OpenPoseFullExtraLimb
Pose control for when a character has a third arm
- IMAGE
- MASK
Here's the niche one. OpenPoseFullExtraLimb is the CUP loader for the openpose_full_Extra_Limb control type - and if you've never heard of that, you've probably never needed a character with more than two arms. The Extra Limb variant of the openpose_full control was built for exactly that: multi-limbed characters. Deities, demons, spider-mecha, the occasional very unfortunate anatomy demo. Standard pose detectors cap out at two arms; this condition type gives you more.
It's the same frame loader as its siblings - an OpenPoseBase subclass with posetype = "openpose_full_Extra_Limb", reading <image-dir>/openpose_full_Extra_Limb/ImageNNNN.png. No detection happens in ComfyUI; the Blender addon (ComfyUI-BlenderAI-node) renders the extra-limb skeleton, and this node fetches it frame by frame. Which is honestly the sensible division of labor - posing a six-armed character by hand in Blender is fiddly, but it's a thousand times more reliable than hoping a diffusion model invents the arms consistently.
How it works
- image (STRING) - directory of rendered control images, filled by the Blender addon.
- frame (INT, default 0) - which frame's skeleton to load.
Outputs: IMAGE (the extra-limb skeleton → Apply ControlNet with an openpose_full_Extra_Limb model) and MASK (from the PNG's alpha, optional). Note the word "full": like the regular OpenPoseFull, this condition bundles body + face + hands, so all those extra arms get hand keypoints too.
Install and gotchas
Part of ComfyUI-CUP - Manager search ComfyUI-CUP, or:
cd ComfyUI/custom_nodes
git clone https://github.com/AIGODLIKE/ComfyUI-CUP.git
Restart after. No models to download, no pip deps beyond ComfyUI's own.
Real talk: this is the most niche node in the pack. Most people who search for it are just googling the name out of confusion - and the answer to "do I need this?" is: only if your character legitimately has extra limbs AND you're using the Blender pipeline with an Extra Limb model on the ControlNet side. Without that model, the condition does nothing. And of course the usual rules apply: it's a loader, it's inert without the Blender addon, and a missing frame silently returns a 64×64 black placeholder.
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| image | STRING | — | |
| frame | INT | 0-2147483648–2147483648 | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| IMAGE | IMAGE | — |
| MASK | MASK | — |