ComfyUI-ArchiGraph

Flip a binary image's blacks and whites in one click

The cheap box blur before everything else

The soft blur that detection pipelines want before Canny

Kill salt-and-pepper noise without smearing the edges

Canny edges on CPU, thresholds in 0-to-1, no model download

Count the blobs in your binary image, with their stats

Mark the detected centers on your image

Overlay your detected outlines on the original

Generate your batch filenames before you've saved anything

Feed a whole folder into a loop, one file at a time

Trace the outlines of everything in a binary image

The math version of line detection — rho and theta, not endpoints

Find the straight lines in your edge map, endpoints included

Size your images by the long edge, not the square

A black & white slider rack, the way Photoshop means it

Walk a whole folder through your pipeline, one pass at a time

The ComfyUI loop that re-runs the whole graph for you

Pull your saved graph data back out of a .json

Persist your graph's data between runs

Bring a saved numpy array back into the OpenCV world

Save your numpy arrays to disk — the cache node that makes expensive preprocessing optional

See what's actually in that nparray — without committing it to disk

The debug node that tells you what's actually flowing through that wire

Save a PNG exactly where you want — no output-folder roulette

Turn your OpenCV result into a real PNG in the output folder

Grab a chunk of any list — the slice node for the rest of your graph

One slice, applied to every sublist — for cutting each clip down to size

Edge detection the OpenCV way — the node that finds your lines

Turn a grayscale mess into a clean binary mask

The bridge back — from OpenCV arrays to normal ComfyUI images

The bridge in — feed ComfyUI images to the OpenCV half of the graph