AD Image Concatenation

Glues two images together into one - either side by side or stacked top-to-bottom. That's the whole job. If you've ever wanted a before/after pair in a single frame, a two-panel comparison, or a reference image welded next to your subject to feed an editing model, this is the node that does it without any fuss.

Why you'd reach for it

ComfyUI's built-in image handling is great at making pictures and oddly clumsy at arranging them. Concatenation is one of those small plumbing operations you end up needing constantly once you're doing anything beyond a single generation: comparison grids, diptychs, or the increasingly common trick of handing an edit model two images at once (a reference and a target) as a single stitched input. AD Image Concatenation is a clean one-node answer instead of a mask-and-composite dance.

How it works

You give it two images and a direction. horizontal puts them left-to-right, vertical stacks them. The catch with joining images is that they rarely share dimensions, so the node has match_size (on by default) to resize one to fit the other along the shared edge - heights must agree for a horizontal join, widths for a vertical one. When it resizes to make that happen, method picks the resampling filter.

The inputs that matter

• direction - horizontal or vertical. This is the one you'll actually change.
• match_size - leave it on unless your two images already match exactly. With it off, mismatched sizes will not line up cleanly.
• method - the resize filter: lanczos, bicubic, bilinear, or nearest. lanczos is the sharp, sensible default; leave it there unless you have a reason. nearest is only for pixel art where you want hard edges.

The output is a single IMAGE - the combined result - which wires into anything: a Preview, a saver, or straight into an img2img or edit node if you're doing the two-images-as-one-input trick.

Common issues

The thing that trips people up is turning match_size off and then wondering why the join looks wrong or errors out. If the two images don't share the relevant dimension, they can't be concatenated cleanly - either keep match_size on, or resize both to a common size upstream first. Beyond that there's not much to break; it's a deliberately simple node. If your stitched image comes out softer than expected, that's the resize step doing its thing - feed it images that are already close in size so there's less resampling to do, and keep method on lanczos.

Installing ComfyUI-Addoor

ComfyUI-Addoor (葵花宝典) is a grab-bag utility pack from developer Eagle-CN - mostly batch and file-plumbing nodes rather than models or samplers. ComfyUI Manager: Install Custom Nodes → search "ComfyUI-Addoor" → install → restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/Eagle-CN/ComfyUI-Addoor.git
cd ComfyUI-Addoor
pip install -r requirements.txt

Restart and find it under 🌻 Addoor / image.