Hi everyone,
it´s really great what you guys are doing ! Thank you so much!
I am a Solutions Architect working in the heavy machinery manufacturing sector. We are currently evaluating an architectural shift to establish OpenUSD and MaterialX as a interoperable 3D backbone between engineering and downstream commercial applications.
Our Core Stack:
-
Engineering & PLM: Dassault Systèmes (CATIA, 3DEXPERIENCE, 3DEXCITE)
-
Business/BOM: SAP (for product variants and configurations)
-
Downstream Targets: WebGL 3D Configurators (via glTF), Spatial Computing / AR (Apple Vision Pro via USDZ), and High-End CGI (tools used: pixyz, Blender, VRED)
My Vision: I would like to stop duplicating data and materials for different endpoints. The goal is to use OpenUSD for non-destructive scene composition (merging CATIA geometry with SAP variant logic) and MaterialX as our single source of truth for materials across Dassault, CGI tools, and Web/XR.
I’d love to connect with anyone who has tackled a similar enterprise pipeline. Specifically, I am looking for insights on:
-
Dassault 3DEXCITE & USD: Has anyone successfully utilized the recent OpenUSD and MaterialX export capabilities (e.g., DELTAGEN R2025x) in a production pipeline? How clean is the exported scene graph regarding hierarchies (Kinds) and instances?
-
SAP to VariantSets: What is your approach for mapping complex SAP configuration logic (BOMs) into USD
VariantSetswithout bloating the master stage? -
MaterialX for Web & XR: What are the current best practices for authoring MaterialX networks that need to translate well both into high-end renderers (like VRED or Unreal) and into lightweight formats (like glTF for WebGL and USDZ for Vision Pro)?
-
Handling Massive Assemblies: Our models easily hit 1M+ parts. Are you heavily relying on Payloads and Scenegraph Instancing before processing the data for lightweight endpoints?
Any shared experiences, whitepapers, or pointers to reference architectures would be highly appreciated. Thanks in advance!