Object Drop Named Image

Unlike a lot of the "drop" nodes in this pack, Object Drop Named Image actually works fine outside a real drag-and-drop - it takes a file path directly. It wraps bpy.ops.object.drop_named_image(), the operator Blender runs when you drag an image file into the 3D viewport: it creates a new Empty object with that image displayed on it, positioned wherever you tell it to.

How it works

Give it a filepath and it loads that image as a data-block, then adds an Image-type Empty in the scene at your chosen location/rotation/scale, aligned per align. Functionally that's close cousin to what Object_LoadReferenceImage does - both put an image-backed Empty into the Blender scene, which is why the output here is BPY_OBJ and not something like IMAGE. The difference is which Blender operator you're calling: LoadReferenceImage is the dedicated reference-image tool the README's documented character-loading workflow uses (with its own file-browser filter options); this one is the more general drag-and-drop image operator, useful when you want direct control over exactly where and how the image gets placed rather than following the reference-image flow specifically.

The inputs and outputs that matter

• filepath (string) - path to the image file. This is the one setting a beginner actually needs to set.
• name (string) - the data-block name to give the loaded image; leave blank and Blender names it after the file.
• relative_path (bool, default true) - store the path relative to the .blend file rather than as an absolute path. Matters mainly for portability if you're moving files around; leave it on unless you have a reason not to.
• align (enum: WORLD, VIEW, CURSOR) - how the new Empty is oriented on creation.
• location / rotation / scale - its starting transform.
• session_uuid - drag-session plumbing carried over from Blender's UI operator; harmless to leave at its default 0 when calling this directly rather than via a real drag.
• BPY_OBJ (optional) - for chaining onto existing scene context.

Output: a single BPY_OBJ, the new image Empty.

How to install it

ComfyUI Manager: search "Avatar Graph", install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/avatechai/avatar-graph-comfyui
cd avatar-graph-comfyui && python -m pip install -r requirements.txt

Same dependency wall as the rest of the pack: bpy only ships for Python 3.10.x. Windows can grab Avatech's prebuilt Python-3.10 ComfyUI zip; macOS/Linux wants a dedicated conda create --name comfyui python=3.10 rather than an existing newer-Python install. Add --enable-cors-header on restart if you want the live avatar preview reaching editor.avatech.ai (Mac also wants --force-fp16); skip it and you still get full functionality through .glb/.gltf export.

Common issues & troubleshooting

Install failures almost always trace to Python version - check for an import error on bpy in the startup console before suspecting anything else.

Filepath not resolving - with relative_path on, a path saved relative to one machine's .blend location can break if the graph runs somewhere else (say, a different ComfyUI install or a cloud executor). If the image doesn't load where you expect, try an absolute path first to rule that out.

Meant to load a character reference, got a generic image Empty instead - that's this node working correctly, just not the node the README's rigging workflow actually uses. Swap to Object_LoadReferenceImage for the documented open-mouth, 768×768-minimum character-portrait flow; reach for this one when you want more direct placement control over an arbitrary image instead.