OpenCV fastNlMeansDenoisingColored_1

You're looking at the same story as fastNlMeansDenoising_1, one function over. The _0 and _1 variants of fastNlMeansDenoisingColored are two auto-generated overloads of the same cv2.fastNlMeansDenoisingColored call, and the generator produced identical schemas for both - same inputs, same single nparray output, same behavior. There is no reason to prefer one. Grab whichever, wire it up, move on.

For the actual content, read the fastNlMeansDenoisingColored_0 page - that's where the algorithm and the knobs live. The one-line summary: it's OpenCV's non-local means denoising for color images, which averages each pixel with similar-looking pixels elsewhere in the frame instead of its neighbors, so it kills grain while keeping edges crisp, and it denoises luminance separately from chroma. Inputs are src (BGR nparray), h (luminance strength, ~10 to start), hColor (chroma strength - keep it low), templateWindowSize (odd patch size, 3–7) and searchWindowSize (the search radius, default 21, and the cost knob). The optional dst out-parameter exists only because OpenCV's signature has it; leave it unplugged.

Installing it

Part of the geroldmeisinger/opencv-comfyui pack - install the pack once, get all ~600 nodes. ComfyUI Manager → search opencv-comfyui, or:

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

then restart. It needs opencv-contrib-python, numpy and torch. If startup dies with Cannot import name 'guidedFilter' from 'cv2.ximgproc', that's the known conflicting-OpenCV problem - consolidate to a single install.

And the pack-wide house rules: this works on nparrays in BGR 0–255 (Image2Nparray in, Nparrays2Image out), and batch size must be 1. If you see "Only images with batch_size==1 are supported", put an ImageFromBatch with length 1 in front.

That's the whole article, and that's fine. There genuinely isn't more to say about a node whose only distinguishing feature is its twin - the useful work is the colored denoise itself, and you'll find it explained properly on the _0 page.