Load Images For Loop

You've got a folder of 200 images and one workflow you want to run over every single one of them. The naive approach - load them as a batch - falls over fast: a 200-image batch of anything heavy will eat your VRAM alive. easy loadImagesForLoop does it the sane way. It's a loop that reads a directory and yields one image per iteration, so your workflow processes image 1, finishes, frees the memory, then processes image 2, and so on. It's the loop-start half of a for-loop that happens to know how to walk a folder.

Why you'd reach for it

Any time you have a batch job where the per-image work is non-trivial: upscale a folder, face-detail a folder, caption a folder, run img2img over a directory of frames. Because each pass handles a single image, memory stays flat no matter how many files are in the folder - the thing a real batch can't promise.

It pairs with easy forLoopEnd, which closes the loop. You wire this node's flow output down through your per-image processing and into the loop-end node; the end node loops back until the folder's exhausted. Values you want to accumulate across iterations (a running list, a counter) ride the initial_value / value sockets around the loop.

The inputs and outputs

• directory (STRING) - the folder path to read images from. This is the one required setting.
• start_index (default 0) - skip the first N files, handy for resuming a partial run.
• limit (default -1) - cap how many images to process; -1 means all of them.
• initial_value1 / initial_value2 (any) - seed values carried into the loop, for accumulating state across iterations.

Per-iteration outputs:

• flow (FLOW_CONTROL) - the loop control signal. Wire this to easy forLoopEnd. This is what makes it a loop, not a one-shot loader.
• index (INT) - which iteration you're on. Useful for naming outputs or driving an index switch.
• image and mask - the current file's image and (if present) alpha mask.
• name (STRING) - the current file's name, so you can save results under matching filenames.
• value1 / value2 - the carried-through loop values.

Installing it

ComfyUI Manager: search ComfyUI Easy Use, install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Use

then install the requirements (install.bat on Windows, else pip install -r ComfyUI-Easy-Use/requirements.txt) and restart.

Common issues

• You must close the loop. This node needs easy forLoopEnd connected via flow, or the loop has no end and won't iterate correctly. A "loadImagesForLoop that only ran once" is almost always a missing or mis-wired loop-end.
• The directory path. It reads from a server-side path - the folder as ComfyUI's process sees it, not your desktop. On a cloud or remote instance that's a path on the remote machine. Get the path right and confirm it actually contains readable images.
• Save outputs inside the loop. Since each iteration overwrites the previous image, wire your Save node into the loop body (use the name output to keep filenames unique) rather than expecting a batch to fall out the end.
• Mixed image sizes are fine here - because you're handling one at a time, the images don't need to share dimensions the way a real batch would.