Simple Load Image Batch

What it is

WAS Node Suite's Load Image Batch node is a workhorse - it's the standard way to feed a folder of images into ComfyUI one at a time across repeated runs, and WAS Node Suite in general is the big general-purpose image-processing toolkit a lot of people have installed for exactly this kind of utility. But it does a lot: extra modes, extra options, extra flexibility you may never touch. The author of this pack was processing something like 10,000 video frames and found that WAS's version was, in their words, a bottleneck - noticeable only at that kind of scale, but real. Simple Load Image Batch is their fix: the same core idea, stripped down to almost nothing, and reportedly about 100x faster for it.

Reach for this when you're iterating over a folder of images frame-by-frame - video-frame processing pipelines, batch upscaling, anything where you're re-running the same graph against the next file in a directory - and you don't need WAS's extra knobs.

How it works

Point it at a folder, and each time the workflow executes it loads the next image in sequence. The tricky part of any "load the next thing" node in ComfyUI is that the engine caches nodes whose inputs haven't changed and skips re-running them - which would mean this node just keeps handing back the same image forever under Auto Queue. The author's fix, by their own account, was adding a SEED output specifically so the node's output looks different each time, which is enough to convince ComfyUI's cache that something changed and the graph should actually advance.

Inputs that matter

• path – the folder to read images from.
• allow_RGBA_output – a true/false choice for whether alpha-channel images pass through with their alpha intact or get flattened.
• reset (BOOLEAN, default false) – restarts the internal position back to the first image in the folder. Flip this when you want to start the batch over rather than continuing from wherever it left off.
• seed (INT, default 0) – exists to force re-execution under Auto Queue, as above. You generally don't need to hand-tune this; it's plumbing, not a creative parameter.

Outputs

Three: IMAGE (the loaded image), SEED (the value that keeps the node from getting cached), and a STRING - almost certainly the filename of the image it just loaded, which is handy to route into a save node so your output filenames trace back to their source frame.

Installing it

Bundled with the rest of the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/Shadetail/ComfyUI_Eagleshadow

Or install "Eagleshadow Custom Nodes" via ComfyUI Manager. No model weights, no heavy dependencies - this is filesystem iteration, not inference.

Common issues

The batch doesn't advance. If you're not running under Auto Queue, or something in your queue setup isn't triggering re-execution, you'll keep getting the same image. Confirm Auto Queue is actually on and that nothing's caching the node's inputs upstream in a way that defeats the seed trick.

You lose the flexibility WAS Node Suite gave you. This node is deliberately barebones - that's the entire point, and the entire performance win. If you relied on some of WAS's extra Load Image Batch options (filtering, sorting modes, whatever your workflow used), you won't find them here. This is a straight speed trade, not a feature-for-feature replacement.

Order isn't guaranteed the way you expect. Folder-iteration nodes generally read in whatever order the filesystem or a sort routine gives them - if frame order matters (and for video frames, it does), name your files with zero-padded numeric prefixes (0001.png, 0002.png, …) so lexicographic sorting matches sequence order, rather than trusting an unpadded scheme like 1.png, 2.png, 10.png.