OpenCV cubeRoot_0

cubeRoot_0 is the most honest node in this pack: it wraps cv2.cubeRoot, takes one FLOAT input, returns one FLOAT output, and does exactly what the name says - the cube root. cubeRoot(27.0) gives you 3.0. cubeRoot(-8.0) gives you -2.0. That's the entire spec. There's no mode, no hidden behavior, no parameter to misconfigure.

Where does it come from? opencv-comfyui is an auto-generated pack - the author parsed every top-level standalone function in OpenCV's Python type definitions and emitted a node for each. Most of those functions are image operations; a few, like this one, are just... math. cv2.cubeRoot exists in OpenCV as a fast C implementation of the cube root used internally by some camera-calibration and motion-estimation code, and the generator wrapped it like everything else. So you get a cube-root node in your ComfyUI graph whether you asked for it or not.

When you'd actually use it

Honest answer: rarely, and almost never as a pixel operation. A couple of realistic cases:

• You're porting a calibration or distortion-correction formula into a graph and it calls cbrt/cubeRoot in the middle - this node lets you replicate the exact math instead of approximating.
• You need a cube-root curve as a gentle tone response (much weaker than gamma - it compresses range without inverting anything). You could apply it per-pixel, though a dedicated curve node is usually cleaner.
• You're doing numeric experiments in a workflow and want the operation as a node rather than inline code.

In every case, this is glue. It exists because the pack has breadth, not because someone thought cube roots were a hot ComfyUI feature.

The spec

• val - FLOAT, the number (or numeric value) to cube-root. Note: this is a scalar, not an array. Unlike most of the pack it doesn't take an NPARRAY; wiring an image here won't work.
• Output: float, the cube root. Negative inputs are fine - the cube root of a negative number is negative, and OpenCV handles it.

Install

Part of opencv-comfyui (geroldmeisinger). ComfyUI Manager → search opencv-comfyui → install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/geroldmeisinger/opencv-comfyui

Restart. Requirements: opencv-contrib-python, numpy, torch. No model files.

Anything to watch?

Not really, and that's the review. The only trap is expecting array support - this is a scalar math node, so a multi-value input won't behave the way an image node would. If you find yourself needing the cube root of every pixel, that's a different operation and this isn't the node for it. For everything else: one input, one output, done. Sometimes the most useful thing a node can do is be exactly as simple as it looks.