OpenPoseLineart

Lineart control is where pose-driven animation gets its sketch feel. Where Canny gives you hard, mechanical edges, lineart produces softer, hand-drawn-looking outlines - the classic companion for anime and stylized work, and a community favorite for keeping a character recognizable across scenes. OpenPoseLineart is the ComfyUI-CUP loader that pulls pre-rendered lineart frames into your graph for the AIGODLIKE Blender pipeline.

The pack naming strikes again: this is an OpenPoseBase subclass with posetype = "Lineart" (capital L, matching the folder it reads). It doesn't extract line art from anything - Blender does that. This node loads <image-dir>/Lineart/ImageNNNN.png for the frame you ask for, converts it to a 0–1 RGB tensor, and hands it to you.

How it works

• image (STRING) - directory of rendered control images, filled by the Blender addon.
• frame (INT, default 0) - which frame's lineart to load.

Outputs: IMAGE (the line drawing → wire into Apply ControlNet with a lineart model) and MASK (from the PNG's alpha, optional). Because lineart conditions are looser than a skeleton or a depth map, the model keeps more freedom on shading and texture while still honoring the contours. That's the sweet spot for stylized shots: structure locked in, rendering left to the model.

Install and gotchas

It's part of ComfyUI-CUP. ComfyUI Manager → search ComfyUI-CUP, or:

cd ComfyUI/custom_nodes
git clone https://github.com/AIGODLIKE/ComfyUI-CUP.git

Restart after. Zero model downloads, zero extra pip dependencies.

Standard CUP warnings: it's a loader, not a detector, and it does nothing in a standalone web workflow - it needs the Blender addon to be rendering Lineart frames. And the silent-failure trap: a frame Blender never rendered gives you a 64×64 black placeholder, which a lineart ControlNet will faithfully turn into "nothing here." If the line style suddenly vanishes, confirm Lineart/ImageNNNN.png actually exists before touching your sampler.