Seed-Nodes: LoadMultipleImages

LoadMultipleImages is the node you reach for when your ComfyUI graph is doing the same thing to a hundred images and you refuse to double-click them one at a time. You point it at a folder, and it hands you every image in there as one batch, ready to feed into img2img, a pixelator, a captioner, whatever. It's the batch-loading half of the tiny ComfyUI-Seed-Nodes pack (by Aerse/"Seed"), and it's a clone-with-a-twist of what you'd get from ComfyUI-Light-Tool.

How it works

You give it a directory path and it lists the folder, keeps only .jpg, .jpeg, .png and .webp files, sorts them by name, and loads each one as its own item in the output batch. Two things to know about the sort: it's plain alphabetical, so img10.png sorts before img2.png. If processing order matters to you, name your files with zero-padding (a 01, 02 … style) or don't care.

The keep_alpha_channel toggle (default on) decides whether images with an alpha channel stay 4-channel RGBA or get flattened to RGB. Either way you also get a MASK list per image, built as 1 - alpha, so fully opaque pixels read as black - the same inverted-alpha convention ComfyUI uses everywhere.

The inputs that matter

• directory - the absolute path to your folder. There's no file picker here; you type the path.
• keep_alpha_channel - leave it on unless downstream nodes are choking on RGBA input.

Outputs are IMAGE (a list) and MASK (a list). Note the "list" part - this node emits a list of images, not a single batched tensor, so whatever you plug it into has to accept a list. That trips people up more than anything else here: plain IMAGE inputs will refuse to connect, and you'll be hunting for the "use it in a batch" variant of your target node.

How to install it

This is part of the ComfyUI-Seed-Nodes pack, so it's a pack install, not a single node. Easiest route is ComfyUI Manager - search "ComfyUI-Seed-Nodes" and click install. Or do it by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/Aerse/ComfyUI-Seed-Nodes
# restart ComfyUI

One honest warning before you do: the README in the repo gives a stale clone URL (github.com/Seed/ComfyUI-Seed-Nodes.git) that doesn't exist. Use the Aerse/ComfyUI-Seed-Nodes path above. Dependencies are just pillow, numpy, torch and torchvision - no model downloads, nothing heavy, and the LoadMultipleImages node itself has no extra deps beyond those.

Common issues

• "Directory cannot be found" error - the path is checked as-is, so a wrong or relative path throws. Give it the full absolute path.
• No new files being picked up - the node's change-detection is keyed to its inputs, not the folder contents. If you add images to a folder and the graph won't re-run, nudge the directory string (add a trailing space, then remove it) to force a refresh.
• Node won't connect downstream - see the "it's a list" gotcha above. That's the usual culprit, not a broken install.

For a genuinely tiny pack, this is the node with the most daily-use mileage in it. If you batch anything, this is the one you'll keep.