Object Call Override Hierarchy Create

This one lives inside Avatar Graph, Avatech AI's pack for building real-time interactive 2D avatars from a single character portrait - blinking eyes, lip sync, driven by Blender's Python engine (bpy) under the hood plus Avatech's own "Shape Flow" runtime. The README highlights a short list of nodes you'd actually reach for while building a workflow - Create Mesh Layer, Create Shape Flow, Avatar Main Output. Object Call Override Hierarchy Create isn't on that list; it's from the large ObjectCall_* family that exposes Blender's bpy.types.Object API roughly one method per node, generated wholesale rather than curated for the docs.

It's the "do the whole tree at once" version of Override Create. Where a plain library override touches a single linked object, object.override_hierarchy_create(scene, view_layer, do_fully_editable=False) creates overrides for the object and everything hanging off it - child objects, the parenting relationships, dependent data - in one call, within the context of a given scene and view_layer. do_fully_editable (default false) controls how permissive the resulting overrides are: left off, only the specific properties Blender's override system flags as commonly-edited get made editable; set it true and more of the hierarchy opens up for local tweaking.

This is genuinely niche even by this node family's standards - it's the kind of call you'd script if a workflow linked in a whole rigged character (armature, mesh, constraints, the works) from a shared library file and needed to make the entire rig locally editable at once instead of overriding pieces one at a time. If your Avatar Graph workflow builds its mesh and shape keys from the pack's own nodes rather than linking external .blend assets, you're unlikely to ever need this one.

Schema-wise, this is one of the more parameter-heavy ObjectCall_* nodes: BPY_OBJ (optional per the schema, effectively required), scene, a field typed " view_layer" - leading space and all, one more instance of the auto-generator naming a type after its field when it couldn't map Blender's real type cleanly - and do_fully_editable, a plain BOOLEAN defaulting false. scene and " view_layer" both need a value wired in from elsewhere in the graph rather than a widget you can type into. Output is BPY_OBJ, the same object passed through for chaining.

Installing it means installing the whole pack - ComfyUI Manager (search "Avatar Graph") or manually: cd ComfyUI/custom_nodes && git clone https://github.com/avatechai/avatar-graph-comfyui, then cd avatar-graph-comfyui && pip install -r requirements.txt, restart. The constraint to plan around: bpy needs Python 3.10.x exactly, so the README's advice to build a dedicated conda create --name comfyui python=3.10 env instead of reusing an existing Flux/SDXL setup is worth heeding. Restart with --enable-cors-header (--force-fp16 on macOS too) for the live avatar preview.

Troubleshooting: expect a Python-version mismatch - bpy failing to import because ComfyUI isn't running exactly 3.10.x - to be the failure you actually hit; it's a common, recurring category of pain across Blender-Python-based ComfyUI nodes broadly. No --enable-cors-header just costs you the live preview panel, not the graph's ability to run or export via Avatar Main Output. With zero documentation for this node in the repo and a genuinely advanced Blender feature underneath it, Blender's own manual pages on library overrides are the place to actually understand what do_fully_editable changes before you flip it.