Object Simulation Nodes Cache Delete

This is a small, specific housekeeping node, and it only matters if you're using one particular feature: Blender's geometry-node simulations. If an object in your Avatar Graph scene has a simulation zone attached (physics-like behavior driven through Blender's geometry nodes - cloth-ish motion, a settling effect, anything computed frame-by-frame and cached), Blender bakes and caches those frames so it doesn't recompute the whole simulation every time you scrub or re-render. Object Simulation Nodes Cache Delete clears that cache.

How it works

It's a direct wrap of Blender's own simulation-cache-clearing operator. The reason this exists as its own node rather than something that "just happens" automatically: once a simulation is baked, Blender will keep reading the cached result even after you change an upstream input that should have altered the simulation's outcome. If you tweak a mesh, a modifier, or a value feeding into a simulation zone and the result on screen (or in a render) doesn't change, a stale cache is the first suspect - and this node is the fix, forcing a fresh simulation the next time it's evaluated.

The inputs and outputs that matter

Both inputs are optional at the schema level, but one actually changes behavior:

• BPY_OBJ - the object whose simulation cache you're clearing.
• selected - boolean, defaults false. When true, the operation applies to whatever's currently selected in the scene rather than only the object you passed in directly - useful if you want to clear caches across several related objects at once rather than wiring each one through individually.

Output is a single BPY_OBJ, the object passed through so you can keep chaining after the cache clear.

How to install it

ComfyUI Manager, search "Avatar Graph" and install, or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/avatechai/avatar-graph-comfyui

then pip install -r requirements.txt and restart. This is a bpy-backed node, so Blender's Python module requirement applies in full - Python 3.10.x specifically, nothing newer. On macOS/Linux that means a dedicated conda environment for this pack rather than folding it into an existing install; Windows users can use Avatech's prebuilt Python-3.10 ComfyUI zip and install the pack through Manager on top.

Common issues & troubleshooting

If this node itself fails to appear or load, that's the pack's Python version mismatch, not a simulation-specific issue - check ComfyUI's console at startup for an import error and confirm you're on a clean Python 3.10.x environment with requirements.txt fully installed.

The more common "problem" people run into isn't a bug in this node, it's not knowing when to reach for it: if you've built a simulation-driven effect in your graph and changes upstream stop reflecting downstream, that's almost always a stale cache rather than something wrong with your logic - clear it here before you spend time debugging a simulation that was actually working, just reading old data.