opposite boolean value that <x> evaluates to. An error results if <x> does not evaluate to a boolean value.

eq(<x>, <y>)

Compares the evaluated values of <x> and <y>. Returns true if values are equal, false otherwise. An error results if <x> and <y> evaluate to different types. For strings, comparison is a lexicographic comparison.

Examples:

eq(12, ${INT_VAR})  # returns true if INT_VAR is defined, is a int64 variable, and evaluates to 12
eq("hello", "HELLO") # returns false

neq(<x>, <y>)

Compares the evaluated values of <x> and <y>. Returns true if values are not equal, false otherwise. An error results if <x> and <y> evaluate to different types. For strings, comparison is a lexicographic comparison.

lt(<x>, <y>)

leq(<x>, <y>)

gt(<x>, <y>)

geq(<x>, <y>)

Compares the evaluated values of <x> and <y>. Returns true if <x> is less than (lt), less than or equal (leq), greater than (gt), or greater than or equal (geq) to <y>. An error results if <x> and <y> evaluate to different types. For strings, comparison is a lexicographic comparison. For booleans, true is considered greater than false.

Examples:

lt(10,12) # returns true
gt("Abc", "abc") # returns false, capital letters are 'earlier' lexicographically
gt("xyz", "abc") # returns true, xyz is 'later' than abc
gt("applecart", "apple") # returns true
gt(true, false) # returns true

contains(<list_or_string>, <value>)

Returns true if <list_or_string> contains <value>, false otherwise. This function works on both lists and strings. If <list_or_string> is a list, this function checks if <value> is an element in the list. If <list_or_string> is a string, this function checks if <value> is a substring in that string.

An error results if <list_or_string> does not evaluate to a list or string. If <list_or_string> is a list, and <value> is a different type than the list element type, this results in an error. Similarly, if <list_or_string> is a string and <value> is not a string, this results in an error.

Note that lists cannot contain the None value, as described in Defining Expression Variables in a Layer:None.

Examples:

# defined in layer expressionVariables:
#   string[] assetList = ["assetsLow/extraAsset.usda", "assetsMedium/extraAsset.usda", "assetsHigh/extraAsset.usda"]

contains(${assetList}, "assetsExtraHigh/extraAsset.usda") # returns false
contains("12345678", "45") # returns true

at(<list_or_string>, <index>)

Returns the value at the (0-based) <index> in <list_or_string>. This function works on both lists and strings. If <list_or_string> is a list, this function returns the list element at index <index>. If <list_or_string> is a string, this function returns a string that contains the character at index <index> in the string.

Note that negative index values are supported. Negative index values behave like negative indices in Python, accessing elements by counting from the end of the list or string rather than from the beginning.

An error results if:

  • <list_or_string> does not evaluate to a list or string.

  • <index> does not evaluate to an int64.

  • <index> is not a valid index into <list_or_string>.

Examples:

# defined in layer expressionVariables:
#   string[] assetList = ["assetsLow/extraAsset.usda", "assetsMedium/extraAsset.usda", "assetsHigh/extraAsset.usda"]
#   int64 newIndex = 1

at(${assetList}, 0) # returns "assetsLow/extraAsset.usda"
at(${assetList}, ${newIndex}) # returns "assetsMedium/extraAsset.usda"
at(${assetList}, -1) # returns "assetsHigh/extraAsset.usda"
at(${assetList}, 15) # results in Index of of Range error
at("123456",-2) # returns "5"

len(<list_or_string>)

Returns the number of elements in <list_or_string>. An error results if <list_or_string> doesn’t evaluate to a list or string. Can be used to sanity check that lists or strings are not empty before accessing them with at() or contains(). Depending on how lists are used in your scene, checking the length of a list could also provide more insight into asset structure (e.g., a “versions” list with length greater than 1 could imply additional versions of an asset exist).

Examples:

# returns number of elements in assetList, or error if assetList undefined or not a list or string
len(${assetList})

# returns 5
len("hello")

# sanity check assetList has at least one element, and if so, return the 0th index element
# otherwise return a default path string
if( gt(len(${assetList}), 0), at(${assetList},0), "defaultAssetPath")

Examples

This section provides additional examples using variable expressions for various use-cases.

Flexible Variant Selections

Variable expressions can provide flexible ways of specifying variations for logical groups of prims that are controlled via a single expression variable.

For example, consider an asset with a large crowd of models, each with variants controlling some aspect of their look, like a color palette. A crowds artist could select models that should have the same color palette and set their variant selections to corresponding variables, like:

Crowd.usda

#usda 1.0
(
    expressionVariables = {
        string COLOR_GROUP_1 = "regular"
        string COLOR_GROUP_2 = "regular"
    }
)

def "Crowd"
{
    def "Model_1" (
        variantSelection = {
            string palette = "`${COLOR_GROUP_1}`"
        }
    )
    {
    }

    def "Model_2" (
        variantSelection = {
            string palette = "`${COLOR_GROUP_2}`"
        }
    )
    {
    }

    # etc., etc. ...
}

When another artist references this crowd asset into their own stage, they can change the color palettes for all of the prims in the groups determined by the original crowd artist just by overriding the expression variables:

Shot.usda

#usda 1.0
(
    expressionVariables = {
        string COLOR_GROUP_1 = "rainbow"
        string COLOR_GROUP_2 = "monochrome"
    }
)

def "Crowd" (references = @Crowd.usd@</Crowd>)
{
}

The artist can still change the palette variant on individual models if they desire simply by overriding the variant selection on the specific prim.

Asset-valued Texture File Attribute

Different renderers may require different texture file formats. Rather than using alternate layers that specify different texture files for each renderer, you can use variable expressions to conditionally pick the desired texture file. The following example demonstrates this use-case by using variable expressions to configure the inputs:file asset-valued attribute on a UsdUVTexture Shader.

#usda 1.0
(
    expressionVariables = {
        string TEXTURE_DIR = "textures"
        bool USING_PRMAN_RENDERER = true
        string DEFAULT_TEXTURE_FILE = "pattern.png"
        string PRMAN_TEXTURE_FILE = "pattern.tex"
    }
)

...

def Material "boardMat"
{
    def Shader "diffuseTexture"
    {
        uniform token info:id = "UsdUVTexture"
        asset inputs:file = @`if( ${USING_PRMAN_RENDERER}, "${TEXTURE_DIR}/${PRMAN_TEXTURE_FILE}", "${TEXTURE_DIR}/${DEFAULT_TEXTURE_FILE}")`@
        float2 inputs:st.connect = </boardMat/stReader.outputs:result>
        float3 outputs:rgb
    }

    ... other USD Preview Material Shaders omitted...
}

Conditionally Include Sublayers

If you use multiple render passes when rendering your scene, you may need to use different sublayers for specific render passes. For example, you might have a render pass for FX that includes a sublayer that deactivates non-effects-related geometry to speed up rendering. You can use variable expressions to conditionally specify which sublayers to include for a given render pass. This might look something like:

#usda 1.0
(
    expressionVariables = {
        string RENDER_PASS = "fx"
    }

    subLayers = [
        @`"render_pass_${RENDER_PASS}.usd"`@,
        ...
    ]
)