OpenCV find4QuadCornerSubpix_1

find4QuadCornerSubpix_1 is the second auto-generated overload of cv2.find4QuadCornerSubpix, and it is identical to find4QuadCornerSubpix_0 in every way you can observe from the graph: same inputs (img, corners, region_size), same outputs (bool success flag plus the refined nparray of corners), same underlying function. The _0/_1 split is an artifact of the pack generator numbering OpenCV's overloads; when both schemas land the same, you get twins. No workflow will ever behave differently on one versus the other.

The substance lives on the find4QuadCornerSubpix_0 page. The short version: it refines coarse corner guesses on a quadrilateral image to subpixel accuracy by fitting local gradients - the step that makes camera-calibration numbers trustworthy. Feed it img (grayscale works best; convert with cvtColor code 6), corners (initial guesses, ideally from findChessboardCorners), and region_size as a literal tuple like (5, 5). Check the bool output before trusting the corners; silent refinement failure is the failure mode.

Installing it

It's part of geroldmeisinger/opencv-comfyui, which installs as one ~600-node pack. ComfyUI Manager → search opencv-comfyui, or:

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

then restart. Needs opencv-contrib-python, numpy, torch. If startup throws Cannot import name 'guidedFilter' from 'cv2.ximgproc', conflicting OpenCV installs are colliding - consolidate to one.

Pack-wide rules: nparrays in and out (Image2Nparray / Nparrays2Image), batch size 1 only.

And that's the complete story of this node: a faithful twin with nothing to add. If you need subpixel corner refinement, use either variant and read the _0 page for how.