Vector 3D (Blender)

About as simple as this pack gets: three number fields bundled into one typed socket. It's part of Avatar Graph, avatech.ai's node pack for rigging character illustrations into interactive avatars inside ComfyUI, and it exists purely as plumbing - a way to hand a Blender-style 3D vector to any node in the pack that needs one, without that node having to expose three raw float widgets of its own.

What it's for

A lot of what Blender's mesh and object operators expect isn't a single number but a coordinate - a location, a scale factor, an axis. This node packages exactly that: x, y, and z, combined into a single B_VECTOR3 output. It's the equivalent of ComfyUI's own primitive/constant nodes, just typed specifically for this pack's Blender-flavored sockets instead of a generic float or string. You'll use it wherever another node in this pack takes a B_VECTOR3 input rather than three separate scalar fields.

Inputs and outputs

Three required fields, all floats, all defaulting to 0 with a fine 0.01 step for precise adjustment:

• x
• y
• z

The output is a single B_VECTOR3 socket carrying all three values as one unit. There's no other logic here - no clamping, no normalization, just three numbers packed together.

Installing it

ComfyUI Manager: search avatar-graph-comfyui. 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

This particular node doesn't touch bpy directly - it's just three floats - but it ships and only makes sense as part of a pack that's entirely built around Blender's Python module, and bpy is picky about its host: Python 3.10.x, nothing newer. That's the real prerequisite for getting the pack running at all: a dedicated conda environment on Mac/Linux (conda create --name comfyui python=3.10), or the README's prebuilt Python 3.10 ComfyUI zip on Windows. Restart afterward with --enable-cors-header if you plan to use the live avatar preview panel.

Troubleshooting

There's genuinely not much that can go wrong with this node in isolation - if the vector you're producing doesn't seem to be affecting anything, the problem is almost always on the receiving end: check that whatever node you've wired this into actually expects a B_VECTOR3 at that input, and that you haven't mixed it up with a similarly-shaped B_VECTOR4 input elsewhere in the pack. Beyond that, this is too small and self-contained a node to have generated any community discussion of its own - Avatar Graph's real troubleshooting lore, such as it is, centers on the pack's Python/bpy install requirements rather than its primitive utility nodes.