ComfyUI Node Runs on cloud

Image Stitch

Glue two images together with a feathered seam

By WASasquatch·Created 3 years ago·Updated about a year ago· 1,812
Image Stitch

      Sometimes you just want to put two images next to each other and get one image out. Image Stitch does that - the README calls it "stitch images together on different sides with optional feathering blending between them." Pick a side, and it welds the second image onto the first's top, bottom, left, or right, with an optional soft blend across the join so it doesn't look like two rectangles taped together.

      What it's for

      The honest answer is: comparisons and layouts. Before/after pairs. Side-by-side variations of the same prompt. A cheap two-panel or four-panel (chain a few together). Building a wide reference sheet. It also shows up as a plumbing step in bigger workflows - stitching a generated region back next to its source, or assembling tiles. It's not a panorama-stitcher in the computer-vision sense (no feature matching, no perspective warp); it's a straightforward "place image B on side X of image A."

      How it works

      Two image inputs, a side to attach the second one (top / bottom / left / right), and a feathering amount that controls how softly the two blend where they meet. Feathering at zero gives you a hard edge - a crisp panel border. Turn it up and the seam fades, which is what you want when the two images share content and you're trying to hide the join. The output is the combined image.

      That's the whole node. Its simplicity is the appeal.

      Watch the dimensions

      The catch is the same one that bites every stitch-style node: the two images should match on the axis they're joined along. Stitching left-to-right, they need the same height; top-to-bottom, the same width. Mismatch and you'll get padding, stretching, or an error depending on the sizes. Send the odd one out through Image Resize first so the shared edge lines up. If your stitched result has a weird gap or a squished panel, that's the dimension mismatch talking.

      One more note worth knowing: modern ComfyUI core (and several other packs) now ship their own image-stitch/concat nodes, so if you already have one you like, there's no need to pull in all of WAS just for this. But if WAS is already in your graph - and it often is - this does the job fine.

      Installing it

      Image Stitch rides along with WAS Node Suite. Install the pack through ComfyUI Manager (search was-node-suite-comfyui, install, restart) or clone it:

      cd ComfyUI/custom_nodes
      git clone https://github.com/WASasquatch/was-node-suite-comfyui
      

      install requirements.txt, restart. No model downloads for this one.

      If it won't load

      The pack's been in maintenance-only mode since late 2023. The standard headache is the whole suite failing to import after a ComfyUI update - a dependency clash, opencv most often, because WAS holds some older pins. Reinstall the requirements into the right venv (activate it, or use install.bat) and it's back. Nothing about this node itself is fragile; the friction is always the pack-wide dependency reconcile.

      Categoryx

      Inputs (0)

      No inputs

      Outputs (0)

      No outputs