USD Guide

USD (Universal Scene Description) support for DCC-MCP-Core.

Overview

Provides:

Core USD types: SdfPath, VtValue, UsdPrim, UsdStage
USDA serialization: Export stages as human-readable .usda text
JSON transport: Serialize/deserialize stages for MCP IPC
DCC bridge: Convert between SceneInfo JSON and UsdStage
Pure Rust: No dependency on OpenUSD C++ library

WARNING

This crate provides a compatible data model and serialization format for lightweight scene description exchange. It does not link against the OpenUSD C++ library.

SdfPath

USD scene graph path (e.g. /World/Cube).

Creating Paths

python

from dcc_mcp_core import SdfPath

# Create from string
path = SdfPath("/World")
print(path)  # /World

# Child path
child = path.child("Cube")
print(child)  # /World/Cube
print(child.name)  # Cube
print(child.is_absolute)  # True

# Parent path
parent = child.parent()
print(parent)  # /World

Path Properties

python

path = SdfPath("/World/Cube")

print(path.is_absolute)  # True
print(path.name)         # Cube
print(path.parent())     # SdfPath("/World")

VtValue

USD variant value container (bool, int, float, string, vec3f, etc.).

Factory Methods

python

from dcc_mcp_core import VtValue

# Scalars
v_bool   = VtValue.from_bool(True)
v_int    = VtValue.from_int(42)
v_float  = VtValue.from_float(3.14)
v_string = VtValue.from_string("hello")
v_token  = VtValue.from_token("normal")
v_asset  = VtValue.from_asset("/path/to/texture.png")

# Vectors (x, y, z as separate floats)
v_vec3 = VtValue.from_vec3f(1.0, 2.0, 3.0)

Getting Values

python

# Convert back to Python primitive
print(v_bool.to_python())    # True
print(v_int.to_python())     # 42
print(v_float.to_python())   # 3.14
print(v_vec3.to_python())    # (1.0, 2.0, 3.0)

# Type name
print(v_vec3.type_name)  # float3

UsdStage

Main stage container for USD scene description.

Creating Stages

python

from dcc_mcp_core import UsdStage, SdfPath, VtValue

# New stage with name
stage = UsdStage("my_scene")
print(stage.name)  # my_scene
print(stage.id)    # UUID

Defining Primitives

python

# Define prims with path string
stage.define_prim("/World", "Xform")
stage.define_prim("/World/Cube", "Mesh")
stage.define_prim("/World/Sphere", "Mesh")
stage.define_prim("/World/Lights", "Scope")

Setting Attributes

python

# Set various attribute types
stage.set_attribute("/World/Cube", "extent", VtValue.from_vec3f(1, 1, 1))
stage.set_attribute("/World", "timeCodesPerSecond", VtValue.from_float(24.0))
stage.set_attribute("/World", "name", VtValue.from_string("World"))

Querying Stage

python

# Check if prim exists
print(stage.has_prim("/World/Cube"))  # True

# Get prim
prim = stage.get_prim("/World/Cube")
if prim:
    print(f"Path: {prim.path}")
    print(f"Type: {prim.type_name}")
    print(f"Active: {prim.active}")
    print(f"Name: {prim.name}")

# Get attribute
val = stage.get_attribute("/World/Cube", "extent")
if val:
    print(val.to_python())

Traverse Primitives

python

# Traverse all prims
for prim in stage.traverse():
    print(f"{prim.path} ({prim.type_name})")

# Get prims of specific type
for mesh in stage.prims_of_type("Mesh"):
    print(f"Mesh: {mesh.path}")

Stage Properties

python

# Default prim
stage.default_prim = "World"
print(stage.default_prim)  # World

# Up axis
stage.up_axis = "Y"
print(stage.up_axis)  # Y

# Units
stage.meters_per_unit = 0.01  # cm
print(stage.meters_per_unit)  # 0.01

# FPS
stage.fps = 24.0
print(stage.fps)  # 24.0

# Time range
stage.start_time_code = 1001.0
stage.end_time_code = 1050.0

Remove Primitives

python

stage.remove_prim("/World/Sphere")  # True if removed
print(stage.has_prim("/World/Sphere"))  # False

Metrics

python

metrics = stage.metrics()
print(f"Prim count: {metrics.get('prim_count', 0)}")

Serialization

Export to USDA

USDA is the human-readable text format:

python

usda = stage.export_usda()
print(usda)

Output:

usda

#usda 1.0
(
    defaultPrim = "World"
)

def Xform "World"
{
    def Mesh "Cube"
    {
        float3 extent = [(1, 1, 1)]
    }
}

JSON Format

JSON is compact for IPC:

python

# Export to JSON
json_str = stage.to_json()

# Import from JSON
stage2 = UsdStage.from_json(json_str)

DCC Bridge

Convert between dcc-mcp-protocols SceneInfo JSON and UsdStage.

SceneInfo to UsdStage

python

from dcc_mcp_core import scene_info_json_to_stage

# From protocol scene info JSON
scene_info_json = '{"name": "my_scene", "prim_count": 100}'
usd_stage = scene_info_json_to_stage(scene_info_json, "maya")

UsdStage to SceneInfo

python

from dcc_mcp_core import stage_to_scene_info_json

# Convert USD stage to protocol scene info JSON
scene_info_json = stage_to_scene_info_json(stage)
print(scene_info_json)  # {"name": "my_scene", ...}

Unit Conversion

python

from dcc_mcp_core import units_to_mpu, mpu_to_units

# Convert unit string to metersPerUnit
print(units_to_mpu("cm"))   # 0.01
print(units_to_mpu("m"))    # 1.0
print(units_to_mpu("inch")) # 0.0254

# Convert metersPerUnit to unit string
print(mpu_to_units(0.01))   # cm
print(mpu_to_units(1.0))    # m

Complete Example

Creating a Simple Scene

python

from dcc_mcp_core import UsdStage, VtValue

# Create stage
stage = UsdStage("sample_scene")
stage.default_prim = "World"

# Define scene structure
stage.define_prim("/World", "Xform")
stage.define_prim("/World/Geometries", "Scope")
stage.define_prim("/World/Geometries/Cube", "Mesh")
stage.define_prim("/World/Geometries/Sphere", "Mesh")
stage.define_prim("/World/Lights", "Scope")

# Set cube attributes
stage.set_attribute("/World/Geometries/Cube", "extent", VtValue.from_vec3f(1, 1, 1))

# Export
usda = stage.export_usda()
print(usda)

Loading from JSON

python

# Export
json_str = stage.to_json()

# Load back
restored = UsdStage.from_json(json_str)
print(f"Restored: {restored.name}")
print(f"Prims: {len(restored.traverse())}")

Best Practices

1. Validate Paths

python

# Always check prim exists
if stage.has_prim("/World/Cube"):
    prim = stage.get_prim("/World/Cube")
    print(prim.type_name)

2. Use Appropriate Value Types

python

# Use from_vec3f for positions (takes x, y, z separately)
position = VtValue.from_vec3f(1.0, 2.0, 3.0)

# Use from_int for counts
count = VtValue.from_int(42)

# Use from_string for names
name = VtValue.from_string("Sphere")

3. Batch Operations

python

# Define multiple prims
prims = [
    ("/World/Cube", "Mesh"),
    ("/World/Sphere", "Mesh"),
    ("/World/Cylinder", "Mesh"),
]

for path, type_name in prims:
    stage.define_prim(path, type_name)

4. Use JSON for IPC

python

# For network transmission, use JSON
json_str = stage.to_json()
send_over_ipc(json_str)

# For human-readable output, use USDA
usda = stage.export_usda()
save_to_file(usda)

Limitations

Pure Rust/USD-compatible data model, not a full OpenUSD implementation
No C++ USD library dependency required
Some advanced USD features may not be available

USD Guide ​

Overview ​

SdfPath ​

Creating Paths ​

Path Properties ​

VtValue ​

Factory Methods ​

Getting Values ​

UsdStage ​

Creating Stages ​

Defining Primitives ​

Setting Attributes ​

Querying Stage ​

Traverse Primitives ​

Stage Properties ​

Remove Primitives ​

Metrics ​

Serialization ​

Export to USDA ​

JSON Format ​

DCC Bridge ​

SceneInfo to UsdStage ​

UsdStage to SceneInfo ​

Unit Conversion ​

Complete Example ​

Creating a Simple Scene ​

Loading from JSON ​

Best Practices ​

1. Validate Paths ​

2. Use Appropriate Value Types ​

3. Batch Operations ​

4. Use JSON for IPC ​

Limitations ​

USD Guide

Overview

SdfPath

Creating Paths

Path Properties

VtValue

Factory Methods

Getting Values

UsdStage

Creating Stages

Defining Primitives

Setting Attributes

Querying Stage

Traverse Primitives

Stage Properties

Remove Primitives

Metrics

Serialization

Export to USDA

JSON Format

DCC Bridge

SceneInfo to UsdStage

UsdStage to SceneInfo

Unit Conversion

Complete Example

Creating a Simple Scene

Loading from JSON

Best Practices

1. Validate Paths

2. Use Appropriate Value Types

3. Batch Operations

4. Use JSON for IPC

Limitations